Dystocia in dairy cattle breeding [:with special attention to sire evaluation for categorical traits]

Field data recorded in the early seventies showed that dystocia (calving difficulty) in the main Dutch cattle breeds occurred at a rate of 15-17% in heifers and 5-6% in cows. Corresponding stillbirth rates were 9-12% and 2.5-3.5% respectively. In the sixties and seventies many research papers on dystocia and stillbirth appeared which dealt with frequencies in various breeds and breed crosses, causes, associated factors, economic consequences and factors with systematic impact. These papers are reviewed in chapter 1. Important conclusions can be summarized as follows.

1. Dystocia is a major cause of stillbirth: 40 to 60% of the stillbirth cases encountered in field data are associated with dystocia. Additional negative consequences of dystocia are increased labour and veterinary costs, reduced fertility and milk yield, and an increased culling rate.

2. First-calf heifers and older cows differ both in frequency and in cause of dystocia. The frequency of calving difficulty in eifers is generally three to four times as high as in cows. In heifers, the most frequent cause of dystocia is feto-pelvic-incompatibility, i.e. a disproportion between the size of the calf and the pelvic dimensions of its dam. Deviations in the presentation of the calf account for a second substantial part. In cows, feto-pelvic- incompatibility is encountered less frequently, whereas other causes such as weak labour, uterine torsion and insufficient cervical dilation are rela tively more important.

3. Within parity, the probability of dystocia is positively related, i.e. increases, with the weight and dimensions of the calf at birth. The (phenotypic) relationships show a threshold character, i.e. the probability of dystocia increases strongly when the size of the calf exceeds a certain threshold value. The probability of dystocia is positively related with gestation length as well, probably mediated by calf size. With respect to the dimensions of the pelvic inlet of the dam a negative relationship is found. The size of the calf and the dimensions of the dam's pelvic inlet are the basic elements in feto-pelvic-incompatibility.

4. Apart from parity of the dam, factors with systematic influence on the frequency of calving difficulty are sex of calf and season of calving. In male calves the frequency is about twice as high as in females which is mainly due to the larger average size of the males. In the North-Western part of Europe dystocia tends to be more frequent in autumn and early winter than in spring and summer. This seasonal pattern is often, but not always, found to be associated with a seasonal fluctuation in birth weight.

5. Dystocia, stillbirth, as well as the associated calving traits birth weight and gestation length, are influenced by the genotype of both the calf and its dam, which results in a complicated genetic model. With a view to sire selection, the influence of the genotype of the calf is generally referred to as sire-of-calf effect, direct genetic effect or mate effect. Similary, the influence of the genotype of the dam is referred to as maternal-grandsire effect or indirect genetic effect. The mode of action of the direct genetic effect is supposed to be relatively simple: the sire transfers genes to the calf which are expressed in the calf phenotype (e.g. size, gestation length) which may affect the ease of birth. The maternal grandsire is supposed to act upon calving traits in a more complicated way. First, the maternal-grandsire effect contains a direct genetic component since the sire of the dam also transfers genes to the calf. This part of the maternal-grandsire effect is similar to the sire-of- calf effect but half of the size. Besides, the maternal grandsire may in fluence calving traits by transferring genes to the dam which actual ly find expression in the dam, for instance in the pelvic dimensions or by means of a maternal effect on the size of the calf. This is referred to as the maternal genetic component. The direct and mater nal genetic component are suspected to interact. Estimates of rele vant parameters are summarized in tables III to VI of chapter 1.

6. Difficulties are encountered in the genetic-statistical analysis of data on ease of calving and stillbirth because of their categorical nature. At least in theory, the non-normality, the location dependent variance and the non-additivity of-effects involved reduce the value of the usual estimation and prediction techniques based on the linear model and on normality of traits.

In the sixties and seventies AI-organisations in the Netherlands gradually started to apply the results from the research on calving traits, those on the effect of parity and the sire-of-calf effect in particular, by introducing a so-called heifer-bull-programme. This involves the testing of young bulls for direct genetic effect on calving traits and the recommendation of low-risk bulls with respect to dystocia and stillbirth for heifer matings. At that time, the data collection procedure was not uniform over AI-organisations, whilst the method of sire evaluation was not optimized. At the end of the seventies the research project which underlies this thesis was started, with two main objectives:

1. Optimization of the heifer-bull-programme including data collection and sire evaluation.

2. Examination of alternative sire-selection strategies to reduce dystocia, with emphasis on the impact of maternal genetic effects and direct maternal-interaction.

Sire evaluation for categorical traits

The optimization of data collection within the framework of the heiferbull-programme is no subject-matter of this thesis, but one of the main results, a mark-sense card for birth recording which is used nation-wide now, is shown as figure 1 in chapter 4. In the optimization of sire evaluation, emphasis was laid on a comparison between the method of Best Linear Unbiased Prediction (BLUP), which has optimal properties with normally distributed data, and a pseudo- Bayesian nonlinear method of sire evaluation for categorical variates based on the "threshold', concept, when applied to binary and ordered tetrachotomous reponses.

Chapter 2 refers to a comparison of these two methods with simulated data with a family structure consisting of half-sib groups of equal or unequal size. Simulations were done at varying levels of heritability and incidences of the binary trait. The methods were compared for their ability to elicit responses to truncation selection. The nonlinear procedure, which is theoretically more appealing, proved to be superior in a mixed model setting when the responses were binary, the layout was highly unbalanced, heritability was moderate to high and the incidence was below 25%. When a one-way model was tenable or when responses were tetrachotomous, the results by both methods were very similar.

In a similar simulation study, it was examined whether the incorporation of heteroscedastic residual variance into the BLUP-model would improve the sire-ranking ability of this method with categorical variates (chapter 3). Such an "adjustment" was proposed in the literature. The results indicated however that the ability of BLUP to rank sires correctly for categorical traits is impaired by adjustments of this kind, particulary when responses are binary, in spite of a reduction in apparent prediction error variance. This is consistent with deductions from the theory of threshold models.

In a third study (chapter 4) BLUP, with homogeneous residual variance, and the nonlinear method were compared in estimating genetic merits for ease of calving and stillbirth under field conditions, i.e. with a data sample from the birth recording programme in the Netherlands. Since only minor differences in sire ranking were observed between the two methods, it is concluded that there is no clear advantage in applying the nonlinear prcedure, which entails more complicated and time consuming computations than BLUP, in sire evaluation for dystocia and stillbirth.

The results of these studies lead to the conclusion that BLUP with homogeneous residual variance, which is a standard method of sire evaluation for dairy traits in many countries, is fairly robust against the departures from normality and linearity arising with categorical data. Because of the computational expenses involved in the nonlinear procedure, particulary with large data sets, an appropriate BLUP-model which should at least correct for parity, sex of calf, herd and season of calving is advised for the evaluation of sires for dystocia and stillbirth in the Netherlands.

Sire selection for dystocia

Sire selection for direct genetic effect on dystocia may not be very appealing under the Dutch market conditions because dystocia is very highly cor related, in the genetic sense, with birth weight (r _g  ≈0.9) as shown in chapter 4. A reduction in level of dystocia by this type of selection would almost inevitably entail a reduction in birth weight which will lower the proceeds from the sale of young calves from the dairy farm. Since maternal effects are assumed to be involved in ease of birth, selection for traits contributing to maternal calving performance might be an alternative. With the exception of the pelvic inlet dimensions, which are difficult to measure and hence unsuitable for al large scale progeny test, such traits had not been identified. Chapter 5 refers to a study into the phenotypic association between ease of calving and a number of, mainly morphologic, dam traits in Holstein, Dutch Red and White and Dutch Friesian heifers. However, the only factor clearly associated with ease of calving was the size of the calf.

Since the attempt to identify dam traits with significant effect on calving performance failed, alternatives left apart from the heifer- bull-programme, are selection for direct genetic effect, with the disadvantage discussed, or selection for maternal-grandsire effect. Because of the suspected negative covariance between the direct and maternal genetic effects, it was felt necessary to examine the sign and the size of the covariance between the direct genetic effect and the maternal-grandsire effect. This necessity was underlined by a reluctancy among herdsmen to use the low-risk bulls identified in the heifer-bull-programme because they suspected that these bulls would sire small calves, which develop into small heifers and which run an increased risk of dystocia. In chapter 6 this hypothesis is examined in a two-way sireselection trial with purebred Holstein, Dutch Red and White and Dutch Friesian cattle. Although the results were not entirely consistent over breeds, this study yielded additional evidence that direct and maternal-grandsire merits for ease of birth as well as for birth weight and gestation length are correlated positively. It was conjectured that the maternal- grandsire effect on ease of birth is dominated by direct inheritance of calf size. This conclusion implies that selection of sires for ease of birth of their progeny as calves is expected to yield a favourable response with respect to the ease of calving of their daughters and vice versa.

Selection programmes

In the final chapter, the monetary (discounted) returns from three schemes of breeding bull selection for calving traits (score for calving ease, birth weight, gestation length), viz. identification of lowrisk bulls for heifer matings among breeding bulls selected for dairy traits, selection on direct genetic merit for calving traits in addition to selection for dairy traits and selection on maternal-grandsire merit for calving traits in addition to selection for dairy traits are estimated for several calf price levels and incidences of dystocia. The first scheme, i.e. the heifer-bull-programme, proved to increase the returns from a breeding programme in dairy cattle, particularly when the incidence of dystocia or the calf price is high. Within such a scheme sire evaluation for birth weight and gestation length in addition to the evaluation for a score for calving ease contributes only marginally to the accuracy of low-risk-bull identification, at least when the latter is based on about 200 progeny per sire. Selection indices for direct genetic or maternal-grandsire effects on calving traits reduced the incidence of dystocia only at low calf prices (upto about Dfl 5.00 per kg). In the normal calf price range in the Netherlands (Dfl 7.50 - Dfl 15.00 per kg) either type of selection, if advisable at all, should be for an increase in birth weight in spite of a corresponding rise in incidence of dystocia. This is due to the highly positive genetic correlation between dystocia and birth weight. Because of a reduced intensity of selection for dairy traits, the impact of additional selection for birth weight on the total returns from the selection of breeding bulls in dairy cattle is either negative, or only marginally positive however. Consequently, the heifer-bull-programme is the most attractive breeding strategy to reduce dystocia.