211 q 2004 The Society for the Study of Evolution. All rights reserved. EVOLUTION INTERNATIONAL JOURNAL OF ORGANIC EVOLUTION PUBLISHED BY THE SOCIETY FOR THE STUDY OF EVOLUTION Vol. 58 February 2004 No. 2 Evolution, 58(2), 2004, pp. 211–221 ADAPTATION AND SPECIES RANGE JOEL R. PECK 1,2 AND JOHN J. WELCH 1 1 Centre for the Study of Evolution, School of Life Sciences, University of Sussex, Brighton BN1 9QG, East Sussex, United Kingdom 2 E-mail: j.r.peck@sussex.ac.uk Abstract. Phase III of Sewall Wright’s shifting-balance process involves the spread of a superior genotype throughout a structured population. However, a number of authors have suggested that this sort of adaptive change is unlikely under biologically plausible conditions. We studied relevant mathematical models, and the results suggest that the concerns about phase III of the shifting-balance process are justified, but only if environmental conditions are stable. If environmental conditions change in a way that alters species range, then phase III can be effective, leading to an enhancement of adaptedness throughout a structured population. Key words. Adaptation, epistasis, group selection, shifting balance, species range. Received June 3, 2003. Accepted September 15, 2003. Some genotypes confer a relatively high level of fitness. However, because of epistasis (fitness interactions among loci) the alleles that constitute a high-fitness genotype may not confer high fitness when they occur in combination with other alleles. In this type of situation, we say that the high- fitness genotype is ‘‘coadapted.’’ Sewall Wright spent much of his working life investigating the evolutionary consequences of coadapted genotypes (Wright 1931, 1932, 1982). He suggested that superior co- adapted genotypes can arise and spread through a population by means of a three-phase process, which he called the shift- ing-balance process. The shifting-balance depends on the population being subdivided into a number of demes (local breeding populations). In phase I of the process, a coadapted genotype with a relatively high fitness arises in one of the demes by means of genetic drift. (For brevity, we will simply call this genotype the ‘‘high-fitness genotype.’’) In phase II, selection causes the high-fitness genotype to become common within the deme where it arose. Finally, a process of inter- demic selection sees the high-fitness genotype become com- mon throughout the entire metapopulation (i.e., throughout the entire collection of demes). Recently, Wright’s theory has been the focus of substantial negative criticism (Barton 1992; Gavrilets 1996; Coyne et al. 1997, 2000). The critics conclude that problems with phase III, the process of interdemic selection, constitute ‘‘the cen- tral weakness of the theory’’ (Coyne et al. 1997). This echoes the thoughts of Haldane (1959) in one of the earliest critiques of the shifting-balance process. The central difficulty with phase III, as recognized by Wright, is that individuals with a coadapted high-fitness ge- notype invading an area where other genotypes are common will find that, when they mate, their genomes are broken apart by segregation and recombination (Wright 1931, 1932, 1949, 1970, 1977, 1982; Barton and Hewitt 1989; Coyne et al. 1997). Thus, as a consequence of their genetic coadaptation, migrants with the high-fitness genotype may produce low- fitness offspring. This can prevent the spread of the high- fitness genotype. In a defence of phase III, Crow et al. (1990) used a model designed to be unfavorable to the efficacy of the shifting- balance process. They concluded that phase III was remark- ably successful, and that quite low levels of migration from the deme with the high-fitness genotype were sufficient to ensure its spread, even in the face of reverse migration. How- ever, these results were reanalyzed and reinterpreted by Bar- ton (1992). Barton’s results suggested that, unless selection is very strong or the number of loci is very small, the outcome of evolution depends only weakly on any fitness advantage of the coadapted genotype, and more strongly on factors un- connected with adaptation. The most important of these fac- tors are the dominance relationships of the alleles that com- prise the coadapted genotype, and the relative rates of mi- gration between the demes. Barton showed that excessive migration from a deme could lead to the spread of the ge- notype it contained, even if this genotype conferred a selec- tive disadvantage. Crow et al.’s results, he suggested, were an example of migration overwhelming recombination and selection. Barton’s conclusion was that while a phase III type of process could occur, the result was not likely to be adaptive (Barton 1992; Phillips 1993; Rouhani and Barton 1993). In Wright’s original formulation of the shifting-balance