BRIEF COMMUNICATIONS E\'0/1/1iOIl,
52(3). 1998. pp. 910-915
TEMPORAL CHANGE OF MITOCHONDRIAL DNA HAPLOTYPE FREQUENCIES AND FEMALE EFFECTIVE SIZE IN A BROWN TROUT (SALMO TRUITA) POPULATION LINDA LAIKRE,I PER ERIK JORDE,2 AND NILS RYMAN
Division of Population Genetics, Stockholm University, S-106 91 Stockholm, Sweden 'Esmail:
[email protected] Abstract.-We report data on genetic drift of mitochondrial DNA (mtDNA) haplotypes in a natural brown trout (Salmo trutta) population in Sweden. Large temporal frequency shifts were observed over the 14 consecutive year classes studied. The observed rate of drift was used to estimate the effective size of the population. This effective size applies to the female segment of the population as mtDNA is maternally inherited. The magnitude of mtDNA haplotype frequency change is compared with the corresponding allele frequency changes at 14 allozyme loci in the same population. The female effective size is estimated as 58, which is approximately half the effective size of 97 for the total population (both sexes) previously obtained from the shifts of allozyme allele frequencies. Key words.-Effective population size, genetic drift, mtDNA, temporal variation.
Received December 2, 1997.
Large amounts of data have been generated on the geographical distribution of allele frequencies in natural populations since the introduction of molecular genetic techniques in the mid-1960s (Avise 1994). Despite its importance for understanding the mechanisms behind the observed spatial patterns, relatively little information is available on the temporal dynamics of allele frequencies. Although numerous studies have compared gene frequencies from populations sampled two or three times (e.g., Krimbas and Tsakas 1971; Begon et al. 1980; Ryman and Stahl 1980; Ryman 1983; Burns and Zink 1990; Thomas et al. 1990; Waples and Teel 1990; Hedgecock et al. 1992; Hansen and Loeschcke 1996; Nielsen et al. 1997), only a few studies exist that systematically follow genetic changes within a population through extended periods of time (e.g., DeSalle et al. 1987; Jorde and Ryman 1996). More typically, population genetic investigations include sampling on only one occasion, and no information on the temporal stability of the observed allele frequencies is available. Hence, it is largely unclear to what extent allele frequencies are stable over time (e.g. Burns and Zink 1990). Quantification of the amount of temporal gene frequency shifts over extended periods of time is of interest for understanding the dynamics of genetic change in natural populations and because it also provides a means for estimation of the genetically effective population size, N e . This important quantity has been conspicuously difficult to estimate for populations living under natural conditions. This difficulty is due to the fact that a wide range of demographic parameters affect N, and measurements of these parameters require detailed information on individual members of the populations and their survival and reproduction (e.g., Wood 1987; Grant and Grant 1992). Research has recently focussed on the possibility of estimating the effective population size from temporal change of allele frequencies. The so-called temporal method is based on the logic that when genetic drift is the 2 Present address: Division of Zoology, Department of Biology, University of Oslo, P.O. Box 1050 Blindern, N-0316 Oslo, Norway.
Accepted January 28, 1998.
only cause of allele frequency change over time N, can be estimated from empirical observations on temporal change in those frequencies (Krimbas and Tsakas 1971; Nei and Tajima 1981a; Pollak 1983; Waples 1989; Jorde and Ryman 1995). Here we examine temporal shifts of mtDNA haplotypes in a natural brown trout (Salmo trutta) population in Sweden. We quantify the extent of genetic drift over 14 consecutive cohorts (year classes) and find large temporal frequency changes. Based on the observed temporal shifts we estimate the average effective size. Because of the maternal inheritance of mtDNA this effective size refers to females only and is denoted female effective size (Nej ) . The amount of mtDNA haplotype frequency change is compared with the corresponding allele frequency changes at 14 allozyme loci for the same population and cohorts (Jorde and Ryman 1996). We compare the effective sizes estimated from mtDNA and allozymes, corresponding to the effective number of females and the total effective size, respectively, and find that the estimate for females is approximately half that of the total. This result is in accordance with theoretical predictions for populations with an even sex ratio among breeding individuals. MATERIALS AND METHODS
Data on mtDNA haplotype frequencies were obtained from 704 brown trout collected from 1979 through 1993 from the natural population of Lake Blanktjarnen, a small lake (about 10 ha) in the mountain range of the Province of Jamtland in central Sweden. These fish represent a subset of those collected within an ongoing longitudinal genetic study of natural and introduced fish populations in this area and include samples of a minimum of 40 fish from each of 14 consecutive cohorts (year classes). In a previous report from this study, Jorde and Ryman (1996) analyzed temporal change of allozyme frequencies at 14 protein-coding loci in about 1400 fish and obtained an estimate of 97 for the total effective size (Ne ) of this brown trout population. For details on sampling
910 © 1998 The Society for the Study of Evolution. All rights reserved.
911
BRIEF COMMUNICATIONS
methods, tissue collection, age determination, and demographic analyses we refer to that paper.
1. Demographic statistics for the female segment of the natural brown trout population in Lake Blanktjarnen, Data are from Jorde and Ryman (1996). TABLE
Mitochondrial DNA Analysis Mitochondrial DNA haplotype frequencies were obtained through amplification and restriction enzyme cleavage of a segment containing a known polymorphic site identified during a pilot screening for mtDNA variation in this brown trout population. This preliminary screening included 204 fish from which total DNA was isolated from muscle or liver tissue according to the protocol of Taggart et al. (1992). Two different segments of size 2.0 and 2.5 kilobases (kb), respectively, were amplified using the polymerase chain reaction (PCR). The 2.0-kb segment contains NADH dehydrogenase gene-l (ND-l) and the 16s rRNA gene, and the 2.5kb segment includes ND-5/6 (Hall and Nawrocki 1995). Both segments were cut with five restriction enzymes previously known to detect genetic variation in these mtDNA segments in brown trout (Hall and Nawrocki 1995): AvaIl, Alul, HaeIII, HpaIl, and HinjI for the 2.0-kb region, and Taql instead of Alul for the 2.5-kb region. In addition, a subs ample of 10 fish were screened by an additional five enzymes (Hhal, Mval, Neil, Rsal, and Sau 3A) in both mtDNA regions. These initial screenings revealed only two haplotypes in this population. The remaining 500 fish were analyzed in the 2.0-kb region usingHaeIlI only, which was sufficient to distinguish the two haplotypes. Estimation of N ef The assessment of effective population size from temporal shifts of allele frequencies is not straightforward when dealing with populations with overlapping generations, as is the case with the brown trout. When generations overlap, the amount of temporal shift in allele and haplotype frequencies depends not only on the effective size but also on the demographic characteristics of the population. Hence, these characteristics must be measured and taken into account when applying the temporal method for estimating the effective population size (Jorde and Ryman 1995). The estimation procedure follows three steps. First, the amount of allele frequency change between cohorts is quantified. Second, the demographic attributes affecting the amount of allele frequency shift is summarized in adequate parameters. Third, the effective size is estimated from the components obtained under step one and two (see the Appendix). From the haplotype frequency data from 14 consecutive cohorts we obtain 13 pairwise consecutive measurements of haplotype frequency change. The unweighted mean (F'; eq. A2) of these 13 values was used for estimating N ef (eq. A3). Throughout this paper we denote the F- and F'-values obtained from mtDNA data F ml and F'ml' respectively, and the corresponding values for allozymes are denoted by F nuel and F' nuclDemographic statistics for the female segment of the brown trout population of Lake Blanktjarnen are presented in Table 1. The age-specific female survival rates, Ii, were assessed from the number of female trout in each age class using the Chapman-Robson method (Robson and Chapman 1961; Youngs and Robson 1978). This method takes into account
Age
Age-specific
class,
survival rate,
Age
i
t,
hi
Probability of the mtDNA of an individual being inherited from a female ofage t, Pi
0+ 1+ 2+ 3+ 4+ 5+ 6+ 7+ 8+ 9+ 10+ 11+ 12+ 13+
1 2 3 4 5 6 7 8 9 10
1.0000 0.4800 0.2304 0.1106 0.0531 0.0255 0.0122 0.0059 0.0028 0.0014 0.0006 0.0003 0.0001 0.0001
0.0 0.0 0.0 0.0 0.0 1.9 26.8 43.7 65.5 73.7 72.7 72.7 72.7 72.7
0.0000 0.0000 0.0000 0.0000 0.0000 0.0484 0.3270 0.2578 0.1834 0.1032 0.0436 0.0218 0.0073 0.0073
II
12 13 14
Birth rate,
that young age classes may be underrepresented in the catch due to gear selectivity, and it is assumed that the probability of fish surviving from one year to another is the same at all ages. No fish older than 13 years (corresponding to age class 14) was found, and survival beyond that age was presumed to be zero, The assumption of constant survival was necessary because we do not have data for the juvenile (immature) age classes. However, as discussed by Jorde and Ryman (1996, pp. 1379-1380), the present estimation procedure appears quite robust to assumptions regarding juvenile survival rates. The female birth rates, b.; were estimated from the observed proportions of female breeders in various age classes (i) multiplied by the average weight of females of that age. The products where normalized to result in a constant population size (i.e., 'ilibi = I), The resulting birth rates (b i; Table 1) appear quite high because of the high overall mortality rate, but are consistent with direct observations on brown trout (cf. Jonsson 1977; Elliott 1994). To establish if there is a statistical difference in the amount of drift observed for mtDNA versus nuclear genes, we performed a randomization test of the null hypothesis of equal means for the F' nucl" and F'mt-values. The test was done by randomly distributing the observed F' -values (13 values for mtDNA and 124 for allozymes: Table 3) into two new sets (of 13 and 124 measurements, respectively) and comparing the difference of the means of these two new sets with the actually observed difference between F'mt and F'nucl- This procedure was repeated 100,000 times. The proportion of times in which the random distribution resulted in a difference between the means that was larger than or equal to the difference actually observed was interpreted as the probability of the null hypothesis of equal means being true. The precision of the estimated female effective size was assessed from the confidence interval (CI) for F under the assumption that the largest uncertainty in Nef arises from sampling errors when determining the amount of drift, rather than when measuring demographic factors. Previous methods for estimating CI for F have relied on various assumptions
912
BRIEF COMMUNICATraNS
TABLE 2. Amount of haplotype/aIlele frequency change (F and F'; eqs. AI, A2) between consecutive cohorts at mtDNA and aIlozyme loci, respectively. The aIlozyme data originate from Jorde and Ryman (1996). The frequency of the most common mtDNA haplotype is presented in the p column, and 11 is the number of sampled fish. P refers to the probability from Fisher's exact tests; the probability for the total was obtained from a 2 X 14 contingency chi-square test for homogeneity (df = 13). mtDNA data Cohort
1975
p
n
0.640
50
1976
0.880
50
1977
0.694
49
1978
0.800
50
1979
0.596
52
1980
0.621
66
1981
0.700
50
1982
0.675
40
1983
0.620
50
1984
0.760
50
1985
0.735
49
1986
0.680
50
1987
0.660
50
1988
0.521
48
Total/mean
0.680
704
Allozyme data
r;
r;
p
0.316
0.276
0.009
0.206 0.060 0.197 0.003 0.028 0.003 0.013 0.092 0.003 0.014 0.002 0.080 0.078
0.166 0.019 0.158 -0.032 -0.007 -0.042 -0.032 0.052 -0.037 -0.026 -0.038 0.039 0.038
regarding the distribution of this quantity: either it is assumed to follow a normal distribution (Nei and Tajima 1981a) or a chi-square distribution (Waples 1989). The normal assumption is expected to hold only when F for each interval is based on a large number of loci and/or alleles, which is clearly not the case for the mtDNA data considered here. However, computer simulations indicate that the chi-square assumption may lead to an overly conservative estimate of the CI. Here, we bypass the problems associated with the poorly known distribution of F through estimating it by means of the bootstrap technique (Efron and Tibshirani 1993). This was done by randomly resampling individuals (i.e., haplotypes) with replacement from each of the 14 cohorts and calculating new sets of 13 F-values and their mean, F, from these bootstrap samples. This procedure was repeated 5000 times, and the distribution of these 5000 F-values was then used as an approximation for the (unknown) distribution of Fm t • RESULTS
The observed mtDNA frequencies among cohorts are given in Table 2. There are significant shifts in mtDNA haplotype frequencies among cohorts (total X2 = 25.38, df = 13, P = 0.02). The amount of drift in mtDNA (F'mt = 0.038) is also markedly greater than that for the nuclear loci (F f nucl = 0.009), and the randomization test rejects the null hypothesis of equality of and F' nucl (P = 0.03). Also, the relation between the amount of drift observed in the two sets of genes
r:
No. loci
n
3
63
3
109
3
67
7
69-132
8
60-122
8
118-164
8
97-102
14
54-63
14
94-101
14
121-123
F nud
0.028 0.254 0.032 0.850 0.433 0.822 0.661 0.194 0.820 14
118
14
57
14
54
14
81
0.660 1.000 0.218 0.02
3-14
1356
F /nud
0.030
0.017
0.039
0.026
0.023
0.011
0.001
-0.008
0.005
-0.004
0.009
0.001
0.042
0.029
0.011
-0.003
0.Ql8
0.009
0.012
0.004
0.017
0.004
0.034
0.016
0.040
0.025
0.021
0.009
(0.038/0.009 = 4.22) agrees well with the ratio of 4.00 expected when the two sexes are equally represented in the breeding segment of the population (Birky et al. 1983). Based on the observed value of and the demographic parameters the point estimate of the female effective size is 58. The estimate for the total population including both sexes is 97 (Table 3). Although the point estimate of N ef is subject to considerable sampling error, the statistically significant difference between r ; and F' nucl indicates that N ef is significantly smaller than N e • The standardized frequency shifts between cohorts born at various number of years apart are depicted in Figure 1 for mtDNA haplotypes (F'mt) and allozymes Under the assumption that random drift is the main source of temporal change, a comparatively smaller value of F' is expected for cohorts that are born one generation apart (Waples and Teel, 1990; Jorde and Ryman 1995). Such a minimum is observed both for the allozyme and the mtDNA data. Apparently, female generation length is somewhat longer than for both sexes considered together (8.26 yr vs. 7.82 yr; Table 3), and this difference may account for the observation that the lowest value of is found one year later than for the allozymes (at 8 and 7 yr, respectively; Fig. 1).
r.:
.r;».
r;
DISCUSSION
To our knowledge, the present study is the first that reports longitudinal mtDNA data from a natural population covering
913
BRIEF COMMUNICATIONS
TABLE 3. Observed frequency shifts of mtDNA haplotypes and allozymes in a natural brown trout population. C is the correction factor for overlapping generations and G is the mean generation length. F (eq. AI) is the observed amount of temporal allele frequency change averaged over all measurements (combinations of loci and consecutive cohorts) and F' (eq. A2) is the corresponding quantitycorrected for the expected contributionfrom sampling.The resulting estimates of effective population size, N, (eq. A3) and their 95% confidence intervals (CI) are also given. Females: mtDNA
data
Demographic data Temporal change
C G
Number of Fvalues
F F'
Estimated effective size
Ne CI
0.12 mitochondrial DNA
0.08
F'
Both sexes: allozyme data'
18.14 8.26 13
13.17 7.82 124
0.078222 0.038127 58 16-502
0.020752 0.008708 97 56-195**
nuclearDNA
0.10
0.06
0.04
'" '"
0.02
0
, From Jorde and Ryman 1996. " Calculated according to Waples (1989).
several years and cohorts; the main observations may be summarized as follows. First, we have detected changes in mtDNA haplotype frequencies over an evolutionarily short period in a natural population of brown trout. Second, it has been possible to detect these frequency changes from a limited number of sampled individuals. Third, we have obtained an estimate of the effective population size corresponding to the observed haplotype frequency changes. This effective size estimate refers to the females of the population only, as mtDNA is maternally inherited. Fourth, we have been able to compare the patterns of haplotype frequency changes of the mtDNA with the allele frequency changes at nuclear markers. Finally, it has been possible to compare the estimates of effective size for to the total population with the estimate obtained for the females. An important finding of the present study is the large temporal shifts in haplotype frequency among cohorts. While the magnitude of such differences vary considerably among pairs of cohorts, presumably reflecting the stochastic nature of genetic drift, even cohorts born in consecutive years differ in haplotype frequency by as much as 0.24 (Table 2). With sample sizes of about 50 individuals, this difference is highly significant (P = 0.009), demonstrating that there are real genetic changes occurring within the population over a very short time scale. Considering cohorts born at somewhat longer intervals, the difference in haplotype frequency may be even larger, in this study up to 0.359 over a 12-year period, which corresponds to about 1.5 generations. The finding of large temporal frequency shifts within a single biological population shows that caution must be exercised when interpreting gene-frequency data in general. Gene frequencies at nuclear loci and mtDNA are frequently used for characterizing populations genetically, and differences among populations are used for drawing conclusions regarding the degree of reproductive isolation and demographic autonomy of local populations. Clearly, such conclusions may be quite misleading when no information exists on the temporal stability of the gene frequencies. To reduce
2
3
4
5
6
7
8
9
10
Years apart 1. Amount of temporal allele-frequency change between cohorts born at different number of years apart. The change is measured by F' (eq. A2), and F' is the average over cohorts and loci. Vertical lines represent the generation interval (G) estimated for the total population and its female segment, respectively (7.82 vs. 8.26 yr; Table 3).
FIG.
the risk of such errors, it is imperative that samples of individuals for genetic analyses are collected in an appropriate manner. In particular, when sampling age-structured populations, such as the brown trout considered here, samples should include as wide a range of age classes as possible. Alternatively, individuals should be aged to check the temporal stability of the gene frequencies. The effective number of females estimated for this brown trout population is clearly quite small (Nej = 58; Table 3), although the estimate is subject to considerable sampling error (see below). The point estimate is close to half that of the total effective size (Ne = 97; Jorde and Ryman 1996), as expected if the effective numbers of the two sexes are the same. The finding of similar effective numbers of the two sexes is in rough agreement with direct observations on the relative numbers of females and males in the population (a female:male ratio of 0.46 was observed for the total population and 0.38 among breeding individuals). It has not been possible to compare the effective numbers with the total number of fish in this lake because we lack reliable estimates for the latter quantity. However, indirect observations indicate that the total number of individuals is likely to be considerably larger than the estimated effective size (Jorde and Ryman 1996). Although it has been possible to obtain a point estimate of N ej it is evident that this estimate may be fairly imprecise as judged from the large confidence interval (Table 3). The lower 95% confidence limit for Nejindicates that the effective number of females is not smaller than 16, which seems reasonable relative to the point estimate of 58. The upper limit of 502 is quite high, however. With this estimation technique the upper confidence limit for N, is frequently very sensitive to even minor variations in the point estimate of F. In contrast to the confidence limits for N ej , the estimate of total effective
914
BRIEF COMMUNICATIONS
size (N e ) is much less uncertain. This discrepancy may appear somewhat counterintuitive, considering the larger shifts expected for mtDNA (which is also observed in this study), but it can be shown that there is a theoretical basis for this observation. In short, the F'nucl used to obtain Ne is based on a larger number of independent observations than the used for Ne! , which results in a higher precision. The relatively large temporal genetic changes observed in the mtDNA genome resulting in a low estimated female effective size is an interesting observation from the perspective of evolutionary biology. Small effective sizes imply that stochastic processes may have a profound impact on the genetic composition of natural populations. Despite the small N e , however, the amount of genetic variation at allozyme loci measured as average heterozygosity is not particularly low in this population. For mtDNA the average haplotype diversity (h; Nei and Tajima 1981b) is 0.436. Although mtDNA data for brown trout is not yet available from the other nearby lakes, comparison with the haplotype diversity found in brown trout populations in other parts of Europe (Hynes et al. 1996) does not suggest that the diversity observed here is particularly low. One possible explanation for the relatively high levels of genetic variation in the present population, in spite of its low effective size for both nuclear and mtDNA markers, is that some gene flow exists from neighboring populations. At present, however, empirical data on this issue are lacking.
r;
ACKNOWLEDGMENTS
This study was supported by grants to NR from the Swedish Natural Science Research Council and the Swedish Environmental Protection Agency. PEl was supported by a postdoctoral grant from the Research Council of Norway (no. 109332/410). LITERATURE CITED AVISE, 1. C. 1994. Molecular markers, natural history and evol ution. Chapman and Hall, New York. BEGON, M., C. B. KRIMBAS, AND M. LOUKAS. 1980. The genetics of Drosophila subobscura populations. Heredity 45:335-350. BIRKY, C. W., T. MARUYAMA, AND P. FUERST. 1983. An approach to population and evolutionary genetic theory for genes in mitochondria and chloroplast, and some results. Genetics 103 :513527. BURNS, K. J., AND R. M. ZINK. 1990. Temporal and geographic homogeneity of gene frequencies in the fox sparrow (Passerella iliaca). Auk 107:421-425. DESALLE, R., A. TEMPLETON, 1. MORI, S. PLETSCHER, AND J. S. JOHNSTON. 1987. Temporal and spatial heterogeneity ofmtDNA polymorphisms in natural populations of Drosophila mercatorum. Genetics 116:215-223. EFRON, B., AND R. J. TlBSHIRANI. 1993. An introduction to the Bootstrap. Chapman and Hall, New York. ELLIOTT, J. M. 1994. Quantitative ecology and the brown trout. Oxford Univ. Press, New York. FELSENSTEIN, J. 1971. Inbreeding and variance effective numbers in populations with overlapping generations. Genetics 68:581597. GRANT, P. R., AND B. R. GRANT. 1992. Demography and the genetically effective sizes of two populations of Darwin's finches. Ecology 73:766-784. HALL, H. J., AND L. W. NAWROCKI. 1995. A rapid method for detecting mitochondrial DNA variation in the brown trout, Salmo trutta. J. Fish BioI. 46:360-364.
HANSEN, M. M., AND V. LOESCHCKE. 1996. Temporal variation in mitochondrial DNA haplotype frequencies in a brown trout (Salmo trutta L.) population that shows stability in nuclear allele frequencies. Evolution 50:454-457. HEDGECOCK, D., V. CHOW, AND R. S. WAPLES. 1992. Effective population number of shellfish broodstocks estimated from temporal variance in allelic frequencies. Aquaculture 108:215-232. HYNES, R. A., A. FERGUSON, AND M. A. MCCANN. 1996. Variation in mitochondrial DNA and post-glacial colonization of north western Europe by brown trout. J. Fish BioI. 48:54-67. JONSSON, B. 1977. Demographic strategy in a brown trout population in western Norway. Zool. Scr. 6:255-263. JORDE, P. E., AND N. RYMAN. 1995. Temporal allele frequency change and estimation of effective size in populations with overlapping generations. Genetics 139: 1077-1090. - - - . 1996. Demographic genetics of brown trout (Salmo trutta) and estimation of effective population size from temporal change of allele frequencies. Genetics 143:1369-1381. KRIMBAS, C. B., AND S. TSAKAS. 1971. The genetics of Dacusoleae. V. Changes of esterase polymorphism in a natural population following insecticide control-selection of drift? Evolution 25: 454-460. NEt, M., AND R TAJIMA. 1981a. Genetic drift and estimation of effective population size. Genetics 98:625-640. - - - . 1981b. DNA polymorphism detectable by restriction endonucleases. Genetics 97: 145-163. NIELSEN, E. E., M. M. HANSEN, AND V. LOESCHCKE. 1997. Analysis of microsatellite DNA from old scale samples of Atlantic salmon Salmo salar: a comparison of genetic composition over 60 years. Mol. Ecol. 6:487-492. POLLAK, E. 1983. A new method for estimating the effective population size from allele frequency changes. Genetics 104:531548. ROBSON, D. S., AND D. G. CHAPMAN. 1961. Catch curves and mortality rates. Trans. Am. Fish. Soc. 91: 181-189. RYMAN, N. 1983. Patterns of distribution of biochemical genetic variation in salmonids: differences between species. Aquaculture 33: 1-21. RYMAN, N., AND G. STAHL. 1980. Genetic changes in hatchery stocks of brown trout (Salmo trutta). Can. J. Fish. Aquat. Sci. 37:82-87. TAGGART, J. B., R. A. HYNES, P. A. PRODOHL, AND A. FERGUSON. 1992. A simplified protocol for routine total DNA isolation from salmonid fishes. J. Fish BioI. 40:963-965. THOMAS, W. K., S. PAABO, R X. VILLABLANCA, AND A. C. WILSON. 1990. Spatial and temporal continuity of kangaroo rat populations shown by sequencing mitochondrial DNA from museum specimens. J. Mol. Evol. 31:101-112. WAPLES, R. S. 1989. A generalized approach for estimating effective population size from temporal changes in allele frequency. Genetics 121:379-391. WAPLES, R. S., AND D. J. TEEL. 1990. Conservation genetics of Pacific salmon. 1. Temporal changes in allele frequency. Conservo BioI. 4: 144-156. WOOD, J. W. 1987. The genetic demography of the Gainj of Papa New Guinea. 2. Determinants of effective population size. Am. Nat. 129:165-187. YOUNGS, W. D., AND D. S. ROBSON. 1978. Estimation of population number and mortality rates. Pp. 137-164 ill T. Bagenal, ed. Methods for assessment of fish production in fresh water. IBP handbook no. 3. 3d ed. Blackwell Scientific, Oxford. Corresponding Editor: R. DeSalle
ApPENDIX The magnitude of haplotype frequency shift between consecutive cohorts was measured by Pollak's (1983) F-value,
F
= _1_
i
a - I i~ I
where
Xi,1
(Xi,t (Xi.1
Xi,t+I}2 ,
(Al)
+ Xi,1+ I }/2
is the observed frequency of the ith haplotype in the
915
BRIEF COMMUNICAnONS
sample from cohort t, and the summation is over all a haplotypes at the locus; in this case a = 2 because only two haplotypes were detected. This standardized variance of temporal frequency shifts is corrected for the expected contribution from sampling under the assumption that the number of newborn individuals of a cohort is large. The new measure is ,
F mt == F mt
I -
-
-
n,
(A2)
-
n., I
where n, is the number of individuals in cohort t examined for mtDNA variability. The female effective size was estimated as
.
C
where F'mt is the unweighted mean of the 13 F'mt values (Table 2). C is a correction factor that accounts for the contribution to F"" that arises because generations are overlapping and each cohort only represents a fraction of the total population (Jorde and Ryman 1995). and G is the generation length. The C correction factor is determined exclusively by the li- and bi-values (Table O. and it is therefore independent of population size and allele frequency (for computational formulas see Jorde and Ryman 1996). The female mean generation length, G, is calculated from the data in Table 1 according to Felsenstein (1971) as
(A3)
Nef = GF' • mt
k
G =
2: p.i,
(A4)
;=1
where k is the number of age classes and p; = lib; is the estimated probability that the haplotype in an individual was inherited from a female of age i,
Evolution, 52(3). 1998. pp. 915-920
DNA HETEROZYGOSITY AND GROWTH RATE IN THE ATLANTIC COD GADUS MORHUA (L) GRANT I Department
H.
POGSON 1 AND SVEIN ERIK FEVOLDEN 2
of Biology and Institute of Marine Sciences, University of California, Santa Cruz, California 95064 E-mail: pogsontirdarwin.ucsc.edu 2University of Tromso, Nonvegian College of Fishery Science, N-9037 Tromse, Nonvay E-mail: sveinftisnfh.uit.no
Abstract.-Relationships between growth rate and the degree of individual heterozygosity at ten nuclear RFLP loci were examined in two populations of the Atlantic cod. Gadus morhua, from northern Norway. A highly significant positive correlation was observed between growth rate and DNA heterozygosity in one population (Balsfjord) but not in the other (Barents Sea). Our results provide support for an important prediction of the associative overdominance hypothesis that heterozygosity-fitness correlations can be detected at neutral genetic markers and suggest that environmental conditions might playa dominant role in the manifestation of the correlation.
Key words.-Associative overdorninance, Gadus morhua, growth rate. heterozygosity, nuclear RFLP loci. Received September 26. 1997.
Although correlations between the degree of individual heterozygosity and fitness-related characters have been commonly described in natural populations of many species. the genetic basis of these associations remains poorly understood (see reviews by Mitton and Grant 1984; Zouros and Foltz 1987; Houle 1989; Zouros and Pogson 1994; Hedgecock et al. 1996). Progress towards understanding the genetic basis of the heterozygosity-fitness correlation has been hampered by several factors. First. when a correlation is observed it typically accounts for a small proportion of the observed phenotypic variance (usually 3-6%) making the phenomenon difficult to detect and interpret (see Britten 1996). Second. the correlation often exhibits low repeatability, even in the same population over consecutive years (e.g., Gaffney 1990; Pecon Slattery et al. 1993; David and lame 1997). Third, and perhaps most importantly, it has proven extremely difficult to distinguish between competing explanations for the phenomenon that either treat the genetic markers scored in these studies as causative agents of the correlation (called here the "selection" hypothesis), or as neutral markers of linked deleterious genes (the "associative overdominance"
Accepted March 2. 1998.
hypothesis). The inability to distinguish between the selection hypothesis and the associative overdominance hypothesis is primarily due to the fact that both explanations make a number of remarkably similar predictions (Charlesworth 1991; Zouros 1993; Houle 1994; however. see David 1997). A test of the associative overdominance hypothesis has been recently proposed by Pogson and Zouros (1994) which is based on comparing the effects of heterozygosity at the allozyme and DNA levels on a fitness-related character. All selection hypotheses predict that the correlation should only be expressed at markers having the potential to directly influence fitness characters (i.e .• allozymes) and should not be expressed at presumably neutral markers (i.e .• restriction site polymorphisms occurring in noncoding DNA regions). In contrast. the associative overdominance explanation predicts that the correlation should not be restricted to any particular type of genetic marker, provided that these have the potential to assess inbreeding levels or be linked to deleterious recessive genes. In the only application of this test to date, Pogson and Zouros (1994) examined the effects of heterozygosity at eight nuclear RFLP loci in a cohort of juvenile scallops (Pla-
