Nix ist bewiesen, es gibt nur bestimmte Hinweise und Vermutungen.
Das Ganze ist noch nicht vollständig verstanden worden.
Hier ein Auszug aus dem Originaltext:
Aus
http://www.pnas.org/content/108/46/18726.full
With a combination of high resolution data on group movements and function fitting, we have tested the assumptions made by models of the collective motion of fish. We have identified some key similarities with many model assumptions—namely, that fish responded to the position of their neighbors through short-range repulsion and longer-range attraction rules. These rules have been proposed by a number of models (7, 24, 25, 26). There are also some inconsistencies between the classical models of fish motion and our results. For example, we do not find evidence that fish aligned specifically with their neighbor’s orientation; a rule adopted by some models (7, 26). Instead, it appears that group alignment is achieved in some other way, possibly through attraction and repulsion rules and by fish following individuals in front of themselves. Models have already shown that such interactions can produce highly polarized groups (23, 27). However, this is not to say that alignment rules are never adopted by this or other species. Under simulated predation threat, it is clear that alignment with neighbors can allow information to be transmitted rapidly through fish schools (28). Whether fish change their adopted rules depending on context is an intriguing possibility that warrants further investigation.
Another difference is the importance of acceleration in response to a neighbor’s position. Unlike many self-propelled particle models, where speed is fixed and interactions are mediated through changes in direction (1, 6), mosquitofish actively changed their speed in order to avoid or move toward neighbors. Other studies on bird flocks and fish schools have also found strong evidence that individuals’ speeds fluctuate in groups (29–31). Clearly, therefore, it is important for future models to incorporate changes and differences in speeds, which is likely to affect a number of group level properties.
Perhaps the most surprising finding of our study is the extent to which the single nearest neighbor dominates social interactions. Previous observations on birds have identified interacting neighbors using correlations in movement (1, 15) or anisotropy of the position of the nearest neighbors (10). These methods have typically identified multiple interacting neighbors. Given that we find only one interacting neighbor using our function fitting method, it would be interesting to see if similar results hold for birds and other species. Indeed, if we apply a purely correlational approach to our data, we find that the turning angle of a focal fish is as correlated with its second and third neighbors as it is with its first, suggesting multiple interacting neighbors (Fig.*S4). These correlations likely arise through multiple interactions between the fish over time and do not necessarily reflect their immediate rules of motion.
Previous experiments on mosquitofish have shown that information about the position of predators can be transmitted rapidly between group members (32). A question remains as to whether these nearest neighbor interactions are sufficient to explain these observations or if the fish change their interactions in risky situations. Early models of collective motion suggested that interactions with more than one neighbor are required for collective motion (33). More recent models have shown that single neighbor interactions can produce complex schooling patterns (5). Further modeling work is required to understand the group-level outcome of the rules we have identified. Mosquitofish form relatively small shoals in freshwater, unlike the larger pelagic shoals formed by some marine species, so we would expect models based on the rules we have identified to produce only a limited range of collective patterns.
There is a qualitative difference between paying attention to one neighbor or to multiple individuals. Responding to two or more neighbors is likely to bear significant costs in information processing time because it requires integrating positional information over the whole visual field (34). Integrating larger quantities of information may be a strategy affordable to animals with larger brains such as birds or mammals but less suitable to animals such as small shoaling fish. However, we are beginning to appreciate that even humans might rely on relatively simple rules to effectively navigate their environment (35). A model on retinal information processing shows that having a threshold mediated response plus sensitivity to particular information can allow individuals to filter out unimportant information from multiple neighbors while responding to strong directed motion of others (36). Such a mechanism is indeed compatible with our findings and may explain how individuals can disregard information from multiple neighbors.
Future studies should attempt to identify the interaction rules between individuals in a comparative way across species. Under different selection pressures different interaction rules are likely to be selectively advantageous (2, 37, 38). Our aim should be to compare the interaction rules adopted by different species, classify the group-level properties generated by these rules, and discuss these in the light of evolution and phylogenetic considerations. If the rules adopted by individuals are similar across distantly related species, then this will show their effectiveness in driving coordinated group movement under different evolutionary pressures. On the other hand, universality in the patterns generated by groups may mean that each species has a very distinct set of rules, each leading to the same set of global patterns (1). Our current study shows us that a detailed analysis of animal interactions of a particular species can provide a different picture of collective motion than we imagine when we concentrate solely on building models.
LG Tisch