emergence of old drivers, it also greatly hinders the invasion OF previously unseen drivers. Drivers are typically associated with chromosome regions in which Sirs, there is an absence of recombination and hence they Crow's discussion(') of why Mendelian segregation is exist in very tight The frequency of crossing the rule and not the exception places great emphasis on over has been used to account for, at least in part, the the importance of suppressors of meiotic drivers. Crow relative abundance of X Y drive above autosomal shows that driver-specific suppressors readily evolve to drive('). evict the meiotic cheat from a population. Though we We also expect to find other general devices agree with what Crow argues, we believe that his emp1oye.d to protect meiosis from drivers. Though sure viewpoint does not represent the whole qtory. We evidence is wanting, there is a sizeable body of would like to elaborate on Crow's view on two counts. circumstantial evidence indicating that invasion of First, we question what happens to Crow's models if drivers might be prevcntable by the evolution of devices there exists a cost of suppression. Second, we draw which guarantee mutually assured destruction to attention to meiotic mechanisms which can be indrivers. That is, if one chromosome attempts to drive terpreted as general mechanisms to hinder the invasion and this is detected, then thc non-driving chromosomes of drivers. cause complete meiotic breakdown. To use a sporting Crow's model takes no account of any potential cost mctaphor. when in a rugby scrum the opposition's pack of suppression. This is a key restrictive assumption. We is forcing your retreat, it can be in your interests to have shown analytically that, as one might intuitively collapse thc scrum and give away a penalty rather than expect, if the possession of a suppressor involves a cost, the more costly try. The sex vesicles (Mammals)@)and the driver is not evicted and the modifier is not fixed but sex parachutes (Coleoptera)('), which surround the sex rather they are held in stable polymorphism('). It has chromosomes during meiosis, can be interpreted as been shown in SL, that insensitivity is costly(3) and it is such detectors of drive and guarantors of collapse. highly likcly that suppression i s as well. If so, specific Other mechanisms rcstrict the ability of chromomodifiers of drive alone cannot explain the presumed somes to label or damage the opposite pole during predominance of Mendelian segregation. meiosis. For example, the heterochromatic nature of Two possibilities remain. First, that Mendelism is not chromosomes during meiosis makes them transcriptionthe rule but the exception and/or second, that other ally inert and so unable to easily seed homologues explanations are required. Support for the first proposition has been provided by H i r a i ~ u m i ( ~ ) . during pairing. Further, McKee(') has conjectured that pairing of chromosomes at meiosis is the signal for Detailed examination of the segregation frequencies of inactivation. It is thus significant that pairing abnorchromosome 2 homologues in Drosophiln (without SD) malities do result in drive('). Similarly, up until the revealed small but significant departure from 1:1 point of release of a spermatozoan numerous spermafrequencies in the heterozygous condition, attributable tids are maintained within a syncytium. Thc production to pre-fertilisation events. The segregation frequency of a RNA transcript in one of the spermatids results in was also a function of the age of the individual all spermatids receiving the transcript(''). Such a concerned. Some small but significant deviations from syncytium ensures that for a driver to work it must not 1:l have also been re orted for human sex chromosomes in spermatozoa(-) (but see Martin(')). However, only seed its homologue with toxin but it must also be immune to the toxin. It is possibly this which underlies it is generally assumed that fair meiosis is the rule the fact that driver systems always involve at least two although rigourous demonstration is wanting. loci and hence are faced with the problem of remaining There is also support for the second possibility. The in linkage disequilibrium. view we favour is that meiosis is a delicate process A consequence of these 'clamps' on deviation from seeded with general anti-meiotic drive devices. Crow(') fair segregation is the difficulty breeders find in touched briefly on one such anti-drive mechanism, selecting on the sex ratio("). If therc were numerous namely crossing-over. Typically drivers are multi-locus specific modifiers present, that changed the degree of effects, requiring linkage disequilibrium between distortion, selection on the sex ratio should be as driver(s) and insensitivity loci. This linkage disequilistraightforward as selection on any polygenic trait. brium is maintained by an absence of crossing over However, attempts to modify primary sex ratios have between the relevant loci. It has been suggested that only been able to cause minute change("). crossing over evolved and is maintained as an anti-drive Another corollary of our view that meiosis is a mechanism(7). If this view is correct then crossing over delicate process sceded with general anti-drive devices can be viewed, at least in part, as one of Crow's is that rather than seeing drive cxposed in hybrids, as suppressors of drive. However, unlike Crow's implicit Crow presumes, one might witness sterility. This view of driver-specific suppressors, this rncchanism can be maintained for its general anti-drive capability rather argument, aligned with the absence of recombination than its ability to suppress one particular driver. The between X and Y making X / Y drive far more probable prescnce of crossing over not only prcvents the rethan X / X drive, has been put forward to explain why

Maintaining Mendelism: Might Prevention be Better than Cure?

P

when one sex is disturbed in h brid crosses, it is 1. typically the heterogametic Thus in sum, although we agree with Crow that the invasion of a particular driver will typically be repressed by a particular set of suppressors we feel that this does not fully explain why segregation is usually fair, if indeed it is. Rather, we maintain that for a driver to invade it must first evade a number of general anti-drive mechanisms. The cure is only necessary if the preventative medicine did not work. These general devices are probably in place because of the long history of problems with drivers. Suppressors of both the general and specific type we expect to be continually coevolving with drivers. LAURENCE D. HURST Department of Zoology, South Parks Road, Oxford, OX1 3PS UNTTED KINGDOM

2

ANDREW POMIANKOWSKI Department of Genetics and Biometry University College London London, NW1 2HE UNTTED KINGDOM References 1 CROW,J . F. (1991). Why is Mendelian segregation so exact? BioEssuys 13; 305-312. 2 HURST, L. D. AND POMIANKOWSKI, A. (1991). Causes of sex ratio bias may account for unisexual sterility in hybrids: A new explanation of Haldane's Kule and related phenomena. Genetics, in press 3 WII. C-I., TRUE,J. R. AND JOHNSON, N. (1989). Fitness reduction associated with the deletion of a satellite DNA array. Nature 341, N8-251. S. (1969). Paternal age and segregation 4 HIRAIZCMI, Y. A N D GROVE,J. frequency in Drosophila melanogaster. Jap. J. Genet. 44 (Suppl. 1). 193-199 5 BEAN,B. (1990). Progenitive sex ratio among functioning sperm cells. Am. J. Hum. Genet. 47. 351-353.

6 MAXIIN,R. H. (1990). Sex ratio among sperm cells. Am. J. Hum. Gmct. 47, 349-351. 7 HAIG, D. AKD GRAFEY, A . (1991). Genetic scrambling as a defense against I . Theor. Diol., in press meiotic drive. . 8 WHITE,M. J . D. (1973). Animal Cyfology and Evolution. Ed. 3 Cambridge University Prcss. Canihridge. 9 MCKEE, B. (1991). X- Y pairing, meiotlc drive and ribowmal DNA in Drosophila melanogustcr males. Am. /Vat. 137. 332-339. 10 C A T ~ W F .K. L LA. . AXD HANDEL. M.A. (1991). Protamine transcript sharing among postmeiotic spermatids. Proc. h'afllh a d . Sci. USA 88, 24W-2411. 11 MAYNARD SMILH,J . (1978). The evolution of sex. Cambridge University Preas, Cambridge. 12 FRANK. S. A. (1991). Divergence of meiotic drive-suppression systems as an explanation for sex biased hybrid sterility and inviability. Evolution 45,262-267

James F. Crow replies: Hurst and Poiniankowski suggest that I have given too much emphasis to one mechanism, drive-reducing modifiers, for keeping Mendelian ratios at, or close to, 1:l. They may indeed be correct. Any process as ubiquitous and important as Mendelian segregation is expected to have evolved a number of protective devices. Therefore, without necessarily assigning importance to all that these writers suggest, 1welcome the mention of possible mechanisms additional to the onc 1 discussed. It is often very hard to learn the 'real' function of a generalized mechanism. For example, crossing over adds to the difficulty of establishing some drive systems and thus acts as a preventative. But crossing over has evolutionary advantages on its own, and chiasmata are useful for preventing nondisjunction. So, is drive prevention a major reason for crossing over, or a side effect? There is plenty of room for further ideas and research, and Hurst and Pomiankowski have provided food for thought.