168

Update

TRENDS in Ecology and Evolution

Vol.19 No.4 April 2004

different selection pressures in different geographic regions, and (ii) the case where an allele originated in one region but, helped by its selective advantage, spread geographically? Both processes can give rise to an allele frequency cline but would have very different evolutionary causes. In human evolutionary genetics such questions arise routinely, for example in relation to (infectious) disease, diet or environmental factors and their inuence on the genetic make-up of populations. Mathematical models in evolutionary biology The paper by Galvani and Slatkin serves as a useful illustration of how and when to use mathematical models in biology. Useful mathematical models enable us to represent our knowledge effectively or to make predictions and test hypotheses (for a similar example see reference [10]): for example, whether plague could account for the selection pressure on the CCR5-D32 allele. The plague connection was an excellent story to tell, but the analysis indicates that the temporal correlation between the estimated allele age and the historical plague epidemic was probably coincidental. Although the mathematical model on its own would not hold too much sway, it is the combination of geographical and temporal data and clinical information with the mathematical model that is persuasive. In light of the combined evidence, smallpox emerges as a plausible source of the selection pressure on the CCR5-D32 allele. Strictly speaking, there is probably still a continuum of possible selection pressures to be evaluated and, generally, we will not be able to do more than rule out a model of neutral evolution. But Galvani and Slatkin have raised the bar considerably: if an alternative hypothesis (e.g. polio or any combination of diseases) is to replace smallpox, then it
| Letters

has to explain the population dynamics, correlation of spatial distribution with historical disease and the population genetics of the CCR5-D32 allele better than smallpox does; and it has to be biologically more plausible.
Acknowledgements
We thank A. Galvani, P. Harvey, M. Lipsitch, O. Pybus, M. Slatkin and two anonymous referees for their comments about the article.

References
1 Mummidi, S. et al. (1998) Genealogy of the CCR5 locus and chemokine system gene variants associated with altered rates of HIV-1 disease progression. Nat. Med. 4, 786 793 2 Schliekelman, P. et al. (2001) Natural selection and resistance to HIV. Nature 411, 545 546 3 Galvani, A.P. and Slatkin, M. (2003) Evaluating plague and smallpox as historical selective pressures for the CCR5-Delta32 HIV resistance allele. Proc. Natl. Acad. Sci. U. S. A. 100, 15276 15279 4 Stephens, J.C. et al. (1998) Dating the origin of the CCR5-Delta32 AIDS-resistance allele by the coalescence of haplotypes. Am. J. Hum. Genet. 62, 1507 1515 5 Libert, F. et al. (1998) The deltaccr5 mutation conferring protection against HIV-1 in Caucasian populations has a single and recent origin in Northeastern Europe. Hum. Mol. Genet. 7, 399 406 6 Slatkin, M. (2001) Simulating genealogies of selected alleles in a population of variable size. Genet. Res. 78, 49 57 7 Sullivan, A.D. et al. (2001) The coreceptor mutation CCR5Delta32 inuences the dynamics of HIV epidemics and is selected for by HIV. Proc. Natl. Acad. Sci. U. S. A. 98, 10214 10219 8 Enard, W. et al. (2002) Molecular evolution of FOXP2, a gene involved in speech and language. Nature 418, 869 872 9 Mead, S. et al. (2003) Balancing selection at the prion protein gene consistent with prehistoric kurulike epidemics. Science 300, 640 643 10 Lipsitch, M. and Sousa, A.O. (2002) Historical intensity of natural selection for resistance to tuberculosis. Genetics 161, 1599 1607
0169-5347/$ - see front matter q 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.tree.2004.02.001

Democracy in animal groups: a political science perspective

Christian List
Department of Government, London School of Economics, London, UK, WC2A 2AE

In a recent paper [1], Conradt and Roper investigated democratic decisions among animals. Some red deer herds move when 62% or more of the animals stand up. Honeybees reach consensus on nest sites in coordinated dances. Voting behaviour includes body postures, movements and vocalizations. Why does democracy give animals an evolutionary advantage? Conradt and Roper suggest it is because democracy produces less extreme decisions than despotism. Drawing on the jury theorem of the 18th-century French thinker the Marquis de Condorcet, I suggest a further explanation: Democracy is good at pooling information from different individuals.
Corresponding author: Christian List (c.list@lse.ac.uk).
www.sciencedirect.com

Conradt and Ropers model addresses group decisions about starting or stopping some synchronous activity. Individuals vote on the basis of their activity budgets. Condorcets model can represent group decisions under uncertainty whose payoffs (costs and benets) depend on some feature of the environment: whether a predator is nearby, whether food exists at some site, whether some travel route is optimal. Individuals vote on the basis of noisy, but partially reliable signals: Some might have noticed a predator or food source (a positive signal), whilst others might have not (a negative signal). Condorcet proved the following [2,3]: suppose each signal is correct (i.e. it matches the relevant feature) with a probability p above 0.5 but below 1, and different signals are mutually

Update

TRENDS in Ecology and Evolution

Vol.19 No.4 April 2004

169

Correct majority decision (probability)

independent. Then the probability that a majority among n signals (binomially distributed) is correct equals ! X n k p 1 2 pn2k kln=2 k As illustrated in Figure 1, this probability exceeds p (except for small even n, where majority ties are frequent) and approaches 1 as n increases, by the law of large numbers. A democratic decision among n individuals is more likely to be correct than a despotic decision by one individual. Condorcets model has testable implications for animal group decisions. Suppose payoffs of correct and incorrect decisions are symmetrical for different equally probable states of the environment [4]. The best nest site might be in one location or another, other things equal. Here, majority voting is optimal: It maximizes the probability of a correct decision. Suppose payoffs are asymmetrical or one state is more probable than another [4,5]. When a group nds a modest food source in a harsh environment, false positive decisions (searching further when no better food source exists) are more costly than false negative ones (not searching although a better one exists). Here, special majority voting is optimal: Requiring strong support for abandoning a food source makes false positives less probable. When there may be predators, false negative decisions (not moving when a predator is present) are more costly than false positive ones (moving although none is present). Here, submajority voting is optimal: Allowing a few warning signals to alert the group makes false negatives less probable. Similarly, special majority voting is used in criminal trials to protect the innocent, and submajority voting in decisions on considering legislative initiatives to avoid overlooking valid initiatives. Democracy still outperforms despotism under several generalizations of Condorcets model, including differentially reliable signals among individuals [3,6], certain dependencies between individual signals [3,7], nonbinary decisions [8], and some but not all cases of strategic voting [9]. Democratic information pooling has been noted, but not formally explained, in relation to the choices of nest sites by honeybees [10], where Conradt and Ropers activity synchronization model does not apply. I suggest the importance of Condorcets model for explaining the use of democratic methods in animal group decisions under uncertainty.

0.9 0.8 0.7 0.6 0.5

10

20

30

40 50 60 Group size

70

80

90

100

TRENDS in Ecology & Evolution

Figure 1. Probability of a correct majority decision (y-axis) for groups of size n 1 to 100 (x-axis), where each individual signal has a reliability p 0.6. A correct majority decision is slightly more probable for odd n (top curve) than for even n (bottom curve) because majority ties are impossible for odd n but possible for even n.

Acknowledgements
I thank Brett Calcott, Ben Jeffares, Peter McBurney and Kim Sterelny for helpful discussions.

References
1 Conradt, L. and Roper, T.J. (2003) Group decision-making in animals. Nature 421, 155 158 2 Grofman, B. et al. (1983) Thirteen theorems in search of the truth. Theory Dec. 15, 261 278 3 Borland, P.J. (1989) Majority systems and the Condorcet jury theorem. Statistician 38, 181 189 4 Ben-Yashar, R.C. and Nitzan, S.I. (1997) The optimal decision rule for xed-size committees in dichotomous choice situations: the general result. Int. Econ. Rev. 38, 175 186 5 List, C. On the signicance of the absolute margin. Br. J. Philos. Sci. (in press) 6 Nitzan, S. and Paroush, J. (1982) Optimal decision rules in uncertain dichotomous choice situations. Int. Econ. Rev. 23, 289 297 7 Ladha, K. (1992) The Condorcet jury theorem, free speech and correlated votes. Am. J. Polit. Sci. 36, 617 634 8 List, C. and Goodin, R.E. (2001) Epistemic democracy: generalizing the Condorcet jury theorem. J. Polit. Philos. 9, 277 306 9 Feddersen, T. and Pesendorfer, W. (1999) Elections, information aggregation, and strategic voting. Proc. Natl. Acad. Sci. U. S. A. 96, 10572 10574 10 Seeley, T.D. and Buhrman, S.C. (1999) Group decision making in swarms of honey bees. Behav. Ecol. Sociobiol. 45, 19 31

0169-5347/$ - see front matter q 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.tree.2004.02.004

www.sciencedirect.com