Structure of a dynamic simulation model for beef cattle production ...

Structure of a dynamic simulation model for beef cattle production systems H. Pang1,2 , M. Makarechian1,3, J. A. Basarab2, and R. T. Berg1 1Department of Agricultural, Food and Nutritional Science, University of Alberta, Edmonton, Alberta, Canada T6G 2P5; 2Alberta Agriculture, Food and Rural Development, Edmonton, Alberta, Canada T6H 5T6. Received

16 March 1999, accepted 21 June 1999. Pang, H., Makarechian, M., Basarab, J. A. and Berg, R. T. 1999. Stucture of a dynamic simulation model for beef cattle production systems. Can. J. Anim. Sci. 79: 409–417. A dynamic deterministic model for simulating beef cattle production systems is developed to evaluate the effects of production traits and management strategies on the bioeconomic efficiency of beef production systems. The model, named Alberta Beef Production Simulation System (ABPSS), is composed of four major submodels: herd inventory, nutrient requirement, forage production, and economic submodels. The herd inventory submodel is used to simulate population dynamics and feed requirements in the herd. The nutrient requirements submodel is mainly based on the 1996 version of the National Research Council (NRC). It is used to evaluate nutrients and feed requirements for calves and cows depending on their physiological status (maintenance, growth, lactation and gestation) and the climatic condition. The forage production submodel is used to predict forage growth rate, cattle grazing rate, available forage biomass and total hectares required for grazing. The economic submodel measures bioeconomic efficiency, as net return per cow, by subtracting total cost from total return. The nutrient requirements predicted by ABPSS were compared with those recommended by the NRC for testing. The results that were predicted by the NRC model and ABPSS model were similar, as expected. Sensitivity analyses showed that cow mature weight, milk production, calf weaning weight and feed prices were the most critical input parameters in the model. It must be noted that the model was developed based on available experimental results and data from the literature and, due to the unavailability of a suitable data set, the model could not be validated. We suggest that the ABPSS has the potential for providing a useful method for simultaneous consideration of many factors in an integrated system, which could be helpful to beef cattle extension specialists and cow-calf production managers for assessing the potential effects of different management and selection strategies on bioeconomic efficiency. Key words: Beef cattle, simulation and modelling, production system, optimization Pang, H., Makarechian, M., Basarab, J. A. et Berg, R. T. 1999. Structure d’un modèle en simulation dynamique applicable aux systèmes de production de bovins à viande. Can. J. Anim. Sci. 79: 409–417. Nous avons mis au point un modèle déterministe dynamique de simulation des systèmes de production de bovins à viande, pour évaluer les effets des caractères de production et les protocoles de gestion sur l’efficience bioéconomique de ces systèmes. Désigné par le nom de «Alberta Beef Production Simulation System» (ABPSS), il comprend quatre grandes composantes : inventaire du troupeau, besoins en nutriments, production fourragère et économie. La première composante est utilisée pour simuler l’évolution des effectifs et les besoins en aliments du troupeau. La composante «besoins nutritionnels», essentiellement basée sur la version 1996 des normes du NRC, sert à évaluer les besoins en nutriments et en aliments des veaux et des vaches selon leur état physiologique (entretien, croissance, lactation et gestation) et les conditions climatiques. La composante «production fourragère» sert à prédire le taux de croissance fourragère, le rythme de consommation au pâturage, la biomasse fourragère disponible et la surface totale requise en pâturage. La composante «économie» mesure l’efficience bioéconomique d’après le revenu net par vache, c.-à-d. les recettes totales moins les dépenses totales. Les besoins en nutriment prédits par le modèle ABPSS ont été confrontés avec ceux recommandés par le National Research Council (NRC). Comme on s’y attendait, les chiffres prévus par les deux modèles, NRC et ABPSS étaient concordants. L’analyse de sensibilité a fait ressortir le poids des vaches adultes, la production de lait, le poids des veaux au sevrage et les prix des aliments comme les paramètres d’intrants les plus critiques dans le modèle. Il faut rappeler que le modèle ABPSS a été construit à partir de résultats expérimentaux et des données publiées disponibles et que, en raison de l’absence d’un jeu de données actuelles convenables, il a été impossible de le valider. À notre avis, le modèle a ce qu’il faut pour la prise en compte simultanée de nombreux facteurs en un système intégré, laquelle serait susceptible d’aider les vulgarisateurs spécialisés en bovins à viande ainsi que les exploitations de naissage, dans l’évaluation des effets éventuels des diverses stratégies de gestion et de sélection sur l’efficience ou sur le rendement bioéconomique. Mots clés: Bovins à viande, simulation, modélisation, système de production, optimisation

The beef cattle production system is a complex aggregate of subsystems, each somewhat dependent on the others. Maximizing the effect of one particular sector may reduce the functional potential of the total system. Fitzhugh (1978) indicated that production efficiency should be evaluated at the integrated system level rather than at the subsystem or individual animal level. Amer et al. (1994) stated that all aspects of production must be taken into account in order to optimize production. The goal is to replace the traditional

Abbreviations: ABPSS, Alberta Beef Production Simulation System; ADG, average daily gain; BCS, body condition score; BWt, calf birth weight; CNCPS, the Cornell net carbohydrate and protein system; DM, dry matter; DMI, dry matter intake; GR, grazing rate; LCT, lower critical temperature; ME, metabolizable energy; MP, metabolizable protein; MWt, cow mature weight; NE, net energy; NEm, net energy requirement for maintenance; NRC, National Research Council; SD, standard deviation; WWt, calf weaning weight 409

410

CANADIAN JOURNAL OF ANIMAL SCIENCE

single output production function with a complex simulation model. Evaluation of the system’s component interactions should lead to a greater understanding of the total system. Research is needed to develop the necessary methodology to evaluate and compare alternative management strategies under a set of available resources. It is often difficult and expensive to collect all the data required to develop a system for evaluating bioeconomic efficiency. Simulation provides an excellent tool for the integration of quantitative knowledge from a variety of disciplines. It is possible to alter a subsystem and study its effect on the whole system (Spreen and Laughlin 1986). Computer simulation requires substantially less time and expenses than experimental work with a true biological system; however, it is limited by the designer’s ability to describe the system and its components mathematically. If the components of a system can be quantified and presented properly in a mathematical form, a simulation model can provide detailed and mechanistic studies to allow a thorough understanding of the total system (France and Thornley 1984). Simulation modeling has received considerable attention during the past 20 years. The widely recognized simulation models of cattle production can be divided into two categories: 1) simulation models used on mainframe computers, such as the Texas A&M beef cattle simulation model (Sanders and Cartwright 1979; Notter et al. 1979; Sullivan et al. 1981; Doren et al. 1985; Cartwright and Doren 1986; Bourden and Brinks 1987) and the Kentucky beef-forage model (Loewer et al. 1981; Loewer et al. 1983; Smith et al. 1985; Loewer and Smith 1986); 2) simulation models used on PC computers, such as The Cornell net carbohydrate and protein system (CNCPS) for evaluating cattle diets (Fox and Black 1984; Fox et al. 1988; Fox et al. 1992; Fox et al. 1995) and the NRC model (NRC 1984; NRC 1996). Specific questions being addressed by these simulation studies vary considerably; however, all can be described as involving the interface between animal performance and livestock production systems. The Texas A&M beef cattle simulation model and the Kentucky beef-forage model were used on a mainframe computer which are not user friendly. Although the Cornell CNCPS model and the NRC model are user friendly, they are mainly used as beef ration balancers and predict only nutrient requirements of cattle at specific sets of physiological and environmental conditions. They are not dynamic models. With the rapid development of more powerful PC computers, it has become possible to develop complex integrated dynamic simulation models that integrate a variety of disciplines on PC computers. Biological type, which is primarily based on cow size, maturing rate and milk production, is an important parameter in beef cattle production. Different sets of feed, labor and capital input may lead to important biological type x environmental interactions in beef production efficiency. The objective of this study was to develop a dynamic simulation model to aid researchers and producers in predicting the nutrient requirements of cows and calves on daily basis, and to evaluate the effects of production traits and management strategies on the bioeconomic efficiency of beef production systems. It must be noted that the model

was developed based on available experimental results and data from the literature and, due to the unavailability of a suitable data set, the model could not be validated. MATERIALS AND METHODS The present project was a cooperative effort among the University of Alberta, Edmonton, AB, Alberta Agriculture, Food and Rural Development (AAFRD), Edmonton, AB, and Beefbooster Cattle Alberta Ltd., Calgary, AB, Canada. The default input parameters were derived from appropriate published literature as well as unpublished reports of AAFRD, Beefbooster and the University of Alberta Ranch at Kinsella, AB. These inputs can be modified and adapted by the users for any specific beef cattle herd. The model is a dynamic, deterministic simulation and is programmed using STELLA (High Performance System, Inc. 1994) simulation software on PC/Windows or the Macintosh platforms. The STELLA software is designed to build specific dynamic systems and processes. It is a multilevel, hierarchical environment for constructing and interacting with models. The environment consists of two major layers: the high-level mapping and I/O (input/output) layer, and the model construction layer. It has the ability to create interactive user interfaces for sharing modeling insights. On the high-level mapping and I/O layer, some main input parameters are controlled by sliders and graphical input devices that can be easily changed. The output in graphs and tables can also be collected on the high level. Users need not be distracted by the details of the model, as they run simulations and test scenarios (High Performance System, Inc. 1994). Overall, STELLA can be a powerful simulation tool for evaluation of beef production systems. In our model, the numeric values of the various inputs can be updated by the user as new information becomes available. Nutrient requirements and performance of individual animals and their offspring can be simulated in daily or weekly intervals over a number of years as defined by the user. A dynamic deterministic approach is used in this model, except for the daily temperatures, which were generated stochastically based on monthly average temperatures and standard deviations (SD) from normal distribution. The reason for using stochastic variable for daily temperature is because the daily temperature is not predictable, while the stochastic daily temperatures were generated based on the monthly means and SD, which reflect the trend. The default monthly temperatures and their SD are based on the thirty years of records in Edmonton, Alberta (Environment Canada 1995), which can be changed by the user depending on the region’s climatic conditions. The important traits included in the model were: cow mature weight, milk production, body condition score, calf birth weight, weaning weight, pre-weaning and post-weaning ADG, calving and weaning rates, mortality, forage yield and quality, forage growth rate, animal grazing rate, and some other environmental and economical factors. Body condition score (BCS) was assessed on a scale ranging from 1 (emaciated) to 9 (extremely fat) based on the American evaluation system (Richards et al. 1986). The model named Alberta Beef Production Simulation System (ABPSS) is the result of an interdisciplinary effort

PANG ET AL. — A DYNAMIC SIMULATION MODEL FOR BEEF CATTLE Table 1. Age distribution, culling rates and mortalities of cows in different age groups used as default input parametersz Age (yr) Ratio to population (%) Culling rate (%) Mortality (%) zBased

1

2

3

4

5

6

7

8

9+

25 18 3.1

20 18 2.1

14 16 1.7

10 15 1.2

8 16 1.0

6 15 1.0

5 17 1.0

4 17 1.6

8 50 5.2

on Arthur et al. (1992).

that incorporates the interactions of energy and protein requirements, herd dynamics, forage production and economics for range beef cattle production systems. The model is composed of four major submodels: herd inventory, nutrient requirements, forage production, and economic submodels. This manuscript is derived from the senior author’s thesis, which describes the model in detail (Pang 1997). Herd Inventory Submodel Cows in the herd are divided into several categories based on age and physiological status. The herd inventory submodel is used to: 1) Simulate dynamics of age groups in a cow herd. Cows are classified into nine age groups (from 1 to 9 yr and older). At the end of each year, cows in each age group are automatically moved to the next age group. The default ratio for each age group is based on data from the University of Alberta Beef Research Ranch at Kinsella, (Table 1, Arthur et al. 1992). It can be adapted by the user to any specific herd. 2) Predict the number of replacements and culled cows based on expansion or reduction of the herd. Older and open cows are culled at the end of each year to maintain the desired herd size. If needed, first-year heifers can be bought from outside the herd to change the herd size. The default culling and mortality rates in each age group are shown in Table 1. These are input variables, which can be easily changed by the user. 3) Input the calving pattern. The calving season is divided into six periods of 21 d each. The default number of calves born (%) in each period is shown in Table 2 (Mathison 1993), which can be changed by the user. The calving rate and the starting date of calving season are also input parameters. 4) Simulate the number of culled cows and weaned calves available in each age group and in the whole herd. 5) Estimate metabolizable energy (ME) and metabolizable protein (MP) requirements, and dry matter intakes (DMI) for the herd and for each cow age group, based on the outputs of the nutrient requirements submodel, which estimates the nutrient requirements for individual animals. Nutrient Requirements Submodel The ability of an animal to meet its nutritional requirements depends on the amount of dietary energy and protein.

411

Routine evaluation of diets including costs, animal performance and environmental conditions unique to a particular farm are essential to the process of making appropriate decisions with regards to the management of the herd. The nutrient requirements submodel, is mainly based on the Cornell Net Carbohydrate and Protein System for Evaluating Cattle Diets (CNCPS; Fox et al. 1984, 1988, 1992, 1995), and has been modified to accommodate the new version of NRC (1996). It evaluates nutrient and feed requirements of calves and cows depending on their physiological status (maintenance, growth, lactation and gestation) and climatic conditions (temperature, relative humidity and wind speed). The model can also be used as a daily management aid for adjusting nutrient requirements and utilization over wide variations in cattle, feed, management and environmental conditions. The effects of cold stress and lower critical temperature (LCT) on maintenance requirements can also be addressed. Forage Production Submodel Forage growth is a dynamic phenomenon that functions as a self-sustaining process responding to its environment. The forage production submodel predicts forage growth rate, cattle grazing rate (DMI), forage yield and quality, stocking rate and total hectares required for a herd, based on the common forage species in different soil climatic zones in Alberta (Alberta Agriculture 1995). Logistic functions (Richards 1959) were found generally useful to describe the relationship of forage yield to days of growth. The form of the logistic equation is: Y = K / (1 + b * e–rt)

(1)

where Y is the forage yield or biomass (DM kg ha–1); K is an estimate of maximum yield of the forage (kg ha–1); r is the maximum instantaneous rate of growth (kg d–1 ha–1); b is a constant; and t is the period of growth in days. Based on Eq. 1, the forage growth rate (kg d–1 ha–1) can be derived as: dY/dt = r * Y (1 – Y/K)

(2)

where Y, K and r are input parameters. They differ with different pasture species, soil type and environment, and can be changed by the user. Forage biomass (yield) is also affected by the cattle grazing rate (forage intake rate). Grazing rate increases at low forage biomass and reaches a plateau at high forage biomass. However, as forage biomass declines, the amount of forage eaten by each cow also declines because of low forage availability. A saturation of DMI for a cow is reached only when forage biomass is abundant so that the cow can eat all she wants. This is called “functional response.” This

Table 2. Calving distribution of beef cattle during the calving season used as default input parameters z Days from the first calving Days from 1st calving Calves born (%) zBased

1–21 42.5

22–42 33.4

on the survey in Alberta, Canada by Basarab (Mathison 1993).

43–63 16.2

64–84 7.9

85–105 0

106–126 0

412

CANADIAN JOURNAL OF ANIMAL SCIENCE

Table 3. Numbers of replacements and culled cows of different age groups for a stable herd of 100 cows for six years predicted from the Alberta Beef Production Simulation System (ABPSS) Number of culled cows by age group (yr) Year 1 2 3 4 5 6

Replacement

1

2

3

4

5

6

7

8

9+

21 21 21 21 21 20

4 4 4 4 4 4

4 4 3 3 3 3

2 2 2 2 2 2

1 2 2 2 2 2

1 1 2 2 2 1

1 1 1 1 1 1

1 1 1 1 1 1

1 1 1 1 1 1

4 3 3 3 3 3

is described by the Michaelis-Menton equation, Y/(Y+Y50), which approximately estimates the effect of this saturation (Shipley and Spalinger 1992). Therefore the cow grazing rate (GR) is estimated as: GR = DMImax * (Y / (Y +Y50))

(3)

where DMImax is the maximum dry matter intake of a cow from one hectare based on nutrient requirements (kg ha–1); Y is the forage biomass or yield (kg ha–1); and Y50 is the forage biomass at half of the maximum grazing rate (kg ha–1). Here Y50 = 350 DM kg ha–1 (NRC 1987). Economic Submodel Bioeconomic efficiency, measured as net return per cow, is obtained by subtracting total cost from total return.

Total cow cost was the sum of fixed and variable costs. Fixed costs included interest on intermediate and long-term debt, depreciation, property tax, repair and insurance. Variable costs included cow or replacement cost, feed and non-feed expenses for the cow herd and for calves to weaning or to slaughter. Replacement cost was considered only if herd size was increasing and some replacements had to be purchased from outside. Non-feed costs included death loss, interest on operating loan, labor, manure and feed handling, fuel, veterinary service, housing and feed storage, marketing, breeding and miscellaneous costs. The default values in the model for cow fixed costs and non-feed costs were $65.54 and $68.39 per cow for one year, respectively, based on Beefbooster data (MacNeil et al. 1994). Feed costs included pasture and winter feed. The prices for pasture and winter feed were assumed to be $0.03 kg–1 and $0.06 kg–1, respectively (Alberta Agriculture 1987). All the costs and prices are input parameters in the model.

Net return = Total return – Total cost RESULTS AND DISCUSSION Total return is estimated from the sale of culled cows and calves at different market endpoints: (weaned, backgrounded and finished). The average sale prices were based on 10 yr of data (Agriculture and Agri-Food Canada 1995). Cattle sale price is discounted by two factors: body weight and carcass grade. The default weaned calf price was discounted $4 per 45 kg body weight as calf weight increased, starting from 140 kg of calf weaning weight (Alberta Agriculture 1987).

Herd Inventory Submodel The distribution of cows in each age group changes over the years. Table 3 shows the predicted numbers of replacements and culled cows for each age group for a 100 cow herd. The herd inventory submodel was also used to estimate the metabolizable energy requirements for the whole herd and for each age group, based on the outputs of the nutrient requirements submodel. Fig. 1 shows the metabolizable energy requirements (Mcal d–1 herd–1) for a herd of 100 cows.

Fig. 1. Metabolizable energy (ME, Mcal d–1 herd–1) requirement for a 100 cow herd in one year.

PANG ET AL. — A DYNAMIC SIMULATION MODEL FOR BEEF CATTLE

Fig. 2. Daily temperature (°C) in Edmonton, AB, Canada used in this example, and low critical temperature (LCT, °C) for a cow with BCS = 5.

413

Fig. 3. Metabolizable energy (Mcal d–1 cow–1) requirement for maintenance (MEm) and total (MEtotal) for the first calving year of a heifer (input parameters: cow mature weight = 550 kg, peak milk yield = 10.9 kg, cow BCS = 5, calf birth weight = 40 kg, calving at the end of March, weaning at the end of October, and cattle grazing on pasture from late of May to October).

Fig. 4. Metabolizable energy (Mcal d–1 cow–1) requirement for lactation (MElact) for the first calving year of a heifer (for input parameters, see Fig. 3).

Nutrient Requirements Submodel Daily temperature (°C) and low critical temperature (LCT, °C) for a cow with body condition score of 5 are shown in Fig. 2. The default daily temperature was stochastically estimated from a normal distribution of monthly means and standard deviations (SD) based on 30 years of meteorological data in Edmonton, AB (Environment Canada 1995). The monthly temperature means and SD are input parameters. Predicted metabolizable energy (ME, Mcal d–1) requirements for total, maintenance, lactation and pregnancy for a heifer over 1 yr on a daily basis are shown in Figs. 3, 4 and 5. Net energy requirement for maintenance (NEm) increased during the grazing season (late May to October) because of

the grazing activity. Some peaks were observed during winter months, which indicate that extra energy for maintenance is required when temperature drops below the animal’s LCT (–13 to –20°C). The extra NEm required for cold stress during winter is influenced by the cow’s BCS. When a cow’s BCS = 2 (thin), her LCT increases, resulting in extra NEm to combat cold stress. At a high BCS, no extra NEm is required. The model can also be used to predict the nutrient requirements of a cow over several consecutive years. Available body reserve is used when energy intake is below requirement and is replenished when requirement is lowered or feed supply becomes more plentiful. Current BCS of an animal can be used to predict the number of days

414

CANADIAN JOURNAL OF ANIMAL SCIENCE

Fig. 5. Metabolizable energy (Mcal d–1 cow–1) requirement for pregnancy (MEpreg) for the first calving year of a heifer (for input parameters, see Fig. 3). Table 4. Effects of energy reserve and energy balance on body condition scores (BCS) of cows differing in mature weight z Mature weight (kg)

Energy reserve (Mcal score–1)

NE change (Mcal d–1)

Days change to next BCS

New BCS after 10 d

450 550 650

127 155 183

–3 –3 –3

–34 –41 –49

5 Ô 4.73 5 Ô 4.78 5 Ô 4.82

450 550 650

127 155 183

+3 +3 +3

42 52 61

5 Ô 5.21 5 Ô 5.17 5 Ô 5.15

zInitial

condition score was assumed BCS = 5.

Table 5. Monthly changes in cow condition score based on the energy balancez Month MEreq (Mcal d–1)y DMIreq (kg d–1)y DMIact (kg d–1) BCS change (score d–1) BCSnew (score)

Jan

Feb

Mar

Apr

May

Jun

Jul

Aug

Sept

Oct

Nov

Dec

22.11 10.05 9.77 –0.02 4.96

22.81 10.37 9.9 –0.03 4.84

22.58 10.27 10 –0.02 4.75

23.5 10.68 10 –0.06 4.60

24.11 9.65 10 0.02 4.71

24.95 9.98 10.05 0.02 4.68

24.25 10.78 10.94 0.02 4.72

24.25 10.78 11.67 0.06 4.78

24.0 10.67 11.57 0.06 4.90

20.41 9.07 9.72 0.04 5.22

19.33 9.66 9.6 –0.01 5.28

20.99 10.49 9.6 –0.07 5.14

zInput parameters: cow mature weight = 550 kg, peak milk yield = 10.9 kg, cow BCS = 5, calf birth weight = 40 kg, calving at the end of March, weaning at the end of October, and cattle grazing on pasture from May to October. yME req and DMIreq were predicted from the model.

required for one unit change (±) in condition score based on the energy intake relative to the energy requirement of the animal. Table 4 presents the estimated number of days required for one unit change in condition score, when the daily net energy intake is ±3 Mcal d–1 higher or lower than the energy requirement for cows of three different mature weights. Energy reserves for cow size of 450, 550 and 650 kg were 127, 155 and 183 Mcal per score, respectively. As an example, if a ration is deficient by 3 Mcal of NE d–1, then a cow with a BCS of 5 would reach BCS of 4 in 34, 41 and 49 days for cow sizes of 450, 550 and 650 kg, respectively. If a cow at BCS of 5 consumes 3 Mcal of NE d–1 in excess of her requirements, she will move to a BCS of 6 in 42, 52 and 61 days for cow sizes of 450, 550 and 650 kg, respectively.

The model can also be used to predict changes in cow condition score year-round based on the diet and animal energy balance. Table 5 presents an example of the ME and DMI requirements of a cow predicted from the model, her actual DMI, the rate of change in her condition score (BCS change) and her new BCS in each month of the year. This indicates that generally a cow’s BCS increases during the grazing season (May to October), and decreases during winter (November to April) if the winter feeding does not meet her requirement. Forage Production Submodel The total forage yield and available forage yield (kg ha–1) based on the forage quality (protein, %) of smooth brome grass in southern Alberta are presented in Fig. 6. The model

PANG ET AL. — A DYNAMIC SIMULATION MODEL FOR BEEF CATTLE

415

Fig. 6. Estimated total forage yield and available forage yield (DM kg ha–1) for Smooth Brome in southern Alberta, Canada.

Table 6. Nutrient requirements of beef cows based on the Alberta Beef Production Simulation System (ABPSS) and NRC (1996)z Month NE total (Mcal NRC (1996) ABPSS

Jan d–1,

Feb

with grazing activity) 15.5 16.7 15.1 16.4

MPtotal (g d–1) NRC (1996) ABPSS DMI recomm. (kg d–1) NRC (1996) ABPSS

592 603 13.4 13.4

680 690 12.4 12.4

Mar

Apr

May

Jun

Jul

Aug

Sept

Oct

Nov

Dec

19.1 19.0

19.5 19.3

23.5 23.0

21.8 21.2

20.5 20

19.5 19.3

19.2 19.2

16.0 15.9

12.5 12.7

13.9 14.0

879 863 13.0 13.1

968 965 13.1 13.2

919 932 12.7 13.0

829 832 12.4 12.4

740 748 12.6 12.3

665 661 12.3 12.7

610 611 12.5 12.5

480 478 11.9 12.1

501 502 12.4 12.4

536 535 12.4 12.4

zInput parameters: breed = Angus or Hereford; cow mature weight = 590.2 kg; peak milk yield = 9.761 kg; BCS = 5; calf birth weight = 36.32 kg; age at calving = 5 yr old; calf weaning age = 210 d; time peak milk = 8.5 weeks; milk fat = 4.0%; milk protein = 3.4%; milk SNF = 8.3%; breeding date = 140 d; start calving in March, weaning in October, grazing season between May and October; rations: 5 kg brome, 10 kg orchard grass DM (diet ME = 1.87 Mcal kg–1, NEma = 1.04 Mcal kg–1, NEga = 0.5 Mcal kg–1).

can also be used to predict forage growth rate, cattle grazing rate, forage quality, stocking rate and total hectares required for a herd. TEST OF THE MODEL Validation of a model involves testing and an assessment of the performance of the model. A model should mimic the real system satisfactorily to fulfil the objectives for which it is developed. When one is confident that the behavior of the model is satisfactory, the formal validation process is finished (Dent and Blackie 1979). Gaining confidence in the model is generally a slow process and occurs through model construction, validation and application. Most of the equations used in this model have been independently validated for the nutrient requirements submodel (NRC 1984, 1987, 1996; Fox et al. 1984, 1988, 1992, 1995). Comparisons of predicted nutrient requirements for a beef cow using the NRC model (NRC 1996) and ABPSS are presented in Table 6. The results predicted by the two models were similar, as expected. There were no experimental data available for validating the forage growth curves in the forage submodel. The herd inventory and economic submodels are based mainly on

logic rather than on quantitative relationships. It is difficult to validate a logic model (Dent and Blackie 1979). However, the herd inventory and economic submodels were checked and verified to ensure that they were not generating unreasonable results, and were evaluated for robustness. SENSITIVITY ANALYSES Sensitivity analysis is a procedure carried out on the validated model that involves exploring the operation and performance of the model (Dent and Blackie 1979). It is used to identify the parameters to which the system is highly sensitive. Sensitivity analyses were conducted by varying the important input parameters by ±10% of their default values. When the value of one parameter is changed, all other parameters are held constant. The important input parameters analyzed for sensitivity in this study included cow mature weight, calf weaning weight (which also reflects the level of cow milk production), calf birth weight and feed prices (pasture price and winter feed price). The effects of changing cow mature weight, calf weaning weight and calf birth weight by 10% on the required total metabolizable energy of a cow while holding the other parameters constant are shown in Table 7. The model was sensi-

416

CANADIAN JOURNAL OF ANIMAL SCIENCE

Table 7. Change in parameters by 10% for sensitivity analyses –10% Input parameter Cow mature weight, (kg) Calf weaning weight (kg) Calf birth weight (kg) Feed price: Pasture price ($ kg–1) Winter feed price ($ kg–1) Output Cow total ME required (Mcal d–1): Cow mature weight Calf weaning weight Calf birth weight Cow total cost Feed price ($ d–1 herd–1)

unchanged

+10%

495 225 36

550 250 40

605 275 44

0.027 0.054

0.03 0.06

0.033 0.066

22.3 (–5.9)z 21.9 (–8.5) 23.8 (–0.6)

23.7 23.9 23.9

25.1 (+5.7) 26.0 (+9.0) 24.0 (+0.6)

85.4 (–7.4)

92.2

99.0 (+7.4)

zValues

in the parentheses show the difference (%) of the change ±10%. Difference (%) = (value of 10% – value of unchanged) / value of unchanged * 100%.

tive to an increase or decrease in cow mature weight (MWt) and calf weaning weight (WWt), which resulted in a large change in cow total ME requirement, the differences were 6% and 9% for MWt and WWt, respectively. The results demonstrate the importance of cow mature weight, calf weaning weight and cow milk production on cow nutrient and feed requirements. The effect of varying calf birth weight by ±10% on the total required metabolizable energy of a cow was very small, causing a change of only 0.6%. The model was therefore not sensitive to changes in birth weight, and accuracy in the estimation of birth weight was not important with respect to the total nutrient requirements of a cow. The effect of increasing or reducing the feed prices (pasture price and winter feed price) by 10% on total cow cost in a herd, given that the other parameters are held constant, is shown in Table 7. As would be expected, the model was very sensitive to changes in feed price (7.4% change in total cost for 10% change in feed prices). Thus accuracy in the estimation of pasture price and winter feed price was critical in the model. CONCLUSIONS The Alberta Beef Production Simulation System (ABPSS) is a dynamic, deterministic model, composed of four major submodels: herd inventory, nutrient requirement, forage production, and economic submodels. The herd inventory submodel simulates dynamic changes in a herd of nine age groups and predicts the number of culled cows and weaned calves available in each age group and in the herd. The nutrient requirement submodel predicts nutrient requirements (ME, MP) for maintenance, growth, lactation and pregnancy of a cow, a calf, a cow-calf pair and a herd. The model also considers the effects of cold stress and lower critical temperature (LCT) on maintenance requirement. The model can predict changes in cow condition score on a year-round basis, based on the diet and animal energy balance. The forage production submodel can predict forage growth rate, cattle grazing rate, forage yield and quality,

stocking rate and total hectares required for a herd, based on climatic conditions and common forage species in different soil climatic zones in Alberta. The economic submodel can estimate feed cost, total cow cost, return and net return or bioeconomic efficiency of the herd. In stating the conclusions, the authors would like to emphasize the fact that the model was developed based on available experimental results and data from the literature and, due to the unavailability of a suitable data set, the model could not be validated. Sensitivity analyses of the model show that: 1) cow mature weight play an important role in cow nutrient and feed requirements during the winter; 2) calf weaning weight and cow milk production were also among the most important traits affecting the total ME requirement; 3) the cow total ME requirement was not sensitive to changes in calf birth weight; and 4) accuracy in the estimation of pasture price and winter feed price were critical in the model. It is important to note that the model did not consider the effect of a deficient diet on production traits such as milk production and growth. The model assumed that animals always had access to enough feedstuff and could meet their nutrient requirements. In summary, the ABPSS has the potential for providing a useful method for simultaneous consideration of many factors affecting beef production in an integrated system, which could be helpful to beef cattle extension specialists and cowcalf production managers, for predicting the effects of different management and selection strategies on bioeconomic efficiency. Proper use of the model, when the factors influencing bioeconomic efficiency are well defined, can hopefully identify a strategy for improving the probability of success under given environmental conditions. As an example of the application of the ABPSS model, in the next paper, the effects of calving season and weaning age on bioeconomic efficiency of beef cattle production are examined. Producers can use their own input variables in comparing different management strategies. ACKNOWLEDGMENTS The authors gratefully acknowledge the supports of the University of Alberta, Edmonton, AB, Alberta Agriculture, Food and Rural Development (AAFRD), Edmonton, AB, Beefbooster Cattle Alberta Ltd., Calgary, AB, and the Alberta Agricultural Research Institute (AARI). Agriculture and Agri-Food Canada. 1995. Livestock market review. Agriculture and Agri-Food Canada, Ottawa, ON, Canada. Alberta Agriculture. 1987. Beef herd management. Alberta Agriculture, Edmonton, AB, Canada. Alberta Agriculture. 1995. Varieties of perennial hay and pasture crops for Alberta. Agri-Fax. Alberta Agriculture, Edmonton, AB, Canada. Amer, P. R., Kemp, R. A., Buchanan-Smith J. G., Fox G. C. and Smith, C. 1994. A bioeconomic model for comparing beef cattle genotypes at their optimal economic slaughter end point. J. Anim. Sci. 72: 38–50. Arthur, P. F., Makarechian M., Berg, R. T. and Weingardt, R. 1992. Reasons for disposal of cows in a purebred Hereford and two multibreed synthetic groups under range conditions. Can. J. Anim.

PANG ET AL. — A DYNAMIC SIMULATION MODEL FOR BEEF CATTLE Sci. 72: 751–758. Bourdon, R. M. and Brinks, J. S. 1987. Simulated efficiency of range beef production. I. Growth and milk production. J. Anim. Sci. 65: 943–955. Cartwright, T. C. and Doren, P. E. 1986. The Texas A&M beef cattle simulation model. Pages 159–192 in T. H. Spreen, and D. H. Laughlin, eds. Simulation of beef cattle production systems and its use in economic analysis. Westview Press Inc., Boulder, CO. Dent, J. B. and Blackie, M. J. 1979. Systems simulation in agriculture. Applied Science Publishers Ltd., London, UK. Doren, P. E., Shumway, C. R., Kothmann, M. M. and Cartwright, T. C. 1985. An economic evaluation of simulated biological production of beef cattle. J. Anim. Sci. 60: 913–934. Environment Canada. 1995. Canadian climate normals: Temperature and precipitation. Environment Canada, Ottawa, ON. Fitzhugh, H. A. 1978. Animal size and efficiency, with special reference to the breeding female. Anim. Prod. 27: 393–401. Fox, D. G. and Black, J. R. 1984. A system for predicting body composition and performance of growing cattle. J. Anim. Sci. 58: 725–739. Fox, D. G., Sniffen, C. J. and O’Connor, J. D. 1988. Adjusting nutrient requirements of beef cattle for animal and environmental variations. J. Anim. Sci. 66: 1475–1495. Fox, D. G., Sniffen, C. J., O’Connor, J. D., Russell, J. B. and Van Soest, P. J. 1992. A net carbohydrate and protein system for evaluating cattle diets: III. Cattle requirements and diets adequacy. J. Anim. Sci. 70: 3578–3596. Fox, D. C., Barry, M. C., Pitt, R. E., Roseler, D. K. and Stone, W. C. 1995. Application of the Cornell net carbohydrate and protein model for cattle consuming forages. J. Anim. Sci. 73: 267–277. France, J. and Thornley, J. H. M. 1984. Mathematical models in agriculture. Butterworths, London, UK. High Performance System, Inc. 1994. STELLA II: An introduction to systems thinking. High Performance System, Inc., Hanover, NH. Loewer, O. J. and Smith, E. M. 1986. The Kentucky beef-forage model. Pages 63–100 in T. H. Spreen and D. H. Laughlin, eds. Simulation of beef cattle production systems and its use in economic analysis. Westview Press Inc., Boulder, CO. Loewer, O. J., Smith, E. M., Gay, N. and Fehr, R. 1983. Incorporation of environment and feed quality into a net energy model for beef cattle. Agric. Sys. 11: 67–94. Loewer, O. J., Smith, E. M., Benock, G., Bridges, T. C., Welles, L. G., Gay N., Burgess S., Springate, L. and Debertin, D. 1981. A simulation model for assessing alternative strategies of beef production with land, energy and economic constraints. Trans. ASAE 24: 164–173.

417

MacNeil, M. D., Newman S., Enns, R. M. and Stewart-Smith, J. 1994. Relative economic values for Canadian beef production using specialized sire and dam lines. Can. J. Anim. Sci. 74: 411–417. Mathison, G. W. 1993. The beef industry. Pages 35–74 in J. Martin, R. J. Hudson and B. A. Young, eds. animal production in Canada. University of Alberta, Edmonton, AB. Notter, D. R., Sanders, J. O., Dickerson, G. E., Smith, G. M. and Cartwright, T. C. 1979. Simulated efficiency of beef production for a midwestern cow-calf-feedlot management system. I. Milk production. J. Anim. Sci. 49: 70–82. National Research Council. 1984. Nutrient requirements of beef cattle. National Academy Press, Washington, DC. National Research Council. 1987. Predicting the feed intake of food producing animals. National Academic Press, Washington DC. National Research Council. 1996. Nutrient requirements of beef cattle. 7th ed. National Academic Press, Washington, DC. Pang, H. 1997. Simulation studies on the effects of cow size, milk production and calving season on the bioeconomic efficiency of beef production. PhD thesis. University of Alberta, Edmonton, AB. Richards, F. J. 1959. A flexible growth function for empirical use. J. Exp. Bot. 10: 290–300. Richards, M. W., Spitzer, J. C. and Warner, M. B. 1986. Effect of varying levels of post-partum nutrition and body condition at calving on subsequent reproductive performance in beef cattle. J. Anim. Sci. 62: 300–306. Sanders, J. O. and Cartwright, T. C. 1979. A general cattle production systems model. I. Structure of the model. Agric. Sys. 4: 217–227. Shipley, L. A. and Spalinger, D. E. 1992. Mechanisms of browsing in dense food patches: Effects of plants and animal morphology in intake rate. Can. J. Zool. 70: 1743–1752. Smith, E. M., Tharel, L., Brown, M. A., Dougherty, C. T. and Limbach, K. 1985. A simulation model for managing perennial grass pastures. I. Structure of the model. Agric. Sys. 17: 155–180. Spreen, T. H. and Laughlin, D. H., eds. 1986. Simulation of beef cattle production systems and its use in economic analysis. Westview Press, Boulder, CO. Sullivan G. M., Cartwright, T. C. and Farris, D. E. 1981. Simulation of production system in east Africa by use of interfaced forage and cattle models. Agric. Sys. 7: 245–265.