Bio 373: Aggression Lab

Transcription

1 Bio 373: Aggression Lab I. Introduction A. Primary Goal of the Lab: Animal conflicts can vary from ritual displays with no contact to mixtures of contact and display and finally to unabated violence. The goal of this lab is to identify the factors that modulate levels of conflict escalation in common house crickets. This will entail understanding current game theoretical treatments of animal conflicts sufficiently to create a list of possible factors and then designing experiments to test the relative significance of these factors in specific contexts. B. Secondary Goals of the Lab: This lab is the first in which you are to devise your own hypotheses derived from current theory and then test these hypotheses with experiments that you design and execute. Thus learning how to design and execute such studies is an important goal for the lab. In addition, you will learn current game theoretical treatments of aggressive behavior in general, acquire some practice using and modifying a hand-held data acquisition device, and become familiar with typical cricket behaviors and husbandry. C. Dates: Your final report on this work must be turned in by 11 October (check syllabus for policies). D. Venue: We have ordered domestically raised crickets for these experiments. These will be housed in a central colony that will be attended to by all of us. Each lab group will take a subset of these crickets and house them in your own experimental boxes. You are responsible for feeding and watering your animals. Experiments will be conducted in our lab room and data analysis should be conducted in R (the computer room we've been using up to this date will be available thorughout the course but you may also use your own computers if you wish). E. Individual vs Teamwork: We want you to work in groups for this lab. Each group will set up and manage their own cricket experimental boxes, design their experiments, and collect the data. However, each of you is to do your own analysis and writeup of the results of those experiments. F. Background: We shall go over some general theory about animal conflicts and cricket biology during scheduled hours. Print up the handouts for each background topic and have them ready for these presentations. II. Project Design A. Specific Goal of this Lab: Theory suggests that three factors modulate the willingness of a male cricket to escalate a contest: 1) the value to it of winning (V); 2) the costs of engaging in the contest and the cost of losing an escalated fight (K s); and 3) the amount of prior information the cricket has about its chances of losing an escalated fight (Q). The goal of this lab is for you to design, execute, and analyze a series of experiments in which you manipulate V, the K s, and/or Q to test for the relative importance of each of these factors in modulating cricket escalation behaviors.

2 Aggression Lab, p. 2 B. Setups and Husbandry: 1. Setups: We shall focus on male cricket contests. Each student group will be assigned several experimental boxes, each filled with sand, food & water dishes. A large common facility contains a colony of crickets from which you may take subjects. Crickets should be fed special cricket food, which contains all the nutrition and moisture they require (if needed we will switch to cat food and water). You can use hollowed out half-corks as refuges. Habitat box may be divided in half using cardboard partitions if you need it... be creative. Learning how to solve unforeseen problems that may arise during your project is an important goal for this lab. 2. Husbandry: You are responsible for seeing that your crickets are fed and watered. At the beginning of your experiments, select animals from the buckets, weigh them, measure their lengths, and glue a tiny bee numbered tag to their back (we shall show you how. Females are available for manipulations of V. Be sure to feed and check your crickets daily, including weekends. Food dries out and animals can (and probably will) die in the middle of an experiment. 3. Handling: WARNING!! Crickets are fragile animals. You must be very careful when handling them or they will lose legs, break off antennae, etc. Always capture your test animals with caution and try to identify which ones you want before you grab them. When you return animals to the larger facility, gently tip their container so that they can move easily into the larger one. Watch out that they do not jump away from the containers. Any marking, weighing, or surgery on the animals is best done by chilling them first. 4. Isolation Strategies: We have found it difficult to partition the larger plastic boxes sufficiently to prevent isolated animals from getting through the partitions. They crawl under tape and get stuck, chew through cardboard, or dig under barriers. If you decide to separate or isolate participants in the same box, see the TA about the best method to do this. Otherwise, move your animals to separate containers. C. Learning Cricket Behaviors: The first days of the lab, you should spend several hours watching the crickets to learn all of the aggressive, courtship, and appeasement behaviors. Be sure to learn the differences between the types of sounds. Practice scoring contests, identifying the maximum levels or intensities, and timing the durations. You will not be able to do this lab until you are quite fluent with the animal s behaviors. Try using ipads or other recording devices to record some contests. Watch the resulting videos and use imovie to slow them down if you need to. D. Cricket V s: For cricket males that are well fed and watered, the major V s are related to mating and ownership of a burrow. 1. Mates: A male which has just mated with a female will perceive her as a high V since he knows she is receptive and likely to mate several more times. Typically, males remain with females for second or even third matings. Note that a male who has just mated with a female for her first mating will assign her a higher V than a naïve male who then appears because the latter has no evidence that she is receptive. He must

3 Aggression Lab, p. 3 just assume that she is receptive with average (low) probability. All other factors being equal, the male who has just mated with the female is more likely to escalate than the naïve male should a fight ensue over her. However, note that females are only valuable to males for short periods after the males encounter them. If you house a female with a male, he will value her highly for about min. After that, he is likely to have mated with her if she is receptive or found out that she is not receptive. If you test a male after he has spend more than an hour with a female, he is unlikely to act any differently than he did before you added the female. Cricket time scales are a lot shorter than ours!! 2. Burrows: All males seek to control a burrow. This is the basic territorial unit from which calling songs are emitted and where matings occur. In your set-ups, males can protect cork halves or other provided structures as potential burrows. They usually aggregate around them, hide in them, and often fight over them. Females appear to prefer burrows in which the soil is slightly moist. You may want to manipulate burrow suitability by moistening some burrows. 3. Manipulations: You can thus manipulate V s by allowing a male to occupy a cork house or other potential burrow alone for some time: he will then value it more highly than a new male who has not had time to spend time investigating and using the house. You can also manipulate V s by allowing a male to mate with a receptive female. This is a bit more difficult since we cannot always provide receptive females. The best method is to isolate juvenile females (i.e., no wings) and let them become adults in the absence of males so as to ensure they will be receptive when introduced to your focal male. However, just adding a female to a group of males or a single male will often change the levels of aggression and males with continued access to the same female are likely to act defensively towards other males (e.g. if they have been around the female more, they assign her a higher V than a newcomer does). E. Cricket K s: 1. Levels of Costs: Crickets suffer two kinds of contest costs: energetic costs associated with each alternative action performed during a contest, and the injury costs associated with various levels of biting, flipping, and kicking. Hack (1997) measured the energetic costs of most of the alternative aggressive actions listed above. These same actions can be divided into 3 injury risk categories as follows: a) Low Injury Risk (actions that involve no physical contact between opponents); b) Moderate Injury Risk (actions involving light and/or intermittent contact); and c) High Injury Risk (actions involving strong, sustained physical contact). The following table lists most of the actions listed above according to their costs: Action Injury Risk Energetic Cost Cerci Raising Low 0 Rearing Low 0 Stridulate Low 0.02 Mandible Flare Low 0.08/s Shake Low 0.49 Head Butt Moderate 0.37 Foreleg Punch Moderate 0.37 Mandible Spar Moderate 0.45/s

4 Antenna Lash Moderate 0.68 Stridulation Lash Moderate 0.70 Kick High 0.37 Head Charge High 0.61 Mandible Lunge High 0.61 Wrestle High 0.83/s Aggression Lab, p. 4 The energetic costs are listed as ml oxygen/g of body weight. If a /s is appended, then the cost is given as a rate of energy expenditure; the total cost for an action is the product of this rate and how long the animal performed that action. If no /s is appended, then the cost is assumed to be fixed per each repetition of that action. 2. Measuring K s: As noted earlier, all of the models discussed above assume that energetic, time, and injury costs can be combined somehow into an overall K. It is this K that the animals should use to decide whether or not to escalate. This overall K presumably has the form: K = m E + n T + qpd where E is the energy that would be expended to go to the next step, T is the time that would be spent on that effort, D is the damage that might be inflicted by the opponent, P is the probability that such damage will be inflicted, and m,n, and q are all constants that convert energy, time, and damage costs into losses of future reproduction. Presumably, qd can be a big number: damage can have serious effects on fitness. However, the contribution of qd to K must be discounted by the probability, P, that damage will be experienced. If P is small enough, it is possible that me+nt could easily be about the same order of magnitude as qpd. On the other hand, if there are two unequal opponents, the P value for the better fighter will be smaller than the P value for the less effective fighter. This means that the better fighter may be able to afford to spend the energy to escalate to the next level, increasing its E and T costs, but because its qpd values is much lower, its overall K value of escalating could still be lower than its opponent were they to escalate in the same way. The result is the initially paradoxical result that the better fighter might elect to escalate, the lesser fighter would then give up and leave, and the better fighter would have spent more total energy in the fight than the lesser fighter. This does not mean that the better fighter won despite having a bigger K and thus that the theory is wrong, but only that it figured its overall K including the energetic costs would be lower than its opponent if both escalated. We do not have access to the critical m, n, and q values needed to convert everything to a common currency and thus cannot compute overall K values. We can however tally up the total energy spent by each opponent at each stage in a contest and especially the cumulative costs at the end, whether the animals used low, moderate, or high risk actions, etc. We can then ask how winning a contest depends on relative body size, which party spent the most energy, which party performed the most risky actions, where in the sequence the most energetically expensive and the most risky actions were each introduced, etc. How these factors interact can be as interesting as knowing the overall K. 3. Manipulating K s: The injury risk to a given cricket of using contact actions presumably increases the larger the body size of its opponent. Thus we can manipulate the injury risk levels by providing any given cricket with differently sized

5 Aggression Lab, p. 5 opponents (all other things such as V s being equal). The bigger the difference in combatant body sizes, the easier it should be for both animals to assess their relative risks of escalation after each performs only a few low cost actions. Such contests should be short, show little escalation through the Table above, and result in low net expenditures of energy. Animals of similar body sizes will have difficulty assessing their relative injury risks and will consequently have to exchange actions that are more like real fighting and thus more risky themselves. As predicted in the theoretical section, we do see a positive correlation between the risks of injury performing an action and its energetic costs in crickets. Thus, as predicted by the models, and confirmed in the Table above, more informative actions are usually more costly energetically and more risky at the same time. Two animals that are of similar body size should fight longer, move further down the Table of escalated actions before stopping, and spend more energy total than dissimilarly sized animals. As predicted by the models, there should also be higher variability in contest durations when two animals are evenly matched in body size or perceptions of V. F. Cricket Q s: 1. Information Sources: A cricket has access to three sources of prior information about its fighting ability relative to an opponent before it begins a contest: a) it can assume it is average in the population (a good start but not helpful if the opponent also thinks it is average); b) it can assume that its most recent experience is a good guide to its relative status: if it has won a lot of recent fights, it is probably above average, and if it has lost a lot, then it is probably below average; and c) it may recognize its opponent and remember what happened last time it encountered this same individual. It is not clear that wild crickets can recognize other individuals. However, they do seem to be highly responsive to their most recent record of fighting so can do better than just assume they are average. 2. Manipulating Q s: It is clear that crickets do pay attention to their own most recent contest history. Alexander s (1961) classic experiment puts 4 male crickets in a cage and lets them set up a dominance hierarchy where 1 > 2 > 3 > 4. He then puts crickets 1 and 2 in a cage by themselves and crickets 3 and 4 in another cage for several days. Then he recombines the four. In many cases, the hierarchy (using the numbers from the first run) is now 1 > 3 > 2 > 4. The middle two animals have switched positions. This is because 2 s experience as a pair was that it always lost and never won; and 3 s experience was that it always won and never lost. Putting them back together, 3 now escalates more easily than 2 and the hierarchy is reversed. One can now repeat the isolation phase with 1 and 3 together and 2 and 4 together and upon recombination one restores the original order. Usually 1 and 4 retain their original ranks throughout because whatever factors caused them to be top and bottom in the first place remain unchanged given their experiences. This suggests a number of experiments in which you can vary the recent experience of future contestants in different ways: a) isolate a cricket for 3-4 days which allows it to reset its image of itself to a high level as it has been dominated by nobody (this is the erase prior history option); b) put a focal animal in a cage with one that is much larger or much smaller than it to insure that it is either frequently beaten or frequently wins respectively. Then add this into a group with different histories and observe what happens.

6 Aggression Lab, p Restricting Information Acquisition: As we have seen, stridulation is used at many stages to threaten or declare oneself a winner. One way to break the information transfer chain is to deafen a focal animal by gently puncturing the eardrum on its front legs or covering them with glue. It will then be unable to hear the stridulations of opponents and will assume that it is not being threatened when in fact it is. This is one way to get a small animal to escalate more readily. F. Standard Setups There are a large number of potential experiments that you can design to study the relative roles of V, K, and Q in modulating the escalation levels of cricket contests. Depending upon which experiments you select, you are likely to use one of the following three standard protocols: 1. Preconditioning: In this process, you will select experimental animals from the lab pool and expose them to a specific preconditioning phase. Preconditioning may include isolating an animal for 3-5 days so that it will forget all its recent wins and losses and start your experiment with the assumption that it is average. Alternatively, one might put a focal animal in with a much larger or much smaller cricket for several days so that their most recent experience indicates that they are below or above average in fighting ability respectively. One may condition V values by putting a cricket into a container with a burrow or house or a virgin or receptive female. Finally, one might deafen a focal animal and let it recover a day from the handling before experimenting. 2. Contest Arena: In this type of experiment, you will take two animals carefully selected for relative size and/or some preconditioning experience, and put them simultaneously into a clean contest box. You will then record every interaction between them for 20 minutes. This will be our maximum contest time. Usually, the two crickets will encounter each other 2-3 times fairly soon and one will have won by the second or third bout. Once a winner is clear, you can terminate the experiment. If after 20 minutes, there has been no contest, terminate the experiment and find another pair. During the encounter, you can use your ipad's Time Stamp app to record each of the contest behaviors (and retreat) listed above for each of the two animals. If you find these occur too quickly, video tape the encounters and go back and score them. The use of ipads will be explained in the second day of class on this lab. Depending upon the experiment, you will probably want to record the following: a. The ID numbers for the two animals and who won and who lost b. The amount of time spent in low, moderate, and high risk behaviors c. The sequence of new actions introduced by each animal d. The maximum level action performed before the loser retreated e. The overall duration of the contest f. The total number of actions performed in the contest. g. An estimate of the total energy expended by each party G. Possible Experiments:. We would like you to perform at least 2 related experiments with your crickets. The list below will give you a few ideas but is not exhaustive. It will be up to you to design something exciting and useful. Clear your plans with the instructors BEFORE getting started. Note that to perform any statistics you will need

7 Aggression Lab, p. 7 sufficient replication and that the more complicate your experimental design, the harder it will be to collect more samples. Think about controls if you are planning to manipulate any variable. Try to pick experiments that are sufficiently related that your report can tell some sort of coherent story (not just two isolated vignettes). Here are some suggestions: 1. Deafen a male by dabbing a spot of epoxy glue on each ear and letting it harden. Plot all contests by this male on plots of contest duration vs relative size and contest intensity vs relative size and compare them to those of non-deafened males. 2. Let 2 males set up a dominance hierarchy in a box for a couple of days. Add a virgin female and monitor what happens. Record the fraction of time each male makes courtship songs. Which male mated with the female: the dominant or the subordinate? Did the dominant sing more? Deafen a virgin female and repeat the experiment. Do you get the same results? What might this indicate? 3. Take 2 males of very similar weight. Then pit one against a succession of much larger opponents. Pit the other simultaneously against a series of much smaller opponents. Then pit each against a new opponent of size very similar to their own. Are there any differences in the outcomes of these last fights as a function of each male s most recent history? 4. See whether males will vary their effort depending upon the suitability of females (e.g. vary V). For example, pit two similarly sized males who have not before met into a new box with a female. Monitor the duration of the contests and the maximum level of escalation before one wins. Try this experiment on a number of fresh, paired males, but alternate between virgin females and already mated females. Try also varying the weight of females: males may prefer larger (more gravid) females. 5. Take several pairs of males of given size differences and place each pair together until one completely dominates the other. Then separate them for varying times (2 hours, 6 hours, 12 hours, 2 days, etc.) before recombining them. Can you plot the level of aggression in the recombination contests as a function of time separated to get an estimate of the cricket forgetting curve? III. Possible Problems: A. Sample Size: One of the most common problems with this type of experiment is that you do not collect large enough sample sizes to uncover statistically significant effects. The solution to this is more time in the lab, more samples, longer samples per setup, etc. Take data soon after forming groups. Do not put it off until the weekend after which some of your animals may have died (see next problem). B. Death: The other common problem with these experiments is death of critical crickets before you can finish sampling. Do not put off taking your data. Remember that being housed in a small place with other male crickets can be stressful. Try to undertake very assiduous husbandry to keep your crickets alive through the study (warm humid conditions are ideal be be aware that extreme humidity can cause disease). Note that crickets do not have long lives anyway. After a certain age, they get old and die just like

8 Aggression Lab, p. 8 people. So again, be sure you have enough backup groups for any experiment so that if crickets in one group die, you have ongoing replicates. IV. Analysis and Writeup: Your report should include Introduction, Methods, Results, Discussion, and Summary sections. All tables and graphs in the Results must be clearly labeled and explained, and all statistical tests should be properly applied and accurately reported. Take time to discuss and interpret your experiments in the Discussion. Interpretations invoked to explain one experiment must be compatible with interpretations invoked to explain other experiments using the same animals. Beware of ad hoc explanations that fit one experiment but not another! Aim at logical consistency in your report: what is the overall picture and how do the various experiments fit together to generate that overview? V. Relevant References: Adamo, S.A. and R.R. Hoy Agonistic behaviour in male and female field crickets, Gryllus bimaculatus, and how behavioural context influences its expression. Anim. Behav. 49: Alexander, R.D Aggressiveness, territoriality, and sexual behaviour in field crickets (Orthoptera: Gryllidae). Behaviour 17: Hack, M. A Assessment strategies in the contests of male crickets, Acheta domesticus. Anim. Behav. 53: