The effect of breeding season duration on production and feed consumption in grazing beef cattle in the south of Israel

Introduction

In Amatzia, a cooperative village in the south of Israel, the beef cattle production cycle is planned to take advantage of the limited green-pasture season for pre weaning calf growth and for the lactating dam. The calves are weaned at the end of the green pasture season, irrespective of age. It follows, therefore, that weaning weights will be greatly influenced by time of calving.

In grazing cattle, a 12-month reproduction cycle is considered optimal in order to coincide with the pasture production cycle. A spread of calvings will eventually ensue, even if, initially, calvings took place within a limited period. Nevertheless, calvings can be restricted to a specified period by curtailing the breeding season.

In Amatzia, the breeding season has been reduced over the past three years, from 6 months (November to April, inclusive) to 4.5 months (November to mid-February), resulting in higher conception rates, higher weaning weights, shorter calving intervals (non pregnant cows are culled) and a concentration of calvings (85%) in the first two months of the season. The aim is to reduce the season further, to 2.5 months, by eliminating the stragglers.

A simulation study of the system was made, using the model developed by Kahn (1982), in order to examine the long-term effects of this policy and its repercussions on other factors.

Methods

The model developed by Kahn (1982) was revised in accordance with the findings of Kahn and Spedding (1984) and Kahn and Le89hrer (1984). Other modifications made recently are:

1. The death routines for both calves and adults were completely revised.

2. The optimal lactation curve no longer decreases after plateauing at the age of 7 years.

The model was run to investigate the long-term effects of 2-, 4- and 6-month breeding periods (beginning in November in each case) on the overall performance of the system. The 10-year runs were replicated five times. Annual averages were calculated for only the last five years of each run in order to eliminate the effects of the initial state of the system. The integration time interval was set at 30 days (10 days for calves); the run year was 360 days.

The run-year was set to commence after weaning on 1 May. At weaning, performed after integration for April but before the next integration time-step, all non-pregnant caws were culled and replaced by heifer calves from the weaned calf crop. Any cows which died during the run-year were similarly replaced. This method resulted in an almost constant herd size throughout the run period.

Herd size was set at 30 head. The initial weights and approximate pregnancy status of the cows were derived from a sample of the May 1983 weight and autumn 1983 calving data in the Amatzia herd. Accordingly, 12 caws calved in August, 12 in September, 2 in October and 4 in November, in the first run-year. In trial runs it was found that different initial pregnancy settings (2, 16, 6 and 6 calvings in August, September, October, November, respectively) had no significant effects on subsequent yearly (year 6 to year 10) averages in the 4- and 6-month treatments but did affect runs with a shorter season (3 month).

An attempt was made to simulate average conditions in Amatzia as far as possible. Green pasture was therefore set for three months—February, March and April—with 0.75, 0.7 and 0.65 digestibility values, respectively. May, an intermediate month, was set at 0.55; the remaining months were set at 0.45 (until November), 0.4 (December) and 0.35 (January).

Supplementation, per head and quality-wise, was as given in Amatzia, from June 1983 to January 1984, inclusive. However, replacement heifer calves were given 1kg of concentrate supplement per day over and above the recycled feed (if and when the latter was allocated) until halfway through their second lactation. Calves were given an allowance of concentrate feed in accordance with their weight, amounting to about 175 g/day in October and rising to 950 g/day in January. These amounts are similar to those given in Amatzia. Only the calves and heifers were given concentrate feed; cow supplementation consisted of recycled feed only (average enemy concentration of 9.74 MJ ME/kg DM).

No economic analysis of the system was attempted at this stage. However, supplementary feeds and calf/cow sales were represented by averaged values to enable an overall appreciation of the differences between the systems.

The prices were:

Recycled feed                                $0.06/kg LM

Concentrate feed                            $0.22/kg LM

Weaned calves                               $2.25/kg liveweight

Culled cows                                    $1.25/kg liveweight

Results

The key values of the simulation runs, representing the biological performance of the system, are presented in Table 1.

Table 1. Various consumption and production variable values for model nuns simulating 2-, 4-and 6-month breeding seasons. (All values are for a 30-cow herd plus unweaned calf crop.)

	Breeding season
Variable	2 month	4 month	6 months
Pasture consumption (t DM /year)	72.3*	73.3	74.1
Green pasture consumption (t DM /year)	34.7b	35.8a	35.9a
Recycled supplements (t DM/year)	43.1	43.0	42.7
Concentrate supplements (t DM/year)	6.1a	4.4b	3.9c
ME (from supplements) (1000 MJ/year)	512a	486b	473c
Cost of recycled feed ($1000/year)	2.6	2.6	2.6
Cost of concentrate feed ($1000/year)	1.3a	1.0b	0.9c
ME supps/kg cow+calf sold (MJ)	70	67	66
Annual option rate (%)	84c	90b	93a
No. weaned (N/year)	21.8b	24.6a	24.6a
Ave. wean. wt. (kg)	288a	274b	269b
Ave. wean. age (years)	0.73a	0.69b	0.67c
Ave. cow age (years)	6.3b	9.0a	9.5a
No. replaced (N/5 years)	27.2a	16.8b	14.4b
Cow mortality (N/5 years)	3.8	2.2	3.4
Calf mortality (N/5 years)	10.8	8.4	9.6
Calf sales (t/year)	4.9b	5.9a	6.0a
Culled cow sales (t/year)	2.5a	1.5b	l.1c
Cow+calf sales ($1000/year)	1.5	1.5	1.5
Feed cost: sales (ratio)	0.28a	0.24b	0.23b

*Within row, values with different superscripts differ significantly at the 5% level.

Discussion

The simulated long-term effects of length of breeding season on production variables showed that concentrating calvings in November–December significantly increased average calf age and weaning weight (7%), compared with the two longer-season treatments. However, as a result of the lower conception rates in the 2 month treatment (84 vs. 90 and 93% in the other two treatments), the number of calves weaned was reduced by 12%. Despite higher weaning weights, calf sales in this treatment were 13.5% lower than in the other two treatments, since those heifer calves reared as replacements (5.5 calves per year, almost twice as many as in the other treatments) are not included in the sales figure. The high replacement rate also accounts for the considerably higher concentrate consumption figures (39 and 56% higher than in the 4-and 6-month treatments, respectively), since heifers receive 1 kg concentrate/day during two 9-month periods. These simulated long-term effects are in contrast to the short-term (2 year) records in Amatzia, as well as in Kibbutz Sha'alvim (Lehrer and Schindler, 1984), in which reducing the spread of calvings was associated with an increase in the annual conception rate.

Nevertheless, the overall picture is less clear. The proceeds from culled cows in Treatment 1 are 1.7 and 2.3 higher than in the other treatments so that there are no differences at all between the treatments in total income from livestock sold. Consequently, the feed-cost:livestock-sales ratio was only 17 and 22% higher than in the 4- and 6-month treatments, respectively. The difference between the latter treatments is not significant, in keeping with most of the other differences between these treatments.

These results suggest that, mainly as a result of the progressively lower conception rate associated with the shorter breeding seasons, concentrating calvings in a short season entails some loss of profit, which may be significant when the season is reduced to as little as two months. Interpretation of the simulation results therefore hinges largely on the credence placed on the simulated, long-term reproduction performance.

Kahn and Lehrer (1984) critically examined the reproduction probability equations, adopted by Kahn (1982) from Sanders and Cartwright (1979), and suggested several major modifications. These were validated with field data from Israel. The data used related to hens with 6-month breeding seasons, i.e no attempt had been made to systematically improve the reproductive performance of the herd. On the assumption that curtailing the breeding season weeds out the stragglers, viz. those cows with long or irregular calving intervals, herd conception rate should rise with the more stringent culling policy, a phenomenon observed during the past two seasons in Amatzia. The reproduction probability equations used in the model do not allow for an improvement in overall herd performance; their upper limits under optimal conditions are p=0.85 for post-partum oestrus and p=0.75 for conception, given oestrus, values obtained by Sanders (1974) from extensively-managed beef-cattle herds. If a reduction in the breeding season and associated culling substantially increase herd reproductive performance, as was shown by Warnick and Fields (1976) in Florida, the threshold values for the probabilities should be modified accordingly.

The 2-year data from Amatzia show prima facie that reproductive performance does improve with reduced length of breeding season. However, considerably more data are required before this can be conclusively established, if the variance of the simulation results are any indication of the actual variance of the system. For instance, in the 2-month treatment, November conceptions ranged between 50 and 90% between years and had SD's of 1.2 to 12.2 (ave. for 10 years and 5 replications = 73%). The Amatzia records may represent erratic results which are not repeatable over the years, especially if the effect of cow age is taken into account (see Khan and Lehrer, 1984).

Another attribute of the probability equations, as expressed in the present model runs, may have distorted the results, especially those for the 2-month treatment. The original Sanders and Cartwright (1979) equations were quantified and adjusted for a 30-day integration step, while taking into account the 21-day reproductive cycle of cows. This device may be inappropriate when the breeding season is reduced to two months, a hypothesis which can be tested by running the model under different time steps (Kahn and Lehrer, 1984).

Concentrating calvings in a short season allows a more specifically production targeted supplementation policy. For instance, the January supplementation could be reduced to a point where lactation level is not severely jeopardised, if January is no longer in the breeding season. In the model runs, allocations of recycled feed supplements were identical in all treatments. However, the potential saving would not be large since the January allocation is about one sixth of the annual total and probably could not be reduced by more than 40%.

Despite the potential sources of error in the model, enumerated above, and perhaps others still undiscovered, the model results do indicate that pushing the herd to ever-shorter breeding seasons may have its dangers. However, as long as reproductive performance of the herd is not impaired, the process can be continued, provided that the system is monitored closely.

References

Kahn, Hava E. 1982. The development of a simulation model and its use in the evaluation of cattle production systems. Ph.D. thesis, Univ. of Reading, UK.

Kahn, Hava E and Spedding C R W. 1984. A dynamic model for the simulation of cattle herd production systems: II. An investigation of various factors influencing the voluntary intake of dry matter and the use of the model in their validation. Agric. Systems 13:63–82.

Kahn, Hava E and Lehrer A R. 1984. A dynamic model for the simulation of cattle herd production systems: III. Reproductive performance of beef cows. Agric. Systems 13:143–159.

Lehrer A R and Schindler H. 1984. Adjustment of the breeding cycle of beef herds to the vegetational cycle of the range in Israel. In: The reproductive potential of cattle and sheep. INRA, Paris, France.

Sanders J O. 1974. A model of reproductive performance in the bovine female. M.Sc. dissertation, Texas A & M Univ., USA

Sanders J O and Cartwright T C. 1979. A general cattle production systems model: II. Procedures used for simulating animal performance.Agric.Systems 4:289–309.

Warnick A C and Fields M J. 1976. Reproduction and fertility in beef cattle. In: Beef Cattle in Florida. Institute of Food and Agricultural Sciences, Gainesville, Florida, USA.

Discussion

Question – Since this is a l0-year model and since you have benefits which accrue, have you worked out the present value of the costs and sales?

Answer – No, this is not an economic model. The cost calculation is done just to give a rough idea of the costs of inputs and value of output.

Question– Is the incorporation of stochastic elements really worth the effort? Could not one carry out much simpler calculations just using average conception rates and average costs etc. and arrive at as good an answer?

Answer – The TAMU model was built on class averages, and so it needed a hundred or a thousand classes in order to account for cows of different age and different conception status and so on. Cows would move from class to class but you could not follow cow performance throughout its life.

Comment – But in fact, from a management point of view, it is the herd that one is concerned with, not with individual cows.

Reply – It would be difficult, if not impossible to analyse the reproduction and mortality aspects of the herd with a deterministic model. It is also much simpler to write and to conceptualise the model as composed of individual cows. It makes it easier to communicate with non-modellers. That is a very important aspect of the modelling process.

Comment – If you are interested in keeping the herd as individual animals, then you must deal with stochastic analysis. But if you are interested in management problems, then you do not use the results of individual cows, but need estimates of the mean value.

Comment – The purpose of this exercise is to provide some practical advice on management regimes. It would help to do a cost-benefit analysis on a stochastic versus a deterministic model. It would be interesting to see whether in fact, after a run of years, the model came up with any different final conclusion after eliminating the stochastic element and just using the average conception and mortality rates for a particular age class of animals. When you get a result which is counter-intuitive, then you want to figure out why you got it. It is very difficult to do so with a stochastic model.

Comment – I would say that the importance of a stochastic element in a pastoral model becomes very small, because there are so many other factors that restrict conception rate.

Reply – I tend to agree with you, because where there are all sorts of stress factors that reduce the conception rate, then they will be dominant. Where stress factors are excluded the only factor that remains is the time since calving. Then you get more random variation, rather than less variation.

Comment – The discussion seems to be confusing two issues, the source of the variation and the size of the variation. There is less variation in this kind of system than in a pastoral model. Therefore, to use the stochastic model in this type of system would have less effect, because the coefficient of variation on a calving rate of 90% is probably less than 10%. In the pastoral system with a mean calving rate of 50% and a much higher coefficient of variation, and probably a highly positive skewed distribution a stochastic model might have a much larger effect irrespective of what causes the calving rate to vary so much.

Comment – My question rose from the fact that if at the end of the day you just want to end up with average results, the average cost per calf, the average economic performance, then is it worth tracking out all of this stochastic variation? It may well be that one ought to be using a more sophisticated analysis for decision-making which takes account of this variability. But if you are not interested in measuring the variability, then a priori it would seem wasteful to model it. Whether it is large or small is irrelevant.

As far as reproductive performance is concerned, working with distribution rather than averages is essential if you want to learn about the situation of the herd.

If you take the average probability of performance on a herd basis it would be the same as taking the stochastic probability for the individual animal performance, adding them up and getting the same expected value. The advantage of working on individual animals would be in studying dynamics of a herd that is undergoing major changes in structure. In a stable herd, the expected value should be the same in both cases, whether the model is stochastic or deterministic.

Question – Have you used independent input data to validate your equations?

Answer – I used one set of data, which was collected by a colleague, M. Weitz, and I charged the functions of the TAMU model accordingly. Then I validated on a different set of data also from Israel and from a quite different herd. In the present model I am using the revised, not the original, equations. The biggest difference between the equations was that there was very little effect of weight change on conception if condition score was above 0.9.

Cows start the breeding season in fairly good condition, above 0.9 and then even if they lose weight throughout the breeding season, they attain a very high conception rate. This is in contrast to the widely held view that for high conception rates the cows should be either in weight balance or gaining weight.