Heredity 54 (1985) 85—98 The Genetical Society of Great Britain Received 12 June 1984 Genetic variation in recombination in Drosophila. II. Genetic analysis of a high recombination stock

Brian Charlesworth and School of Biological Sciences, , Deborah Charlesworth Brighton BN1 9QG, UK.*

The difference in Gl-Sb recombination frequency between two stocks of Drosophila melanogaster, H and C, has been analysed genetically. (The H stock was obtained by artificial selection for increased recombination.) Chromosome substitution experiments show that all three major chromosome contain genes which contribute to the H-C difference, and that there are significant interactions between the genes on different chromosomes. In particular, the H 2nd chromosome increases recombination on a background of homozygous H and X and 3rd chromosomes, but reduces it if either the X or 3rd is made homozygous for C. The H X chromosome exhibits partial or complete dominance over C, depending on the genetic background, while the H 2nd and 3rd chromosomes have dominant effects. Mapping experiments show that at least two recombination genes are located on chromosome 3, one between h and Gl, the other to the right of st, probably between Standsr. Previous data suggest that the X and 2nd chromosome effects are each contributed by single genes or tightly-linked gene complexes. These results are discussed in relation to previous work on the genetics of recombination, and to the patterns of response to selection described in the accompanying paper.

INTRODUCTION to localise the 3rd chromosome genes affecting the recombination effect is described. The results are Inthe accompanying paper (Charlesworth and discussed in relation to previous work on the Charlesworth, 1985), we described the results of genetics of recombination. artificial selection on recombination in female Drosophila melanogaster. We also presented the results of some genetic studies of the difference MATERIALSAND METHODS between a line (H), which had responded to selec- tion for increased recombination between the 3rd Allexperimental procedures were as described chromosome markers Gl and Sb,andan unselected in the accompanying paper (Charlesworth and control stock (C). The genetic background of each Charlesworth, 1985), except that some experiments of these stocks was derived largely from an outbred involved the measurement of recombination population (IV), founded from wild-caught flies. frequencies using vial-reared progenies. For such These genetic studies suggested that all three major experiments, 2-pair matings were set up in 3" x 1" chromosomes contributed to the difference in re- vials, and the emerging progeny were counted three combination frequency. A small reciprocal effect times over an 8-day period. was shown to be controlled by the grandmaternal nuclear genotype. In this paper, we extend these genetic studies Chromosomesubstitutions by examining the individual contributions of the Thestocks used in the assays of individual chromo- X, 2nd and 3rd chromosomes to the differences some effects described in the third section were between the H and C stocks, both in homozygous constructed from the isogenic H (high) and C and heterozygous states. Interactions between (control) stocks described in the fourth section of these chromosomes are also studied. An attempt Charlesworth and Charlesworth (1985), using the *Presentaddress: Department of Biology, University of balancers described in that paper. Details of the Chicago, 1103 E. 57th St., Chicago, IL 60637, U.S.A. breeding procedure will be omitted here, but can 86 B. CHARLESWORTH AND D. CHARLESWORTI-l be supplied on request. The genotypes of the stocks Table 1 Chromosome substitution stocks are shown in table 1. Each stock was tested for The following isogenic chromosome substitution stocks were secondary non-disjunction of the X chromosome constructed using the H H Gl11, H H SbH, C C G1 and C C Sb isogenic stocks described by Charlesworth and as described in the fourth section of Charlesworth Charlesworth (1985). The constitutions of the X, 2nd and and Charlesworth (1985), with negative results. 3rd chromosomes, with respect to their origin from the This indicates that the X chromosome of each stock high recombination (H) or control (C) stocks, are shown can be reliably regarded as being derived from the in order; GI and Sb 3rd chromosomes are distinguished from each other. In each case, chromosome 3 is balanced H or C stocks as appropriate. over the TM6 multiply-inverted 3rd chromosome. Each stock was isolated in two independent experiments, in order to guard against balancer breakdowns and other Mappingexperiments accidents; the two isolates are designated as El and E2 Theseinvolved the construction of recombinant respectively chromosomes between the GIchromosome3 and Is! chromosome substitutions the multiply-marked third chromosome rucuca, supplied by Dr D. Ish-Horowicz of the Imperial H C Gl, H C Sb(, C HGlF, C HSbH Cancer Research Fund. This contains the markers 2nd chromosome substitutions ru (OO), h (265), th (432), ct (44O), Cu (5OO), C H G1, C H Sb, H C G1H, H C Sb0 sr(620), e (707) and ca (1007). The wing mutant cu was not followed in these experiments, as it 3rd chromosome substitutions proved difficult to classify reliably. In order to C C G10, C C Sb!!, I-I H Gl, construct recombinant chromosomes, rucuca was H H Sb transferred onto a background of H X and 2nd chromosomes. Individual recombinant chromo- somes carrying GI were isolated from crosses 0 26 41434446586271 101 I I LI I I between this rucuca stock and H H GlH, using the ru h Glthst Sbsr e ca following mating scheme (which shows the isola- tion of an ru h GI chromosome) Figure 1 Approximate map positions of the 3rd chromosome markers used in the mapping experiments are shown GI TM6 rucuca rucuca in the text and tables were calculated directly from the corresponding sample variances of the untrans- TM6 ruhGl formed data, except where otherwise stated. Sb rucuca

TM6 TM6 EFFECTSOF WHOLE CHROMOSOMES ruhGlx ruhGl Inthis section, we shall describe the results of The recombinant chromosomes were maintained measurements of the frequency of recombination balanced over TM6 in the usual way, and their between GI and Sb in stocks differing with respect frequencies of Gl-Sh recombination measured by to their constitution for various combinations of crossing to H H Sb females and scoring the pro- chromosomes from the H and C stocks. Except genies of vial matings of these with IV males. The where otherwise indicated, Gl-Sbrecombination linkage relationships of the markers used in these was assayed in bottle crosses, as described by experiments are shown in fig. 1. Charlesworth and Charlesworth (1985, second section). Statisticalprocedures Homozygous effects of the different Testsof significance were carried out using the chromosomes: methods angular transforms of recombination frequencies. Comparisons of means of different genotypes were The homozygous effects of the different chromo- made by 1-tests (individual culture values were not somes were determined by comparing the frequen- weighted by numbers of flies per culture, since cies of Gl—Sh recombination in all eight possible there was evidence for a substantial non-binomial combinations of H versus C X, 2nd and 3rd component to the error variance). The standard chromosomes, using the original isogenic H and errors attached to the untransformed means given C lines and the substitution stocks of table 1. The GENETIC VARIATION IN RECOMBINATION IN DROSOPHILA. II 87

Table2 Effects of the individual chromosomes on GI-Sbrecombination

Genotype Cf females Cy females Chr: 1 2 3 Mean±S.E. No. of cultures Mean±S.E. No. of cultures

C C C 0130±0005 22 0l60±0007 26 H C C — — 0210±0•0l4 10 C H C 0095±0004 9 0151±0004 16 C C H 0152±0007 13 0l85±0006 19 H •H C 0151±0024 7 0l72±0008 18 H C H — — 0l89±0009 8 C H H* 01l0±0005 14 0165±0009 18 H H H 0l70±0027t II 0259±0010 29 * Theseresults may be artefactual. See text for explanation. This, unfortunately, has an unusually high sampling variance, and the low mean is probably attributable to a statistical fluctuation. The results of previous measurements suggest a value of 0.21 as more representative. results of these experiments, which were carried between Gl and Sb. Any effect of chromosome 2 out in the autumn of 1980, are shown in table 2. can be excluded, since recombination was (Note that the designation C or H for the 3rd measured in flies heterozygous for the highly chromosome means that recombination was efficient SM I balancer. There was no evidence that measured in Gl/Sb or GIH/Sbfl females respec- the X chromosome had acquired any markers from tively.) With the exception of the H C C stock, the FM7 balancer during the breeding programme. results from the independently extracted substitu- Furthermore, as we will describe in the next paper tion lines (El and E2) agreed, and have been in this series, the X chromosome from the El Sb pooled. Reciprocal crosses (i.e., Gi x Sb d versus stock increases recombination uniformly along Sb x Gi d) have also been pooled, as no differen- chromosome 3, and its effect is similar to that of ces were detected. the X chromosome from the H C H stock. These Recombination in females heterozygous for the observations are inconsistent with the possibility 2nd chromosome balancer SM 1 (which carries the that the X chromosome in the H C C stock was marker Cy), and in females homozygous for wild- segregating for a inversion derived from FM7, type, was measured separately. These females are since this would have specifically increased rec- designated Cy and Cy in table 2. It is well known ombination in the centromeric region of chromo- that heterozygosity for an inversion on one some 3 (Lucchesi and Suzuki, 1968). Since the H chromosome increases the frequency of crossing versus C comparison using the E2 stock was only over in other chromosomes in Drosophila barely significant statistically, and could not be (Lucchesi and Suzuki, 1968), and this is seen confirmed in later experiments, it seems reasonable clearly in table 2. The ratio of the recombination to take the El lines as representative of the true frequency in Cy to that in Cy females varies from effect of the H X chromosome on a C background, l59 to l2l, with a mean of l36. It is noteworthy and this is the value given in table 2. that two stocks, H C C and H C H, are sterile (this was found to be due to both male and female chromosome effects: results sterility), so that recombination could be measured Homozygous only in Cy females in these stocks. This sterility Itis clear from table 2 that the different chromo- presumably is due to an interaction between the somes show considerable deviations from additiv- H X and the C 2nd chromosomes, such that Gl/Sb ity of effects. The following features of interest flies are sterile on an H C background. Curiously, may be noted. TM6/Gl and TM6/Sb flies are fertile on this (a) Both the H X and H 3rd chromosomes background. increase recombination when present on a C back- The line H C C (El) had a significantly higher ground for the other chromosomes. However, recombination frequency than its E2 counterpart H C H shows no increases above H C C or C C H (0.210±0.014 versus 0.164±0.007), suggesting (only C'y females can be used for this comparison, that some error had occurred in the breeding pro- of course). gramme. The most likely error is a crossover (b) Conversely, both the C X and C 3rd between the 3rd chromosome balancer TM6 and chromosomes have a strong effect in reducing re- the Gi or Sb chromosomes in the E2 line, introduc- combination when placed on an H background for ing a rearrangement which reduces crossing over the other chromosomes, Paradoxically, C H H 88 B. CHARLESWORTH AND D. CHARLESWORTH appears to have as low a value as C C C, although ground of C 2nd and 3rd chromosomes. The cross this may be an artefact as discussed below. involving H C G1 XC C Sb(.(Cyfemales) gives a (c) The H 2nd chromosome reduces recombi- significantly higher recombination frequency than nation on a background of C X orC3rd chromo- the reciprocal cross (t25 =313,p < 00l). This may somes, but increases recombination when both X be a statistical fluke, since no such effect is apparent and 3rd chromosomes are H. These effects are both in non-Cy females, where there is nevertheless a apparent in Cy females, suggesting that the 2nd significant effect of H/C compared with C/C. The chromosome effect is due to a dominant gene or average over both classes of H/C flies suggests genes. intermediate rather than full dominance for the H X chromosome effect in Cv flies. Heterozygouseffects of the different chromosomes Heterozygouseffect of the X chromosome on an H background Thelaborious nature of the measurements of re- combination frequencies precluded an analysis of Asimilar set of experiments was carried out on a the effect of each chromosome when heterozygous background of H 2nd and 3rd chromosomes, with (H/C) in all possible backgrounds. Instead, a the results shown in table 4. These are somewhat series of experiments were performed in the spring difficult to interpret, as the results of different of 1981, in which a limited range of genotypes was crosses which should yield the same genotypes are measured, using the same methods as above. not consistent. In the first place, the El and E2 C H H genotypes differ significantly, E2 being higher than El. A similar difference was found in Heterozygouseffect of the X chromosome on a C background the earlier results reported in table 2, but was not statistically significant. Secondly, H/C genotypes Thegenotypes assayed, and their recombination in which the G! chromosome is derived from frequencies, are shown in table 3. The clearest H H H (crosses B) tend to have significantly higher results are for Cv females, where it is apparent recombination frequencies than when G! comes that H/C shows an increase over C/C on a back- from C H H (crosses A).

Table 3Heterozygous effects of the X chromosome on G1-Sbrecombinationon a C background

Cy females Cy females Genotype Mean±S.E. No. of Cultures Mean±S.E. No. of Cultures

CCC 0103±0005 10 0114±0004 13 HCC — 0178±0006 12 H/C C C 0125 10 0136±0005 IS (H C SbxCC G!() H/CCC 0119±0005 10 0161±0006 12 (H C Gl(.xCCSb()

Table 4 Heterozygouseffect of the X chromosome on G/-Sbrecombinationon an I-I background Cffemales Cy females Genotype Mean No. of cultures Mean±S.E. No. of cultures CHH(EI) 0083±0003 27 — — CHH(E2) 0113±0007 9 — C/H H H (EIA) 0094±0003 27 — — C/HHH(EIB) 0141±0005 12 — — C/H H H (E2A) 01150005 8 0164± 0005 18 C/H H H (E28) 0127±0007 5 0234±0006 25 HHH 0147±0005 20 0222±0011 16 El and E2 refer to the independently constructed C H H lines; A and B refer to genotypes from crosses of the type C H GixHH Sb11andC H SbHxHH G111respectively. The tests on the Cyfemaleswere carried out one year after the tests on the Cvfemales. GENETIC VARIATION IN RECOMBINATION IN DROSOPHILA. II 89 The most likely interpretation is that the GI for the X chromosome and simultaneously either chromosome in both the C H H El and E2 stocks Cy/CorCy/Hforthe 2nd chromosome, on a acquired some recombination-restricting genetic background of H 3rd chromosomes. No significant material from TM6 during the course of the con- diflerence was detectable between Cy/C and struction of these stocks, to a different extent in Cy/H flies, and the pooled recombination the El and E2 lines. If this is so, the assays on the frequency for the X chromosome heterozygotes C H H lines are worthless, and their unusually low was O252±OOO67 (15 cultures), compared with values (seen in both tables 2 and 4) are explained. O229±OOO7 (10 cultures) and O247±O0l2 for C The appropriate test for the heterozygous effect of and H X chromosome homozygotes. This again the H X chromosome is thus to compare H H H suggests a dominant effect of the H X chromosome. with C/H H H (B) in table 4; the Cy and Cy results suggest almost complete dominance of the H X gene or genes over those of the C chromosome. Heterozygouseffect of 3rd chromosome This is consistent with the results for the effect on (C background) the C background described above. Table6 displays the relevant data, where flies heterozygous for SbH and Gl/Sb have been Heterozygouseffect of 2nd chromosome distinguished from each other. As suggested by the (C background) results of Charlesworth and Charlesworth (1984, Thiswas tested as shown in table 5, using non-Cy section 4) on the Fl's between the isogenic H and females only. It is clear that the H 2nd chromosome C stocks, the SbH chromosome does not appear to is effectively dominant in reducing crossing over have any effect, and the recombination frequency on this background. Time unfortunately did not of Gln/SbH is the same as that of G1H/Sbc. permit a study of this effect on an H background, but the results of table 2 strongly suggest a effect of 3rd chromosome dominant effect of the H chromosome 2, since it Heterozygous (H background) TableS Heterozygous effect of the 2nd chromosome on G1-Sb Similarexperiments using a background of H X recombinationon a Cbackground and 2nd chromomes completely confirm this con- Genotype MeanS.E. No. of cultures clusion. The mean of Gl./Sb is 0l6l and that of GIH/Sbc is 0246±0005, compared CCC 0104±0013 6 with values of 0l5l±O005 for Gl/Sb and CHC 0079±0004 12 0228±0O07 for GlH/Sbf (all these values are for CC/HC 0082±0.003 28 Cy females).

Table 6Heterozygous effects of the 3rd chromosome on GI—Sbrecombinationon a C background

Cy females Cy females Genotype Mean±S.E. No. of cultures Mean±S.E. No. of cultures CCGI/Sb 0104±0004 9 0122±0005 10 CCG1H/SbK 0120±0004 13 0l5l±0006 13 CCGIC/SbH 0103±0004 19 0126±0004 18 CCGl11/Sh 0128±0006 17 0151±0005 17 significantlyincreased recombination when Further analysis of the 3rd chromosome effect heterozygous with Cy on an H X and 3rd chromo- (a)A possible explanation of the apparent some background. (This effect was confirmed in absence of effect of the SbH chromosome is that some later experiments, which we will not report the stocks carrying Sb also contain a dominant in detail.) factor that reduces crossing over, as a result of some error in the breeding programmes. This is Jointheterozygous effects of X and 2nd extremely unlikely, since the difference between chromosomes the G1H and the Sb chromosome is apparent in Anexperiment was carried out to measure the the Fl's between the isogenic H H H and C C C recombination frequency of flies which were H/C stocks, and in both C C H/C and H I-I H/C flies. 90 B. CHARLESWORTH AND D. CHARLESWORTH A direct test for this possibility seemed desirable, causes a much higher frequency of st-sr recombi- however, and was carried out by intercrossing nation than does SbH. Since the interval st-sr over- H H SbHwithH H G1}.J, and backcrossing the laps Gl-Sb (44-62 and 41—58, respectively) this TM6/G1 progeny male to H H Sb.. The re- gives an indication of the extent to which the G1.-Sh combination frequencies ..i theresultant G1/ Sb recombination frequency is affected by GIH. female progeny were thcompared to those of A similar experiment was carried out to com- G11.4/ Sb females from the s 1 between H H Sb and pare the G1. and Sb chromosomes, by intercross- H H G1. The respective mean recombination ing the H H rucuca stock with H H Gl and frequencies were 0156 * 0008 and 0 160 0009. H H Sb respectively. The results are shown in the This excludes the hypothesis of some artefact due bottom half of table 7,andindicate that G1 and to other chromosomes. Sb(. do not differ. These crosses were carried out over a month later than the previous set, so that the apparently low recombination frequencies Table7Effects of different 3rd chromosomes on si-sr re- combination in Cyfemales compared with SbH may not mean much.

Genotype Mean±S.F. No. of cultures MAPPING EXPERIMENT H H G111/rucuca 0270±0016 13 HHSbN/rucuca 0202±0009 13 Methods H H Gi/rucuca(El)01890015 14 HHSbc/rucuca(El)0183±0010 4 An attempt was made to localise the high recombi- HHGi/rucuca(E2)0172±0012 8 0176±0012 10 nation gene or genes carried on chromosome 3, HHSb(./rucuca(E2) using recombinants between the G111 and the multi- ply marked rucuca third chromosomes, as described in "Materials and Methods". The (b) The properties of the Sb}1 and Gl 3rd frequency of Gl-Sb recombination was assayed in chromosomes were further compared by examin- females heterozygous for SMI, on a background ing their effects on recombination between the of H X and 2nd chromosomes, for each recom- neighbouring 3rd chromosome recessive marker binant chromosome. Owing to the necessity of genes si and Sr. To this end, the multiply-marked scoring numerous replicates of each recombinant 3rd chromosome rucuca was crossed onto a back- chromosome, only a limited number of recom- ground of H X and 2nd chromosomes derived from binants could be characterised reasonably accur- the H H GIH stock. This stock was then crossed to ately. The assays of recombination extend over the H H Gl and H H SbH respectively, and st-sr rec- period from December 1982 to September 1983; ombination in Cy females was assayed in vial most recombinant chromosomes, together with two cultures by backcrosses to rucuca males from the control chromosomes (Gl. and GIH on a back- original, unstandardised rucuca stock. The results ground of H X and 2nd chromosomes), were are shown in table 7(top).It is clear that GIH assayed several times over this period; there was

Table8 Gi.-Sh recombinationfrequencies for Gi recombinant chromosomes Mean recombination No. of Mean recombination No. of Chromosome frequency (+S.F..) cultures Chromosome frequency (±S.E.) cultures ru h Gi th st I 02851 003 I 16 GIsr e 31 02460022 25 ru h GI 35 0260±0013 79 Gi sr e Ca 33 0242±0018 39 ru h Gi 70 0239±0042 7 Gisre ca 71 0235±0045 6 ru h Gi 31 0238 0016 45 Gi th st 72 0258±0013 75 ruhGi 10 0233±0014 61 Githst7l 0239±0014 60 ru h Gil 0230±0016 49 Gi th Stcr71 0230±0013 65 ru h Gi 19 0222±0016 48 Gi th St sr e CaI 0229±0013 71 ru h GI 14 0208±0014 54 ,-u h GI 7 0220±0048 5 Gist sr C CO 72 0199±0015 46 ruhGi22 0202±0018 34 Gi4l 0254±0018 41 ru h Gi 34 0193±0018 31 G/76 0260±0028 16 ru h Gi 16 0185±0032 10 Gi51 0266±0035 II h GI 72 0222±0030 48 G1. 0226±0007 205 GENETIC VARIATION IN RECOMBINATION IN DROSOPHILA. II 91

no evidence for systematic variations between on the untransformed scale for a chromosome with different dates of assay for the same chromosome a mean of y on the transformed scale is given by in these data. multiplying the standard error on the transformed scale by 2 sin y cos y, and the 95 per cent con- Results fidence interval is found by multiplying further by 196. Themeans and standard errors of the Gl-Sbre- combination frequencies of the recombinant chromosomes, together with the controls, are Crossoversbetween h and GI shown in table 8. A graphical display of the results Ifthis class of recombinant chromosomes, which for the chromosomes where more than 10 replicates comprises the ru h Gi, h GI and ru h GI th Strecom- were assayed is given in fig. 2. An anomaly in the binants, is examined there is evidence for sig- nificant heterogeneity. Using the values for "IC chromosomes with more than 10 replicates, the 0I 01 between-chromosome mean square on the trans- CI Inn formed scale is 003570, which gives an F value of op In 2 11 (9 and df) when compared with the error GI tfl*t mean square (00l69l), p <005. If the ru h Gi th st GI.I,' I chromosome is omitted, F drops to I 92 (8 and df), which is just below the 5 per cent critical value. If ru h Gi 35 is omitted, which is the next —---. ,flGI highest chromosome, F=091, which is clearly non-significant. — — -—_-_-. ..hGIIh.t The status of ru h Gi 35 is therefore not entirely clear. Neither it nor ru h Gi th St 1 15 20 25 30 35 differ sig- Figure2 Means (dots) and 95 per cent confidence intervals nificantly from the G1H and the GI recombinant (bars) for the Gl-Shrecombinationfrequencies of each chromosomes, whereas the group of other chromo- recombinant chromosome for which more than ten repli- somes have a significantly (p

increased recombination mimics the model of re- variation in recombination is often due to recessive combination modification in a fluctuating environ- high recombination alleles, in contrast to recessive ment studied by Charlesworth (1976). Theoretical mutant alleles which usually decrease recombina- studies of these cases have shown that selection tion (Baker et a!., 1976); examples include on genetic modifiers of recombination is generally Schizophyllum (Stamberg, 1969; Tang and Chang, weak unless there is close linkage between loci 1974) and Neurospora (Catcheside, 1977, Chap. 4). subject directly to selection and the modifiers The existence of such recessive factors for of recombination (Nei 1967; Kidwell, 1972b; inceased recombination is extremely interesting, Charlesworth, 1976). In view of the evidence that since they are unexpected from the molecular point both X and 2nd chromosomes contribute sig- of view (Catcheside, 1977). Furthermore, the nificantly to the increase in recombination between "hitch-hiking" class of models for selection for the 3rd chromosome genes GI and Sb (table 2), it increased recombination frequently predict a is thus not surprising that consistent responses to higher rate of change in gene frequency for recess- selection have not been obtained. The failure to ive high recombination alleles than for dominant obtain a renewed response to selection on its ones (Strobeck et a!., 1976; Charlesworth et a!., resumption after relaxation (Charlesworth and 1977; Maynard Smith, 1978, Chap. 7). (This is due Charlesworth, 1985, third section) could also be to the fact that a selectively-favoured recombinant due to chance rather than lack of variation. gamete will always give a hitch-hiking effect to an The strong epistatic interactions observable in allele for increased recombination if the latter is table 2 also probably limit the efficiency of selec- homozygous, whereas it has only a 50 per cent tion; for example, the effect of the high chromo- chance of being associated with the favoured some 3 genes on a low background of X and 2nd gamete when heterozygous.) A predominance of chromosomes is only about 002, whereas they recessive high recombination alleles in natural have an effect of about OO6 on a high background. population would therefore provide evidence in The high 2nd chromosome reduces recombination favour of hitch-hiking models. (when homozygous) except when the X and 3rd Unfortunately, both the results obtained here chromosomes are both high. Inspection of table 2 and the results of earlier studies rule out such a in fact shows that it is impossible to go from the generalization about dominance. For example, C C C state to one in which three C chromosomes intermediacy of Fl's between high and low re- have successively been replaced by H chromo- combination stocks has been observed in D. somes, and to achieve a net increase in recombina- melanogaster (Chinnici, 197la, b) and in silk- tion, by a path in which each replacement of a C worms (Turner, 1979, Ebinuma and Yoshitake, by an H chromosome results in an increase in 1981). The interpretation of the Fl between our H recombination. and C stocks with respect to Gl-Sb recombination in complicated by the existence of the reciprocal effect, since the cross ? C x H d is often only Asymmetriesin responses and dominance slightly higher in recombination frequency than relations the value for the C stock, whereas the cross ? H x C is always substantially higher (fourth section Severalworkers have reported asymmetrical of Charlesworth and Charlesworth, 1985). Studies responses to upward and downward selection, not- of the dominance effects of individual chromo- ably Dewees (1975) on Tribolium and Kidwell somes on controlled genetic background (third (1972a) on D. melanogaster, who both obtained section of this paper) clarify the situation, however. significant upward but non-significant downward The X chromosome from the H stock is partially responses using family selection designs. (These dominant on both H and C homozygous back- are less subject to the limitations imposed on the grounds with respect to the 2nd and 3rd chromo- efficiency of selection by the linkage considerations somes (tables 3 and 4). The G! chromosome 3 from described above). Both Dewees and Kidwell found the H stock is capable of increasing recombination that the recombination value of the Fl between when heterozygous for a C chromosome 3, on both selected and unselected stocks was close to the H and C homozygous backgrounds of X and 2nd value for the latter, and suggested that the asym- chromosomes (third section and table 6). This is metry of the selection response might have been fully consistent with the data on the Fl's between caused by reliance on low frequency, recessive the H and C stocks and on the female backcrosses, alleles for high recombination. There is evidence reported by Charlesworth and Charlesworth (1985, from several other species that naturally occurring fourth section). Clearly, there is no evidence for 94 B. CHARLESWORTH AND D. CHARLESWORTH consistently recessive effects of high recombination present in different lines. A similar phenomenon alleles on the X and 3rd chromosome. was noted by Kidwell (1972b),whosuggested that The situation with respect to the 2nd chromo- the lethals had accumulated by random drift within some is confused by the existence of strong epi- the chromosomal region around the markers stasis, as mentioned above. On a homozygous that is maintained permanently heterozygous as a background of C X and 3rd chromosomes, the 2nd result of the selection procedure (cf. fig.I of chromosome gene or genes has a dominant effect Charlesworth and Charlesworth, 1985). Nei (1970) of reducing recombination (table 5), whereas it has modelled the process of random accumulation appears to have a dominant effect of increasing of deleterious recessive factors in permanent chro- recombination on a homozygous background of mosomal heterozygotes, and it seems likely that H chromosomes (table 2). The male backcross this explains the results of Kidwell and ourselves. data, however, suggest that the H 2nd chromosome has a negative, dominant effect on recombination or is underdominant when the X and 3rd chromo- Genetic basis of the H-C difference somes are heterozygous H/C (Charlesworth and Number of genes involved Charlesworth, 1985,table4). In this table, the comparison of 2H/2Hwith2H/2c on an H/C back- Where sufficiently detailed genetic studies of the ground is given by twice the contrast of genotype basis of responses to selection on recombination (4) with H XC, and is equal to 0049(non- have been carried out, it is usually found that significant). The contrast of 2H/2(with2/2( several genes are concerned, (e.g.,Chinnici,1971 b; (obtained from twice the difference between C x H Turner,1979; Ebinuma and Yoshitake, 1981). An and genotype (5)) is equal to —0050 (p<00l). exception is provided by the work of Valentin This negative effect of chromosome 2 when (l973a, h),whoshowed that a recessive 3rd heterozygous on a heterozygous background of X chromosome gene mci-Iwasresponsible for the and 2nd chromosomes explains the comparatively change in recombination frequency in his line. The low value of the Fl recombination frequency in results obtained by us clearly indicate that there the CC x H d cross, despite the dominant or par- are genes on all three major chromosomes con- tially dominant effects of the H X and 2nd chromo- cerned in the difference in Gl-Shrecombination s omes. between the H and C stocks. The number of genes The apparent asymmetry in selection response on each chromosome is less certain. The mapping observed here (only upward selection for Gl-Sb experiment described in the fourth section shows recombinationbeing significant) cannot be readily that there are at least two 3rd chromosome factors, explained by these findings on dominance. In par- one located between handGi,andthe other either ticular, the fact that the 2nd chromosome gene or between standsrorto the right of sr(cf.figs I genes reduce recombination except when the X and 2). The data on the h- G!recombinant chromo- and 3rd chromosomes are homozygous for high somes is consistent with there being only a single recombination alleles suggests that it would be factor in this region, but some minor additional quite easy to obtain a downward response to selec- factors cannot be excluded. The generally low Gi— tion if the population were initially predominantly Sb recombinationvalues of these chromosomes C for the X and 3rd chromosome genes. It seems indicate that the high recombination gene or genes most likely that the differences between the selec- are located closer to h than to Gi,whichis perhaps tion lines are due to chance factors such as random surprising in view of the fact that the intensity of loss and fixation of alleles by drift (cf. Falconer, selection is greatest for modifiers that are closely 1981, p. 191.) linked to the marker genes (fifth section): formally, the high recombination gene is placed about 11 Accumulationof lethals in downward selection map units to the left of GI.Ifthe other gene is lines located between Standsr,thenit could be close to Sb(fig.I): its formal position is about 5 units Asdescribed by Charlesworth and Charlesworth to the left of Sb.(Thelack of crossing over in male (1985, third section), the downward selection lines Drosophila means that the effective recombination for G1-Shrecombinationshowed evidence of hav- frequencies for the markers and the modifiers are ing accumulated recessive lethal genes, which were half these values.) linked to the markers and present in high frequen- There are no mapping data for the X and 2nd cies within lines, although different lethals were chromosome genes, but some conclusions about GENETIC VARIATION IN RECOMBINATION IN DROSOPHILA. II 95

the numbers of factors on these chromosomes may Differences between the GIH and SbH be drawn. In the first place, the highly specific chromosomes pattern of interaction between the 2nd chromo- some and the X and 3rd chromosomes, with the Asdescribed in the third section, the GI and Sb reversal of 2nd chromosome effects described chromosomes of the isogenic H stocks differ con- above, is very hard to understand if there is more siderably in their effects on recombination; there than one 2nd chromosome gene. Secondly, tests is in fact no evidence that the ShHchromosome for the segregation of the X and 2nd chromosome carries any genes for increased Gl—Sb recombina- genes are provided by comparisons of the tion. Since the G1HandSbH chromosomes are each appropriate genotypes of the male and female descended from just a single chromosome of the backcrosses (Charlesworth and Charlesworth, appropriate type in the mass-cultured H stock, it 1985, tables 3—6). Due to the absence of crossing is not necessarily the case that the high recombina- over in males, all chromosomes segregate as units tion 3rd chromosome genes were associated only in the male backcrosses, whereas recombination is with the Gi chromosomes of the original selection permitted in the X and 2nd chromosomes in the line. The partial decline of the 3rd chromosome female backcrosses. On the hypothesis of a single effect in the mass-cultured H stock (Charlesworth gene on each of the X and 2nd chromosomes, it and Charlesworth, 1985, third section) unfortu- is easily seen by inspection of tables 3 and 5 of nately makes it difficult to determine the composi- Charlesworth and Charlesworth (1985) that the tion of this stock with respect to high recombina- value of genotype 1 of the 1st generation female tion 3rd chromosome genes. However, indirect backcrosses should equal the mean of the values evidence that Sb chromosomes from the H stock for genotypes (6) and (8) of the male backcross; at one time carried high recombination genes is the values are 0l28±0005 and 0113±0006 provided by the absence of a difference between respectively, which are not significantly different. SbH/Glc and G1H/Sbc flies in the F! between the Similarly, the mean of the values of genotypes 6 mass-cultured H and C stocks (Charlesworth and and 8 of the 1st generation female backcrosses Charlesworth, 1985, fourth section, and in the should equal the mean of genotypes (1) and (3) male backcross experiment (Charlesworth and of the male backcrosses; the values are 0194± Charlesworth, 1985, fourth section). Together with 0005 and 0190±0008 respectively. Unfortu- the evidence for a decline in the 3rd chromosome nately, tests cannot be made using the other values after relaxing selection, this suggests that genotypes in these crosses because of the fact that the mass-cultured GI and Sb stocks derived from sometimes no distinction was made between SbH the selection lines originally contained high re- andGl in the male backcross experiments, combination genes associated with both the Gl and whereas a clear difference between the two was Sb genes, but that the stock was segregating for found in the female backcrosses and other experi- them. The failure to include any such genes on the ments on the isogenic H and C stocks (see next Sb chromosome isolated in order to form the section below). Finally, genotypes I and 5 in the isogenic stock is presumably fortuitous. (The only 2nd female backcrosses generation have sig- partial nature of the decline in value of the 3rd nificantly smaller values than genotypes I and 5 chromosomeincidentallyprovides further of the 1st generation (table 6 of Charlesworth and evidence for the existence of more than one high Charlesworth, 1985). In view of the evidence, recombination locus on this chromosome.) described above, that 2/2 has a considerably lower value than 2/2 on an H/C or C/C back- ground of X and 3rd chromosomes, the differences Geneticbasis of the reciprocal effect between the 1st and 2nd female backcross gener- Asdescribed by Charlesworth and Charlesworth ations for genotypes I and 5 must provide an (1985, fourth section), there is a significant underestimate of one-quarter of the H/C-C/C reciprocal difference in the Fl between the H and effect for the X chromosome; the joint estimate of C stocks, both mass-cultured and isogenic. On the this effect is 0096OO2,which substantially larger scale of recombination frequency, the value of this than the estimate obtained from table 3 (00l9± effect is estimated as being about OO15. The female 0006 for Cy females and OO34±OOO7 for Cy backcross experiment suggests that it is controlled females). There is certainly no suggestion here that by a grandmaternal effect of the nuclear genotype, the segregation of the X chromosome has been most probably due to chromosome 3. The factors obscured by recombination. responsible for this reciprocal effect seem to be 96 B. CHARLESWORTH AND D. CHARLESWORTH present in both the ShEI and G1H isogenic H stocks, tion when used as a maternal parent, in Fl crosses since it is observed in Sb x G1. as well as GIH X with the other stocks, although the highest level Sb Fl crosses. In view of the fact that the SbH of male recombination obtained here is much lower chromosome 3 from the isogenic H stock lacks than that reported for P-M dysgenic crosses by high recombination genes, this evidence suggests Kidwell et a!., (1977): 1-3 x l0— compared with strongly that the chromosome 3 factors responsible 7-2 x l0. The data in table 9 suggest that the G1H for the reciprocal effect are distinct from those stockproduces increased recombination in Fl detected in the assays of genic effects on recombi- males when used as the female parent with either nation (tables 2, 6, 7). Sb or Sb11; the frequency of male recombination The grandmaternal nuclear control of the in the cross Sb x G1 is also possibly increased reciprocal effect is reminiscent of the mode of compared with the CXC crosses (p

Table 9Frequencies of male recombination between G/ and Sb in crosses between the C and H isogenic stocks

No. of No. of Frequency of F1 Cross Recombinants Individuals Recombination

9 d CCG1xCCSb 0 9311 0 CCSbcXCCG! 0 9089 0 HHGI}IXHHSbH 10 10108 989x104 l-1HSbHXHHGlF 2 4423 452xl04 CCGl.xHHSb11 1 11850 8.44Xl0s H H Sb11XCC G1(. 2 8868 225 x10 CCShCXHHGIH I 10734 932xl05 HHGI}xCCSb( 13 9876 l•32Xl03 The tests were carried Out by crossing single F1 G1/Sb males to en bw; ne females in vials. Most of the putative recombinant progeny were tested by crossing to en bw; ri e to check that they segregated for the markers they carried. All individuals tested appeared to be genuine recombinants. Clusters of more than one recombinant per culture were scored as a single event (these were very rare). is due to a dysgenic effect. It is hard to test this other three stocks are P. This is confirmed by tests hypothesis rigorously, because of the small size of using Engels' sn' allele, which is unstable in dys- the reciprocal effect, but some tests of the mass- genic crosses, mutating to wild-type and strong sn cultured and isogenic H and C stocks with respect alleles with high frequency (Engels, 1979b). to P-M dysgenesis have been carried out. Measure- Crosses of the four stocks of table 9 to sn" on a ments of male recombination frequencies in G1+ P nuclear background (stock supplied by Dr A. J. /+Sb males in the mass-cultured C and H stocks Leigh Brown) yielded instability of sn only in the yielded values of 14/22,885 (61 xl04) and case of G1H mothers. The P-M cytotypes of the 12/22,237 (5.4 x 10) respectively. This does not stocks thus do not correlate perfectly with the suggest any significant contribution of factors pattern of reciprocal effects on female recombina- inducing male recombination to the difference in tion, although it is clear that crosses involving G! female recombination between these stocks, but as female parent will be dysgenic with P strains. does not exclude a contribution of hybrid dys- It is thus not possible to reach a firm conclusion genesis to the reciprocal effect in the Fl, with respect to the role of hybrid dysgenesis in The results of male recombination tests of the causing the reciprocal effect on recombination in isogenic stocks and crosses between them are Fl females detected here; the evidence suggests shown in table 9. These suggest that the G1H stock that dysgenesis may be occurring in some crosses, induces elevated frequencies of male recombina- but that it is not necessarily causally involved with GENETIC VARIATION IN RECOMBINATION IN DROSOPHILA. II 97 the determination of the reciprocaleffect. CHARLESWORTH, B. 1976. Recombination modification in a fluctuating environment. Genetics, 83, 181-195. Reciprocal effects on recombination have CHARLESWORTH, B. AND CHARLESWORTH, D. 1985. Genetic previously been reported for D. melanogaster by variation in recombination in Drosophila. I. Responses to Thoday and Boam (1956), Lawrence (1958), Kid- selection and preliminary genetic analysis. Heredity, 54, well (l972a), and Lüning (1983), and for Neuro- 7 1—83. CHARLESWORTH, D., CHARLESWORTH, B. AND STROBECK, spora by Lamb (1971) and Landner (1974). In the C.1977. Effects of selfing on selection for recombination. studies of Thoday and Boam (1956) and Lüning Genetics, 86, 213-226. (1983), cytoplasmic inheritance was implicated; CHINNICI,.i.p1971a. Modification of recombination in Lüning also had evidence for non-cytoplasmic Drosophila. I. Selection for increased and decreased cross- reciprocal effects. The other cases have not been ing over. Genetics, 69, 71—83. CHINNICI, .i.p. 1971b. Modification of recombination in investigated in detail. Drosophila. II. The polygenic control of crossing over. Genetics, 69, 85—96. DETLEFSEN,J. A. AND ROBERTS, E. 1921.Studies on crossing Conclusions over. I. The effect of selection on crossovervalues. J. Expt!. Theresults reported here demonstrate that the IV Zoo!., 32, 333-354. DEWEES,A. A. 1975.Genetic modification of recombination natural population contains genetic factors that rate in Tribo!ium castaneum. Genetics, 81, 537—552. modify the frequency of recombination in the GI- EBINUMA, H. AND YOSHITAKE, N. 1981. The genetic system Sb region of chromosome 3. In the final paper of controlling recombination in the silkworm. Genetics, 99, this series, we will show that other regions of the 23 1-245. ENGELS,W. R. 1979a.Hybrid dysgenesis in Drosophila genome are also affected by these factors, so that melanogaster: rules of inheritance. Gene!. Res., 33, 219— the genes concerned are not highly specific for a 236. very limited part of the Drosophila genome. We ENGELS,w.R.19796.Extrachromosomal control of mutability have shown that there are at least four genes con- in Drosophila melanogaster. Proc. Nat. A cad. Sci. USA, 76, cerned in the difference between the high selected 4011-4015. FALCONER,D. 5.1981. Introduction to Quantitative Genetics. stock and the control, and that they interact in a (2nd ed). Longman, London. complicated fashion. There is no evidence for HALDANE,J. B. 5. 1962.The selection ofdoubleheterozygotes. recessivity of the high recombination genes. A J.Genet.,58, 125—128. small part of the selection response (approximately KIDWELL,M. . 1972a.Genetic change ofrecombinationvalue in Drosophila melanogaster. 1. Artificial selection for high OOl5, out of a response of O'lO, on the scale of and low recombination and some properties of recombina- recombination frequency) is due to a grandmater- tion-modifying genes. Genetics, 70, 4 19—432. nat effect of chromosome 3. The results of this, KI DWELL,M.G. I 972b. Genetic change of recombination value and other studies of natural variation in recombi- in Drosophila melanogaster. II. Simulated natural selection. Genetics, 70, 433-443. nation, show that the genetic control of recombina- KIDWELL, M. G. 1977. Reciprocal differences in female rec- tion in eukaryotes is highly complex, and that the ombination associated with hybrid dysgenesis in information obtained from mutant screening pro- Drosophila melanogaster Genet. Res., 30, 77—88. cedures provides only a partial picture of the range KIDWELL, M.G., KIDWELL, J. F. AND SVED, J. A. 1977. Hybrid of genetic control. dysgenesisin Drosophila melanogaster: a syndrome of aber- rant traits including mutation, sterility and male recombi- nation. Genetics, 86, 8 13—833. LAMB,B. C. 1971.Some details and effects of the premeiotic Acknowledgements We thank Martyn Stenning, Joan West, controls of recombination frequencies in Neurospora and Doris Williams for technical assistance. This work was crassa. Genet. Res., 18, 255—264. supported by grants from SERC (GR/A/1253.3) and the LANDNER, L. 1974. Genetic control of recombination in Nuffield Foundation. Neurospora crassa. III. Selection for increased and decreased recombination frequency. Hereditas, 78, 185— 200. LAWRENCE, NI. i. 1958. Genotypic control of crossing-over on the first chromosome of Drosophila melanogaster. Nature, REFERENCES 182, 889-890. LUCCI-JESI. J. C.ANDSUZUKI, D. T. 1968. The interchromo- ABDULLAI-I, N. F. AFD CI-IARLESWORTI-I, B. 1974.Selection somal control of recombination. Ann. Rev. Genet., 2, 87- forreduced crossing over in Drosophila melanogaster. 120. Genetics, 76, 447—451. LUNING, K. G. 1983. Genetics of inbred Drosophila ACTON, A. B. 1961. An unsuccessful attempt to reduce recombi- melanogaster. X.Maternaland cytoplasmic effects on re- nation by selection Amer. Nat., 95, 119-120. combination. Hereditas, 99, 57-68. BAKER, B. S., CARPENTER, A. T. C., ESPOSITO, M.S.,ESPOSITO, MAYNARD SMITH, i. 1978. The of Sex. Cambridge R.E.AND SANDLER, L. 1976. The genetic control of University Press, London. meiosis. Ann. Rev. Genet., 10, 53—134. MOYER. s. E. 1964. Selection for modification of recombination CATCHESII)E, D. G. 1977. The Genetics of Recombination. frequency of linked genes. Ph.D. , University of Min- Arnold, London. nesota. 98 B. CHARLESWORTH AND D. CHARLESWORTH

NFl, M. 1967.Modification of linkage intensity by natural THODAY, J.M.AND BOAM, T. it.1956. Apossibleeffect of the selection. Genetics, 57,625—64I. cytoplasm on recombination in Drosophila melanogaster. NFl, M. 1970. Accumulation of nonfunctional genes on shel- J. Genet., 54, 457-461. tered chromosomes. Amer. Nat., /04, 311-322. TURNER, J. R. G. 1979. Genetic control of recombination in the PARSONS, P. s,1958.Selection for increased recombination in silkworm I. Multigenic control of chromosome 2. Heredity, Drosophila melanogaster. Amer. Nat., 92, 255-256. 43, 273-292. SOKAL, R. R. AND ROHLF. F. j.1981.Biometry (2nd ed.) Free- VALENTIN, .j.l973a.Selection for altered recombination man, San Francisco. frequency in Drosophila melanogaster. Hereditas, 74, 295-. STAMBERG, j.1969.Genetic control of recombination in 297. Schizophyllum commune: the occurrence and significance VALENTIN, .j.l973b.Characterization of a meiotic control gene of natural variation. Heredity, 24, 361—368. affecting recombination in Drosophila melanogaster. STROBECK,C., MAYNARDSMITH, .1.ANI) CHARLESWORTH, Hereditas, 75, 5-22. B.1976. The effects of hitch-hiking on a gene for recombina- tion. Genetics, 82, 547—558. TANG, C. Y. ANDCHANG,S. T. 1974. Variation in recombination frequencies in Schizophyllum commune and its genetic con- trol. Aust.J.Rio!. Sci., 27,103—110.