7
8
6
9
tests were adjusted by Bonferonni correction. Differences between the strains in the
average number of alleles per locus and mean expected (H
e
) and observed (H
o
)
heterozygosities were tested using ANOVA (loci as blocks). Loci with null alleles
were not used in these tests.
2.3.1. Predicted distributions of relatedness (population allele frequencies)
The observed allele frequencies were used to generate 2500 pairs of individuals for
each of three relatedness classes (full-sib, half-sib, unrelated) for each strain with a visual
basic program (Blouin, unpublished). The program calculated relatedness for each of these
pairs using the following equation of the relatedness coefficient (r) of Queller and
Goodnight (1989):
r
X
x
X
k
X
l
P
y
P*
X
x
X
k
X
l
P
x
P*
where x indexes individuals (two in pairwise comparisons), k indexes loci and l indexes
allelic position. P
x
is the frequency of an allele at allelic position l at locus k in individual
x. P
y
is the frequency of that allele in individual y and P* is the frequency of the allele in
the population. This calculation is considered symmetric as both individuals are consid-
ered as x and y (individual allele frequencies, corrected for their occurrence in the
population, for each allelic position for each locus for each individual are summed in both
the numerator and the denominator prior to division) (Queller and Goodnight, 1989). Di-
allelic loci cannot be used with this equation as the relatedness between a pair of
heterozygotes will be undefined. This occurs because the denominator (and the numerator)
will always sum to zero (Queller and Goodnight, 1989). As mentioned previously, loci
were not used if they contained null alleles or were di-allelic (Table 1).
The calculated values of r from the generated pairs were plotted to produce
expected distributions for each relatedness class within each of the three strains (see
Blouin et al., 1996 for details of the approach). Expected rates of error were
determined for specific cut-off values of r as the proportion of related pairs whose
r-value falls to the left of (less than) the cut-off (probability of assigning a related pair
as unrelated). This was repeated for r values ranging from 0 to 0.3 for both the
half- and full-sib distributions. This was chosen as a useful range for minimizing
inbreeding because while the probability of inadvertently selecting a related pair will
continue to decrease with lower values of r, few unrelated pairs will be available for
selection. Conversely higher cut-off values, while allowing for more unrelated pairs to
be available, will increase the error rate to unacceptable levels. In practice, this range
could be adjusted to meet the needs of the breeder in obtaining enough breeding pairs
while minimizing the inclusion of related pairs.
2.3.2. Actual distributions of relatedness (progeny of known parents)
A second set of distributions for pairs of progeny based on known parental
genotypes was created for each of the three strains in order to determine how
G.J. McDonald et al. / Aquaculture 233 (2004) 6578 70
accurate these distributions of r would be for the selection of unrelated couples. While
actual progeny had been produced from the adult fish used here (McDonald, 2001),
their numbers were too small to allow for the determination of distributions of r in the
populations. This limitation was overcome as follows. For each strain, 12 males and
12 females were randomly selected from the 48 individuals used to estimate allele
frequencies. Two simulated progeny genotypes were produced for all possible male by
female parent combinations within each strain PROBMAXG (Danzmann, 1997). This
program uses entered parental genotypes to generate possible progeny genotypes
according to Mendelian expectations. r was then calculated for all pairs of simulated
progeny within each strain using the software package, RELATEDNESS (version 5.6)
(Queller and Goodnight, 1989). The allele frequencies used in the calculations were
the same as those used in the previous analysis and were based on 48 individuals in
each strain. While the simulated progeny could be used to provide sufficient
relatedness comparisons to produce r distributions for both unrelated and half-sib
classes, not enough full-sib comparisons could be made. Thus, 10 additional progeny
were generated for each full-sib family and r was calculated for all pairs of
individuals within those families.
In order to standardize the number of pairwise comparisons per distribution, r values
for 2500 pairs were randomly selected from all those that were available for each
relatedness class within each strain. The potential error rate of assigning related pairs as
unrelated was again determined as the proportion of related pairs whose calculated r value
falls to the left of r values ranging from 0 to 0.3.
3. Results
The three strains differed significantly in the average number of alleles per locus
( P= 0.0015) and marginally in average H
e
( P= 0.057) (Table 1). No differences were
detected for H
o
( P>0.05). The degree of conformity of genotype proportions to Hardy
Weinberg expectations also differed among strains. Strain O showed significant deviations
from HardyWeinberg proportions at eight loci while strains G and B showed no
deviations after Bonferroni correction.
Table 2
Mean (standard deviation) for r distributions of different classes of relatedness in three strains of rainbow trout
Strain Distribution Unrelated Half-sib Full-sib
B Predicted 0.002 (0.148) 0.248 (0.155) 0.498 (0.148)
Actual 0.046 (0.180) 0.277 (0.160) 0.526 (0.146)
G Predicted 0.017 (0.172) 0.255 (0.172) 0.500 (0.170)
Actual 0.012 (0.185) 0.218 (0.182) 0.451 (0.183)
O Predicted 0.008 (0.180) 0.255 (0.181) 0.500 (0.167)
Actual 0.017 (0.189) 0.154 (0.187) 0.347 (0.185)
Predicted distributions were determined with randomly generated pairs of individuals based on allele frequencies.
The actual distributions were generated from genotypes of known individuals in the strains. Each distribution is
comprised of 2500 pairwise comparisons.
G.J. McDonald et al. / Aquaculture 233 (2004) 6578 71
Fig. 1. (AC) Relatedness coefficient (r) distributions for three aquaculture strains of rainbow trout (B, G, O).
Each plot shows r values for three different relatedness classes (xunrelated, nhalf-sib and .full-sib).
Solid lines represent distributions obtained from actual population comparisons (between generated progeny of
known parents) while dashed lines represent predicted distributions based on pairs of individuals randomly
generated from population allele frequencies.
G.J. McDonald et al. / Aquaculture 233 (2004) 6578 72
Fig. 2. (AC) The probability of assigning either full- (
.
) or half- (n) sib pairs as unrelated (type II error under
null hypothesis that pair is unrelated) at or below differing values of r for each of the three strains of rainbow trout
examined. Dashed lines indicate values based on the predicted distributions while solid lines represent the rates
from distributions derived from the actual populations.
G.J. McDonald et al. / Aquaculture 233 (2004) 6578 73
The predicted distributions of r (those calculated by generating pairs of each relatedness
class using empirically determined allele frequencies) had means corresponding to those
expected (Fig. 1; Table 2). Unrelated, half-sib and full-sib distributions showed mean
values of approximately 0, 0.25 and 0.5, respectively, in all three strains. However, the
strains differed in the degree of overlap among the distributions of the three relatedness
classes (i.e. in the standard deviations). For instance, Strain B showed the least degree of
overlap (Fig. 1) and consequently the standard deviations for the distributions are the
lowest (Table 2). Strains G and O showed similar standard deviations. Thus, the predicted
error rate (probability of assigning a half-sib or full-sib pair as unrelated based on r values)
is expected to be lowest in strain B (Fig. 2).
The distributions of r produced by the simulated progeny from known parents
within each of the three strains differed from those based on strain allele frequencies
(predicted distributions). For strains O and G, the mean r values generated from
known parents were significantly lower (Table 2) than those based on strain allele
frequencies (MannWhitney U; P< 0.0001 for both strains at all three levels of
relatedness). This was characterized by a shifting to the left of all distributions
compared to those predicted (Fig. 1). The most marked left shift was observed for the
half- and full-sib distributions of strain O. These left shifts would increase the amount
of error (assignment of related pairs as unrelated) for a given value of r over that
predicted above (Fig. 2). Conversely, strain B showed a more modest but highly
significant (MannWhitney U; P< 0.0001) right shift for the curves derived from the
simulated progeny (Fig. 1; Table 2). The result of the right shift would be a reduction
in the amount of related pairs assigned as unrelated for a given value of r.
4. Discussion
The application of Queller and Goodnights (1989) relatedness statistic to the estima-
tion of relatedness had varied success in the three strains of rainbow trout examined here.
In two of the three strains (O and G), the r distributions derived from potential progeny of
known parents were lower than those predicted using pairs of individuals created from
strain specific allele frequencies. This was most pronounced for strain O. The net result of
these left shifts in the distributions is an underestimation of relatedness. An underestima-
tion of relatedness for pairs of known pedigree has been observed in several studies using
this technique (e.g. Altmann et al., 1996; de Ruiter and Geffen, 1998; Hansson et al., 2000;
Kays et al., 2000) and has been attributed to a high degree of relatedness among the
individuals used to estimate population allele frequencies.
The existence of left shifts in the distributions of r and the consequent underestimation
of relatedness among pairs has implications to captive breeding programs of fishes. The
selection of a cut-off value of r from predicted distributions (based on pairs generated with
allele frequencies) thought to represent an acceptable probability of assignment of related
pairs as unrelated might result in higher error than anticipated. This can be visualized as a
greater proportion of the r distributions, produced from the real populations, falling to the
left of the cut-off than predicted. The inclusion of a greater number of related pairs would
result in an increase in inbreeding.
G.J. McDonald et al. / Aquaculture 233 (2004) 6578 74
The results reported here for three rainbow trout strains suggest that the accurate
estimation of relatedness demonstrated by Norris et al. (2000) is not expected to be
universal. The approach may be more suitable in strains with certain genetic attributes and
breeding histories. For example, strains G and B appear to have been derived from
relatively large numbers of males and females and have not experienced intense selection
(McKay, personal communication). In contrast, strain O has undergone strong selection
with associated reductions in genetic effective population size during its culture history.
This strain has experienced inbreeding due to the use of a limited number of broodstock in
each generation (e.g. one to four females). Moreover, the individuals used to estimate
population allele frequencies arose from gametes produced by three females and an
unknown number of males. The relatedness among individuals in strain O is expected to
be high and can explain the severe bias and left shift in r distributions observed here. Thus,
strain O is not a good candidate for the estimation of relatedness among fish when
pedigree information is lacking.
Another factor that may have contributed to the bias observed in these strains is that the
individuals used to estimate allele frequencies may be the product of spawnings conducted
over a limited period of time (grandparents of the progeny examined here). Embryos
produced at the same time tend to hatch within a day or so of each other and within all three
strains synchronized hatching was observed (McKay, personal communication). It is
possible that females spawning at a particular time tend to be more related to each other
(Leary et al., 1989) than those spawning at different times due to the high heritability
reported for the trait of spawning time (Su et al., 1999). If this is the case, the progeny which
were used to estimate the allele frequencies would also be more related to each other.
Distinct spawning sub-populations have been identified in aquaculture populations of
rainbow trout using both microsatellites (Fishback et al., 2000) and mtDNA (Ferguson et
al., 1993; Danzmann et al., 1994) and therefore sampling within a sub-group may not
provide an accurate estimate of strain allele frequencies.
Aside from contributing towards bias, the short period of spawning for females could
further influence the success of this approach in selecting unrelated pairs. At any cut-off
level of r, only a proportion of the expected distribution of all unrelated pairs in the
population is available for breeding. Additionally, female rainbow trout have a very
restricted ovulation period (less than a week) and may not be sexually mature when needed
(based on their r values with available males) and thus the number of available pairs is
further reduced. Therefore, the chosen cut-off may be dependant more on the provision of
enough breeding pairs to produce another generation rather than an acceptable error rate,
unless the population is very large.
The particular relatedness estimator used in the present analysis has been criticized as
being biased in that the estimated means of relatedness are not equal to the true value of
relatedness of the pairs being tested (Van de Casteele et al., 2001). This observation should
not affect the approach being investigated here. To be successful, the estimated distribu-
tions based on individuals generated according to allele frequencies should be identical to
the distributions produced from actual pairs within the population being investigated. As
long as this is true, even if the distributions do not centre exactly around the true value of
relatedness (e.g. 0.5 for full-sibs), it should not affect the accuracy of unrelated pair
selection.
G.J. McDonald et al. / Aquaculture 233 (2004) 6578 75
A comparison of the results reported here with that of Blouin et al. (1996) and Norris et
al. (2000) illustrates the effect of locus number and relative polymorphism on the potential
success of such an analysis. While Blouin et al. (1996) reported mean heterozygosities
similar to those of strain B, their reported standard deviations for relatedness distributions
were less (0.1140.122) than those observed for the predicted distributions in this study.
This is due to their use of a greater number of loci in r calculations (resulting in more
accurate pairwise values of r and thus a reduction in distribution width for relatedness
classes). Norris et al. (2000) achieved similar standard deviations (0.130.16) to those
observed for the predicted distributions of strain B, but only required eight loci to achieve
this. This was due to a greater mean heterozygosity of those markers (0.85) (Norris et al.,
1999). These studies coupled with the results reported here suggest that the most efficient
approach for relatedness estimation is to choose fewer numbers of highly variable loci. It
may be difficult to identify such loci in aquaculture strains that have experienced severe
bottlenecks in their genetic histories as they will have substantially lower amounts of
variation overall. Of course the application of this approach in these strains will be
precluded anyway due to the high background relatedness that would result in inaccurate
relatedness estimation.
Acknowledgements
This work was supported by funds from the Ontario Ministry of Agriculture, Food and
Rural Affairs (Program 42-Aquaculture). We thank Alastair Wilson for commenting on the
manuscript and Dr. Michael Blouin for providing software used in the data analysis.
References
Altmann, J., Alberts, S.C., Haines, S.A., Dubach, J., Muruthi, P., Coote, T., Geffen, E., Cheesman, D.J., Mututua,
R.S., Saiyalel, S.N., Wayne, R.K., Lacy, R.C., Bruford, M.W., 1996. Behaviour predicts genetic structure in a
wild primate group. Pop. Biol. 93, 57975801.
Banks, M.A., Rashbrook, V.K., Calavetta, M.J., Dean, C.A., Hedgecock, D., 2000. Analysis of microsatellite
DNA resolves genetic structure and diversity of chinook salmon (Oncorhynchus tshawytscha) in Californias
Central Valley. Can. J. Fish. Aquat. Sci. 57, 915927.
Bardakci, F., Skibinski, D.O.F., 1994. Application of the RAPD technique in tilapia fish: species and subspecies
identification. Heredity 73, 117123.
Bernatchez, L., Duchesne, P., 2000. Individual-based genotype analysis in studies of parentage and population
assignment: how many loci, how many alleles? Can. J. Fish Aquat. Sci. 57, 112.
Blouin, M.S., Parsons, M., Lacaille, V., Lotz, S., 1996. Use of microsatellite loci to classify individuals by
relatedness. Mol. Ecol. 5, 393401.
Caballero, A., Toro, M.A., 2000. Interrelations between effective population size and other pedigree tools for the
management of conserved populations. Genet. Res. 75, 331343.
Danzmann, R.G., 1997. Probmax: a computer program for assigning unknown parentage in pedigree analysis
from known genotypic pools of parents and progeny. J. Heredity 88, 33.
Danzmann, R.G., Ferguson, M.M., Heculuck, D.M., 1994. Heterogeneity in the distribution of mitochondrial
DNA haplotypes in female rainbow trout spawning in different seasons. Can. J. Fish. Aquat. Sci. 51
(Suppl. 1), 284289.
de Ruiter, J.R., Geffen, E., 1998. Relatedness of matrilines, dispersing males and social groups in long-tailed
macaques (Macaca fascicularis). Proc. R. Soc. Lond., B 265, 7987.
G.J. McDonald et al. / Aquaculture 233 (2004) 6578 76
Estoup, A., Gharbi, K., SanCristobal, M., Chevalet, C., Haffray, P., Guyomard, R., 1998. Parentage assignment
using microsatellites in turbot (Scophthalmus maximus) and rainbow trout (Oncorhynchus mykiss) hatchery
populations. Can. J. Fish Aquat. Sci. 55, 715725.
Falconer, D.S., MacKay, T.F.C., 1996. Introduction to Quantitative Genetics, 4th ed. Longman Group, Essex.
464 pp.
Ferguson, M.M., Danzmann, R.G., 1998. Role of genetic markers in fisheries and aquaculture: useful tools or
stamp collecting? Can. J. Fish. Aquat. Sci. 55, 15531563.
Ferguson, M.M., Danzmann, R.G., Arndt, S.K.A., 1993. Mitochondrial DNA and allozyme variation in Ontario
cultured rainbow trout spawning in different seasons. Aquaculture 117, 237259.
Fishback, A.G., Danzmann, R.G., Sakamoto, T., Ferguson, M.M., 1999. Optimization of semi-automated micro-
satellite multiplex polymerase chain reaction systems for rainbow trout (Oncorhynchus mykiss). Aquaculture
172, 247254.
Fishback, A.G., Danzmann, R.G., Ferguson, M.M., 2000. Microsatellite allelic heterogeneity among hatchery
rainbow trout maturing in different seasons. J. Fish Biol. 57, 13671380.
Goodnight, K.F., Queller, D.C., 1999. Computer software for performing likelihood tests of pedigree relationship
using genetic markers. Mol. Ecol. 8, 12311234.
Hansson, B., Bensch, S., Hasselquist, D., Lillandt, B.G., Wennerberg, L., Von Schantz, T., 2000. Increase of
genetic variation over time in a recently founded population of great reed warblers (Acrocephalus arundi-
naceus) revealed by microsatellites and DNA fingerprinting. Mol. Ecol. 9, 12591538.
Harris, A.S., Bieger, S., Doyle, R.W., Wright, J.M., 1991. DNA fingerprinting of tilapia, Oreochromis niloticus,
and its application to aquaculture genetics. Aquaculture 92, 157163.
Herbinger, C.M., Doyle, R.W., Pitman, E.R., Paquet, D., Mesa, K.A., Morris, D.B., Wright, J.M., Cook, D.,
1995. DNA fingerprint based analysis of paternal and maternal effects on offspring growth and survival in
communally reared rainbow trout. Aquaculture 137, 245256.
Kays, R.W., Gittleman, J.L., Wayne, R.K., 2000. Microsatellite analysis of kinkajou social organization. Mol.
Ecol. 9, 743751.
Leary, R.F., Allendorf, F.W., Knudsen, K.L., 1989. Genetic differences among rainbow trout spawned on differ-
ent days within a single season. Prog. Fish-Cult. 51, 1019.
Lynch, M., Ritland, K., 1999. Estimation of pairwise relatedness with molecular markers. Genetics 152,
17531766.
McConnell, S.K.J., OReilly, P., Hamilton, L., Wright, J.M., Bentzen, P., 1995. Polymorphic microsatellite loci
from Atlantic salmon (Salmo salar): genetic differentiation of North American and European populations.
Can. J. Fish. Aquat. Sci. 52, 18631872.
McDonald, G.J., 2001. Relatedness determination and the detection of spawning time QTL in rainbow trout
(Oncorhynchus mykiss). MSc thesis, Department of Zoology, University of Guelph, Guelph, ON, Canada.
McKay, L.R., McMillan, I., 1997. Growth, maturation and spawning times in crosses of three strains of rainbow
trout. Bull.-Aquac. Assoc. Can. 97, 1517.
Morris, D.B., Richard, K.R., Wright, J.M., 1996. Microsatellites from rainbow trout (Oncorhynchus mykiss) and
their use for genetic study of salmonids. Can J. Fish. Aquat. Sci. 53, 120126.
Nei, M., 1978. Estimation of average heterozygosity and genetic distance from a small number of individuals.
Genetics 89, 583590.
Norris, A.T., Bradley, D.G., Cunningham, E.P., 1999. Microsatellite genetic variation between and within farmed
and wild Atlantic salmon (Salmo salar) populations. Aquaculture 180, 247264.
Norris, A.T., Bradley, D.G., Cunningham, E.P., 2000. Parentage and relatedness determination in farmed Atlantic
salmon (Salmo salar) using microsatellite markers. Aquaculture 182, 7383.
OConnell, M., Danzmann, R.G., Cornuet, J.M., Wright, J.M., Ferguson, M.M., 1997. Differentiation of rainbow
trout populations in Lake Ontario and the evaluation of the stepwise mutation and infinite allele mutation
models using microsatellite variability. Can. J. Fish. Aquat. Sci. 54, 13911399.
OReilly, P.T., Hamilton, L.C., McConnell, S.K., Wright, J.M., 1996. Rapid analysis of genetic variation in
Atlantic salmon (Salmo salar) by PCR multiplexing of dinucleotide and tetranucleotide microsatellites. Can.
J. Fish Aquat. Sci. 53, 22922298.
OReilly, P.T., Herbinger, C., Wright, J.M., 1998. Analysis of parentage determination in Atlantic salmon (Salmo
salar) using microsatellites. Anim. Genet. 29, 363370.
G.J. McDonald et al. / Aquaculture 233 (2004) 6578 77
Painter, I., 1997. Sibship reconstruction without parental information. J. Agric. Biol. Environ. Stat. 2, 212229.
Pante, M.J.R., Gjerde, B., McMillan, I., 2001a. Effect of inbreeding on body weight at harvest in rainbow trout,
Oncorhynchus mykiss. Aquaculture 192, 201211.
Pante, M.J.R., Gjerde, B., McMillan, I., 2001b. Inbreeding levels in selected populations of rainbow trout,
Oncorhynchus mykiss. Aquaculture 192, 213224.
Queller, D.C., Goodnight, K.F., 1989. Estimating relatedness using genetic markers. Evolution 43, 258275.
Quinton, C.D., 2001. Growth rate and spawning time in diallele crosses of three strains of rainbow trout
(Oncorhynchus mykiss). MSc thesis, Department of Animal and Poultry Science, University of Guelph,
Guelph, ON.
Raymond, M., Rousset, F., 1995. GENEPOP (Version 1.2): population genetics software for exact tests and
ecumenicism. J. Heredity 86, 248249.
Sakamoto, T., 1996. Study for the introduction of positional cloning method using DNA markers in fisheries
science. PhD thesis, Tokyo University of Fisheries, Tokyo, Japan.
Sakamoto, T., Okamoto, N., Ikeda, Y., 1994. Dinucleotide repeat polymorphism of rainbow trout, FGT5. J. Anim.
Sci. 72, 2768.
Sanchez, J.A., Clabby, C., Ramos, D., Blanco, G., Flavin, F., Vasquez, E., Powell, R., 1996. Protein and
microsatellite single locus variability in Salmo salar L. (Atlantic salmon). Heredity 77, 423432.
Scribner, K.T., Gust, J.R., Fields, R.L., 1996. Isolation and characterization of novel microsatellite loci: cross-
species amplification and population genetics applications. Can. J. Fish. Aquat. Sci. 53, 841883.
Smith, B.R., Herbinger, C.M., Merry, H.R., 2001. Accurate partition of individuals into full-sib families from
genetic data without parental information. Genetics 158, 13291338.
Su, G.S., Liljedahl, L.E., Gall, G.A.E., 1996. Effects of inbreeding on growth and reproductive traits in rainbow
trout (Oncorhynchus mykiss). Aquaculture 142, 139148.
Su, G.S., Liljedahl, L.E., Gall, G.A.E., 1999. Estimates of phenotypic and genetic parameters for within-season
date and age at spawning of female rainbow trout. Aquaculture 171, 209220.
Thomas, S.C., Hill, W.G., 2000. Estimating quantitative genetic parameters using sibships reconstructed from
marker data. Genetics 155, 19611972.
Van de Casteele, T., Galbusera, P., Matthysen, E., 2001. A comparison of microsatellite-based pairwise related-
ness estimators. Mol. Ecol. 10, 15391549.
G.J. McDonald et al. / Aquaculture 233 (2004) 6578 78
Mult mai mult decât documente.
Descoperiți tot ce are Scribd de oferit, inclusiv cărți și cărți audio de la editori majori.
Anulați oricând.