Anda di halaman 1dari 9

J. Anim. Breed. Genet.

ISSN 0931-2668

ORIGINAL ARTICLE

Estimation of genetic parameters and genome scan for 15


semen characteristics traits of Holstein bulls
T. Druet1, S. Fritz2, E. Sellem2, B. Basso2, O. Gerard2, L. Salas-Cortes2, P. Humblot2, X. Druart2,4 &
A. Eggen3
1 INRA, UR337 Station de Genetique Quantitative et Appliquee, Jouy-en-Josas, France
2 Union Nationale des Cooperatives agricoles dElevage et dInsemination Animale, Paris, France
3 INRA, UR339 Laboratoire de Genetique biochimique et Cytogenetique, Jouy-en-Josas, France
4 INRA, UMR85 Physiologie de la Reproduction et des Comportements, Nouzilly, France

Keywords Summary
Genetic parameters; QTL detection; semen
characteristics. A QTL detection experiment was performed in French dairy cattle to
search for QTL related to male fertility. Ten families, involving a total of
Correspondence 515 bulls, were phenotyped for ejaculated volume and sperm concentra-
S. Fritz, Station de Genetique Quantitative et tion, number of spermatozoa, motility, velocity, percentage of motile
Appliquee, INRA Domaine de Vilvert, 78352
spermatozoa after thawing and abnormal spermatozoa. A set of 148
Jouy-en-Josas, France. Tel: +33 1 34 65 21 95;
Fax: + 33 1 34 65 22 08;
microsatellite markers were used to realize a genome scan. First, genetic
E-mail: sebastien.fritz@jouy.inra.fr parameters were estimated for all traits. Semen production traits were
found to have moderate heritabilities (from 0.15 to 0.30) while some of
Received: 02 June 2008; the semen quality traits such as motility had high heritabilities (close to
accepted: 18 October 2008 0.60). Genetic correlations among traits showed negative relationships
between volume and concentration and between volume and most
quality traits such as motility or abnormal sperm while correlations
between concentration and these traits were rather favourable. Percent-
ages of abnormal sperm were negatively related to quality traits, espe-
cially with motility and velocity of spermatozoa. Three QTL related to
abnormal sperm frequencies were significant at p < 0.01. In total, 11
QTL (p < 0.05) were detected. However, the number of QTL detected
was within the range of expected false positives. Because of the lack of
power to find QTL in this design further analyses are required to con-
firm these QTL.

making the situation critical for both farmers and


Introduction
breeding companies. Female fertility is now taken
Artificial insemination (AI) plays a major role in into account in breeding goals while less importance
dairy cattle because it is directly related to fertility is usually given to male fertility. However, the con-
and it is the main reproductive tool which is used ception rate with AI depends on the characteristics
for genetic improvement by introducing genes from of the semen provided by AI centres. The quality of
superior sires into herds. the semen depends on different factors such as the
Fertility is considered as one of the most important number of motile spermatozoa, sperm motility,
economic traits in dairy cattle. Despite its impor- sperm viability, abnormal spermatozoids and also
tance, conception rate in the Holstein breed has some genetic factors associated to the bull. The male
decreased dramatically in the past 20 years in most used for AI has been reported to be a low source of
countries (e.g. Lucy 2001; Jamrozik et al. 2005) variation for fertility explaining less than 4% of total

2009 Blackwell Verlag GmbH J. Anim. Breed. Genet. 126 (2009) 269277 doi:10.1111/j.1439-0388.2008.00788.x
QTL detection for male fertility T. Druet et al.

variation, but up to 10% differences for non-return Twelve quality traits were observed on a semen
rate (Humblot et al. 1991). sample after thawing. Three straws were mixed and
In addition to the importance of semen quality for analysed for each ejaculate. Individual motility (per-
fertility, sperm production of bulls in AI studs needs centage of motile sperm, MOT) and motility score on
to be optimized for economical reasons, especially a scale from 0 to 5 (MSCO) were evaluated under
because the demand for some AI sires is very high the microscope by a single technician. Percentage of
(Ducrocq & Humblot 1995). living sperm with an intact cytoplasmic membrane
To optimize management and selection for both (LIV) was first observed after thawing. The same
conception rate and sperm production, genetic measurement was made after exposure of sperm to
parameters need to be estimated between all the hypo-osmotic solutions (corresponding to 50, 40 and
involved traits. These parameters can indicate 30% of the isomolality) which measures the sperm
which traits can be efficiently selected, which traits resistance to osmotic stress (RES) considered as an
have a high repeatability (for which successive indicator of the quality of the plasma membrane.
records can be predicted with high precision), etc. The measurement was replicated twice for the three
The traits which are difficult or costly to measure solutions and the mean of the six measures was used
or to select (due to low heritability), can be in statistical analyses.
selected indirectly based on correlated traits. The Percentage of motile spermatozoa (MOTH), aver-
latter ones should have high correlation with traits age path velocity (VAP), amplitude of linear head
of interest and higher heritability and they should displacement (ALH) and percentage of progressive
be easier to measure. Marker assisted selection spermatozoa (PROG) were also determined by using
(MAS) is particularly expected to improve selection a computer assisted sperm analysis (CASA sys-
for traits difficult to measure or to select by con- tem; HTM-IVOS, http://www.hamiltonthorne.com/
ventional methods. However, MAS requires the research/spermanalysis/ivos.htm). VAP was mea-
implementation of a QTL detection program to sured twice for each animal and the mean was used
identify the QTL which could be used in the MAS for estimation of genetic parameters.
process. Finally, the percentage of total abnormal sperm
The objectives of this study were therefore (i) to (ABNO) and percentages of sperm with abnormal
estimate the genetic parameters among the traits head (HEAD), tail (TAIL) or presenting cytoplasmic
related to semen characteristics and (ii) to perform a droplet (DROP) were recorded.
genome scan for these traits.
Genotyping data
Materials and methods A total of 148 microsatellite marker genotypes were
used for linkage mapping. These markers were
Animal material
selected previously in two different experimental
Data were collected from 515 Holstein bulls born designs: the French MAS program (e.g. Guillaume
between 2000 and 2003 from 10 sire families et al. 2008) and a Normande-Holstein F-2 design
belonging to four breeding companies. The number implemented in an experimental farm (Larroque
of bulls per family ranged from 40 to 66 and pheno- et al. 2007). The mean number of markers per chro-
typic performances were directly recorded on these mosome was 5.1, ranging from 2 to 10. The size of
bulls (equivalent to a daughter design for QTL anal- the region covered by the markers varied from 11 to
ysis). The pedigree file included 2131 animals. 127 cM per chromosome with a mean of 86.7 cM
Semen production and semen quality traits were (the average spacing between adjacent markers was
recorded. Production traits included volume (VOL), 21.1 cM). The proportion of heterozygous sires aver-
sperm concentration (CONC) and number of sperm aged across all markers was 0.63 ranging from 0.47
(NSP) for the first ejaculate of the day. Total number to 0.72 depending on the family. Map distances used
of sperm per ejaculate was calculated by multiplying in this study were based on those published by Ihara
the concentration with the volume of semen. These et al. (2004).
traits were recorded between 12 and 18 months of
age at six semen production centres. The mean
Method
number of recorded ejaculate per bull was 9.1.
Semen collection strategies were rather homogenous Mixed models were used to analyse the effects of
within breeding company. various factors on sperm production and sperm

270 2009 Blackwell Verlag GmbH J. Anim. Breed. Genet. 126 (2009) 269277
T. Druet et al. QTL detection for male fertility

quality traits. For semen production traits, the model The genetic parameters were estimated with a
was: modified version of the airemlf90 program of Misztal
et al. (2002).
y ijklmn spci  ysj bck  agel pm um eijklmn 1 For QTL detection, gametic effects were added to
the models:
where yijklmn is the nth record associated to bull m
from semen production centre i from the breeding y ijklmn spci  ysj bck  agel pm um vpm
3
company k recorded in season-year j at age l, vm
m eijklmn
spci*ysj is the fixed effect of the interaction between
the semen production centre, the season and the and
year, bck*agel is the fixed effect of the interaction y iom spci dlo pm um vpm vm
m eiom 4
between the breeding company and the age class,
pm is the permanent environment associated to ani- where vpm and vm m are the effects of the paternal and
mal m, um is the polygenic effect of animal m and the maternal alleles for animal m. The variance of
eijklmn is the residual effect associated to the record the vector of QTL allelic effects (v) is equal to
ijklmn. varv Qr2v where Q is the gametic relationship
For quality traits measured at the laboratory, the matrix and r2v is the allelic variance. Fernando &
model was: Grossman (1989) described the rules to construct the
gametic relationship matrix and its inverse. The
y iom spci dlo pm um eiom 2
computation of the probability of transmission of
where dl is the diluter used for semen preservation gametic effects from parents to offspring was done
in the straw, and the other effects were described with a method similar to Pong-Wong et al. (2001).
above. At a given position, the presence of one QTL can
A multitrait model was applied with the follow- be tested by comparing the maximum likelihood
ing (co)variance structure for all traits: estimated by REML under a polygenic model with
var(u) = A  G, where u = (u1, u2, , un) and ui no QTL fitted (L(H0)) with the maximum likelihood
is the vector of random polygenic effects for trait i, under the one-QTL model (L(H1)). The resulting
A is the additive relationship matrix and G is the likelihood ratio test statistic is (George et al. 2000):
matrix of genetic variances and covariances LH0
between traits; var(p) = I  P, where p = (p1, p2, k 2 ln 5
LH1
, pn) and pi is the vector of random permanent
environment effects for trait i, I is an identity The distribution of this test is not known, but
matrix and P is the matrix of variances and covari- Grignola et al. (1996) showed that this distribution is
ances between traits for the permanent environ- intermediate between the 1 and the 2 d.f. chi-square
mental effects; var(e) = I  R, where e = (e1, e2, distribution for a one-position test.
, en) and ei is the vector of random residual In this study, the QTL significant at p < 0.05 with
effects for trait i and R is a matrix of residual vari- a 2 d.f. chi-square distribution are presented. As this
ances for all traits. It is not possible to distinguish test was performed for all positions at 1 cM intervals
permanent environment effects from residual effects across all positions in the genome, it must be cor-
with a single observation. The permanent environ- rected for multiple testing along the genome and for
ment effect describes an environmental effect which 15 correlated traits (corresponding to approximately
is associated to all the records of an animal while seven independent traits). The number of indepen-
the residual effect is related to an effect affecting a dent tests is comprised between the number of chro-
specific record. In the present study, the environ- mosomes (29) and the number of markers (148)
mental covariance between two records of an ani- multiplied by the number of independent traits (7).
mal for two different traits was modelled through For any significance threshold, the expected number
the permanent environment covariance. The resid- of false positives under the null-hypothesis can be
ual covariances were assumed to be 0 except for estimated as the number of independent tests multi-
semen production traits. Indeed, several ejaculates plied by the corresponding p-value (Benjamini &
were recorded for these traits and the residual co- Hochberg 1995). The false discovery rate (Benjamini
variances modelled additional environmental covari- & Hochberg 1995) is the ratio between the expected
ances between records for different traits measured number of false positives and the number of QTL
on the same ejaculate. detected above the threshold.

2009 Blackwell Verlag GmbH J. Anim. Breed. Genet. 126 (2009) 269277 271
QTL detection for male fertility T. Druet et al.

The heterozygous status of a sire at a tested posi- genetic correlation (0.13) while the genetic correla-
tion was estimated by a t-test which tested the dif- tion between DROP and HEAD or TAIL was negative
ference between paternal and maternal alleles. The ()0.43) and close to 0 (0.06) respectively. As DROP
corresponding standard error was equal to prediction has little influence on ABNO and a negative correla-
error variance of this difference (obtained from the tion with HEAD, it is not surprising that DROP also
inverse of the mixed model equations). had a negative genetic correlation with ABNO
()0.15).
Estimated heritabilities for the remaining quality
Results traits ranged from 0.13 for PROG to 0.59 for RES.
MOT, MSCO, ALH and RES presented high heritabil-
Genetic parameters
ities (between 0.40 and 0.60) while moderate herit-
The mean and standard deviation for all traits are abilities were observed for MOTH, LIV and VAP
presented in Table 1 and the corresponding heritabil- (between 0.20 and 0.30).
ities and correlations are presented in Table 2. For According to the estimated genetic correlations,
clarity, the traits were divided into three groups: two groups of traits may be constituted. The first
production traits, abnormalities percentages and group would contain MOT, MSCO, PROG, MOTH,
other quality measures. Application of an AI-REML LIV and RES which are highly positively correlated
algorithm offers the possibility to obtain standard with each other. For instance, genetic correlations
errors of the estimates. In the present study, stan- between PROG and the other traits of the group ran-
dard errors for the estimated genetic correlations ged from 0.71 to 0.94. The largest genetic correla-
were large, ranging from 0.08 to 0.67. tions estimated within this group were found
Production traits had moderate heritabilities rang- between PROGMOTH (0.94), PROGLIV (0.89),
ing from 0.09 to 0.22. Estimated repeatabilities were MOTMSCO (0.88), MOTHLIV (0.86), PROGRES
0.33, 0.32 and 0.24 for VOL, CONC and NSP, respec- (0.79) and LIVRES (0.78). ALH and VAP would
tively. The VOL and CONC traits were found to be constitute a second group (rg = 0.64) and these two
antagonistic, (rg = )0.55) while NSP showed large traits presented a moderate negative genetic correla-
positive genetic correlations with these two traits. tion with most of the traits of the first group, show-
Traits related to sperm abnormalities also had mod- ing that the speed and the amplitude of the
erate heritabilities, ranging from 0.19 for DROP to movement of individual spermatozoa seem to be
0.35 for HEAD. The total percentage of abnormalities negatively associated to other characteristics.
was strongly correlated with HEAD and TAIL Genetic correlations between production traits and
(rg = 0.71 and 0.78 respectively). This was expected quality traits indicate that VOL shows a positive
as these two categories were the most common (unfavourable) relationship with percentage of abnor-
abnormalities. HEAD and TAIL had a small positive mal spermatozoa and a negative relationship with the

Table 1 Phenotypic means of the 15 analy-


Traits Abbreviation Unit Mean sed dairy cattle semen traits
Sperm volume VOL ml 4.07  1.46
Sperm concentration CONC 109 ml 1.08  0.47
Number of spermatozoa NSP 109 4.38  2.45
Motility MOT % 55.90  7.69
Motility score MSCO 2.64  0.28
Percentage of progressive spermatozoa PROG % 26.38  7.22
Amplitude of linear head displacement ALH lm 6.68  0.94
Motility measured by Hamilton MOTH % 46.56  9.98
Percentage of living sperm after thawing LIV % 55.89  9.59
Average path velocity VAP lm s 112.71  9.73
Percentage of living sperm after osmotic stress RES % 35.93  9.98
Abnormal sperm percentage ABNO % 21.40  6.28
Percentage of spermatozoa with abnormal head HEAD % 12.88  4.40
Percentage of spermatozoa with abnormal tail TAIL % 7.94  4.79
Percentage of spermatozoa with abnormal DROP % 1.65  1.99
cytoplasmic droplet

272 2009 Blackwell Verlag GmbH J. Anim. Breed. Genet. 126 (2009) 269277
T. Druet et al.

Table 2 Heritabilities (on diagonal), genetic correlations (above diagonal) and phenotypic correlations (below diagonal) for the 15 measured semen characteristics traits (standard errors are indi-
cated in parenthesis)

VOL CONC NSP MOT MSCO PROG ALH MOTH LIV VAP RES ABNO HEAD TAIL DROP

VOL 0.22 (0.05) )0.55 (0.18) 0.47 (0.18) )0.20 (0.19) )0.17 (0.19) )0.53 (0.45) 0.08 (0.22) )0.46 (0.25) )0.47 (0.26) )0.00 (0.23) )0.33 (0.20) 0.23 (0.26) )0.12 (0.27) 0.43 (0.35) 0.32 (0.28)
CONC )0.02 (0.02) 0.19 (0.05) 0.46 (0.18) 0.12 (0.20) )0.01 (0.20) 0.35 (0.36) 0.27 (0.22) 0.23 (0.25) 0.29 (0.26) 0.28 (0.23) 0.14 (0.20) )0.34 (0.24) )0.23 (0.24) )0.33 (0.30) )0.09 (0.28)
NSP 0.61 (0.03) 0.71 (0.04) 0.09 (0.04) )0.12 (0.25) )0.24 (0.29) )0.22 (0.67) 0.34 (0.32) )0.28 (0.40) )0.18 (0.38) 0.24 (0.31) )0.23 (0.29) )0.09 (0.38) )0.38 (0.36) 0.14 (0.54) 0.33 (0.43)
MOT )0.03 (0.03) 0.01 (0.03) )0.03 (0.02) 0.43 (0.08) 0.88 (0.06) 0.74 (0.30) )0.21 (0.19) 0.76 (0.12) 0.58 (0.15) )0.27 (0.21) 0.38 (0.14) )0.55 (0.19) )0.56 (0.18) )0.24 (0.24) 0.13 (0.23)
MSCO 0.01 (0.03) 0.02 (0.03) 0.01 (0.03) 0.66 (0.02) 0.50 (0.09) 0.71 (0.27) 0.05 (0.17) 0.75 (0.13) 0.61 (0.15) )0.15 (0.21) 0.44 (0.13) )0.40 (0.18) )0.54 (0.17) )0.02 (0.23) )0.07 (0.21)

2009 Blackwell Verlag GmbH J. Anim. Breed. Genet. 126 (2009) 269277
PROG )0.05 (0.03) 0.06 (0.03) 0.02 (0.03) 0.48 (0.03) 0.40 (0.03) 0.13 (0.13) )0.23 (0.41) 0.94 (0.10) 0.89 (0.17) )0.05 (0.42) 0.79 (0.27) )0.38 (0.38) )0.39 (0.41) )0.15 (0.57) )0.08 (0.44)
ALH 0.03 (0.03) 0.05 (0.03) 0.04 (0.03) 0.02 (0.04) 0.24 (0.04) 0.12 (0.04) 0.52 (0.12) )0.31 (0.25) )0.23 (0.25) 0.64 (0.17) )0.26 (0.18) )0.12 (0.25) )0.20 (0.23) 0.01 (0.31) )0.38 (0.24)
MOTH )0.08 (0.03) 0.03 (0.03) 0.03 (0.03) 0.49 (0.03) 0.40 (0.03) 0.89 (0.01) 0.00 (0.04) 0.27 (0.09) 0.86 (0.08)) )0.31 (0.29) 0.68 (0.12) )0.19 (0.29) )0.29 (0.27) 0.05 (0.38) )0.10 (0.29)
LIV )0.10 (0.03) 0.04 (0.03) )0.03 (0.03) 0.40 (0.03) 0.31 (0.03) 0.70 (0.02) 0.03 (0.04) 0.74 (0.02) 0.21 (0.08) )0.16 (0.27) 0.78 (0.10) )0.23 (0.29) )0.44 (0.28) 0.10 (0.39) 0.05 (0.30)
VAP 0.04 (0.03) 0.06 (0.03) 0.06 (0.02) 0.26 (0.03) 0.34 (0.03) 0.41 (0.03) 0.42 (0.03) 0.15 (0.04) 0.12 (0.03) 0.27 (0.08) 0.18 (0.19) )0.31 (0.24) )0.10 (0.24) )0.35 (0.29) )0.38 (0.25)
RES )0.12 (0.03) 0.01 (0.03) )0.07 (0.03) 0.37 (0.03) 0.31 (0.04) 0.62 (0.02) )0.03 (0.04) 0.67 (0.02) 0.75 (0.02) 0.15 (0.04) 0.59 (0.12) )0.29 (0.21) )0.34 (0.19) )0.07 (0.27) )0.04 (0.24)
ABNO 0.00 (0.03) )0.07 (0.03) )0.05 (0.03) )0.20 (0.03) )0.16 (0.04) )0.24 (0.04) )0.05 (0.04) )0.23 (0.04) )0.20 (0.04) )0.09 (0.04) )0.22 (0.04) 0.25 (0.10) 0.71 (0.18) 0.78 (0.16) )0.15 (0.34)
HEAD 0.02 (0.03) )0.02 (0.03) )0.02 (0.03) 0.17 (0.04) )0.15 (0.04) )0.17 (0.04) 0.01 (0.04) )0.18 (0.04) )0.15 (0.04) )0.02 (0.04) )0.18 (0.04) 0.61 (0.03) 0.35 (0.12) 0.13 (0.39) )0.43 (0.30)
TAIL 0.00 (0.03) )0.06 (0.03) )0.03 (0.03) )0.11 (0.03) )0.05 (0.04) )0.17 (0.04) )0.04 (0.04) )0.14 (0.04) )0.12 (0.04) )0.06 (0.04) )0.15 (0.04) 0.77 (0.02) 0.02 (0.04) 0.19 (0.12) 0.06 (0.37)
DROP 0.01 (0.03) )0.08 (0.03) )0.05 (0.02) )0.07 (0.03) )0.14 (0.03) )0.13 (0.04) )0.16 (0.04) )0.11 (0.03) )0.11 (0.03) )0.18 (0.03) )0.05 (0.04) 0.36 (0.03) 0.12 (0.04) 0.15 (0.04) 0.19 (0.08)

VOL, sperm volume; CONC, sperm concentration; NSP, number of spermatozoa; MOT, motility; MSCO, motility score; PROG, percentage of progressive spermatozoa; ALH, amplitude of linear head
displacement; MOTH, motility measured by Hamilton; LIV, percentage of living sperm after thawing; VAP, average path velocity; RES, percentage of living sperm after osmotic stress; ABNO, abnor-
mal sperm percentage; HEAD, percentage of spermatozoa with abnormal head; TAIL, percentage of spermatozoa with abnormal tail; DROP, percentage of spermatozoa with abnormal cytoplasmic
droplet.
QTL detection for male fertility

273
QTL detection for male fertility T. Druet et al.

remaining group of quality traits consisting of MOT, respectively. The same sire was heterozygous
MSCO, PROG, MOTH, LIV and RES. The genetic (p < 0.05) for these four traits. On chromosome 27,
correlation between VOL and ALH or VAP was close a QTL was detected at the same position for LIV and
to 0. For CONC, the relationship with the other traits MOT and the same two sires had the highest t-tests.
were rather opposite: favourable (negative) with The expected number of false positives under the
percentage of abnormal spermatozoa and positive null-hypothesis for a p-value equal to 0.05 was esti-
(with moderate correlations) with most of the remain- mated using the number of chromosomes (29)
ing quality traits. Genetic relationships between traits multiplied by seven independent traits as an approx-
related to percentage of abnormal spermatozoa and imation of the number of independent tests. The
the remaining quality traits were negative, indicating corresponding false discovery rate was equal to 0.87
that high percentages of abnormalities were associated for a chi-square distribution with 2 d.f. (and 0.25
to sperm with a low motility. assuming a chi-square distribution with 1 d.f.). This
Phenotypic correlations were globally in the same means that with a 2 d.f. chi-square distribution, the
direction as genetic correlations but with generally number of expected false positives is roughly equal
smaller values. to the number of detected QTL whereas 3 out of 11
QTL would be false positives with a 1 d.f. chi-square
distribution.
QTL detection

The detected QTL at significance p < 0.05 (Likelihood


Discussion
Ratio Test (LRT) > 5.99) are presented in Table 3.
QTL were found for seven traits (with two significant The estimated heritabilities for VOL and CONC are
QTL per trait for VOL, LIV, MOT and DROP) for a total lower than those reported previously by Ducrocq &
of 11 QTL. Only three were significant at p < 0.01 Humblot (1995). These published results in the Nor-
(LRT > 9.21). Most QTL were detected for quality mande breed were obtained on mean values from
traits. There were two QTL detected for semen pro- more than 10 records (increasing the heritabilities in
duction and percentage of abnormalities each and consequence). The genetic correlation between VOL
seven QTL for quality measures. Up to three sires were and CONC was similar to the estimation of Ducrocq
estimated heterozygous for one QTL but for most QTL & Humblot (1995). For VOL, Mathevon et al. (1998)
only one sire appeared heterozygous. found a similar value (0.24) while for CONC and
The eleven significant QTL were located on seven NSP, their estimates were larger (0.52 and 0.38).
different chromosomes. On chromosome 23, two However, it should be noted that the authors admit-
QTL affecting RES and LIV were detected at the ted a possible confounding between the permanent
same location and in the same region, LRT values of environmental effect and the additive genetic effect
5.45 and 5.66 were found for CONC and HEAD in their study. Lower estimates of heritabilities were

Table 3 QTL detected with LRT > 5.99 for 15 semen traits and number of heterozygous sires

Location of Number of
LRT maximum Confidence Closest heterozygous
Trait BTA value p-value (in cm) intervala marker sires

LIV 23 9.72 0.0078 14 +848 IOBT528 1


LIV 27 9.35 0.0093 58 4958+ INRA027 2
MOT 27 9.22 0.0100 58 4658+ INRA027 1
DROP 21 7.96 0.0187 41 3849 ILSTS103 1
MOT 11 7.91 0.0192 20 1158 INRA177 1
VOL 22 7.39 0.0248 1 028 CSSM026 1
VOL 15 7.25 0.0266 22 +1445 SPS115 3
DROP 20 7.05 0.0295 62 4078+ BM4107 1
VAP 25 6.61 0.0367 56 2463+ INRA222 1
MOTH 7 6.43 0.0402 34 +2571 RM6 1
RES 23 6.10 0.0474 8 +835 IOBT528 2

VOL, sperm volume; MOT, motility; MOTH, motility measured by Hamilton; LIV, percentage of living sperm after thawing; VAP, average path veloc-
ity; RES, percentage of living sperm after osmotic stress; DROP, percentage of spermatozoa with abnormal cytoplasmic droplet.
a
A + indicates that the confidence interval limit corresponds to the first or the last marker of the map.

274 2009 Blackwell Verlag GmbH J. Anim. Breed. Genet. 126 (2009) 269277
T. Druet et al. QTL detection for male fertility

also found in previous studies: 0.13 for VOL (Lang cally linked. Moderate favourable relationships were
et al. 1988), 0.13 for CONC (Knights et al. 1984) and found between concentration and all the variables
0.05 for NSP (Taylor et al. 1985). In sheep, David describing the movement of spermatozoa and nega-
et al. (2006) observed heritabilities of 0.22, 0.24 and tive correlations were found between all those vari-
0.18 for VOL, CONC and NSP, respectively and the ables and the total percentage of abnormalities. The
reported genetic correlations were in agreement with negative impact of percentage of abnormalities on
our results: )0.24 between VOL and CONC, 0.84 the remaining quality traits is understandable as
between VOL and NSP and 0.30 between CONC and these abnormalities will potentially reduce the sur-
NSP. In addition, our estimate for heritability for vival of spermatozoa and may influence mobility.
total abnormalities was quite high (0.25) but very The reported favourable association between
close to the estimate (0.19) reported for the Nor- sperm concentration and some quality traits was
mande population (Ducrocq & Humblot 1995). The confirmed and extended for more quality traits in
heritability of MSCO (0.50) was larger than herit- this study. Selection on concentration should indi-
abilities reported in previous studies. Estimates for rectly improve these traits but the correlations are
heritability of motility score were 0.23 and 0.24 in only moderate. In addition, selection based on sperm
the Normande breed (Ducrocq & Humblot 1995) concentration would reduce sperm volume.
while Mathevon et al. (1998) reported a heritability As selection on concentration would achieve only
of 0.31 in Holstein cattle. As a comparison, David moderate gains in other variables, the use of QTL
et al. (2006) reported a 0.18 heritability for motility affecting quality traits should be beneficial. Such a
score in sheep. selection strategy including QTL would add informa-
The heritability estimates for semen production tion and improve the precision of the selection pro-
and quality traits confirm the results from earlier cess making it more efficient when compared to a
studies and together with the favourable relation- selection based only on concentration measures.
ships between concentration and quality traits also Marker assisted selection is usually interesting for
reported by Ducrocq & Humblot (1995) indicate that traits costly and difficult to measure. Only a few QTL
selection for such traits can be efficient. These semen were detected and most of them were related to
production traits are relatively easy to measure and sperm quality (to the first group of quality traits).
several records are usually available for each bull as The small number of QTL detected is certainly
most of the semen production centres routinely because of the low power of the design.
record these performances. Therefore, selection for According to Grignola et al. (1996), the distribu-
these traits should not be difficult to implement. On tion of the LRT statistic is intermediate between the
the contrary, most of the quality traits are costly to 1 and the 2 d.f. chi-square distributions. Using the 2
measure and breeding values are based on a small d.f. chi-square distribution, the number of expected
number of records for each young bull. The use of false positives is close to the number of QTL detected
records from correlated traits or of QTL affecting whereas less false positives are expected with a 1 d.f.
those traits would allow selection without measure- chi-square distribution. However, more false posi-
ment of quality traits. However, the use of informa- tives might be expected if the number of indepen-
tion from correlated traits requires precise estimation dent tests is underestimated (for instance, we
of the genetic relationships between the different assumed only one independent test per chromo-
traits. Very little information was available from the some). These observations stress the importance of
literature on the genetic parameters for most of the improving the power of the design. Unfortunately, a
semen quality traits measured in this study. larger design with more sires per families would be
For the semen production traits, estimates of very difficult to implement due to the cost associated
genetic correlations published previously showed an to the phenotyping for quality traits. Moreover, it is
antagonistic correlation between semen volume and also more difficult to determine which sires are het-
concentration. In consequence, these two traits show erozygous when few progeny have both phenotype
the opposite relationship with all the remaining and genotype information. With a lower significance
traits. Most sperm abnormalities were due to head or threshold, more QTL would have been detected, but
tail defects and both traits had a favourable relation- many of them would be false positives (as indicated
ship with concentration. Motility traits and percent- by the expected number of false positives). For many
ages of living spermatozoa (MOT, MOTH, MSCO, positions, the maximum LRT was between 4 and 6,
RES, LIV, PROG) were highly correlated with each indicating the possibility for more QTL to exist. Such
other which makes sense as these traits are biologi- LRT values were often observed for several traits at a

2009 Blackwell Verlag GmbH J. Anim. Breed. Genet. 126 (2009) 269277 275
QTL detection for male fertility T. Druet et al.

common position suggesting some pleiotropic QTL. had a negative relationship and selection of concen-
This is the case for instance for those found on chro- tration should indirectly improve most of the qual-
mosomes 7, 23 and 27. They affected several traits at ity traits. In addition, 11 QTL affecting semen traits
the same position and the same sires appeared het- were detected. Their use in marker assisted selec-
erozygous. Multivariate techniques (e.g. Gilbert & tion may improve selection efficiency but a fine-
Leroy 2003) would probably improve the power of mapping of these QTL would be required before
the design for QTL detection and could fit well with using them. It is likely that more QTL were segre-
the present design as we measured 15 correlated gating in the design but they were not detected
traits. As for many positions, LRT values are just due to the low power of the design. To improve
below the significance threshold, the simultaneous the power of the analysis, multitrait techniques and
use of a group of traits affected by the same QTL denser marker maps should be used in future
should confirm this QTL or discard the false posi- studies.
tives. In addition, this technique would allow to use
for selection some important QTL affecting simulta-
Acknowledgements
neously several semen quality traits.
The use of a denser marker map would certainly This work was performed in the framework of the
be more efficient to increase the power of the design programme GENANIMAL (N03 P 316-317). The
by providing the possibility to use both linkage and authors gratefully acknowledge the ministry of
linkage disequilibrium information. For instance, research FRT-ANR and APIS-GENES for financial sup-
information of maternal haplotypes would then also port. The authors also thank GENOE, MIDATEST,
be used. UNECO and URCEO for providing the semen samples
The size of the design affected also the precision of and semen production records. We acknowledge the
the estimated genetic parameters. Globally, variation anonymous reviewer for the useful comments and
in results between this study and prior studies were James Koltes for English correction of the manuscript.
moderate. The relationships between traits were sim-
ilar and the few differences observed may be due to
References
differences in the studied populations, different col-
lection methods and accuracy of estimates. Indeed, Benjamini Y., Hochberg Y. (1995) Controlling the false
confidence intervals of estimates were rather large. discovery rate: a practical and powerful approach to
Finally, the use of a variance component (VC) multiple testing. J. R. Stat. Soc. B., 57, 289300.
approach in this study presented several advantages David I., Bodin L., Lagriffoul G., Manfredi E., Robert-Gra-
and disadvantages of this last method. The method nie C. (2006) Genetic parameters of ram semen traits
does not allow data permutation to obtain signifi- using an animal model accounting for serial correla-
cance thresholds and it is difficult to determine tions. In: Proceedings of the eighth World Congress on
which results are significant. This problem was par- Genetics Applied to Livestock Production, Belo Hori-
zonte , Brazil, 1318 August 2006.
ticularly important in this study as the power of the
Ducrocq V., Humblot P. (1995) Genetic characteristics
design was relatively low and many LRT values were
and evolution of semen production of young Nor-
between the significance threshold values corre-
mande bulls. Livest. Prod. Sci., 41, 110.
sponding to chi-square distributions with 1 and 2
Fernando R.L., Grossman M. (1989) Marker-assisted
d.f. Nevertheless, VC methods allow the use of com-
selection using best linear unbiased prediction. Genet.
plex pedigree information and account for all rela- Sel. Evol., 21, 467477.
tionships between sire families. Pedigree information George A.W., Visscher P.M., Haley C.S. (2000) Mapping
slightly enhanced the ability to detect QTL effects. quantitative trait loci in complex pedigrees: a two-step
Moreover, the VC approach allowed simultaneous variance component approach. Genetics, 156, 2081
inclusion of several fixed and random effects impor- 2092.
tant to the analysis of this data. Gilbert H., Leroy P. (2003) Comparison of three multitrait
methods for QTL detection. Genet. Sel. Evol., 33, 281304.
Grignola F.E., Hoeschele I., Tier B. (1996) Mapping
Conclusions
quantitative trait loci in outcross populations via resid-
The design implemented to detect QTL affecting ual maximum likelihood. I. Methodology. Genet. Sel.
semen production and quality traits was first used Evol., 28, 479490.
to estimate genetic parameters among the 15 Guillaume F., Fritz S., Boichard D., Druet T. (2008) Esti-
recorded traits. Semen volume and concentration mation by simulation of the efficiency of the French

276 2009 Blackwell Verlag GmbH J. Anim. Breed. Genet. 126 (2009) 269277
T. Druet et al. QTL detection for male fertility

marker-assisted selection program in dairy cattle. Genet. Detection de QTL infleuncant laptitude a` la transfor-
Sel. Evol., 40, 91102. mation fromage`re des laits dans un croisement Hol-
Humblot P., Decoux G., Dhorne T. (1991) Effects of the stein-Normande. In: Proceedings of 14eme Rencontres
sire and district of AI on cow fertility. Reprod. Domest. Recherches Ruminants. 56 December 2007, Paris,
Anim., 26, 225234. France.
Ihara N., Takasuga A., Mizoshita K., Takeda H., Sugimoto Lucy M.C. (2001) Reproductive loss in high-reproducing
M., Mizoguchi Y., Hirano T., Itoh T., Watanabe T., dairy cattle: where will it end? J. Dairy Sci., 84,
Reed K.M., Snelling W.M., Kappes S.M., Beattie C.W., 12771293.
Bennett G.L., Sugimoto Y. (2004) A comprehensive Mathevon M., Buhr M.M., Dekkers J.C.M. (1998) Envi-
genetic map of the cattle genome based on 3802 mi- ronmental, management, and genetic factors affecting
crosatellites. Genome Res., 14, 19871998. semen production in Holstein bulls. J. Dairy Sci., 81,
Jamrozik J., Fatehi J., Kistemaker G.J., Schaeffer L.R. 33213330.
(2005) Estimates of genetic parameters for Canadian Misztal I., Tsuruta T., Strabel T., Auvray B., Druet T., Lee
Holstein female reproduction traits. J. Dairy Sci., 88, D.H. 2002: BLUPF90 and related programs (BGF90).
21992208. In: Proceedings of the 7th World Congress on Genetics
Knights S.A., Baker R.L., Gianola D., Gibb J.B. (1984) Applied to Livestock Production, 1923 August 2002,
Estimates of heritabilities and genetic and phenotypic Montepellier, France.
correlations among growth and reproductive traits in Pong-Wong R., George A.W., Woolliams J.A., Haley C.S.
yearling Angus bulls. J. Anim. Sci., 58, 887893. (2001) A simple and rapid method for calculating iden-
Lang H., Preisinger R., Kalm E. (1988) Analysis of data tity-by-descent matrices using multiple markers. Genet.
on semen quality in Angeln cattle obtained from a Sel. Evol., 33, 453471.
breeding program. Zuchthygiene (Berlin), 23, 1018. Taylor J.F., Bean B., Marshall C.E., Sullivan J.J. (1985)
Larroque H., Gallard Y., Delacroix-Bucher A., Bach C., Genetic and environmental components of semen pro-
Ogier J.C., Mahe M.F., Miranda G., Gautier M., Grohs duction traits of artificial insemination Holstein bulls. J.
C., Legros H., Druet T., Colleau J.J., Boichard D. 2007: Dairy Sci., 68, 27032722.

2009 Blackwell Verlag GmbH J. Anim. Breed. Genet. 126 (2009) 269277 277