Introduction
An experiment comparing results of crossbreeding purebred Holstein cows with Fleckvieh sires vs.
breeding with Holstein sires was performed on 8 Dutch dairy farms. The experiment was carried out by
WUR Livestock Research and funded by Genetic Austria. On the participating farms between august 2005
and august 2007 Holsteins cows in the herds were inseminated with semen of either Fleckvieh or
Holstein bulls a selectively. It was aimed to obtain on average 25 heifer calves for both groups on each
farm, these animals were the animals of which the performance was monitored. For all cows inseminated
the age and season corrected productions before insemination did not differ (p= 0.893) between the
animals inseminated with Holstein or Fleckvieh. This indicates that the matings with either Holstein or
Fleckvieh were indeed performed unselectively with regard to the production of the cows. The purebred
group was sired by 150 different sires, the crossbred group by 19 sires.
The heifer calves born from these matings, the experimental animals, were monitored until September
2012 for milk production, fertility and culling. The farmers followed their own policies regarding breeding
and culling of these animals. Insemination dates (including the identification of the sire) and calving
dates were recorded. For each calving the farmers recorded estimates of the birth weight (in classes of 5
kg ranging from 20 to 70 kg) and estimates of the ease of calving (easy, normal or difficult) in their
management programs. It was recorded whether the calf was born alive and survived the first 24 hours
after birth, was born alive but died within 24 hours or was stillborn. Moreover, sales prices of calves sold
in the mating phase and for experimental animals when culled were recorded. For each farm production
figures were obtained from regular milk recording. Pedigrees of the sires of the experimental animals
(sire and maternal grandsire) were obtained from the national database. The results of the trial are
evaluated for the mating period preceding the birth of the animals monitored, the rearing period and
productive period of these animals separately.
Mating period
The evaluation of the mating period was based on birth weight of the calves, sales prices, gestation
length and calving ease. Calving ease was scored on a 3 point scale: 1 = easy, 2 = normal and 3 =
difficult. Twin calvings were excluded, after the mating period 953 single pregnancies were obtained that
were included in the trial. It was attempted to analyse stillbirth as well, but this did not result in useful
estimates. Of the 953 single calvings (536 bull calves and 417 heifer calves) 40 calves (16 Holsteins and
24 crossbreds, 20 from each sex) were born dead or died within 24 hours after birth. It is assumed that
the survival rate was not different between herds or breed groups. The 436 Holstein calves were
offspring of 149 bulls, the 517 Holstein*crossbred calves were offspring of 18 bulls. The majority of the
bull calves was sold and the majority of the heifer calves was kept. These animals are the animals that
were used to compare the performance of purebred and crossbred dairy cows.
The data for the rearing period were analysed with ASReml with herd, breed group (Holstein or
crossbred) and their interaction, sex of calf and its interaction with breed group and age class of the cow
(1st, 2nd or higher parity) as fixed effects and sire and a spline adjusting for year and month of calving as
random effects. A relationship matrix based on sire and maternal grandsire was included in the analysis.
Predicted means and p-values of the fixed effects are presented in Table 1 and discussed below.
Table 1 Predicted means and p-values of fixed effects for the mating period
Trait
Birth weight
calf (kg)
Sale price calf
(€)
Gestation
length (days)
Calving ease
mean
41.9
±0.4
147 ± 6
283 ±
0.5
1.8 ±
0.1
herd
fixed effect
Herd*breed
Sex of
group
calf
0.002
<0.001
<0.001
Breed
group
<0.001
Breed
group*sex
0.145
Age
class
<0.001
<0.001
<0.001
0. 246
n.a.
n.a.
0.225
0.227
0.007
0.962
0.060
0.035
0.235
<0.001
0.023
0.906
<0.001
0.040
<0.001
Birth weight
Both herd and breed group and their interaction were significant. The significance of the interaction
points out that the difference in birth weight between the Holstein calves and the crossbred calves varied
between herds. This is illustrated in Figure 1.
Figure 1 herd*breed group estimates for birth weight in insemination period
The differences between herds can at least partly be due to the fact that on most farms the weights are
not real measures but farmers estimates. Overall the crossbred calves were significantly heavier (42.8 ±
0.4 kg) than the Holstein calves (41.0 ±0.6 kg). On average the reported weights of the bull calves were
3 kg higher than those of the heifer calves (43.4 ± 0.4 vs. 40.4 ± 0.4 kg respectively). There were also
significant differences in birth weight between different age classes of cows. Calves of heifers weighed
significantly less (40.6 ± 0.6 kg) than those of 2nd calvers (42.1 ± 0.4 kg) and higher parity cows 43.1 ±
0.4 kg).
Sale price of bull calves
Because only incidentally heifer calves were sold the sale price was analysed for the bull calves only. The
analysis revealed that only herd and breed group were significant factors. Predicted herd averages varied
between 122 ± 7 and 182 ± 7 euros per bull calf. The predicted average sale price was €160 ± 7 for the
crossbred bull calves and €135 ± 6 for the purebred Holstein bull calves, so an average difference of
€25. This extra profit has to be weighed against the price of the straws. Herd effects indicate that the
farmer has substantial influence on the sale prices, but the absence of a significant interaction indicates
that the difference between Holstein and crossbred is not affected.
Gestation length
Both breed group and interaction of breed group and sex of the calf were significant factors for gestation
length, this indicates that the difference in gestation length between bull and heifer calves was different
for Holstein (281.9 ± 0.6 and 282.2 ± 0.6 days respectively) and crossbred calves (285.2 ± 0.8 and
283.6 ± 0.8 days respectively). The main finding is that the crossbred calves had a longer gestation
(284.4 ± 0.7 d) than the Holstein calves (282.1 ± 0.5 d).
Calving ease
Higher figures for calving ease indicate more difficult calvings and less easy calvings. Calving ease scores
differed significantly between herds, this could be affected by the subjectivity of the trait. The scores
were also affected by breed group, sex of the calf and the interaction of these factors. The interaction is
due to a larger difference in calving ease scores between bull and heifer calves for the crossbred calves
(2.01 ± 0.07 and 1.80 ± 0.07 respectively) than for the Holstein calves (1.81 ± 0.06 and 1.74 ± 0.06
respectively). In general for the crossbreds there were less easy calvings and more calvings classified as
difficult. Moreover calving ease scores were higher for 1st parities (2.08 ± 0.08) than for 2nd or higher
parities (1.72 ± 0.06 and 1.72 ± 0.05 respectively).
To summarise, crossbreeding on average resulted in heavier calves (~2 kg), longer gestation (~2 days),
somewhat less easy calvings and higher sale prices (25€) of bull calves compared to breeding with
Holstein bulls.
Rearing period
Of the 386 calves born 4 were sold as young calf (3 Holsteins and 1 crossbred) and 4 were sold as
pregnant heifers (2 Holsteins and 2 crossbreds) before their 1st calving. Data from these animals was
ignored for the analysis of the rearing period, while their figures were incomplete. Of the remaining 163
Holsteins and 215 crossbreds 18 (11%) and 19 (9%) died or were culled before a first calving. The main
reason for culling was infertility for both groups. Because of the low numbers culling during the rearing
period was not analysed statistically, but the figures suggest there was no difference between the two
groups. For the 341 animals that had a first calving the following figures for the rearing period were
analysed: number of inseminations, age at 1st calving, non return 56 days after first insemination and
age at 1st calving. These traits were analysed with ASReml. Fixed effects in the model were: herd.year,
breed group of bull used for insemination (HF, FLV or other) and breed group of animal (HF or FLV).
Random effects were sire and a spline for month of birth. Again a relationship matrix based on sire and
maternal grandsire was included in the analysis. The predicted means and p-values for the fixed model
terms are in Table 2.
Table 2 Predicted means and P-values of fixed effects for analysis of the rearing period
Trait
number of inseminations
age at 1st insemination
non return 56 days after 1st
insemination
age at 1st calving
herd.year
fixed effect
bull breed group
1.59 ±0.08
472 ± 4
0.74 ± 0.04
<0.001
<0.001
0.061
0.011
0.903
0.191
animal breed
group
0.359
0.065
0.629
782 ± 7
<0.001
0.124
0.926
mean
Interactions of herd and breed group were tested but were not significant for these traits. There was a
marked difference in the selection of bulls between the two groups: 13% of the crossbreds were
inseminated with a Fleckvieh sire, 36% with a Holstein sire and 51% with a third breed bull, the figures
were 8%, 81% and 11% for the Holsteins respectively. Animals inseminated with a bull of another breed
had less inseminations (1.26 ± 0.13) than those inseminated with a Holstein (1.76 ± 0.11) or Fleckvieh
(1.56 ± 0.20) bull. The analysis reveals there were no significant differences in fertility or age at 1st
calving between the two breed groups, despite that differences in age at 1st insemination (481 ± 6 days
for Holsteins and 469 ± 5 days for crossbreds) were nearly significant between the two experimental
groups.
Productive period
Milk production
The 196 crossbreds that calved at least once were daughters of 15 sires, the 145 Holsteins were
daughters of 73 sires. Of these, 14 and 71 have Dutch breeding values for production traits respectively.
Weighted averages of the breeding values for a number of traits for the two groups of sires are in
Table 3.
Table 3 Weighted average Dutch breeding values for sires of the experimental groups
Holstein
mean
total merit index
Fleckvieh
sd
min
max
mean
sd
min
max
55
66
-125
246
-55
53
-138
15
100.7
4.1
87
110
104.9
2.5
100
111
fertility
97.8
4.2
86
110
108.1
2.7
104
114
milk yield
348
618
-1248
2230
-943
252
-1502
-265
udder health
fat yield
2
22
-52
58
-39
16
-65
-8
13
17
-33
57
-23
6
-40
-11
98.7
4.1
88
112
110.1
3.1
105
116
100.9
4.1
87
111
105.1
3.0
98
113
meat index
98.6
4.4
86
111
117.5
3.7
108
121
body weight
101.5
3.5
95
108
111.9
3.5
105
116
protein yield
calving interval
scc
From Table 3 it can be concluded that the Fleckvieh sires had higher breeding values for udder health,
fertility and meat production, but also lower milk production BV’s and total merit indexes. Moreover, the
variation in milk production BV’s was bigger for the Holstein sires, probably because part of them were
unproven at the time of insemination or because selection criteria for the Holstein bulls were more
diverse between the farmers.
The following production figures calculated from test day yields were analysed: 305 d milk, fat and
protein yield, fat and protein percentages and lactation value (which is the 305-day production corrected
for age and season at calving expressed as a percentage of herd average). Figures from all lactations
longer than 100 days (also including running lactations) with test day information were included in the
analysis. This dataset comprised of 299 1st lactations, 244 2nd lactations, 175 3rd lactations and 59 4th
lactations. Data were analysed with ASReml. Fixed effects in the model were: hys of calving(herd-yearseason), parity, breed group of animal (HF or FLV) and interaction of herd and breed group. 8
year/season classes were defined: 1 = before 1/1/2009, 2= 1st half of 2009, .., 8 = 1st half of 2012.
Random effects were cow (to account for repeated records of the same cow) and sire of cow. Again a
relationship matrix based on sire and maternal grandsire was included in the analysis. The predicted
means and p-values for the fixed model terms are in Table 4.
Table 4 Predicted means and P-values of fixed effects for 305 day production figures
Trait
Milk yield (kg)
Fat percentage
Protein percentage
Fat yield (kg)
Protein yield (kg)
Fat + protein yield
(kg)
Lactation value
mean
8351 ±121
4.45 ± 0.05
3.55 ± 0.02
371 ± 4
296 ± 4
666 ± 7
98 ± 1
Hys
fixed effect
Parity
Breed group
<0.001
<0.001
<0.001
<0.001
<0.001
<0.001
<0.001
0.715
0.036
<0.001
<0.001
<0.001
0.116
0.525
0.412
0.277
0.085
0.127
Herd*breed
group
0.047
0.428
0.615
0.073
0.060
0.052
0.011
0.202
0.115
0.075
Table 4 shows that there were no significant differences between the two breed groups with respect to
milk production. The significance of the interaction of herd and breed indicates that differences between
Holsteins and crossbreds differed between herds, this is illustrated in Figure 2.
Figure 2 herd*breed group estimates for 305 day milk yield
This figure also indicates the range in milk yield on the 8 farms. Differences between hys classes were
significant for all traits but not further investigated, as were differences between parities that were
significant for most production traits. Analysis per parity revealed similar results as the overall analysis.
Predicted means of production traits for both breed groups were similar: 8292 ± 147 kg of milk with 370
± 5 kg of fat and 294 ± 4 kg of protein for the crossbreds and 8410 ±168 kg of milk with 371 ± 6 kg of
fat and 297 ± 5 kg of protein for the Holsteins. It can thus be concluded that production figures of
Holsteins and Holstein * Fleckvieh crossbreds were not significantly different for lactations 1 -3, despite
the differences in breeding values of the sires. A difference of 647 kg of milk would be expected between
the two groups as a result of the sire breeding values, which is much more than the realised production
difference. This is probably the result of heterosis.
Culling
From the 341 animals with at least one calving 171 were culled at the end of the data collection period,
overall culling rates were comparable between the breed groups (51% for Holstein and 49% for
crossbreds). For these animals the following traits were analysed: last standardised production value
(“lactatiewaarde”), number of inseminations since last calving, days in lactation, age at culling, culling
price and log(scc) for the last test day. Data were analysed with ASReml. Fixed effects in the model
were: season of culling (1st, 2nd, 3rd or 4th 3-month period of a calendar year), herd, breed group of
animal (HF or FLV) and interaction of herd and breed group. Random effect was sire of cow. Again a
relationship matrix based on sire and maternal grandsire was included in the analysis. The predicted
means and p-values for the fixed model terms are in Table 5.
Table 5 Predicted means and P-values of fixed effects for culling variables
Trait
Last standardised
production
Number of
inseminations
Days in lactation
Age at culling
Culling price
Log(scc)
fixed effect
Herd
Breed group
mean
season
89 ± 2
0.274
0.333
0.008
Herd*breed
group
0.096
1.46 ± 0.20
0.340
0.002
0.974
0.447
206 ± 17
1322 ± 39
571 ± 41
4.61 ± 0.12
0.294
0.540
0.442
0.066
0.014
0.001
<0.001
0.004
0.306
0.764
0.003
0.945
0.343
0.114
0.456
0.307
For most of the culling traits the differences between herds were significant, but there were no significant
interactions between herd and breed group. Last standardised productions differed significantly between
breed groups, with higher values for HF (93 ±3) than for crossbreds (85 ±3). This suggests that for the
Holsteins low production was less often a reason for culling than for crossbreds. Breed groups also had a
different sale price, with higher returns for crossbreds (694 ± 58 euro’s) than for Holsteins (459 ± 48
euros).
Calving ease, birth weight and stillbirth
Until the end of the data collection period the animals had 878 calvings (335 1st, 255 2nd, 197 3rd, 81 4th
and 10 5th parities) for which calving ease (1 = easy, 2 = normal, 3 = difficult; scored by the farmer),
birth weight and stillbirth could be analysed. Data were analysed with ASReml. Fixed effects in the model
were: herd, breed group of animal (HF or FLV), interaction of herd and breed group, parity (parities 4
and 5 were grouped together), breed group of sire of cow and sex of calf. Random effects were cow (to
account for repeated records of the same cow) and sire of cow. A relationship matrix based on sire and
maternal grandsire of the sires was included in the analysis. The predicted means and p-values for the
fixed model terms are in Table 6.
Table 6 Predicted means and P-values of fixed effects for calving traits
Trait
Calving ease
Birth weight
(kg)
Stillbirth (%)
mean
1.60 ±
0.02
41.6 ±
0.2
4.2
±1.9
Herd
fixed effect
Herd*breed
Parity
group
<0.001
<0.001
<0.001
Breed
group
0.016
Breed of
sire
0.768
Sex of
calf
<0.001
<0.001
0.002
0.008
0.001
0.676
<0.001
0.789
0.125
0.613
<0.001
0.445
0.367
For calving ease there were significant differences between breed groups, but also a significant
interaction between herd and breed group, indicating that the difference was not the same for all herds.
Higher scores indicate more difficult calvings and less easy calvings. Overall the score was slightly higher
for Holstein (1.64 ± 0.04) than for crossbreds (1.56 ± 0.03), thus the crossbreds did have more easy
calvings. Effects of parity and sex again pointed out that 1st calvings were less easy than those for higher
parities, and heifer calves were born easier than bull calves. The calves of the crossbred animals on
average were heavier (42.4 ± 0.2 kg) than those of the Holsteins (40.8 ±0.3 kg), but the significance of
the interaction between herd and breed group points out that the difference varied between herds. This
is similar to the mating period. Effects of parity and sex also were similar to those for the mating period.
Stillbirth was not significantly different between breed groups, despite that the predicted mean was
considerably lower for the crossbreds (1.8%) than for the Holsteins (6.7%). The results for breed effects
on this trait were similar when analysed per parity. Probably the number of records is insufficient to
detect differences for a binary trait. Differences between 1st and higher parities were significant (11.3,
1.8, 2.0 and 1.9% respectively for parities 1 - 4).
Average weighed somatic cell count
From the milk recording data average weighed cell counts (avgscc) were calculated for each lactation
that was also included in the analysis of production traits. The logarithms of these data (log(avgscc))
were analysed with ASReml. Fixed effects in the model were: hys (herd-year-season), parity, breed
group of animal (HF or FLV) and interaction of herd and breed group. 8 year/season classes were
defined: 1 = before 1/1/2009, 2= 1st half of 2009, .., 8 = 1st half of 2012. Random effects were cow (to
account for repeated records of the same cow in the overall analysis) and sire of cow. Again a
relationship matrix based on sire and maternal grandsire was included in the analysis. The predicted
means and p-values for the fixed model terms are in
Table 7.
Table 7 Predicted means and P-values of fixed effects for log(avgscc)
Trait
Mean
Hys*
Cccel overall
4.52 ± 0.07
<0.001
Ggcel parity 1
4.30 ± 0.08
<0.001
Ggcel parity 2
4.43 ± 0.07
<0.001
Ggcel parity 3
4.52 ± 0.12
0.061
Ggcel parity 4
n.e.**
0.067
* herd for analysis per parity! ** not estimable
fixed effect
Parity
Breed group
0.030
0.036
0.295
0.046
0.053
0.147
Herd*breed
group
0.054
0.516
0.090
0.031
0.906
Interaction of herd*breed group tended to significance for overall log(avgscc), thus differences between
breed groups tended to differ between herds. For 5 of the farms the averages for Holsteins were higher,
on two farms the averages were not different and on one herd the averages for the crossbreds were
higher. On average, log(avgscc) increased with parity, and crossbreds had lower log(avgscc) (4.37
±0.08) than Holsteins (4.66 ±0.10). When estimated per parity with reduced models differences
between HF and crossbred animals were not significant in 1st parity but were significant for 2nd and
nearly significant for 3rd parities. This despite there was no difference in log(scc) at culling, which
suggests there is no difference in culling with regard to scc. From the difference in breeding values of
the sires of the two groups of around 4 points a difference was to be expected of ±10.000 cells/ml, but
the realised difference was much bigger (79.000 for the crossbreds and 106.000 for the Holsteins.
Therefore it is concluded that there was considerable heterosis in the crossbred group.
Fertility after 1st calving
Finally the fertility figures number of inseminations, interval calving – 1st insemination, non return 56
days after first insemination and calving interval for all 2nd and higher parity calvings were calculated.
This dataset had records for 260 2nd calvings, 202 3rd calvings and 97 higher parity calvings. These
figures were analysed with ASReml. Fixed effects in the model were: herd, parity, breed group of animal
(HF or FLV) and breed group of sire. Random effects were sire of cow and a spline adjusting for year and
month of calving as random effects. Again a relationship matrix based on sire and maternal grandsire
was included in the analysis. The predicted means and p-values for the fixed model terms are in Table 8.
Table 8 Predicted means and P-values of fixed effects for fertility after 1st calving
Trait
Nins
Int1
Nr56
Calving interval
Mean
1.74 ± 0.12
83 ± 2
0.66 ±0.03
396 ± 4
herd.year
0.034
<0.001
<0.001
<0.001
fixed effect
parity
Breed group
<0.001
0.109
<0.001
0.006
0.581
0.089
<0.001
0.027
Breed of sire
0.034
0.382
0.373
0.261
Interactions of herd and breed group were tested but were not significant. Parity effects were highly
significant except for non return, and mainly pointed out that 4th and 5th parities had shorter calving
intervals than 2nd and 3rd calvings. Breed of sire was only significant for number of inseminations,
because the number of inseminations with Holstein was significantly lower (1.48 ± 0.18) than that for
Fleckvieh (1.92 ± 0.19) or other breeds (2.00 ± 0.13). Breed group of the cow was not significant for
number of inseminations and non return percentage, but the predicted values suggested better fertility
for the crossbreds (1.55 ± 0.19 inseminations and 67 ± 3% non return) than for the Holsteins (2.05 ±
0.13 inseminations and 60 ± 4% non return). Intervals between calving and 1st insemination and calving
intervals were significantly longer for the Holsteins (89 ± 3 and 412 ± 5 days respectively) than for the
crossbreds (77 ± 3 and 380 ± 5 days). So these figures indicate that fertility after 1st calving was better
for the crossbred animals, and that this was not dependent on herds. Similar to production and scc, the
difference between the two groups in calving intervals (32 days) was bigger than the 17 days that would
be expected from the difference of the breeding values of their sires. So also for fertility the crossbreds
presumably profit from a considerable heterosis.