Gene Flow James Mallet 16 Calton Laboratory Department of Biology U - PDF document

Gene Flow James Mallet 16 Calton Laboratory, Department of Biology, Univetsity College What is Gene Flow? 'Gene flow' means the movement of genes. In some cases, small fragments of is rare. Most of the time, gene flow is caused by is used probably because of an implicit belief in is still based on simple Two Meanings of 'Gene Flow' We are often taught that 'dispersal does not necessarily lead to gene flow'. sense of a final state of actual EICAR International 2001. Insect Movement: Mechanisms and Consequences (eds I.P. Woiwod., D.R. Reynolds and C.D. Thomas) 338 1f let movement of genes. The tension between actual movement and successful is at the heart of many misunderstandings of the term 'gene stare of populations) and obtain an estimate of N(the product of population size and fraction tells u

s almost nothing about actual fraction of the m. Actual gene flow, m, could be very high or very low, species concept, actual gene flow, As with 'gene flow', the is all too easy to become deluded into N and then to attempt to use this knowledge in m, between populations, is fine provided one realizes at species may not actually be reproductively isolated: Which 'Gene Flow' do we Want to Measure? Both types of gene flow are of course worth measuring. However, the Gene Flow 339 an invading species is expected to be v, =12ra.,,et al., 1993). In contrast, 'gene flow' in the sense of the races or species. The Balance Between Gene Flow and Genetic Drift One of the simplest effects of gene flow is on genetic drift in interconnected Fsr =1 (2,Nrin each generation. N., is the 'effective size' of an N. Beca

use some individuals in a population can is often less than, although of the as, N (Nunney, 1993). Frankham (1995) reviewed data which N, may typically be as low as 5-10% of the value of N. In the Ty, subpopulations would diverge so that a fraction Ty of the populations iy(1 - Ty) when all populations are fixed, i.e. 2 F57— Ti(can reach an equilibrium when drift, which causes divergence of neutral is balanced by the homogenizing N,; migrants can come from anywhere else in the metapopulation or from a Migrant pool q 340 J. Mallet (a) (b) ........... .......... Distance Fig. 16.1. Models of population structure. (a) Wright's island model. A large A fraction m of each the average over all subpopulations. Under this model, the potential for local drift then depends on the combined parameter isolation

by distance model. In this model, population structure so that individuals are distributed continuously in a potential for local drift can be shown to depend on the NE, = 4rrp2, is is again a product of gene flow and population numbers, and it plays a role in Pr Gene Flow 341 local subpopulations (Slatkin and Barton, 1989). If, in addition to local distances, Fis no long-distance dispersal, mutation or weak selection to FsT will increase with distance as drift within Ndrift q = 0 or q = 1, sc, Fs-1- —) 1. Conversely, if populations Nmigration out- —1 0. The Nis therefore a kind of cusp between these two regimes, where Nas a product of population size M, especially by 11 1 Fsr However, it is worth remembering that this 'gene flow' in fact consists of a is formed from a ratio of these probabilities

, is to be proportional sense, then, the interpretation of Non as a number-based form Fsr, rather Kto 'predict' equilib-FsT under neutral drift, but this is of very limited use because FsT is: it is the information from which we FsT from which this 'gene flow' Kwhether based 342 J. Mafia on FsT, rare alleles (Slatkin, 1987) or gene genealogies (Slatkin, 1989, 1993; is symmetrical around the D, much used in ecology et aL , 1993): in fact, D = „ /2 . Assuming that 86.5% of offspring +. , around the birthplace of 95.4% probability integral of p, (Wright, 1969). Drift enters lip,, a probability term with units DISTANCEcrrepresents the probability of movement also with units Nb, although numerically Nb has Nb are neither measures of across which FsT is measured will determine its magnitude (Wright, 196

9; Slatkin, tand Nb, it is hard to infer anything very useful about actual gene Mainly aabb Mainly AABB BB AA Bb AA BB Gene Flow 343 apparently attempted (see the useful reviews by Roderick, 1996; Peterson and FsT do give some idea of FsT is likely to (m or crspecies with low FFsrr is even et al., 1993). I defend this paper on the grounds that we Nm (see below). Multilocus Data and Interlocus Correlations Because dispersal causes a flow of whole genotypes, or 'genotype flow', it FST. In principle, therefore, combined data on A B increases 6 decreases Fig. 16.2. Gene flow and gametic correlations. A pair of populations are assumed ab gametes, but immigrants will bring in mainly AB gametes. If 344 1. Mallet Fsr and linkage disequilibria can be used separately to estimate the gene (Nor NO at

equilibrium: two independent effects are et al., 1999). Ceratitis capitata (Weiclemann) (Davies et al., 1999a,b). However, it is as yet Spatial Information in Gene Flow/Drift Balance Spatial information present in the measurement of gene frequencies under (Fs-r) is controlled by Nb, while the spatial extent of these fluctu-see also Fig. 16.3). The Gene ct,

Height of fluctuations x Fig. 16.3. Drift in continuous populations. In a continuous population, gene Ne,. Gene Row distance at which gene frequency fluctuations in local populations become FsT have been used to Heliothis virescens (F.) we collected all samples within a single generation, and showed that allozyme is restricted at this scale, even though FUSA (Korman et al., 1993). Heliothis is manageable on a local Nwith distance can similarly be used to give an idea of the degree of is possible to use 'spatial autocorrelation' (Sokal et al., 1989) to et al., 1999), but direct 1997). An interesting feature of drift in two-by selection could in Nb and cr.„ (Barton and Clark, Direct Estimates of Dispersal and "Slatkin's Paradox' A

puzzling phenomenon in empirical studies of population structure is known as ‘Slatkinis paradox'. Direct estimates of effective population size and NI, (-0.1-10; e.g. see 1995) which predict high expected equilibrium levels of FST. In FsT at marker loci such as allozymes, especially in sizes (Ni, N10; for reviews of insect cases see McCauley and Eanes, 346 J. Mallet 1987; Roderick, 1996; Peterson and Denno, 1998). For example, the butterfly Euphydryas editha (Boisduval) differs little in allozyme frequencies between disperses across the vast Great Basin desert (Slatkin, 1987). FsT at different loci every locus. FsT values resulted from Pbalacrus substriatus (G-yllenhal) shows clearly that long-distance colonizations cause a much FsT compared with as found by Whitlock (1992) Bolitotberus cornutus (

Panzer). Thus, FsT may escape the Drosophila pseudoobscura Frolova as a = 59-81 m Gene Flow 3 47 that local genetic drift could be quite extensive in this species. However, later D. pseudoobscura females dispersing many tens of kilometres across desert habitat (Jones et al., 1981; Coyne et al., 1982). Of course, dispersal over desert may be enhanced were done. None the less, is a long-distance 'tail' containing significantly more use of cr,is to cause the value of IV itself to be greatly underestimated Gene Flow — Selection Balance One reason why inferring gene flow indirectly is so difficult is that, in rea-m are large, FsT or disequilibria are expected to Heliothis (Korman et at., 1993). et al., 1998). m and s (s is the fractional increase of mortality of genotypes 4, nth, providing that m (Hald

ane, 1930). This has potential applications for the possibility of 348 J. Mallet the evolution of resistance in a treated population until the selection for mis (Comins, 1977). Minimum Size of an Area for Adaptation The balance between selection and gene flow is also tractable when we 'slave of the resistance (or any other adaptation) will where selection on an s) and (1 - s) in two adjacent areas is given by -11(3CF Is) (Fig. 16.4; Slatkin, 1973; Endler, 1977; Roughgarden, 1979). s 5 0.2. For insecticide resistance to evolve, the critical diameter dTr ax ticide resistance can attain a high frequency inside the treated area (Fig. 16.4; Indirect Measurement of Gene Flow in Clines Although theory relates cline width to gene flow s in a is again hard to reverse the equation. We will s separately rath

er than their ratio. An estimate of crxi-gc, being merely a NIf we had direct measures of either crs, we case of chromosomal or butterfly warning Distance \t- Gene Flow 349 underestimate of s, which enters into the cline equation as a square root. Even s are A possible exception occurs when a cline is moving as a wave of advance. (1991) estimated dispersal in Drosophila simu.lans Sturtevant from data on a moving cline of maternally inherited Wolbachia parasites. The Wolbachia infection, which causes unidirectional sexual 175 km yearWolbachia, and laboratory estimates of selection pressures. Again in accordance with on D. pseucioobscura (\Wright, 1969). Nbased on Nfrom F assumes that drift and gene flow have reached equilibrium. However, the Environment a WAA=-s Waa=Environment A .,,q=1 wFig. 16.4.

Selection in continuous populations. Selection, s, in a continuous a on the left of the diagram, and allele A on the right. 350 J. Mallet hiatus in gene frequency across a cline implies either a temporary situation is maintaining allelic divergence (albeit possibly Gene Flow in Multiple Locus Clines Often, multiple locus clines occur together in hybrid zones. For example, in Heliconitis erasto (L.), three loci determine colour pattern differ-Heliconius are warningly of selection suggest that cline widths at equilibrium should stabilize at w V(80s) to w -4126„s), depending on dominance (Mallet and Barton, 1989a). causes the immigration As first such clines of allele frequency R AB between a pair of genes, A B of approximately RAB 4cr,c2 C ABW near the centre of the hybrid zone where gene frequ

encies are 50%. Here, R AB is the interlocus correlation coefficient, wA and wB are the widths of B, and CAB is the recombination rate. The correlation reduces it. By sampling genotypes, we wB as well as RAB directly. Gametic correlations can also, or epistasis. In Heliconius, fitnesses at the different loci were assumed is indeed likely in ax and s, which H. era() led to estimates of 01„ 2.6 km gen.s 0.15-0.33 gen.et al., 1990, 1998a). An estimate of Gene Flow 351 selection based on mark—recapture agreed approximately with the indirect Culex L. has used similar techniques to obtain robust estimates of selection Ester. These loci are linked with recombination rate c 145%. Resistance is against in untreated areas inland, leading to the formation of clines Ester respectively. Gene flow was estimate

d as et al., 1998). Culex were able to show variation in selection et al., 1999; Lenormand and were consistent with laboratory data (Lenormand and et al. (1999) recommend that if the coastal Culex story is perhaps the most complex and informative analysis of ever achieved in the field, and the results are of 352 J. Mallet Gene Flow Between Host Races or Species As we move up the evolutionary continuum from populations through to use mass-action population genetic theory similar to that Rbagoletis pomonella (Walsh), disequilibrium measurements suggest races was about m 20% per generation (Barton et al., 1988; Feder et al., 1988, 1990). m 6% (Feder et al., 1994). However, is something that, under is probably et al., 1998a). Few Drosophila species are melanogaster Meigen, pseudoobsagra Frolova, Hawai

ian picture-wing, and studied (Gupta et al., 1980). It seems Drosophila is as Nifi measures 'gene flow' between Fsr = and N= 0 is inferred. less practical for estimating rates of gene flow. Heliconisis erato and Gene Row 353 Heliconius himera Hewitson, hybrids identified via colour pattern were et aL, 1997). In a direct et al., 1997; McMillan et al., 1997; Mallet et al., 1998b). In another Heliconius charithonia (L.) and Heliconius peruvianus Felder were identified via allozyme genotypic signatures (jiggins As already mentioned, this genotypic identification et al., 1999; Davies et al., 1999b). "We have Zeirapbera Guenee (Emelianov et al., 19%). These estimates are similar to et al., 2000). species ever occur? Hybridization itself is: can genes be transferred via backcrossing between populations c

ases, evidence strongly supports D. melanogaster originated recently from elements common Drosophila willistoni group, and were transferred, perhaps et al., 1991). Some individuals of Drosophila Tsacas & Davis, a species in the simulans group endemic to species D. simulans Sturtevant, while other D. ma tiritiana have a more D. simulans into D. mauritiana, presumably via hybridization (Solignac and Monnerot, 1986). Adh compared with other genes in the D. psettdoobscura group suggest transfer of Adh genes between closely et al., 1997). If these interpretations are correct, 354 J. Mallet Prospects for Estimating Gene Flow The use of selection/gene flow balance and disequilibrium across clines to cases. Apart from et al., 1994). In every case, development of new population genetic theory et al., unpub

lished). Every other example studied so far had a unique mode of et al., 1997). Practitioners of these newer gene flow methods will have uselessness of Hcombined parameter, the slow attainment of equilibrium 'yes'. At minimum, Fsi- must avoid unrealistic since the 1930s, when Dobzhansky's hopes D. pseudoobscura were is especially unwise to N 77i when a cline or other situation clearly Gene Flow 355 Conclusions Estimating gene flow, or its converse, reproductive isolation, is essential for reviewed the somewhat confused current state of the field. In particular, an impression is often based on field observations of local insect N or 'reproductive isolation' in the sense of pre- and post-mating where selection/gene flow balance can these new methods will Summary Insect populations have provided many

of the best studies of gene flow, a diverse as population refers to N (effective popula-N 071 and its continuous population analogue, 'neigh- 356 J. Mallet host races or closely related species that hybridize, albeit rarely. I give beetles, Heliconiess butterflies (Nymphalidae) and other Lepidoptera, Cu/ex Drosophila (Drosophiliclae). These newer Ai ri9 and N. Acknowledgements This work was supported by NERC and BBSRC grants. I am very grateful N.791 and Nb, and for the prototype of Fig. 16.3 which traces to an unpublished grant c. 1984. References Andow, D.A., Kareiva, P.M., Levin, S.A. and Okubo, A. (1993) Spread of invading Evolution of Insect Pests: Patterns of Variation. John Wiley z Sons, New York, Molecular Marken, Natural History and Evolution. Chapman & bilobdturn (Chiroptera: Pyllostomat

idac). Evolution 36,863-866. Population Genetics and Evolution. Springer-Verlag, Berlin, pp. 115-174. Hybrid Zones and the Evolutionary Process. Oxford University Press, The probability of fixation of a new karvotype in a continuous population. Evolution 45,499-517. Genetical Research, Cambridge 75,179-198. Trends in Ecology and 13,202-206. Journal of Theoretical Biology 64,177-197. Gene Flow 357 Ncw methods employing multilocus genotypes to select or exclude populations as Genetics 153,1989-2000. Drosophila. American Naturalist 119,589-595. Ceratitis capitata: source estimation using DNA sequences at multiple intron Genetics 153, 351-360. individuals: multilocus genotyping in nonequilibrium population genetics. Trends in Ecology and Evolution 14,17-21. Zeiraphera diniana (Lepidoptera: Tortricidae

): polymorphism, Evolution (in press). Geographic Variation, Speciation, and Clines. Princeton Evolution 51, 672-681. Genetical Research 73,251-261. Rhagoletis pomonella. Nature 336, Evolution 44, Proceedings of the National Academy of Sciences USA 91,7990-7994. Naturalist 109, 359-368. Genetical Research, Cambridge 66,95-107. Les ProWernes de l'EspVol. I. Societe zoologique de France, Paris, pp. 129-201. Drosophila. Experientia 36,290. Proceedings of the Cambridge Philosophical .Society 26,220-230. Principles of Population Genetics, 2nd edn. Drosophila genes by the mite Proctolaelaps regalts253, 358 I Mallet Ing-varsson, P.K. and Olsson, K. (1997) Hierarchical genetic structure and effective Ph alarms substriatus. Heredity 79,153-161. Heliconius charithonth (Lepidoptera: Nyrnplialidae). Biologica

l Journal of the 64,57-67. Heredity 79,495-505. Drosophila 98,157-173. Ecological Monographs 52,261-282. Heliothis (F.) (Lepidoptera: Noctuidae): an estimate of gene flow. Annals of the of America 86,182-188. Proceedings of the Royal Society of London Series B 265,1985-1990. American Naturalist 155,70-82. case study. Genetics 149,13.83-1392. Culex pipiens. Nature 400, Philosophical 354,1-18. Heliconius erato (Lepidoptera: Ny.mphalidae). Oecologi a 68,210-217. Encyclopedia of Biocliversity, Vol. 5. Academic Press, New York, pp. 427-440. Genetics 122,967-976. Evolution 43,421-431. Heliconius hybrid zones. Genetics 124,921-936. Endless Formes: Species and Speciation. Oxford University Press, New York, Heliconims species in the wild. Evolution 52,503-510. Tetraopes tetraophthalmus (Forster). Heredity 53,193

-201. Gene Flow 359 Proceedings of the National Academy of Sciences USA 94, 8628-8633. Genetics 80, 595-615. Evolution 47, 1329-1341. Genetic Structure and Local Adaptation in Natural Insect Populations. Chapman & Hall, New York, pp. 263-322. (Limenitis: Nymphalidac). Systematic Zoology 39, 131-148. and Evolutionary Biology. University of Chicago If al Review of Entomology 41, 325-352. Theory of Population Genetics and Evolutionary Ecology: an Introduction. Macmillan, New York. F-statistics under isolation by distance. Genetics 145, 1219-1228. Sceloporus grammicus complex Evolution 49, 9-36. Genetics 75, 733-756. Science 236, 787-792. Genetics 121, 609-612. Evolution 47, 264-279. Evolution 43, 1349-1368. Genetics 123, 603-614. Genetics 121, 845-855. Drosophila simulans, D. mauritiana, and D. seche

llia inferred from Evolution 40, 531-539. Papilla (Insecta: Lepidoptera): poor fotrrnal of Zoology 68, Bombina bombin a and B. variegata: comparisons Evolution 45, 237-261. Drosophila. Nature 353, 440-442. 360 I Mallet Vitalis, R. and Couvet, D. (2001) Estimation of effective population size and Genetics (in press). J. and Hey, J. (1997) Gene flow and natural selection in the Drosophib pantdoobscurra and close relatives. Genetics 147,1091-1106. Genetic Data Analysis H. Sinauer Associates, Sunderland, American Naturalist 139,952-970. Evolution 44,1717-1724. Heredity 82,117-125. Evolution and the Genetics of Populations, Vol. 2. The Theory of University of Chicago Press, Chicago, Illinois. Evolution and the Genetics of Populations, Vol. 4. Variability and Among Natural Populations. University of Chi