Microsoft word - paper24.doc

The use of milk progesterone measurements for genetic
improvement of fertility traits in dairy cattle
R F Veerkamp1, J K Oldenbroek1 and T van der Lende2
1Department of Animal Breeding and Genetics, ID-DLO P.O. Box 65, 8200 AB Lelystad, The Netherlands 2Animal Breeding and Genetics Group, WIAS, Wageningen Agricultural University, P.O. Box 338, 6700 AH Wageningen, The Netherlands.
Abstract
Fertility measures based on calving and insemination dates are heavily affected by a farmers decision to cull
or inseminate a cow. This might partially explain the low heritability for these fertility traits and the
unfavourable association between yield and fertility. Progesterone measurements might provide an alternative.
Heritability estimates ranged from 0.14 to 0.20 (s.e. = 0.08) for interval between calving and first luteal
activity (CLA). However, measuring this trait in a progeny testing scheme is not feasible. Therefore, mean
progesterone profiles (transformed to a binomial scale) were investigated for groups of 50 animals differing in
CLA. Using splines clearly different curves were observed, even when only monthly progesterone
measurements were available for each cow. Days between calving till 50% of the daughters is cycling
(progesterone > 3 ng/ml) might provide a good trait for sire selection. Phenotypic association between CLA
and milk yield were unfavourable. If milk yield increased with 1 kg per day over the first 100 days of
lactation, CLA is expected to increase with 1.4 days. Similarly, the regression coefficient of CLA on energy
balance was -0.33 (s.e. = 0.11). Hence, at cow level progesterone measurements provide a heritable trait which
is useful for selection and for investigating the relationship between fertility and production traits. Indications
from this study are that progesterone measures during monthly milk recording might be useful to establish a
mean progesterone profile for a group of animals. Whether this measure of female fertility is heritable and
useful for sire selection in a progeny testing scheme requires a larger data set for further investigation.
1. Introduction
and many other factors. Error variances might be relatively large therefore, rather than that genetic Low fertility is of economic importance for dairy variances for fertility traits are small. For example, enterprises, because it results in higher levels of compared with a genetic coefficient of variation of involuntary replacement, slippage in calving pattern, 9.5% for milk yield, this parameter for calving veterinary invention, hormonal treatment and interval, days to first service and conception to first reduced annual milk production (Esslemont and service are 1.8%, 4.4% and 10.3%, respectively Peeler, 1993). Even when the definition of optimal (Price et al., 1997a). Further indications that errors fertility performance might change (for example the due to management decisions might affect optimal length of the calving interval is still disputed heritability estimates for fertility traits is that for the for the highest productive cows) genetic interval to first heat a relatively high heritability of improvement of fertility remains important: it gives 0.15 has been reported. Furthermore, heritability the farmer control over its fertility management (i.e. estimates coming from records collected on an which heat to inseminate, which month of the year experimental farm, under controlled fertility does he wants most milk, what is the optimal calving management, are also relatively high, i.e. 0.08 for interval). Hence, genetic improvement should give a calving interval, 0.13 for days open and days to first the farmer the option to inseminate cows at the service (Pryce et al. 1997c). Hence, there appears to economically most optimal time period, whatever the be sufficient genetic variation for fertility traits, but time period is. However, genetic selection for recording practises include too much noise with a improved fertility is hindered by the low heritability for fertility traits (Jansen, 1985; Pryce et al. 1997b, The second difficulty with using calving dates and insemination dates to measure a cows fertility One of the reasons for this low heritability might objectively is that association between yield and be that most fertility traits are based on calving and fertility might be biased (Philipsson, 1981; Jansen, insemination dates. These are obviously affected by 1985). Farmers might cull low producing cows the farmers decision to inseminate or cull animals earlier or delay inseminating a high producing cow which both might have an affect on the observed (Monday afternoon and Friday morning). Animals correlation between yield and fertility. with uterine infections or animals which had an To overcome some of these problems, Darwash et abortion were excluded from the data set. For all al. (1997a, b) suggested the use of progesterone other animals (n = 333) days to first luteal activity profiles to measure the start of luteal activity. These (CLA) was determined from the progesterone profile authors demonstrated how early re-establishment of as follows: when two consecutive samples were ovarian activity (CLA), measured using progesterone above 3 ng/ml the date of the first sample was taken profiles, is an important prerequisite for high as start of luteal activity. Since progesterone samples fertility; one day delay in the interval to first ovarian were taken till approximately day 100 in lactation, activity was associated with 0.24 and 0.41 days extra animals not cycling received a CLA of approx. 100 in the intervals to first service and conception, respectively. Each additional 21 days in CLA On a subset of these animals (n = 226), data on progressively reduced the probability of failure at a milk yield and energy balance during the first 100 given insemination to 0.89 of its previous value days of lactation were available. More details of the Darwash et al. (1997a). Furthermore, Darwash et al. recording and data handling are given by Oldenbroek (1997b) reported a heritability of between 0.13 and 0.28 for CLA. Hence, using progesterone profiles, a measure of fertility is available which is unaffected 2.2 Analysis
by management decisions and seems to have a The heritabilities (plus s.e.) were estimated using Because only one estimate of the heritability for VCE (Groenenveld, 1996). Additional to the random CLA is available yet in the literature, the first animal effect (including relationship matrix), year- objective of this study was to estimate the season and age at calving were fitted as fixed effects. heritability of days between calving and first ovarian Because its skew distribution, not only CLA was activity measured using progesterone profiles. analysed, but also two transformations of CLA: the Measuring progesterone profiles on individual cows is still costly and labour intensive as many milk There were obviously not sufficient records to samples (two or three times a week) need to be investigate average profiles for individual bulls. collected and analysed. This might only be feasible However, to investigate differences in mean on a small scale in for example a nucleus herd. As an progesterone profiles for groups of around 50 alternative, using the mean progesterone profile of animals, data were grouped according to CLA. Then, many irregular sampled animals (for example average progesterone profiles were calculated for monthly milk recording) might be sufficient for a each group separately, whereby the actual progeny tested bulls. However, because not progesterone records were transformed in a binomial sufficient progesterone measurements were available trait: al records above 3 ng/ml received the value of to investigate average profiles for individual bulls, 1 and all other records the value 0. This latter trait mean progesterone profiles for groups of around 50 has the advantage that it is easily interpretable animals differing in CLA were investigated herein. compared with actual progesterone levels, i.e. the Finally, because CLA is unaffected by average indicates the proportion of animals with management decisions, the phenotypic association progesterone > 3 ng/ml on a certain day (which between CLA and both milk yield and energy might be indicative for the proportion of animals cycling at a certain date). Also the binary trait simulates a relatively simple ‘yes or no’ 2. Material and Methods
progesterone test. In practise, such a ‘dipstick test’ might have advantages when used on a large scale. 2.1. Data
To investigate mean profiles, curves were created for each group separately. First, by simply averaging Data were available for Holstein Friesian heifers all the records per day of lactation, and secondly by calving at the ID-DLO experimental farm ‘t Gen fitting a smoothing spline in Genstat. The latter from 1994 to 1996. Part of the animals in this approach was applied on all records of individual analysis are among the highest genetic merit animals, cows (two a week; ca. 29 per animal), and on a as these are part of the ‘Delta’ sib testing program. subset of measurements for each cow (1 every The other heifers originate from the ID-DLO farm month; ca. 3 per cow) simulating measurements and were on average about half a standard deviation during normal milk recording. Hereto, each animal below the Delta heifers for INET (the Dutch was assigned a random start day of first test (uniform production index). distribution between 5 and 35 days), and A data set was prepared containing two subsequently for each cow the records closest to four progesterone records per week for each animal A simple linear regression was used to investigate 0.4 or 0.5 seem to be in the most distinctive part of the phenotypic association between CLA and energy The same conclusion can be drawn when only monthly progesterone samples were available for 3. Results
each cow: it was still possible to identify the curves for the six groups clearly (Figure 1c). Although the 3.1. Heritability for CLA
within group variation might be larger for sire progeny groups than for our six groups, similar mean In this study mean CLA was 38 days and the differences in CLA are expected between progeny standard deviation was 25 days. Heritabilities for groups. This is because the genetic standard CLA and its transformations ranged from 0.14 to deviation for CLA is close to 11 days (Table 1). 0.20 (Table 1). A genetic standard deviation of 11 Therefore, it might be possible to estimate smoothed days suggest that differences between animals are mean progesterone profiles for bulls using considerable for the interval between calving and progesterone samples on daughters during normal Table 1. Mean, standard deviation (in days) and
3.3. Milk yield and energy balance
heritability (s.e.) for days till first luteal activity (CLA) based on progesterone profiles, and Another use of progesterone is that it measures loge(CLA) and 100/CLA (n = 333). fertility more objectively, compared with relying on h² insemination dates and recording of the farmer. For this reasons it might provide a tool that enables us to exclude the management effect (i.e. later insemination of high producing cows) on the relationship between yield and fertility. Figure 2 demonstrates that even when fertility is measured 3.2. Average group curves
using progesterone, the phenotypic association between CLA and milk yield is unfavourable. When Mean progesterone profiles for groups of 50 animals daily milk increased with 1 kg a day over the first with all twice weekly progesterone measurements 100 days, CLA is expected to increase with 1.4 days animals are given in Figure 1a. Although curves are (s.e. = 0.3). Similarly, the phenotypic relationship clearly distinguishable between the groups, between between energy balance and CLA is negative: the days there is a lot of noise. This is not unexpected, regression coefficient is -0.33 (s.e. = 0.11). as, besides normal error variation, individual animals were only sampled twice a week and curves of 4. Discussion
individual animals are not synchronised. Whilst a normal cycling cows is expected to have one or two The use of progesterone measurements for the progesterone samples below 3 ng/ml every 21 days, improvement of fertility records (as suggested by and on average about 14 cows are measured each Darwash et al. 1997a and 1997b) is demonstrated in day, large sampling fluctuations in the mean this study. Using twice weekly samples on individual progesterone curves are expected. For the same cows, days till first luteal activity could be reason it is expected also that most curves asymptote established. Heritabilities for CLA and its around 0.7 - 0.8. Given that most of the animals in transformations were not significantly different from this study were not inseminated before day 100, it is the values reported by these authors. Although, on expected of normally cycling cows that, on a given the measured scale our heritability was slightly day, about 1/6 or 2/6 of the progesterone lower (although not significantly different from the measurements is below 3.0 ng/ml, because these The mean value of 38 days is higher than the When splines were fitted on all individual value of 27 days reported by Darwash et al. (1997b) progesterone records, but for each group separately, and the standard deviation was nearly twice the mean profiles became more distinct (Figure 1b). value reported by these authors. There are several Therefore it might be possible that from these possible explanation for this: i) in our data set smoothed curves, traits of interest for selection can progesterone was measured twice a week only and be derived. For example, interval between calving started at day 15 only. Darwash et al. (1997b) had at and the time period that 0.5 of the samples is above least three progesterone measures a week. Given that 3.0 ng/ml seems to be an option to indicate fertility CLA was based on two consecutive samples above 3 performance of a group of animals. The proportion ng/ml, in our study we might have missed some initial small increases in progesterone; ii) animals included in the study reported here were part of a interval to post-partum ovulation and traditional testing program and therefore hardly any animals measures of fertility in dairy cattle. Animal were culled before day 100. In commercial practice, some animals might have been culled before they Darwash, A.O., Lamming, G.E. and Woolliams, J.A. show heat, and in fact not showing heat might be 1997b. Estimation of genetic variation in the among the culling reasons; iii) mean milk yield in interval from calving to postpartum ovulation of our population was close to 30 kg/d, whereas the dairy cows. Journal of Dairy Science 80:1227- mean milk yield in the population reported by Darwash et al. (1997a, b) was about 18 kg/d. If the Esslemont, R.J. and Peeler, E.J. 1993. The scope for regression of CLA on yield demonstrated a causal raising margins in diary herds by improving relationship (Figure 2), then CLA was expected to be fertility and health. British veterinary Journal However, the results in this study confirm that Groenenveld, E. 1996. REML-VCE a multivariate progesterone measures might provide a useful tool multi model restricted maximum likelihood for genetic selection. In herds where intensive (co)variance component estimation package progesterone sampling is practised (i.e. for version 3.2. Users guide. Federal Research Centre management reasons or in a nucleus herd) CLA can be measured for individual animals. In the situation Jansen, J. 1985. Genetic aspects of fertility in dairy where only monthly milk recording is practised, cattle based on analysis of AI data - a review with mean progesterone profile based on a simple emphasis on areas for further research. Livestock ‘dipstick test’ might by sufficient to distinguish fertility performance of large progeny groups. Oldenbroek, J.K., Galesloot, P.A.J., Pool, M.H. and Splines can be used to smooth mean progesterone van der Werf, J.H.J. 1997. Effects of selection for profiles and based on these splines, the interval milk yield on feed intake and metabolism of between calving and days till 0.5 of the samples are heifers in early lactation. Meeting of the EAAP above 3.0 ng/ml provides a simple trait which could be used as trait to distinguish differences between Philipsson, J. 1981. Genetic aspects of female groups. A larger data set with progesterone measured fertility in dairy cattle. Livestock Production is required, however, to establish whether the heritability for this trait is as large as the heritability Pryce, J.E., Nielsen, B.L., Veerkamp, R.F. and Simm, G. 1997a. Genotype and feeding system effects and interactions for health and fertility Acknowledgements
Pryce, J.E., Simm G. and Veerkamp, R.F. 1997b. Henry van der Gaast, Leo Kruijt and farm staff for Genetics of fertility traits. Conference of the assistance with the data collection and looking after the animals. Holland Genetics for support towards Pryce, J.E., Veerkamp, R.F., Thompson, R., Hill, W.G. and Simm, G. 1997c. Genetic aspects of References
fertility traits in Holstein Friesian dairy cattle (In Darwash, A.O., Lamming, G.E. and Woolliams, J.A. 1997a. The phenotypic association between > 3 0.6
og. 0.5
op.
Pr 0.3
Da ys in la cta tion
og. > 3. 0.5
op. pr 0.3
Days in lactation
og. > 3 0.5
op pr
Pr 0.3
Da ys in la cta tion
Figure 1. Mean progesterone profiles (prog >3.0 ng/ml sample is 1 otherwise sample is 0) for cows grouped
according to their CLA. Median CLA values for the six groups are 17, 19, 24, 31, 47 and 85 days,
respectively, and each group contains around 50 records. The top graph is the average of the raw data based on
two progesterone samples a week per cow. In the second graph smoothing splines are fitted through this data.
In the bottom graph monthly milk recording was simulated by using only one sample every 28 days per cow.
Milk during first 100 days
(kg/d)
Energy balance (MJ /
d)

Figure 2. Days till first luteal activity (CLA) measured using progesterone as a function of average a) milk
yield and b) energy balance during the first 100 days of lactation

Source: http://interbull.slu.se/www/bulletins/bulletin18/paper24.pdf