Doi:10.1016/j.tree.2004.07.00

Many wrongs: the advantage of group navigation Department of Biology, College of Natural Sciences, Carleton University, 1125 Colonel By Drive, Ottawa, ON, Canada K1S 5B6 Research into the puzzling phenomena of animal navigation and aggregation has proceeded along two Experimental and observational work has refined our distinct lines. Study of navigation generally focuses on knowledge of a wide array of orientation cues that are the orientation ability of the individual without refer- available to long-distance migrants including geomag- ence to the implications of group membership. A simple netic and solar information in combination with an principle (the ’many wrongs principle’), first proposed internal circadian clock, stellar rotation, geographical by Bergman and Donner in 1964, and developed by topology, olfactory cues and complex interactions or cross- both Hamilton and Wallraff three decades ago, provides calibration of these cues (c.f. However, naviga- a link between these lines of current interest by tional imprecision has many sources. For example, suggesting that navigational accuracy increases with geomagnetic compass precision is reduced near the group size. With unprecedented scope for testing the equator and the poles; stellar rotational cues are unavail- hypotheses it generates, it is now time that the many able for much of the year in polar regions; and solar cues vary with season and location Navigation errorintroduced by limitations that are inherent to the Animal navigation has been a source of fascination to orientation cues themselves is compounded by additionalhazards such as wind drift Correction mechanisms humans for centuries. Navigation directly affects dis- serve to reduce directional bias, but add a further source of persal patterns, which influence conservation efforts as random error. Even if orientation cues were absolutely well as population and evolutionary dynamics. In spite of reliable, flawless navigation would require perfect sensory intense recent research that has improved our under- interpretation and integration of cues by individuals. This standing of the navigational tools used by animals, a inherent individual error is at the heart of the current consensus has yet to be reached on explanations for the controversy over whether orientation mechanisms, as they accuracy with which migration is accomplished In are currently understood, are sufficient to explain the a seemingly unrelated area, scientists continue to seek explanations for why animals tend to form groups, with Error rates measured on individual birds are implicitly recent work focusing on group decision making assumed to result in a corresponding dispersion of and complexity theory . Although migrating animals individuals around the target migration destination.
often occur in groups, the studies of navigation and This scatter is traditionally described (using circular aggregation have persisted as independent lines of distribution statistics) in terms of a mean migration research. A largely unnoticed idea that was first direction and directional concentration. The three focal suggested 40 years ago might hold the key to solving papers show how the failure of orientation systems to the current impasse: it could account for unexplained account for observed navigational accuracy could be the navigational accuracy and simultaneously offer an result of an unnoticed but flawed assumption about how individual error rates should determine directional scatter Bergman and Donner first made the intriguing sugges- tion that we should not expect to account for accuracy ofmigration by studying individual navigational error rates we should expect group error to be lower than that of The many wrongs principle is based on the idea that the individual members because group migration ‘increases pooling of information from many inaccurate compasses the accuracy of the orientation mechanism’ . Hamilton yields a single more accurate compass (because and Wallraff then placed the original suggestion individual orientation error is suppressed by group within a solid theoretical framework. Inadequacies in both cohesion . Renewed attention to the principle the technology for tracking migratory animals and might provide a vital step toward a fuller understanding of datasets available to test the principle might explain why the papers received little attention at the time of The simplest, or null model can be envisioned as one publication. They now offer an appropriate null model and in which there are no innate differences in accuracy general framework on which to base empirical tests of the among individuals within a flock, and individuals contribute equally to the mean flock direction. Underthis null model, expected flock accuracy is a functionof flock size; smaller flocks are expected to miss Corresponding author: Andrew M. Simons ([email protected]).
their target more often (. Recent work on the evolutionary advantage of aggregation posits superior group decisions based on consensus or democratic votecounting , and requires communication of infor- Given that it is the individual rather than the group that has tools fornavigation, it is at this level that studies of navigation have mation among individuals. The elegance of the many traditionally focused. Interindividual variation in flight paths is an wrongs principle lies in its simplicity: the navigational appropriate measure of scatter for birds in solitary flight [2,3], but flocks are cohesive units that share a single path, and groups must be viewed as random samples of n individuals drawn from the This flock size advantage can be generalized to more On the scale of linear measurements, for example, the standard complex scenarios. Variation in navigational abilities deviation of the mean (or the standard error) is given by sffi among individuals is expected, and has several sources.
error decreases as sample size, n, within each group increases.
For example, adult and juvenile raptors differ in their Variation among flight trajectories must be measured on an angular ability to compensate for wind drift Similarly, rather than linear scale because orientation is a compass directionthat is taken from a circular distribution that has neither true zero, nor naı¨ve migrants might orient correctly but only true high or low values. Analogously to the linear scale, it is the experienced individuals can adjust this vector naviga- angular sampling distribution of flock means rather than that of tion if displaced from the correct route. When differ- individuals that determines directional scatter [12,24]. Therefore, all ences in ability are recognized by fellow flock members, else being equal, individuals flying within larger flocks will arrive navigational responsibility might be weighted unequ- more reliably at their destination than will individuals within smallerflocks, because of the declining relationship between flock size and ally among individuals. The effective sample size of a standard error of flight trajectories (Figure I). Thus, the assumption flock for navigation would then be smaller than the that error rates of flight trajectories are given by individual error rates bird count, but the correct expectation for directional is erroneous when individuals navigate in groups constrained by scatter among groups must still consider the general their cohesiveness to follow a single trajectory.
The sampling distribution of flocks converges on that of individuals only if a single individual assumes exclusive navigational responsibility for an entire flock. However, even at this logical extreme, flocks will show little scatter ifexperienced individuals lead the flocks. Thus, the use of individual error to predict directional scatter is reasonable only for solitary flight or under the unreasonable assump-tion that a single individual of average navigational abilityleads.
Tests and implicationsThus far, the scant references to the original papers,with few exceptions, do not directly test the naviga-tional advantage of aggregation. Important exceptionsdo exist. A comparison of orientation ability in homingpigeon Columba livia pairs that were manipulated todiffer in orientation abilities shows that the trajectoryof the flock of two birds is a result of a compromisebetween the paths that the birds would have takenindividually . Randomly assembled flocks of three –six directional scatter and homing times compared withsingle birds although, in another study comparingone- and four-member pigeon flocks, no significantgroup accuracy increases with flock size in the skylark Alauda arvensis providing support for the many Figure I. The advantageous effect of large group size on navigational accu- racy. Shaded areas depict 95% confidence intervals of trajectories for With the development of tracking technology and groups of 1, 10, 100, and 1000 individuals of equal navigational ability.
the accumulation of large datasets, several testable The trajectory of a cohesive group is given by the angular mean of theindividual navigators comprising the group as calculated using angular predictions based on the automatic grouping advantage statistics [24]. The mean angular deviation is given by s ¼ 1808 should provide fruitful avenues for research. The most where r, the angular concentration, is given by the length of the mean vec- basic prediction is that, within species, directional tor. The 95% angular confidence interval of group orientation is then cal-culated as 2s. Here, a fixed value chosen for r results in an angular scatter among groups will decrease with increasing deviation s ¼ 10.368 for animals navigating alone; s ¼ 3.438 when partici- group size. Furthermore, if the ’correct’ direction can pating in groups of size 10; s ¼ 1.078 in groups of 100; and s ¼ 0.348 ingroups of 1000. In a manner analogous to this simultaneous vector sum- be inferred, larger flocks will deviate less from the mation for many individuals, accuracy can also be gained from sequential correct direction than will smaller flocks. Species trajectory summation (c.f. [25]) if a trajectory is chosen only at the begin- characterized by small group sizes are predicted to ning of each stage of a migration that occurs in multiple stages.
either have more efficient navigational tools, or to suffer greater losses during migration. Grouping is migration routes: an attempt to evaluate compass cue limitations and predicted to be more prevalent in danger zones or if required precision. J. Avian Biol. 29, 626 – 636 environmental factors limit the efficacy of orientation 3 Thorup, K. et al. (2000) Can clock-and-compass explain the distri- bution of ringing recoveries of pied flycatchers? Anim. Behav. 60, tools. If overwintering and breeding destinations differ in size, flock size is predicted to differ accordingly 4 Thorup, K. and Rabøl, J. (2001) The orientation system and migration during spring migration compared with migration in pattern of long-distance migrants: conflict between model predictions the autumn. Finally, flight paths will deviate from the and observed patterns. J. Avian Biol. 32, 111 – 119 shortest path if perilous habitat (e.g. open ocean) 5 Ba¨ckman, J. and Alerstam, T. (2003) Orientation scatter of free-flying nocturnal passerine migrants: components and causes. Anim. Behav.
occurs on one side of this path. Smaller flocks will hedge their bets by deviating landward because the 6 Conradt, L. and Roper, T.J. (2003) Group decision-making in animals.
fitness cost of a random deviation oceanward from the 7 Rands, S.A. et al. (2003) Spontaneous emergence of leaders and The interpretation of observational data might be followers in foraging pairs. Nature 423, 432 – 434 8 List, C. (2004) Democracy in animal groups: a political science confounded by factors that are themselves amenable to perspective. Trends Ecol. Evol. 19, 168 – 169 study. First, the size of a group in nature might be 9 Parrish, J.K. and Edelstein-Keshet, L. (1999) Complexity, pattern, determined by the navigational abilities of its mem- and evolutionary trade-offs in animal aggregation. Science 284, bers, in that group size might increase until a critical group orientation threshold is reached. If so, little 10 Bergman, G. and Donner, K.O. (1964) An analysis of the spring migration of the common scoter and the long-tailed duck in southern variation in error will be found among groups in spite of differences in group size. Furthermore, if there is a 11 Hamilton, W.J. III (1967) Social aspects of bird orientation cost to large group size, recognized navigational ability mechanisms. In Animal Orientation and Navigation (Storm, might influence group membership decisions. Although R.M., ed.), pp. 57 – 71, Oregon State University Press difficult, ideal tests of the many wrongs principle would 12 Wallraff, H.G. (1978) Social interrelations involved in migratory follow randomly assembled aggregations of different orientation of birds: possible contribution of field studies. Oikos 30,401 – 404 13 Able, K.P. and Able, M.A. (1995) Interactions in the flexible orientation Early work recognized that the standard error of system of a migratory bird. Nature 375, 230 – 232 group trajectories is lower than the scatter that is 14 Weindler, P. et al. (1996) Magnetic information affects the stellar predicted based on individual error rates The orientation of young bird migrants. Nature 383, 158 – 160 navigational advantage is general in that it applies in 15 Gudmundsson, G.A. and Sandberg, R. (2000) Sanderlings (Calidris alba) have a magnetic compass: orientation experiments during spring principle to all cohesive animal groups that exhibit any migration in Iceland. J. Exp. Biol. 203, 3137 – 3144 degree of random individual orientation error and 16 Alerstam, T. et al. (2001) Migration along orthodromic sun compass subsumes, as special cases, scenarios in which naviga- routes by Arctic birds. Science 291, 300 – 303 tional responsibilities are unequally divided among 17 Lohmann, K.J. et al. (2001) Regional magnetic fields as navigational markers for sea turtles. Science 294, 364 – 366 18 Thorup, K. et al. (2003) Bird orientation: compensation for wind drift in The automatic advantage of group navigation helps migrating raptors is age dependent. Proc. R. Soc. Lond. Ser. B 270, to reconcile the difference in expected and observed migratory accuracy, and provides an additional expla- 19 Perdeck, A.C. (1958) Two types of orientation in migrating starlings nation for the widespread phenomenon of animal Sturnus vulgaris L. and Chaffinches Fringilla coelebs L., as revealed aggregation. Given the current interest in both sub- by displacement experiments. Ardea 46, 1 – 37 jects, the research orientation advocated by Wallraff in 20 Burt de Perera, T. and Guilford, T. (1999) The orientational consequences of flocking behaviour in homing pigeons, Columba 1978 now merits amplification: the ‘influence of social relationships on migratory orientation of birds 21 Tamm, S. (1980) Bird orientation: single homing pigeons compared deserves, in my opinion, more attention than it has with small flocks. Behav. Ecol. Sociobiol. 7, 319 – 322 22 Keeton, W.T. (1970) Comparative orientational and homing perform- ances of single pigeons and small flocks. Auk 87, 797 – 799 23 Rabøl, J. and Noer, H. (1973) Spring migration in the skylark (Alauda arvensis) in Denmark. Influence of environmental factors on the flock I thank M.R. Forbes, J-G.J. Godin, M.M. Krohn, S. Medd, T.N. Sherratt size and correlation between flock size and migratory direction.
and J.E. Yack for comments and suggestions. This paper benefited from comments by H. Wallraff, T. Alerstam and two anonymous reviewers.
24 Batschelet, E. (1981) Circular Statistics in Biology, Academic Supported by a Natural Sciences and Engineering Research Council 25 Alerstam, T. (2000) Bird migration performance on the basis of flight mechanics and trigonometry. In Biomechanics in Animal Behaviour (Domenici, P. and Blake, R.W., eds), pp. 105 – 124, BIOS 1 Mouritsen, H. (1998) Modelling migration: the clock-and-compass model can explain the distribution of ringing recoveries. Anim. Behav.
56, 899 – 907 0169-5347/$ - see front matter q 2004 Elsevier Ltd. All rights reserved.
2 Sandberg, R. and Holmquist, B. (1998) Orientation and long-distance

Source: http://www.swarmagents.cn/bs/files/hufeng2009728222038.pdf