There are two ways in which animals can adjust behaviorally to predictable events in their environment. If environmental events are consistent between generations, alleles that 'hardwire' favorable responses to those conditions can increase in the population through differential reproduction. If specific environmental conditions are less predictable between generations than within generations, the ability to learn from experience may be selected for. Animals in nature usually experience a mixture of conditions, some that are consistent between generations, some that are not. Thus, the behavioral repertoires of animals are often a combination of innate and learned components. Jenny Jenkins, a graduate student in the lab of William Rowland in the Program in Animal Behavior at Indiana University, has explored the relationship between innate behaviors and learning in the three-spined stickleback (Gasterosteus aculeatus). Her research addresses whether or not the stickleback, which provides classic examples of innate responses to sign stimuli, is able to fine-tune its courtship and territorial defense behavior based on experience.

In the spring, male three-spined sticklebacks enter shallow coastal waters and dig nests in the sediment. As nest building progresses, their bellies and throats change from a cryptic grey color to an eye-catching red. After their nests are complete d, the males aggressively defend their breeding territories against rival males and actively court females. Whether they attack, or court, is dependent upon cues present in the conspecifics they encounter. Ter Pelkwijk and Tinbergen (1937) investigated sign stimuli in sticklebacks using fish models and found that a red belly triggers aggression and a swollen belly triggers courtship. It was later observed that a male's reaction to a red-bellied rival also depends on the context of the situation (Rowlan d et al. 1995; see also Brown 1975).

If sticklebacks can adjust their reaction to sign stimuli to specific environmental conditions, do they learn to do so from experience? Jenkins was interested in investigating the potential ecological significance that a specific type of learning, k nown as classical conditioning, holds for the breeding behavior of three-spined sticklebacks. In classical conditioning an animal learns to associate a predictive ('conditioned') stimulus such as a light with a primary ('unconditioned') stimulus such as food. Jenkins asked whether or not male sticklebacks from a population in coastal New York state could learn to associate red or green lights with the appearance of rival males. She examined this question using male sticklebacks in individual aquaria with the following protocol. First, she presented 20 males with 40 pre-training trials in which a red light was displayed 20 times and a green light was displayed 20 times. A rival was not present during any of the pre-training trials. These training trials served to test for differences in the reaction to red and green before conditioning. Following pre-training, 240 trials were conducted in which half of the trials had a light paired with a rival and half of the trials had a light presented alone. The males were divided into two groups for the training phase. Each group was presented with the red light in 50% of the trials and the green light in 50% of the trials. In one group, rivals were paired with the red light and the green light was presented alone. In the other group, rivals were paired with the green light and the red light was presented alone. Jenkins then conducted 160 extinction trials in which signals were presented but no rivals were presented. In all sessions, approach to the end of the tank where rivals appeared and zigzagging swimming patterns were recorded. Zigzagging behavior was recorded because it occurs in both territorial defense contexts and courtship contexts.

In training sessions, males in both treatment groups increased their frequency of both approach and zigzagging when presented with both a light and the rival. Jenkins observed no significant difference between the proportion of approach or zigzaggin g between the two treatment groups, however, so the color red did not serve as a more effective predictive stimulus in training than did green. In the post-training phase, Jenkins observed that males in both treatment groups continued to approach the sti mulus light they were trained with in the absence of rival presentations. The frequency of approach did decline in extinction, but not significantly. Jenkins concluded that once sticklebacks learn about cues that predict the presence of rivals, it may pay to remember them. Interestingly, in extinction trials, the males trained with red as the predictive stimulus showed a less pronounced tendency to approach when the light was present than did the green group. This may be because red elicits conflicting tendencies in sticklebacks, the tendency to display aggression and the tendency to be intimidated (see Rowland et al. 1995; Archer 1988).

This conditioning study revealed that male sticklebacks can learn to associate predictive cues with the appearance of a rival. Jenkins next examined whether male sticklebacks can learn to adjust their selection of females according to prior experience. Ter Pelwijk and Tinbergen (1937) observed that courtship behavior in male sticklebacks was elicited by a swollen belly in a female. Swollen bellies indicate the presence of eggs, and thus a trait for distinguishing between gravid and non-gravid females would be selected for. Males are so focused on a swollen belly as a sign stimulus that they will preferentially select models with unnaturally swollen ('super-normal') bellies (Rowland 1989). Males may be presented with conflicting cues, however, because fish with heavy parasite loads have gravid-looking abdomens. These observations led Jenkins to wonder if male sticklebacks could fine-tune their tendency to court gravid-looking females according to experience in order to avoid parasitized females.

Jenkins tested this idea by choosing a group of males that preferred fish models of super normal females in preference tests. She then divided the males into two groups for conditioning trials. In conditioning trials, each group received 10 presentations of a normally gravid dummy paired with a super-normally-gravid dummy. Males in one group received a mild shock when they courted super-normal models, males in the control group courted the same models but did not receive shock. Jenkins then performed four post-conditioning preference tests by presenting males from both groups with regular and super-normal models, without shock, after 15 minutes, one day, two days, and seven days had elapsed. Experimental males spent a lower proportion of time in both approach and zigzagging in the presence of the supernormal dummy than did control males in conditioning trials and in post-conditioning tests. In post-conditioning tests, males' preferences for the super-normal female steadily returned to original levels. Interestingly, the experimental males continued to maintain high rates of approach and zigzagging during conditioning trials, even though they were being exposed to an aversive stimulus. These results indicate that males are capable of fine-tuning their courtship behavior according to experience, but that they maintain a strong tendency to show innate elicited responses.

The results from the territorial defense experiment established that males can learn to associate predictive cues with the appearance of rivals. This finding has clear ecological significance; we now know that males in the wild are capable of adjusting their territorial defense behavior according to subtle cues they have experienced in the course of a breeding season. The ecological significance of Jenkins' courtship experiment is less clear. The courtship experiment established that males can temporarily learn to avoid females that are associated with an aversive stimulus, but are males in the wild able to assess a cost associated with parasitized females? If males cannot determine that there is a cost to courting parasitized females, Jenkins' experiment does not have direct ecological relevance. It is possible that parasitized females would be less likely to follow through with mating, and this would be a cost that could be learned by males. Heavily parasitized males would also appear gravid and males courting a gravid-looking male may learn that when they do so they suffer the cost of lost fertilization opportunities. Of course, it is also possible that males can identify parasitized females upon initial contact, perhaps by chemical cues; if they could identify parasitized females initially, an innate tendency to avoid parasitized females might confer greater reproductive success than having to learn to avoid them. In addition, it is important to consider that males may court super-normal dummies, and gravid-looking, parasitized females, because the males are so tuned to the sign stimuli of a swollen belly that they continue to respond to the cue due to a design constraint, even when it is unfavorable to do so (Rowland 1989, 1994).

Jenkins has made exciting discoveries about the ability of sticklebacks to tune their stereotypical behavior patterns according to experience. A great deal remains to be learned about stickleback breeding behavior, however. Sticklebacks display chains of elicited responses in their courtship behavior. The behaviors in the chain may have been added to each other, step by step, through evolution. Each step might represent a different behavioral 'mode' in the stickleback's behavioral hierarchy or 'behavior system' (Tinbergen 1951; Timberlake 1990). Are different steps in the hierarchy of courtship behaviors more responsive to learning than others? Are there slight differences in the behavioral hierarchy of stickleback courtship behavior among populations, indicating different evolutionary pathways (see Foster 1994)? If there are differences in behavior among populations, what are the ecological conditions that may be responsible for such variation? What are the constraints on the interaction between learning and innate behavioral traits? These questions will continue to be explored by Jenkins, William Rowland, and their colleagues.
 

LITERATURE CITED

Archer, J. 1988. The Behavioral Biology of Aggression. Cambridge University Press.

Foster, S. A. 1994. Evolution and reproductive behavior of threespine stickleback. Pages 381-398 In: The Evolutionary Biology of the Threespine Stickleback (M. A. Bell and S. A. Foster, eds.). Oxford University Press.

Rowland, W. J. 1989. The effects of body size, aggression and nuptial coloration on competition for territories in male threespine stickleback, Gasterosteus aculeatus. Anim. Behav. 37: 282-289.

Rowland, W. J. 1994. Proximate determinants of stickleback behavior: an evolutionary perspective. Pages 297-344 In: The Evolutionary Biology of the Threespine Stickleback (M. A. Bell and S. A. Foster, eds.). Oxford University Press.

Rowland, W. J., K. J. Bolyard, and A. D. Halpern. 1995. the dual effect of stickleback nuptial coloration on rivals: manipulation of a graded signal using video playback. Anim. Behav. 50: 267 272.

ter Pelwijk, J. J., and N. Tinbergen. 1937. Eine reizbiologische analyse einiger verhaltensweisen von Gasterosteus aculeatus (L.). Z. Tierpsychol. 1:193-200.

Timberlake, W. 1990. Natural learning in laboratory paradigms. Pages 31-54 In: Contemporary Issues in Comparative Psychology (D. A. Dewsbury, ed.). Sinauer Associates.

Tinbergen, N. 1951. The Study of Instinct. Oxford University Press.