- Email: [email protected]
Implications of on-farm segregation for valuable milk characteristics
AGRICULTURAL SYSTEMS Agricultural Systems 85 (2005) 82–97 www.elsevier.com/locate/agsy
Implications of on-farm segregation for valuable milk characteristics A.E. Dooley a
a,*
, W.J. Parker a, H.T. Blair b, E.M. Hurley
c
AgSystems, AgResearch Limited, Ruakura Research Centre, Private Bag 3123, Hamilton, New Zealand b Massey University, Private Bag 11 222, Palmerston North, New Zealand c 26 Springdale Grove, Palmerston North, New Zealand Received 17 April 2003; received in revised form 2 June 2004; accepted 14 July 2004
Abstract Milk composition varies between herds and cows within herds, enabling its segregation on farm, rather than during processing, for the manufacture of specific dairy products. Benefits may include increased product yields, reduced processing costs and greater suitability of differentiated milk for the production of high value niche market products. However, costs are also likely to be greater. An integrated spreadsheet model was developed to determine the break-even premium required for a farmer with a seasonal calving herd to be economically no worse off producing segregated than conventional milk. The model incorporated breeding (quantitative and qualitative traits), cow requirements and feeding, transport, and economic sub-models. Cows were segregated within herd and milk composition was altered over time by genetic selection. Four quantitative trait (‘‘white’’ milk colour) scenarios and two qualitative trait (BB b-lactoglobulin milk) scenarios were considered. The model suggested that ‘‘white’’ milkfat would need to earn 38.4% more at the farm gate than conventional milkfat for the two systems to break even. ‘‘White’’ milk cows produced less than their status quo counterparts due to the reduced selection pressure on production milk traits and this had a considerable impact on the premium, as did the low initial volumes of white milkfat. The difference in production between the B b-lactoglobulin cows and their status quo counterparts was less than for selection on white milkfat only. The high risk to farmers of discontinuing a differentiated milk policy could be moderated by changing the structure of
*
Corresponding author. Tel.: +64 7 838 5914; fax: +64 7 838 5012. E-mail address: [email protected] (A.E. Dooley).
0308-521X/$ - see front matter Ó 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.agsy.2004.07.012
A.E. Dooley et al. / Agricultural Systems 85 (2005) 82–97
83
premium payments over time. Hence, processing companies and farmers will need to work together to facilitate the uptake of milk segregation. This research model could be applied by dairy companies and farmers considering milk segregation policies. Ó 2004 Elsevier Ltd. All rights reserved. Keywords: Dairy industry; Milk characteristics; Milk payment; Modelling; Genetic improvement
1. Introduction New Zealand has traditionally exported bulk commodity dairy products. However, to remain competitive in the international market there has been a steady move towards producing more branded and value-added dairy products aimed at niche markets. These products require greater emphasis on meeting customer specifications for quality and consistency than commodity products (Larsen, 1997; Scrimgeour, 1999). Product type and quality, and the amount of product able to be produced (and for some products the timing of manufacture) are all affected by milk composition and quality, as are dairy industry returns through the product mix made (Fenwick and Marshall, 1991; Hobman, 1992; Mackle and Bryant, 1995; Ryan, 1996). Natural variation exists between cows and herds in the many proteins, fats, carbohydrates, vitamins and micro-elements that constitute milk (Crabtree, 1984; Nickerson, 1995). An opportunity exists for dairy processing companies to take advantage of this natural variation by ensuring milk from different cows, herds or regions is kept separate until processing can commence at the factory. Cows with specialist characteristics could either be aggregated onto particular farms or into different herds on the same farm. Milk has not been segregated on-farm in New Zealand until very recently, and then only on a very small scale for a few niche products. New Zealand has largely supplied commodity products, and farmers have been paid on milk volume, and milkfat and protein production. There has been little work done on altering aspects of milk for traits other than these. It has been assumed that any other changes required in milk can be achieved by various processing techniques (Thomson, 1993). However, greater interest in developing niche markets and better meeting customer requirements has lead to interest in the possibility of on-farm milk segregation. Farmers could modify milk characteristics by management (e.g., calving spread, replacement policies), genetic selection and feeding (Crabtree, 1984; Holmes et al., 1984; Gibson, 1989, 1991, 2000; Holmes, 1989; Sutton and Morant, 1989; Kolver and Bryant, 1992; Nickerson, 1995; Mackle and Bryant, 1995, 1996; Auldist et al., 1997; Kennelly and Glimm, 1998; OÕBrien, 2000). Selection of cows based on a particular milk trait will affect correlated milk characteristics, and environmental factors, such as nutrition, cow age, stage of lactation may further affect these. Opportunities to modify milk also exist through the use of new genetic and reproductive technologies (Karatzas and Turner, 1997; Cunningham, 1998, 1999; Visscher et al., 1998; Wells et al., 1998; LÕHuillier, 1999; Wall, 1999).
84
A.E. Dooley et al. / Agricultural Systems 85 (2005) 82–97
A further on-farm implication is that some cows may be traded on their ability to produce milk of a particular type, rather than solely on breeding worth. To achieve this, cows and sires will need to be tested for the desired milk components, and selection of herd replacements may be based on customised breeding indices. Farmers will require a financial incentive before they supply dairy companies with segregated milk. Extra on-farm costs may be incurred in the production and storage of specialised milk, and greater transport and processing costs may be incurred because of the need for milk tanker re-routing and separation of milk at the factory. The primary purpose of this paper is to investigate the impact of milk segregation policies on farm and discuss implications to the farmer and the industry. The implications of milk separation are numerous and complex, and to systematically assess these, a model of biological and economic effects of milk segregation was developed. This provided a practical and inexpensive means to evaluate the impact on the system, particularly as breeding was involved and changes took considerable time (years). Two example milk traits, B b-lactoglobulin and milk colour, were evaluated with the model. Volumes and average levels for specified milk characteristics for the different lots of milk due to biological variation were defined, and the management and financial implications associated with milk segregation considered. The following sections describe the scenarios modelled and provide a brief description of each of the sub-models i.e., their function and the primary inputs and outputs. A comprehensive description of each of the sub-models, not provided in this paper, is presented in Dooley (2002).
2. Milk segregation scenarios Traits can be altered by breeding (Van Vleck et al., 1987). Traits which involve many loci and a graduation of phenotypes are described as quantitative traits. Traits controlled by few loci, so that one allele affects the phenotype are described as qualitative traits. Qualitative breeding trait genes can affect production directly or act as marker genes for production traits. Four quantitative scenarios (breeding and segregating cows with whiter milkfat colour) and two quantitative scenarios (breeding or selecting cows with a BB b-lactoglobulin genotype) were investigated with the model (Table 1). Although milk is not currently segregated for these traits in New Zealand, they were selected to evaluate the economic implications of milk segregation and demonstrate the model because they have known product advantages. New Zealand milkfat products tend to be more yellow than those of other countries because of our pasture-based farming systems (Keen, 1986) resulting in higher levels of b-carotene in the milkfat. Whiter milkfat (4.5 mg b-carotene/kg milkfat or less (MacGibbon, 2001; pers. comm.)) products (e.g., butter, ghee, white cheeses and whipped toppings) are preferred by consumers in some markets (Keen, 1986; Hobman, 1992; Keen and Wilson, 1993). The model segregated milk from cows with a ‘‘whiter milkfat colour’’ from that of the remaining cows.
A.E. Dooley et al. / Agricultural Systems 85 (2005) 82–97
85
Table 1 Description of the scenarios modelled Scenario Quantitative Scenario 1 Scenario 2 Scenario 3 Scenario 4
Qualitative Scenario 1 Scenario 2
Description Selection on colour only, with an initial average colour of 8.0 mg b-carotene/kg milkfat Selection on colour and milkfat yield, with an initial colour of 8.0 mg b-carotene/kg milkfat Selection on colour only as for scenario 1, but with an initial colour of 6.0 mg bcarotene/kg milkfat As for scenario 1 for the first 10 years, then reverting to status quo values for genetic gain and milk prices for the remaining years
Changing to a BB herd over time by breeding Changing to a BB herd in year one by buying and selling cows
The qualitative scenarios considered breeding and/or segregating cows with a BB b-lactoglobulin genotype. Milk from cows of this genotype is superior to milk from cows with AA or AB b-lactoglobulin genotypes for cheesemaking, milk powder production and UHT milk production (Hill et al., 1997; Ng-Kwai-Hang, 1998). BB b-lactoglobulin genotype cows have higher casein and lower whey protein production (including b-lactoglobulin). Some trials have also shown higher milkfat concentration and production, lower total protein production and lower milk yields in these cows compared to AA b-lactoglobulin cows (Hill et al., 1997). Breeding or management policies could include: initial testing of the cows and potential ABheterozygous/B-homozygous calves for genotype to identify the B-homozygous animals to breed and segregate milk from; testing and sale and purchase of cows in year one so all cows are B-homozygous from year 1; and no testing (use of homozygous bulls only and all milk kept in one lot). The first two scenarios are presented in this paper. The initial amount of milk able to meet specifications and the rate of gain in the desired trait are affected by the initial levels in the population. For the quantitative milk trait, cows were assumed to be Friesians which have whiter milk than Jerseys (Keen and Wilson, 1993; Newman et al., 1994; Winkelman et al., 1999). For the qualitative trait, the herd was assumed to be Jerseys, which have a higher frequency of the B b-lactoglobulin allele than Friesians (Winkelman and Wickham, 1996). Status quo herds were modelled for both quantitative and qualitative scenarios assuming selection on production traits only and industry rates of genetic gain. Selection on other traits (milkfat colour and b-lactoglobulin) reduced genetic gain in the production traits, particularly for the quantitative trait (colour). Breeding (extra NZ$5/cow) and herd testing (extra NZ$2/cow tested for colour, and NZ$24.50/ cow tested for b-lactoglobulin genotype) costs were greater for segregated than status quo milk herds. Animals only needed testing once in the qualitative scenarios.
86
A.E. Dooley et al. / Agricultural Systems 85 (2005) 82–97
The whole herd plus replacements needed to be tested in year 1 for qualitative scenario 1 (1.25 tests/cow allowing for replacements), with some calves only in subsequent years. The second qualitative scenario considered buying and selling cows so all were BB genotype in year one. This required testing 2.14 cows for each animal in the herd to identify sufficient BB cows, and replacing 54% of the herd paying a premium price for those purchased (NZ$200/cow and NZ$150/heifer more than the price received for cows sold). An extra milk vat was purchased in year one (NZ$5000). An average size New Zealand dairy farm (93 ha, LIC, 2000) with Manawatu region pasture growth rates was assumed for all scenarios. Variable per cow costs for animal health, breeding, herd testing, dairy expenses and power were assumed to be NZ$119.76. Feed costs (e.g., silage and hay making, calf rearing, grazing off of young stock, freight to grazing) and stock income would have differed between stocking rates. Status quo value milk prices were average New Zealand values for the 1999/2000, 2000/2001 and 2001/2002 seasons at NZ$2.894/kg and NZ$6.241/kg for milkfat and protein, respectively and a deduction of NZ$0.0415/L milk volume. An additional milk volume charge of one cent per litre was assumed to be incurred on segregated milk.
3. Model description A deterministic simulation model was developed to derive the break-even premium for a dairy farmer, ceteris paribus, to segregate milk for a specific trait or processing characteristic on-farm. The break-even premium calculated required that a farmer be no worse off financially if a segregated milk policy was adopted. A premium was received on the milkfat component which met the criteria value of 4.5 mg b-carotene/kg milkfat or less (MacGibbon, 2001 pers. comm.) for the quantitative trait. For the qualitative trait, a premium was received on both the milkfat and the protein components of the BB b-lactoglobulin milk. The sensitivity of premiums to discount rates and transport costs was explored, as was the effect of redistributing the milk premium through time (Dooley, 2002). Changes required to alter a herdÕs milk composition can take time, often years, particularly if breeding is involved. The model therefore incorporated the changes over a 20-year time frame. A cost-benefit analysis using discounted cash flows was used to calculate the break-even premium over periods of 10 and 20 years, and to infinity. Fig. 1 shows the model design and the main input–output flows between the submodels. The model was developed using Microsoft Excel and Visual Basic software and was run for a 20-year period. Although milk processing was not modelled, this is shown diagrammatically to illustrate how it fits into the overall scheme, and to indicate the input and output variables between this and the other sub-models. All submodels contributed variables to the overall model – the farm economic model. This model was used to calculate the premium required for a range of comparative scenarios.
A.E. Dooley et al. / Agricultural Systems 85 (2005) 82–97 Breeding Models
Proportion of cows in each herd
87
Proportion of cows in each herd
Potential production by cow “type” or herd
Production/cow/day by herd and period
Cow Requirements and Production
Energy value of the feed by period
Feed required per cow by herd and period
Feed Budget
Cow numbers
Cow numbers
Milk Income Model
Cow numbers. Costs of feed made & purchased
Milk income
Farm Economic Model
Stock Reconciliation
Sale & purchase costs. Grazing, freight & calf costs
Production/herd/day
OFF-FARM
Milk income
Transport Cost Model
Volume charge
ON-FARM
Milk Income Model
Volume charge
Processing Model
Pricing of components
Fig. 1. Model design and information flow. The milk income model is shown twice (on-farm and off-farm aspects) to simplify the diagram.
The farm production and economic sub-models incorporated breeding and feeding factors which affect milk quantity and composition, and farm profitability. The milk transport sub-model was used to assess the cost implications of differentiated milk collection logistics. Milk processing, distribution and marketing data are largely subject to commercial sensitivity and therefore unavailable, making inclusion of a processing sub-model impractical.
88
A.E. Dooley et al. / Agricultural Systems 85 (2005) 82–97
3.1. Breeding sub-models Quantitative and qualitative breeding sub-models were developed (Dooley, 2002) to evaluate selection for milkfat colour and b-lactoglobulin, respectively. Inputs for both breeding sub-models included base production levels, herd structure, age adjustments, and changes in genetic gain over time. Annual outputs for both submodels included the proportion of the cows in each herd, the average (across age groups) potential milk, milkfat and protein production, and liveweight for each herd (i.e., the differentiated herd and the standard milk herd). Inputs specific to the quantitative breeding model included phenotypic correlations and standard deviations for selection traits, and the cut-off criterion for milkfat colour (average milkfat colour required in the white herd based on a set lactation length). Quantitative breeding model outputs included the actual cut-off value to meet the criterion (minimum average milkfat colour required for cows to be in the white herd) and the average milkfat colour for each herd. As milkfat colour varies over lactation, the cut-off criterion and actual cut-off values related to the average milkfat colour over the lactation. Cows were assumed to remain in their allocated herd for the entire lactation. Inputs specific to the qualitative breeding model were the frequency of the B allele in the initial population and the genotype of the bulls used. Annual outputs specific to the qualitative breeding model included: the genetic frequency and the change in frequency over time, the production of each milk type allowing for age adjustments and differences in genetic merit between age groups; and if all milk is stored in the same vat, the proportion of B b-lactoglobulin to A b-lactoglobulin in the milk. 3.2. Cow requirements and production sub-model Feed requirements and production per cow, over a year divided into 24 periods, were modelled on the basis of energy using the template of Brookes et al. (1993). The original model was altered to allow for growth in younger cows and changes in milk composition (milk, milkfat, protein and milkfat colour) over the lactation. Stage of lactation and feeding affect milkfat colour over the lactation and these effects are often confounded due the New ZealandÕs seasonal calving pattern and pastoral based feeding system (Keen, 1986; Keen and Wilson, 1993; Winkelman et al., 1999). Feed requirements were adjusted for culling and spread of lactation effects. 3.3. Feed budget sub-model Stocking rate was derived from a pasture-based farm feed budget (Gray and Parker, 1990; Brookes et al., 1993) using per cow requirements as an input. Young stock were assumed to be grazed off or farmed separately from the milking herd. Production per herd (allowing two herds per farm), and per farm was then calculated for each of the 24 periods in the year. Herd inputs specified included the composition of the diet and the energy value of the components, intake per cow (based on production), and the proportion of cows
A.E. Dooley et al. / Agricultural Systems 85 (2005) 82–97
89
in each herd. Farm inputs included pasture growth rate and utilisation; maximum and minimum pasture covers allowed over the year (Holmes et al., 1984); effective area of the milking platform (area of the farm used for the milking herd); area in crops and crop production; and pasture grown from nitrogen applied. When combined, the farm inputs defined the herdÕs feed supply. Model outputs for each of the 20 years included: stocking rate and number of cows, opening cover, grass silage fed and made, hay fed and made, hay, silage and concentrates purchased, hectares (ha) of crop grown, pasture production per ha per year and milk production as described above. Feed and grazing costs were derived within the feed budget model. 3.4. Stock reconciliation sub-model Livestock purchased and sold over the 20 years were derived from an annual stock reconciliation. Inputs included calving percentage, herd age structure, deaths, and specified sales and purchases. Sales (culls) and purchases (to make up numbers) of the different livestock classes were calculated. Livestock related costs and income (e.g., sales of cull cows and bobby calves, calf rearing costs and stock freight costs), and capital value of the stock were derived within the stock reconciliation model. 3.5. Milk income sub-model Annual milk income from a farm was calculated over the 20-year period taking both the milk payment system and time-related adjustment factors into consideration. Prices were entered for the components in the standard New Zealand payment formula, milk yield (NZ$/L,negative value), milkfat (NZ$/kg), and protein (NZ$/kg), and another trait if required (i.e., for differentiated milk). Factors affecting milk price for each component were herd, year, month and criteria value as described in Table 2. The criterion value specified for the differentiated milk trait (e.g., milkfat colour) determined whether this milk received a premium, with milk from the differentiated herd meeting or exceeding this value (e.g., whiter) receiving a premium. This takes into account factors that affect the colour of the differentiated milk over the lactation e.g., stage of lactation, feeding effects. This criterion value may differ from the cut-off Table 2 Factors affecting prices received for milk Factor
Description and inputs
Herd Year
Status quo herd and differentiated milk herd Up to three sets of prices able to be specified over the 20 years e.g., years 1–7, 8–14 and 15–20 For each month, milk is specified as receiving either peak milk prices, or off-peak milk prices Differentiated milk herd only. Criterion value specified for each year e.g., milkfat colour required for milk to receive a premium
Month Criteria to be met for the differentiated trait
90
A.E. Dooley et al. / Agricultural Systems 85 (2005) 82–97
criterion specified in the breeding model, which is the average for the lactation. The breeding model cut-off criterion may be set higher than the criterion value specified here to ensure milk receives a premium over the whole year. If the same value is used in both sub-models, milk will only receive a premium for the part of the year when it is whiter than average. Other inputs included the number, and proportion, of cows in each of the two herds for each of the 20 years, the discount rate and farm size (effective hectares). The production per cow per period for each herd was read in from the cow requirements model. Annual milk income per herd per year was read into the farm economic model. 3.6. Farm economic sub-model A cost-benefit analysis was used to calculate the difference between segregated and status quo scenarios, the present value (PV) of the income less costs for each year and the net present value (NPV) over 10 years, 20 years and infinity (20+ years). Discount rates of 7% and 10% were used. The break-even value required for differentiated milk is the value where the NPV for the segregated and status quo scenarios was the same. All cows in the status quo scenario were assumed to be in one herd. Segregated scenarios could have two herds (a status quo milk type and a differentiated milk type) in all or some years. Income, cow-related variable costs and capital costs were calculated for each of the herds. A partial budget approach was used, assuming fixed costs, not directly affected by the policy, remained constant between the policy options. 3.7. Transport sub-model A transport model was developed to estimate the likely increased cost of milk collection for segregated milk compared to non-segregated milk. The imputed value for segregated milk was entered into the milk income model. Milk collection was modelled where none, 25%, 50% and all farms changed to another milk type over a 20year period. Two contrasting regions (30 farms each) with different farm sizes and roading networks (collection areas) were simulated, at two stages of the spring-calving seasonal lactation (peak milk flow and late lactation). The model was developed using Microsoft Excel spreadsheets, and the Evolver genetic algorithm software (Palisade Corporation, 1998) was applied to search for the order of the farm milk collection which gave an optimal, least cost solution for milk collection for each prescribed set of inputs.
4. Results and discussion The six policies, and the premiums required for these differentiated milk alternatives to break even with the status quo over 10, 20 and 20+ years (infinity) years are summarised in Table 3.
A.E. Dooley et al. / Agricultural Systems 85 (2005) 82–97
91
Table 3 The status quo prices and the premium required per kilogram of protein and milkfat for each scenario, for the NPV to equal zero over 10, 20 and 20+ years Scenario
Component
Break-even period 10 Years
20 Years
20+ Years
Status quo prices
Milkfat Protein
$2.89 $6.24
$2.89 $6.24
$2.89 $6.24
2. Selection on colour and milkfat yield, initial colour 8.0 3. Selection on colour only as for 1, but with initial colour 6.0 4. As for no. 1 for first 10 years, then reverting to SQ values for genetic gain and prices paid
Milkfat Protein Milkfat Protein Milkfat Protein Milkfat Protein
$2.08 – $2.10 – $0.41 – $2.08 –
$1.11 – $1.08 – $0.46 – $5.62 –
$0.92 – $0.87 – $0.47 – $9.66 –
Qualitative scenarios 1. Changing to a BB herd over time by testing and selection 2. Changing to a BB herd in year 1 by testing, and stock purchase
Milkfat Protein Milkfat Protein
$0.113 $0.243 $0.139 $0.300
$0.098 $0.212 $0.119 $0.256
$0.093 $0.200 $0.107 $0.231
Premium required Quantitative scenarios 1. Selection on colour only, initial colour 8.0
Prices are in New Zealand dollars.
The proportion of the total milk attracting a premium increased over time for all scenarios except qualitative scenario 2, with only small volumes of milk attracting a premium in the early years for some of the scenarios (Fig. 2). Quantitative scenario 4 is the same as quantitative scenario 1 up until year 10, after which no milk attracts a premium. The low volumes contributed to the high premium required for some of the scenarios, particularly quantitative scenarios 1, 2 and 4, where only 2.4% of the milk met the specifications for differentiated milk in year one. By contrast, the policy of buying and selling cows in qualitative scenario 2, meant all milk met the requirements in year one. Changes in milk production and composition between the status quo and the scenarios for the qualitative trait were small. Initial cow numbers for all scenarios was 329. Initial milksolids (milkfat plus protein) production for the status quo and scenario 1 was 307.4 kg/cow (307.2 kg/cow for scenario 2), and 1087 kg/ha (all scenarios), with a protein to fat ratio of 0.7557 (0.7502 for scenario 2). In year 20 there was very little difference in production between the status quo and the scenarios (383 kg milksolids cow or 1433 kg milksolids per ha), and no difference in cow numbers (348 cows). Protein to fat ratio was slightly lower for the scenarios than the status quo (0.826 compared to 0.832) after 20 years. In contrast, the quantitative scenarios had greater changes in cow numbers, and production over the 20 years (Table 4). In the first year, all scenarios had 293 cows.
92
A.E. Dooley et al. / Agricultural Systems 85 (2005) 82–97
% of milk attractinga premium
100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 1 2
3 4
5 6 7
8 9 10 11 12 13 14 15 16 17 18 19 20
Year MF8, colour
MF8, colour+MF
MF6, colour
BB milk
Fig. 2. The percentage of milk attracting a premium over the 20 years for three quantitative scenarios (initial milkfat of colour 6 or 8 b-carotene/kg milkfat, selected on colour only or colour and milkfat) and the first qualitative scenario (BB b-lactoglobulin milk).
Milksolids production was 317.6 kg/cow and 1001 kg/ha, with a protein to fat ratio of 0.7867. Milkfat production was 177.8 kg/cow and 560 kg/ha. The reduction in per cow production for the milkfat and protein components (Table 4) contributed to the high premium required under the quantitative scenarios. Shifting some or all selection emphasis to reducing colour slowed genetic progress in the production traits relative to the status quo. Lower production per cow resulted in a higher stocking rate but lower production per ha with the difference increasing over time. For example, under quantitative scenario 1, 5.5% more cows, and 5.1% less milksolids which included 1.1% less milkfat per ha was produced compared to the status quo after 20 years. Therefore, the premium had to compensate for reduced milksolids income as well as increased costs compared to the status quo. The premium required to break even decreased as the time period increased for most scenarios (Table 3) because of the larger volumes of differentiated milk produced and therefore higher returns in the later years. However, quantitative scenarios 3, with an already low premium, required a slight increase in premium to break even (from 14.2% more for a 10 year period, to 16.3% more for 20 and 20+ years), because the increased income from a greater proportion of white milk in later years did not fully compensate for the overall lower production in these years and the higher costs incurred relative to the status quo. For quantitative scenario 4, a large premium was required to break even over longer time periods (Table 3, e.g., 334% more than the status quo for milkfat for the 20+ year period) because of the low volume of differentiated milk produced in the first 10 years. Even though selection changed to the status quo objective after 10 years, it would still take years for the
A.E. Dooley et al. / Agricultural Systems 85 (2005) 82–97
93
Table 4 Cow numbers, and milksolids and milkfat production per cow and per ha for year 20 for the quantitative scenarios SQ
Scenario 1
Scenario 2
Scenario 4
319 +8.87%
336 +14.67% +5.33%
320 +9.22% +0.31%
331 +12.97% +3.76%
380.0 +19.65%
342.6 +7.87% 9.84%
368.18 +15.93% 3.11%
353.8 +11.40% 6.89%
1304 +30.27%
1238 +23.68% 5.06%
1267 +26.57% 2.84%
1259 +25.77% 3.45%
206.1 +15.92%
193.4 +8.77% 6.16%
210.4 +18.32% +2.09%
197.2 +10.91% 4.32%
Production/hectare kg MF/ha in year 20 Diff. over 20 years Diff. from SQ in year 20
707 +26.26%
699 +24.82% 1.13%
724 +29.29% +2.40%
702 +25.36% 0.70%
Protein:fat ratio in year 20
.8437
.7716
.7503
.7942
Cow number Cow number in year 20 Diff. over 20 years Diff. from SQ in year 20 Milksolids (MS) Production/cow kg MS cow in year 20 Diff. over 20 years Diff. from SQ in year 20 Production/hectare kg MS/ha in year 20 Diff. over 20 years Diff. from SQ in year 20 Milkfat (MF) Production/cow kg MF cow in year 20 Diff. over 20 years Diff. from SQ in year 20
The percentage change in these attributes over the 20 years, and between the status quo (SQ) and the other scenarios in year 20 are given. Note: Scenario 3 is the same as scenario 1.
production and stocking rate under scenario 4 to match that of the status quo herd. Therefore the premium received on the low volume of differentiated milk in the first 10 years had to compensate for the lower overall production over the 20+ years as well as the higher costs associated with the differentiated policy compared to the status quo. The results for scenario 4 highlight the downside risk (Hardaker et al., 1997) associated with the future loss of, or reduction in, the premium for differentiated milk because of other technology advances (e.g., a suitable processing technique to produce white milkfat could be developed), reduction in demand because of changes in consumer preferences, or oversupply of differentiated milk. The more risky farmers perceive a policy to be, the lower the likely technology uptake, requiring a higher premium (greater relative advantage) to offset this (Rogers, 1983). While the model
94
A.E. Dooley et al. / Agricultural Systems 85 (2005) 82–97
calculates a break-even premium to cover incremental costs over a given timeframe, in practice the premium is likely to be greater to compensate for some, or all, of the risk associated with a segregated milk policy in order to facilitate farmer adoption. A processing company would need to give these matters careful consideration when structuring its premium payments. In this regard, the amount and timing of supply, and therefore farmer adoption rate, is important because a critical volume of milk will be needed to fulfill market requirements. Company tactics might include identifying and targeting farmers who already produce milk with characteristics similar to that required, or those who are receptive to new technologies. Differentiated milk policies are not readily trialled or observed, may be complex, are not necessarily reversible, and can be risky, and as a result, farmer uptake of differentiated milk policies may be limited (Rogers, 1983). As the scenarios show, changes to a differentiated milk policy may take years to incorporate into the farming system, particularly where breeding is involved or there are extensive changes to management. Farmers will need access to information and decision support to enable them to make decisions and judgements about whether or not the technology is appropriate for them. The involvement of farmers as well as industry representatives, and research and extension researchers is likely to lead to the development of more effective strategies for a differentiated milk policy, and increases the likelihood of adoption (Guerin and Guerin, 1994; Frank, 1997). 4.1. Model application and evaluation The model has the potential to investigate the production and economic consequences of a wide range of scenarios for milk segregation on farm. Slower and expensive field trialling, if required, could then be restricted to ‘‘best bet’’ options. The model could be refined to better predict the effects of feeding on milk composition, and to include stochastic variables to assess the range and probability of different outcomes. Validation of the cow requirements model was by its developers and through onfarm applications. The feed budget, stock reconciliation and partial budget are standard farm management techniques, as is the construction of the genetic selection model. The transport model was assessed relative to actual performance data for milk collection and feedback from a dairy company transport manager. The aggregate model outputs were presented to a small group of processing company representatives and researchers, and they confirmed the reasonableness of the premium estimates and thus the modelÕs utility for developing milk segregation policies.
5. Conclusion On-farm milk segregation conceptually offers advantages to producers, processors and consumers. Implementation of milk segregation, however, is not straightforward as a myriad of factors impact on milk composition and quality, production costs and product returns. The model described in this paper integrates the various on-farm
A.E. Dooley et al. / Agricultural Systems 85 (2005) 82–97
95
components of milk composition and supply, plus any effects on transport costs to a processor. The added incremental costs associated with a milk segregation policy determine the premium required for a farmer to be no worse off relative to still pursuing a status quo milk production system. Scenarios modelled in this research indicate premiums, timeframes to achieve change and risks vary considerably between strategies for supplying specific milks. There are, however, new developments such as nutraceutical milk products, herd genotyping and reproductive technologies which will enhance the prospects of on-farm milk segregation. Acknowledgements We thank the many people who contributed information towards this study, particularly Dr. Ian Brookes and Dr. Nicolas Lopez-Villalobos from Massey University, Mr. Peter Spooner and Mr. Bob Franks from New Zealand Dairy Group, and Dr. Lawrie Creamer and Dr. Alastair MacGibbon from the New Zealand Dairy Research Institute.
References Auldist, M.J., Walsh, B.J., Thomson, N.A., 1997. Seasonal and lactational influences on milk composition in New Zealand. Proceedings of the New Zealand Society of Animal Production 57, 205–206. Brookes, I.M., Morris, S.T., Parker, W.J., 1993. Computer spreadsheets for predicting feed requirements and feed budgeting. Proceedings of the New Zealand Grassland Association 55, 209–210. Crabtree, R.M., 1984. Milk compositional ranges and trends. In: Castle, M.E., Gunn, R.G., (Eds.), Milk Compositional Quality and Its Importance in Future Markets. Occasional Publication No. 9. British Society of Animal Production, Edinburgh, pp. 35–42. Cunningham, E.P., 1998. The potential of new reproductive and genetic technologies. In: Einarsson, S., Rendel, J., (Eds), Impact of Reproductive Technology on Animal Breeding and Genetic Conservation. Acta Agriculturae Scandinavica, Section A, Animal Science, Supplementum 29. Scandinavian University Press, Oslo, Norway, pp. 67–76. Cunningham, E.P., 1999. The application of biotechnologies to enhance animal production in different farming systems. Livestock Production Science 58, 1–24. Dooley, A.E., 2002. Milk Segregation on Dairy Farms. Unpublished Ph.D. Thesis, Massey University, Palmerston North, New Zealand. Fenwick, R.M., Marshall, K.R., 1991. What sort of milk do we need for the 1990Õs? In: Proceedings of the 43rd Ruakura FarmersÕ Conference, pp. 1–7. Frank, B.R., 1997. Adoption of innovations in the North Queensland beef industry. III: Implications for extension management. Agricultural Systems 55 (3), 347–358. Gibson, J.P., 1989. Altering milk composition through genetic selection. Journal of Dairy Science 72, 2815–2825. Gibson, J.P., 1991. The potential for genetic change in milk fat composition. Journal of Dairy Science 74, 3258–3266. Gibson, J.P., 2000. Options for genetic improvement of milk composition. In: Agnew, R.E., Agnew, K.W., Fearon, A.M., (Eds.), Milk Composition. Occasional Publication No. 25. British Society of Animal Science, Edinburgh, pp. 109–118. Gray, D.I., Parker, W.J., 1990. The Planning, Implementation and Control of Pastoral Based Systems. Agricultural and Horticultural Systems Management, Massey University, Palmerston North, New Zealand.
96
A.E. Dooley et al. / Agricultural Systems 85 (2005) 82–97
Guerin, L.J., Guerin, T.F., 1994. Constraints to the adoption of innovations in agricultural research and environmental management: a review. Australian Journal of Experimental Agriculture 34, 549–571. Hardaker, J.B., Huirne, R.B.M., Andereson, J.R., 1997. Coping with Risk in Agriculture. CAB International, Wallingford, UK. Hill, J.P., Thresher, W.C., Boland, M.J., Creamer, L.K., Anema, S.G., Manderson, G., Otter, D.E., Paterson, G.R., Lowe, R., Burr, R.G., Motion, R.L., Winkelman, A., Wickham, B., 1997. The polymorphism of the milk protein – lactoglobulin: a review. In: Welch, R.A.S., Burns, D.J.W., Davis, A.I., Popay, A.I., Prosser, C.G. (Eds.), Milk Composition, Production and Biotechnology. CAB International, Wallingford, England, pp. 173–202. Hobman, P., 1992. The influence of the market on industry R&D. In: Proceedings of the 44th Ruakura FarmersÕ Conference, pp. 57–60. Holmes, C.W., 1989. Effects of feeding on the composition of milk and on the yields of its components. In: Barrell, G.K., (Ed.), Milk Payment and Quality. Animal Industries Workshop, Lincoln College, Canterbury, New Zealand, pp. 43–53. Holmes, C.W., Wilson, G.F., Mackenzie, D.D.S., Flux, D.S., Brookes, I.M., Davey, A.W.F., 1984. Milk Production From Pasture. Butterworths Agricultural Books, Wellington, New Zealand. Karatzas, C.N., Turner, J.D., 1997. Toward altering milk composition by genetic manipulation: current status and challenges. Journal of Dairy Science 80, 2225–2232. Keen, A.R., 1986. Future developments for profitable dairy farming: The colour and flavour of dairy products. Proceedings of the Massey University Dairyfarming Annual 38, 36–39. Keen, A.R., Wilson, R.D., 1993. Effect of breed on colour and flavour. In: New Zealand Dairy Research Institute Milkfat Flavour Forum. Summary of proceedings 3–4 March 1992. New Zealand Dairy Research Institute, Palmerston North, New Zealand, pp. 50–54. Kennelly, J.J., Glimm, D.R., 1998. The biological potential to alter the composition of milk. Canadian Journal of Animal Science 78 (Suppl.), 23–56. Kolver, E.S., Bryant, A.M., 1992. Changing milk composition by feeding. In: Proceedings of the 44th Ruakura FarmersÕ Conference, pp. 61–71. Larsen, W., 1997. Speech to Ruakura farmersÕ conference. In: Proceedings of the 49th Ruakura FarmersÕ Conference, pp. 1–9. LÕHuillier, P., 1999. Genetic modification of livestock for the production of therapeutics and designer foods. Proceedings of the New Zealand Grassland Association 61, 109–115. LIC, 2000. Dairy Statistics 1999–2000. Livestock Improvement Corporation Ltd., Hamilton, New Zealand. Mackle, T.R., Bryant, A.M., 1995. WhatÕs in milk? In: Proceedings of the 47th Ruakura FarmersÕ Conference, pp. 63–70. Mackle, T.R., Bryant, A.M., 1996. Feeding, milk composition and product quality – an overview. Proceedings of the New Zealand Grassland Association 57, 167–172. Newman, S.-A., Knight, T.W., OÕNeill, A.F., Death, A.F., Ridland, M., 1994. Effect of breed on plasma carotene concentration in New Zealand dairy heifers. Proceedings of the New Zealand Society of Animal Production 54, 119–120. Ng-Kwai-Hang, K.F., 1998. Genetic polymorphism of milk proteins: Relationships with production traits, milk composition and technological properties. Canadian Journal of Animal Science 78 (Suppl.), 131–147. Nickerson, S.C., 1995. Milk production: factors affecting milk composition. In: Harding, F. (Ed.), Milk Quality. Chapman and Hall, Glasgow, Great Britain, pp. 3–24. OÕBrien, B., 2000. Composition and processability of bovine milk from herds with different calving patterns. In: Agnew, R.E., Agnew, K.W., Fearon, A.M., (Eds.), Milk Composition. Occasional Publication No. 25. British Society of Animal Science, Edinburgh, pp. 77–86. Palisade Corporation, 1998. Guide to Evolver: The Genetic Algorithm Solver for Microsoft Excel. Palisade Corporation, New York, USA. Rogers, E.M., 1983. Diffusion of Innovations, third ed. The Free Press, MacMillan Publishing Ltd., New York, USA. Ryan, L., 1996. Current research into future milk quality. Proceedings of the Massey University Dairyfarming Annual 48, 191–194.
A.E. Dooley et al. / Agricultural Systems 85 (2005) 82–97
97
Scrimgeour, F., 1999. Future directions for the dairy industry. Proceedings of the Massey University Dairyfarming Annual 51, 9–16. Sutton, J.D., Morant, S.V., 1989. A review of the potential of nutrition to modify milk fat and protein. Livestock Production Science 23, 219–237. Thomson, N.A., 1993. Dairy farm production/manufacturing of dairy products – lets come closer together. In: New Zealand Dairy Research Institute Milkfat Flavour Forum. Summary of proceedings 3–4 March 1992, New Zealand Dairy Research Institute, Palmerston North, New Zealand, pp. 17–23. Van Vleck, L.D., Pollack, E.J., Oltenacu, E.A.B., 1987. Genetics For The Animal Sciences. W.H. Freeman and Company, New York. Visscher, P.M., Van der Beek, S., Haley, C.S., 1998. Marker assisted selection. In: Clark, A.J. (Ed.), Animal Breeding: Technology for the 21st Century. Harwood Academic Publishers, Amsterdam, Netherlands, pp. 119–136. Wall, R.J., 1999. Biotechnology for the production of modified and innovative animal products: transgenic livestock bioreactors. Livestock Production Science 59, 243–255. Wells, D.N., Misica, P.M., Tervit, H.R., 1998. Future opportunities in livestock production and biomedicine from advances in animal cloning. Proceedings of the New Zealand Society of Animal Production 58, 32–35. Winkelman, A.M., Johnson, D.L., MacGibbon, A.K.H., 1999. Estimation of heritabilities and correlations associated with milk color traits. Journal of Dairy Science 82, 215–224. Winkelman, A.M., Wickham, B.W., 1996. Association between milk protein variants and production traits in New Zealand dairy cattle. Proceedings of the New Zealand Society of Animal Production 56, 24–27.
Implications of on-farm segregation for valuable milk characteristics A.E. Dooley a
a,*
, W.J. Parker a, H.T. Blair b, E.M. Hurley
c
AgSystems, AgResearch Limited, Ruakura Research Centre, Private Bag 3123, Hamilton, New Zealand b Massey University, Private Bag 11 222, Palmerston North, New Zealand c 26 Springdale Grove, Palmerston North, New Zealand Received 17 April 2003; received in revised form 2 June 2004; accepted 14 July 2004
Abstract Milk composition varies between herds and cows within herds, enabling its segregation on farm, rather than during processing, for the manufacture of specific dairy products. Benefits may include increased product yields, reduced processing costs and greater suitability of differentiated milk for the production of high value niche market products. However, costs are also likely to be greater. An integrated spreadsheet model was developed to determine the break-even premium required for a farmer with a seasonal calving herd to be economically no worse off producing segregated than conventional milk. The model incorporated breeding (quantitative and qualitative traits), cow requirements and feeding, transport, and economic sub-models. Cows were segregated within herd and milk composition was altered over time by genetic selection. Four quantitative trait (‘‘white’’ milk colour) scenarios and two qualitative trait (BB b-lactoglobulin milk) scenarios were considered. The model suggested that ‘‘white’’ milkfat would need to earn 38.4% more at the farm gate than conventional milkfat for the two systems to break even. ‘‘White’’ milk cows produced less than their status quo counterparts due to the reduced selection pressure on production milk traits and this had a considerable impact on the premium, as did the low initial volumes of white milkfat. The difference in production between the B b-lactoglobulin cows and their status quo counterparts was less than for selection on white milkfat only. The high risk to farmers of discontinuing a differentiated milk policy could be moderated by changing the structure of
*
Corresponding author. Tel.: +64 7 838 5914; fax: +64 7 838 5012. E-mail address: [email protected] (A.E. Dooley).
0308-521X/$ - see front matter Ó 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.agsy.2004.07.012
A.E. Dooley et al. / Agricultural Systems 85 (2005) 82–97
83
premium payments over time. Hence, processing companies and farmers will need to work together to facilitate the uptake of milk segregation. This research model could be applied by dairy companies and farmers considering milk segregation policies. Ó 2004 Elsevier Ltd. All rights reserved. Keywords: Dairy industry; Milk characteristics; Milk payment; Modelling; Genetic improvement
1. Introduction New Zealand has traditionally exported bulk commodity dairy products. However, to remain competitive in the international market there has been a steady move towards producing more branded and value-added dairy products aimed at niche markets. These products require greater emphasis on meeting customer specifications for quality and consistency than commodity products (Larsen, 1997; Scrimgeour, 1999). Product type and quality, and the amount of product able to be produced (and for some products the timing of manufacture) are all affected by milk composition and quality, as are dairy industry returns through the product mix made (Fenwick and Marshall, 1991; Hobman, 1992; Mackle and Bryant, 1995; Ryan, 1996). Natural variation exists between cows and herds in the many proteins, fats, carbohydrates, vitamins and micro-elements that constitute milk (Crabtree, 1984; Nickerson, 1995). An opportunity exists for dairy processing companies to take advantage of this natural variation by ensuring milk from different cows, herds or regions is kept separate until processing can commence at the factory. Cows with specialist characteristics could either be aggregated onto particular farms or into different herds on the same farm. Milk has not been segregated on-farm in New Zealand until very recently, and then only on a very small scale for a few niche products. New Zealand has largely supplied commodity products, and farmers have been paid on milk volume, and milkfat and protein production. There has been little work done on altering aspects of milk for traits other than these. It has been assumed that any other changes required in milk can be achieved by various processing techniques (Thomson, 1993). However, greater interest in developing niche markets and better meeting customer requirements has lead to interest in the possibility of on-farm milk segregation. Farmers could modify milk characteristics by management (e.g., calving spread, replacement policies), genetic selection and feeding (Crabtree, 1984; Holmes et al., 1984; Gibson, 1989, 1991, 2000; Holmes, 1989; Sutton and Morant, 1989; Kolver and Bryant, 1992; Nickerson, 1995; Mackle and Bryant, 1995, 1996; Auldist et al., 1997; Kennelly and Glimm, 1998; OÕBrien, 2000). Selection of cows based on a particular milk trait will affect correlated milk characteristics, and environmental factors, such as nutrition, cow age, stage of lactation may further affect these. Opportunities to modify milk also exist through the use of new genetic and reproductive technologies (Karatzas and Turner, 1997; Cunningham, 1998, 1999; Visscher et al., 1998; Wells et al., 1998; LÕHuillier, 1999; Wall, 1999).
84
A.E. Dooley et al. / Agricultural Systems 85 (2005) 82–97
A further on-farm implication is that some cows may be traded on their ability to produce milk of a particular type, rather than solely on breeding worth. To achieve this, cows and sires will need to be tested for the desired milk components, and selection of herd replacements may be based on customised breeding indices. Farmers will require a financial incentive before they supply dairy companies with segregated milk. Extra on-farm costs may be incurred in the production and storage of specialised milk, and greater transport and processing costs may be incurred because of the need for milk tanker re-routing and separation of milk at the factory. The primary purpose of this paper is to investigate the impact of milk segregation policies on farm and discuss implications to the farmer and the industry. The implications of milk separation are numerous and complex, and to systematically assess these, a model of biological and economic effects of milk segregation was developed. This provided a practical and inexpensive means to evaluate the impact on the system, particularly as breeding was involved and changes took considerable time (years). Two example milk traits, B b-lactoglobulin and milk colour, were evaluated with the model. Volumes and average levels for specified milk characteristics for the different lots of milk due to biological variation were defined, and the management and financial implications associated with milk segregation considered. The following sections describe the scenarios modelled and provide a brief description of each of the sub-models i.e., their function and the primary inputs and outputs. A comprehensive description of each of the sub-models, not provided in this paper, is presented in Dooley (2002).
2. Milk segregation scenarios Traits can be altered by breeding (Van Vleck et al., 1987). Traits which involve many loci and a graduation of phenotypes are described as quantitative traits. Traits controlled by few loci, so that one allele affects the phenotype are described as qualitative traits. Qualitative breeding trait genes can affect production directly or act as marker genes for production traits. Four quantitative scenarios (breeding and segregating cows with whiter milkfat colour) and two quantitative scenarios (breeding or selecting cows with a BB b-lactoglobulin genotype) were investigated with the model (Table 1). Although milk is not currently segregated for these traits in New Zealand, they were selected to evaluate the economic implications of milk segregation and demonstrate the model because they have known product advantages. New Zealand milkfat products tend to be more yellow than those of other countries because of our pasture-based farming systems (Keen, 1986) resulting in higher levels of b-carotene in the milkfat. Whiter milkfat (4.5 mg b-carotene/kg milkfat or less (MacGibbon, 2001; pers. comm.)) products (e.g., butter, ghee, white cheeses and whipped toppings) are preferred by consumers in some markets (Keen, 1986; Hobman, 1992; Keen and Wilson, 1993). The model segregated milk from cows with a ‘‘whiter milkfat colour’’ from that of the remaining cows.
A.E. Dooley et al. / Agricultural Systems 85 (2005) 82–97
85
Table 1 Description of the scenarios modelled Scenario Quantitative Scenario 1 Scenario 2 Scenario 3 Scenario 4
Qualitative Scenario 1 Scenario 2
Description Selection on colour only, with an initial average colour of 8.0 mg b-carotene/kg milkfat Selection on colour and milkfat yield, with an initial colour of 8.0 mg b-carotene/kg milkfat Selection on colour only as for scenario 1, but with an initial colour of 6.0 mg bcarotene/kg milkfat As for scenario 1 for the first 10 years, then reverting to status quo values for genetic gain and milk prices for the remaining years
Changing to a BB herd over time by breeding Changing to a BB herd in year one by buying and selling cows
The qualitative scenarios considered breeding and/or segregating cows with a BB b-lactoglobulin genotype. Milk from cows of this genotype is superior to milk from cows with AA or AB b-lactoglobulin genotypes for cheesemaking, milk powder production and UHT milk production (Hill et al., 1997; Ng-Kwai-Hang, 1998). BB b-lactoglobulin genotype cows have higher casein and lower whey protein production (including b-lactoglobulin). Some trials have also shown higher milkfat concentration and production, lower total protein production and lower milk yields in these cows compared to AA b-lactoglobulin cows (Hill et al., 1997). Breeding or management policies could include: initial testing of the cows and potential ABheterozygous/B-homozygous calves for genotype to identify the B-homozygous animals to breed and segregate milk from; testing and sale and purchase of cows in year one so all cows are B-homozygous from year 1; and no testing (use of homozygous bulls only and all milk kept in one lot). The first two scenarios are presented in this paper. The initial amount of milk able to meet specifications and the rate of gain in the desired trait are affected by the initial levels in the population. For the quantitative milk trait, cows were assumed to be Friesians which have whiter milk than Jerseys (Keen and Wilson, 1993; Newman et al., 1994; Winkelman et al., 1999). For the qualitative trait, the herd was assumed to be Jerseys, which have a higher frequency of the B b-lactoglobulin allele than Friesians (Winkelman and Wickham, 1996). Status quo herds were modelled for both quantitative and qualitative scenarios assuming selection on production traits only and industry rates of genetic gain. Selection on other traits (milkfat colour and b-lactoglobulin) reduced genetic gain in the production traits, particularly for the quantitative trait (colour). Breeding (extra NZ$5/cow) and herd testing (extra NZ$2/cow tested for colour, and NZ$24.50/ cow tested for b-lactoglobulin genotype) costs were greater for segregated than status quo milk herds. Animals only needed testing once in the qualitative scenarios.
86
A.E. Dooley et al. / Agricultural Systems 85 (2005) 82–97
The whole herd plus replacements needed to be tested in year 1 for qualitative scenario 1 (1.25 tests/cow allowing for replacements), with some calves only in subsequent years. The second qualitative scenario considered buying and selling cows so all were BB genotype in year one. This required testing 2.14 cows for each animal in the herd to identify sufficient BB cows, and replacing 54% of the herd paying a premium price for those purchased (NZ$200/cow and NZ$150/heifer more than the price received for cows sold). An extra milk vat was purchased in year one (NZ$5000). An average size New Zealand dairy farm (93 ha, LIC, 2000) with Manawatu region pasture growth rates was assumed for all scenarios. Variable per cow costs for animal health, breeding, herd testing, dairy expenses and power were assumed to be NZ$119.76. Feed costs (e.g., silage and hay making, calf rearing, grazing off of young stock, freight to grazing) and stock income would have differed between stocking rates. Status quo value milk prices were average New Zealand values for the 1999/2000, 2000/2001 and 2001/2002 seasons at NZ$2.894/kg and NZ$6.241/kg for milkfat and protein, respectively and a deduction of NZ$0.0415/L milk volume. An additional milk volume charge of one cent per litre was assumed to be incurred on segregated milk.
3. Model description A deterministic simulation model was developed to derive the break-even premium for a dairy farmer, ceteris paribus, to segregate milk for a specific trait or processing characteristic on-farm. The break-even premium calculated required that a farmer be no worse off financially if a segregated milk policy was adopted. A premium was received on the milkfat component which met the criteria value of 4.5 mg b-carotene/kg milkfat or less (MacGibbon, 2001 pers. comm.) for the quantitative trait. For the qualitative trait, a premium was received on both the milkfat and the protein components of the BB b-lactoglobulin milk. The sensitivity of premiums to discount rates and transport costs was explored, as was the effect of redistributing the milk premium through time (Dooley, 2002). Changes required to alter a herdÕs milk composition can take time, often years, particularly if breeding is involved. The model therefore incorporated the changes over a 20-year time frame. A cost-benefit analysis using discounted cash flows was used to calculate the break-even premium over periods of 10 and 20 years, and to infinity. Fig. 1 shows the model design and the main input–output flows between the submodels. The model was developed using Microsoft Excel and Visual Basic software and was run for a 20-year period. Although milk processing was not modelled, this is shown diagrammatically to illustrate how it fits into the overall scheme, and to indicate the input and output variables between this and the other sub-models. All submodels contributed variables to the overall model – the farm economic model. This model was used to calculate the premium required for a range of comparative scenarios.
A.E. Dooley et al. / Agricultural Systems 85 (2005) 82–97 Breeding Models
Proportion of cows in each herd
87
Proportion of cows in each herd
Potential production by cow “type” or herd
Production/cow/day by herd and period
Cow Requirements and Production
Energy value of the feed by period
Feed required per cow by herd and period
Feed Budget
Cow numbers
Cow numbers
Milk Income Model
Cow numbers. Costs of feed made & purchased
Milk income
Farm Economic Model
Stock Reconciliation
Sale & purchase costs. Grazing, freight & calf costs
Production/herd/day
OFF-FARM
Milk income
Transport Cost Model
Volume charge
ON-FARM
Milk Income Model
Volume charge
Processing Model
Pricing of components
Fig. 1. Model design and information flow. The milk income model is shown twice (on-farm and off-farm aspects) to simplify the diagram.
The farm production and economic sub-models incorporated breeding and feeding factors which affect milk quantity and composition, and farm profitability. The milk transport sub-model was used to assess the cost implications of differentiated milk collection logistics. Milk processing, distribution and marketing data are largely subject to commercial sensitivity and therefore unavailable, making inclusion of a processing sub-model impractical.
88
A.E. Dooley et al. / Agricultural Systems 85 (2005) 82–97
3.1. Breeding sub-models Quantitative and qualitative breeding sub-models were developed (Dooley, 2002) to evaluate selection for milkfat colour and b-lactoglobulin, respectively. Inputs for both breeding sub-models included base production levels, herd structure, age adjustments, and changes in genetic gain over time. Annual outputs for both submodels included the proportion of the cows in each herd, the average (across age groups) potential milk, milkfat and protein production, and liveweight for each herd (i.e., the differentiated herd and the standard milk herd). Inputs specific to the quantitative breeding model included phenotypic correlations and standard deviations for selection traits, and the cut-off criterion for milkfat colour (average milkfat colour required in the white herd based on a set lactation length). Quantitative breeding model outputs included the actual cut-off value to meet the criterion (minimum average milkfat colour required for cows to be in the white herd) and the average milkfat colour for each herd. As milkfat colour varies over lactation, the cut-off criterion and actual cut-off values related to the average milkfat colour over the lactation. Cows were assumed to remain in their allocated herd for the entire lactation. Inputs specific to the qualitative breeding model were the frequency of the B allele in the initial population and the genotype of the bulls used. Annual outputs specific to the qualitative breeding model included: the genetic frequency and the change in frequency over time, the production of each milk type allowing for age adjustments and differences in genetic merit between age groups; and if all milk is stored in the same vat, the proportion of B b-lactoglobulin to A b-lactoglobulin in the milk. 3.2. Cow requirements and production sub-model Feed requirements and production per cow, over a year divided into 24 periods, were modelled on the basis of energy using the template of Brookes et al. (1993). The original model was altered to allow for growth in younger cows and changes in milk composition (milk, milkfat, protein and milkfat colour) over the lactation. Stage of lactation and feeding affect milkfat colour over the lactation and these effects are often confounded due the New ZealandÕs seasonal calving pattern and pastoral based feeding system (Keen, 1986; Keen and Wilson, 1993; Winkelman et al., 1999). Feed requirements were adjusted for culling and spread of lactation effects. 3.3. Feed budget sub-model Stocking rate was derived from a pasture-based farm feed budget (Gray and Parker, 1990; Brookes et al., 1993) using per cow requirements as an input. Young stock were assumed to be grazed off or farmed separately from the milking herd. Production per herd (allowing two herds per farm), and per farm was then calculated for each of the 24 periods in the year. Herd inputs specified included the composition of the diet and the energy value of the components, intake per cow (based on production), and the proportion of cows
A.E. Dooley et al. / Agricultural Systems 85 (2005) 82–97
89
in each herd. Farm inputs included pasture growth rate and utilisation; maximum and minimum pasture covers allowed over the year (Holmes et al., 1984); effective area of the milking platform (area of the farm used for the milking herd); area in crops and crop production; and pasture grown from nitrogen applied. When combined, the farm inputs defined the herdÕs feed supply. Model outputs for each of the 20 years included: stocking rate and number of cows, opening cover, grass silage fed and made, hay fed and made, hay, silage and concentrates purchased, hectares (ha) of crop grown, pasture production per ha per year and milk production as described above. Feed and grazing costs were derived within the feed budget model. 3.4. Stock reconciliation sub-model Livestock purchased and sold over the 20 years were derived from an annual stock reconciliation. Inputs included calving percentage, herd age structure, deaths, and specified sales and purchases. Sales (culls) and purchases (to make up numbers) of the different livestock classes were calculated. Livestock related costs and income (e.g., sales of cull cows and bobby calves, calf rearing costs and stock freight costs), and capital value of the stock were derived within the stock reconciliation model. 3.5. Milk income sub-model Annual milk income from a farm was calculated over the 20-year period taking both the milk payment system and time-related adjustment factors into consideration. Prices were entered for the components in the standard New Zealand payment formula, milk yield (NZ$/L,negative value), milkfat (NZ$/kg), and protein (NZ$/kg), and another trait if required (i.e., for differentiated milk). Factors affecting milk price for each component were herd, year, month and criteria value as described in Table 2. The criterion value specified for the differentiated milk trait (e.g., milkfat colour) determined whether this milk received a premium, with milk from the differentiated herd meeting or exceeding this value (e.g., whiter) receiving a premium. This takes into account factors that affect the colour of the differentiated milk over the lactation e.g., stage of lactation, feeding effects. This criterion value may differ from the cut-off Table 2 Factors affecting prices received for milk Factor
Description and inputs
Herd Year
Status quo herd and differentiated milk herd Up to three sets of prices able to be specified over the 20 years e.g., years 1–7, 8–14 and 15–20 For each month, milk is specified as receiving either peak milk prices, or off-peak milk prices Differentiated milk herd only. Criterion value specified for each year e.g., milkfat colour required for milk to receive a premium
Month Criteria to be met for the differentiated trait
90
A.E. Dooley et al. / Agricultural Systems 85 (2005) 82–97
criterion specified in the breeding model, which is the average for the lactation. The breeding model cut-off criterion may be set higher than the criterion value specified here to ensure milk receives a premium over the whole year. If the same value is used in both sub-models, milk will only receive a premium for the part of the year when it is whiter than average. Other inputs included the number, and proportion, of cows in each of the two herds for each of the 20 years, the discount rate and farm size (effective hectares). The production per cow per period for each herd was read in from the cow requirements model. Annual milk income per herd per year was read into the farm economic model. 3.6. Farm economic sub-model A cost-benefit analysis was used to calculate the difference between segregated and status quo scenarios, the present value (PV) of the income less costs for each year and the net present value (NPV) over 10 years, 20 years and infinity (20+ years). Discount rates of 7% and 10% were used. The break-even value required for differentiated milk is the value where the NPV for the segregated and status quo scenarios was the same. All cows in the status quo scenario were assumed to be in one herd. Segregated scenarios could have two herds (a status quo milk type and a differentiated milk type) in all or some years. Income, cow-related variable costs and capital costs were calculated for each of the herds. A partial budget approach was used, assuming fixed costs, not directly affected by the policy, remained constant between the policy options. 3.7. Transport sub-model A transport model was developed to estimate the likely increased cost of milk collection for segregated milk compared to non-segregated milk. The imputed value for segregated milk was entered into the milk income model. Milk collection was modelled where none, 25%, 50% and all farms changed to another milk type over a 20year period. Two contrasting regions (30 farms each) with different farm sizes and roading networks (collection areas) were simulated, at two stages of the spring-calving seasonal lactation (peak milk flow and late lactation). The model was developed using Microsoft Excel spreadsheets, and the Evolver genetic algorithm software (Palisade Corporation, 1998) was applied to search for the order of the farm milk collection which gave an optimal, least cost solution for milk collection for each prescribed set of inputs.
4. Results and discussion The six policies, and the premiums required for these differentiated milk alternatives to break even with the status quo over 10, 20 and 20+ years (infinity) years are summarised in Table 3.
A.E. Dooley et al. / Agricultural Systems 85 (2005) 82–97
91
Table 3 The status quo prices and the premium required per kilogram of protein and milkfat for each scenario, for the NPV to equal zero over 10, 20 and 20+ years Scenario
Component
Break-even period 10 Years
20 Years
20+ Years
Status quo prices
Milkfat Protein
$2.89 $6.24
$2.89 $6.24
$2.89 $6.24
2. Selection on colour and milkfat yield, initial colour 8.0 3. Selection on colour only as for 1, but with initial colour 6.0 4. As for no. 1 for first 10 years, then reverting to SQ values for genetic gain and prices paid
Milkfat Protein Milkfat Protein Milkfat Protein Milkfat Protein
$2.08 – $2.10 – $0.41 – $2.08 –
$1.11 – $1.08 – $0.46 – $5.62 –
$0.92 – $0.87 – $0.47 – $9.66 –
Qualitative scenarios 1. Changing to a BB herd over time by testing and selection 2. Changing to a BB herd in year 1 by testing, and stock purchase
Milkfat Protein Milkfat Protein
$0.113 $0.243 $0.139 $0.300
$0.098 $0.212 $0.119 $0.256
$0.093 $0.200 $0.107 $0.231
Premium required Quantitative scenarios 1. Selection on colour only, initial colour 8.0
Prices are in New Zealand dollars.
The proportion of the total milk attracting a premium increased over time for all scenarios except qualitative scenario 2, with only small volumes of milk attracting a premium in the early years for some of the scenarios (Fig. 2). Quantitative scenario 4 is the same as quantitative scenario 1 up until year 10, after which no milk attracts a premium. The low volumes contributed to the high premium required for some of the scenarios, particularly quantitative scenarios 1, 2 and 4, where only 2.4% of the milk met the specifications for differentiated milk in year one. By contrast, the policy of buying and selling cows in qualitative scenario 2, meant all milk met the requirements in year one. Changes in milk production and composition between the status quo and the scenarios for the qualitative trait were small. Initial cow numbers for all scenarios was 329. Initial milksolids (milkfat plus protein) production for the status quo and scenario 1 was 307.4 kg/cow (307.2 kg/cow for scenario 2), and 1087 kg/ha (all scenarios), with a protein to fat ratio of 0.7557 (0.7502 for scenario 2). In year 20 there was very little difference in production between the status quo and the scenarios (383 kg milksolids cow or 1433 kg milksolids per ha), and no difference in cow numbers (348 cows). Protein to fat ratio was slightly lower for the scenarios than the status quo (0.826 compared to 0.832) after 20 years. In contrast, the quantitative scenarios had greater changes in cow numbers, and production over the 20 years (Table 4). In the first year, all scenarios had 293 cows.
92
A.E. Dooley et al. / Agricultural Systems 85 (2005) 82–97
% of milk attractinga premium
100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 1 2
3 4
5 6 7
8 9 10 11 12 13 14 15 16 17 18 19 20
Year MF8, colour
MF8, colour+MF
MF6, colour
BB milk
Fig. 2. The percentage of milk attracting a premium over the 20 years for three quantitative scenarios (initial milkfat of colour 6 or 8 b-carotene/kg milkfat, selected on colour only or colour and milkfat) and the first qualitative scenario (BB b-lactoglobulin milk).
Milksolids production was 317.6 kg/cow and 1001 kg/ha, with a protein to fat ratio of 0.7867. Milkfat production was 177.8 kg/cow and 560 kg/ha. The reduction in per cow production for the milkfat and protein components (Table 4) contributed to the high premium required under the quantitative scenarios. Shifting some or all selection emphasis to reducing colour slowed genetic progress in the production traits relative to the status quo. Lower production per cow resulted in a higher stocking rate but lower production per ha with the difference increasing over time. For example, under quantitative scenario 1, 5.5% more cows, and 5.1% less milksolids which included 1.1% less milkfat per ha was produced compared to the status quo after 20 years. Therefore, the premium had to compensate for reduced milksolids income as well as increased costs compared to the status quo. The premium required to break even decreased as the time period increased for most scenarios (Table 3) because of the larger volumes of differentiated milk produced and therefore higher returns in the later years. However, quantitative scenarios 3, with an already low premium, required a slight increase in premium to break even (from 14.2% more for a 10 year period, to 16.3% more for 20 and 20+ years), because the increased income from a greater proportion of white milk in later years did not fully compensate for the overall lower production in these years and the higher costs incurred relative to the status quo. For quantitative scenario 4, a large premium was required to break even over longer time periods (Table 3, e.g., 334% more than the status quo for milkfat for the 20+ year period) because of the low volume of differentiated milk produced in the first 10 years. Even though selection changed to the status quo objective after 10 years, it would still take years for the
A.E. Dooley et al. / Agricultural Systems 85 (2005) 82–97
93
Table 4 Cow numbers, and milksolids and milkfat production per cow and per ha for year 20 for the quantitative scenarios SQ
Scenario 1
Scenario 2
Scenario 4
319 +8.87%
336 +14.67% +5.33%
320 +9.22% +0.31%
331 +12.97% +3.76%
380.0 +19.65%
342.6 +7.87% 9.84%
368.18 +15.93% 3.11%
353.8 +11.40% 6.89%
1304 +30.27%
1238 +23.68% 5.06%
1267 +26.57% 2.84%
1259 +25.77% 3.45%
206.1 +15.92%
193.4 +8.77% 6.16%
210.4 +18.32% +2.09%
197.2 +10.91% 4.32%
Production/hectare kg MF/ha in year 20 Diff. over 20 years Diff. from SQ in year 20
707 +26.26%
699 +24.82% 1.13%
724 +29.29% +2.40%
702 +25.36% 0.70%
Protein:fat ratio in year 20
.8437
.7716
.7503
.7942
Cow number Cow number in year 20 Diff. over 20 years Diff. from SQ in year 20 Milksolids (MS) Production/cow kg MS cow in year 20 Diff. over 20 years Diff. from SQ in year 20 Production/hectare kg MS/ha in year 20 Diff. over 20 years Diff. from SQ in year 20 Milkfat (MF) Production/cow kg MF cow in year 20 Diff. over 20 years Diff. from SQ in year 20
The percentage change in these attributes over the 20 years, and between the status quo (SQ) and the other scenarios in year 20 are given. Note: Scenario 3 is the same as scenario 1.
production and stocking rate under scenario 4 to match that of the status quo herd. Therefore the premium received on the low volume of differentiated milk in the first 10 years had to compensate for the lower overall production over the 20+ years as well as the higher costs associated with the differentiated policy compared to the status quo. The results for scenario 4 highlight the downside risk (Hardaker et al., 1997) associated with the future loss of, or reduction in, the premium for differentiated milk because of other technology advances (e.g., a suitable processing technique to produce white milkfat could be developed), reduction in demand because of changes in consumer preferences, or oversupply of differentiated milk. The more risky farmers perceive a policy to be, the lower the likely technology uptake, requiring a higher premium (greater relative advantage) to offset this (Rogers, 1983). While the model
94
A.E. Dooley et al. / Agricultural Systems 85 (2005) 82–97
calculates a break-even premium to cover incremental costs over a given timeframe, in practice the premium is likely to be greater to compensate for some, or all, of the risk associated with a segregated milk policy in order to facilitate farmer adoption. A processing company would need to give these matters careful consideration when structuring its premium payments. In this regard, the amount and timing of supply, and therefore farmer adoption rate, is important because a critical volume of milk will be needed to fulfill market requirements. Company tactics might include identifying and targeting farmers who already produce milk with characteristics similar to that required, or those who are receptive to new technologies. Differentiated milk policies are not readily trialled or observed, may be complex, are not necessarily reversible, and can be risky, and as a result, farmer uptake of differentiated milk policies may be limited (Rogers, 1983). As the scenarios show, changes to a differentiated milk policy may take years to incorporate into the farming system, particularly where breeding is involved or there are extensive changes to management. Farmers will need access to information and decision support to enable them to make decisions and judgements about whether or not the technology is appropriate for them. The involvement of farmers as well as industry representatives, and research and extension researchers is likely to lead to the development of more effective strategies for a differentiated milk policy, and increases the likelihood of adoption (Guerin and Guerin, 1994; Frank, 1997). 4.1. Model application and evaluation The model has the potential to investigate the production and economic consequences of a wide range of scenarios for milk segregation on farm. Slower and expensive field trialling, if required, could then be restricted to ‘‘best bet’’ options. The model could be refined to better predict the effects of feeding on milk composition, and to include stochastic variables to assess the range and probability of different outcomes. Validation of the cow requirements model was by its developers and through onfarm applications. The feed budget, stock reconciliation and partial budget are standard farm management techniques, as is the construction of the genetic selection model. The transport model was assessed relative to actual performance data for milk collection and feedback from a dairy company transport manager. The aggregate model outputs were presented to a small group of processing company representatives and researchers, and they confirmed the reasonableness of the premium estimates and thus the modelÕs utility for developing milk segregation policies.
5. Conclusion On-farm milk segregation conceptually offers advantages to producers, processors and consumers. Implementation of milk segregation, however, is not straightforward as a myriad of factors impact on milk composition and quality, production costs and product returns. The model described in this paper integrates the various on-farm
A.E. Dooley et al. / Agricultural Systems 85 (2005) 82–97
95
components of milk composition and supply, plus any effects on transport costs to a processor. The added incremental costs associated with a milk segregation policy determine the premium required for a farmer to be no worse off relative to still pursuing a status quo milk production system. Scenarios modelled in this research indicate premiums, timeframes to achieve change and risks vary considerably between strategies for supplying specific milks. There are, however, new developments such as nutraceutical milk products, herd genotyping and reproductive technologies which will enhance the prospects of on-farm milk segregation. Acknowledgements We thank the many people who contributed information towards this study, particularly Dr. Ian Brookes and Dr. Nicolas Lopez-Villalobos from Massey University, Mr. Peter Spooner and Mr. Bob Franks from New Zealand Dairy Group, and Dr. Lawrie Creamer and Dr. Alastair MacGibbon from the New Zealand Dairy Research Institute.
References Auldist, M.J., Walsh, B.J., Thomson, N.A., 1997. Seasonal and lactational influences on milk composition in New Zealand. Proceedings of the New Zealand Society of Animal Production 57, 205–206. Brookes, I.M., Morris, S.T., Parker, W.J., 1993. Computer spreadsheets for predicting feed requirements and feed budgeting. Proceedings of the New Zealand Grassland Association 55, 209–210. Crabtree, R.M., 1984. Milk compositional ranges and trends. In: Castle, M.E., Gunn, R.G., (Eds.), Milk Compositional Quality and Its Importance in Future Markets. Occasional Publication No. 9. British Society of Animal Production, Edinburgh, pp. 35–42. Cunningham, E.P., 1998. The potential of new reproductive and genetic technologies. In: Einarsson, S., Rendel, J., (Eds), Impact of Reproductive Technology on Animal Breeding and Genetic Conservation. Acta Agriculturae Scandinavica, Section A, Animal Science, Supplementum 29. Scandinavian University Press, Oslo, Norway, pp. 67–76. Cunningham, E.P., 1999. The application of biotechnologies to enhance animal production in different farming systems. Livestock Production Science 58, 1–24. Dooley, A.E., 2002. Milk Segregation on Dairy Farms. Unpublished Ph.D. Thesis, Massey University, Palmerston North, New Zealand. Fenwick, R.M., Marshall, K.R., 1991. What sort of milk do we need for the 1990Õs? In: Proceedings of the 43rd Ruakura FarmersÕ Conference, pp. 1–7. Frank, B.R., 1997. Adoption of innovations in the North Queensland beef industry. III: Implications for extension management. Agricultural Systems 55 (3), 347–358. Gibson, J.P., 1989. Altering milk composition through genetic selection. Journal of Dairy Science 72, 2815–2825. Gibson, J.P., 1991. The potential for genetic change in milk fat composition. Journal of Dairy Science 74, 3258–3266. Gibson, J.P., 2000. Options for genetic improvement of milk composition. In: Agnew, R.E., Agnew, K.W., Fearon, A.M., (Eds.), Milk Composition. Occasional Publication No. 25. British Society of Animal Science, Edinburgh, pp. 109–118. Gray, D.I., Parker, W.J., 1990. The Planning, Implementation and Control of Pastoral Based Systems. Agricultural and Horticultural Systems Management, Massey University, Palmerston North, New Zealand.
96
A.E. Dooley et al. / Agricultural Systems 85 (2005) 82–97
Guerin, L.J., Guerin, T.F., 1994. Constraints to the adoption of innovations in agricultural research and environmental management: a review. Australian Journal of Experimental Agriculture 34, 549–571. Hardaker, J.B., Huirne, R.B.M., Andereson, J.R., 1997. Coping with Risk in Agriculture. CAB International, Wallingford, UK. Hill, J.P., Thresher, W.C., Boland, M.J., Creamer, L.K., Anema, S.G., Manderson, G., Otter, D.E., Paterson, G.R., Lowe, R., Burr, R.G., Motion, R.L., Winkelman, A., Wickham, B., 1997. The polymorphism of the milk protein – lactoglobulin: a review. In: Welch, R.A.S., Burns, D.J.W., Davis, A.I., Popay, A.I., Prosser, C.G. (Eds.), Milk Composition, Production and Biotechnology. CAB International, Wallingford, England, pp. 173–202. Hobman, P., 1992. The influence of the market on industry R&D. In: Proceedings of the 44th Ruakura FarmersÕ Conference, pp. 57–60. Holmes, C.W., 1989. Effects of feeding on the composition of milk and on the yields of its components. In: Barrell, G.K., (Ed.), Milk Payment and Quality. Animal Industries Workshop, Lincoln College, Canterbury, New Zealand, pp. 43–53. Holmes, C.W., Wilson, G.F., Mackenzie, D.D.S., Flux, D.S., Brookes, I.M., Davey, A.W.F., 1984. Milk Production From Pasture. Butterworths Agricultural Books, Wellington, New Zealand. Karatzas, C.N., Turner, J.D., 1997. Toward altering milk composition by genetic manipulation: current status and challenges. Journal of Dairy Science 80, 2225–2232. Keen, A.R., 1986. Future developments for profitable dairy farming: The colour and flavour of dairy products. Proceedings of the Massey University Dairyfarming Annual 38, 36–39. Keen, A.R., Wilson, R.D., 1993. Effect of breed on colour and flavour. In: New Zealand Dairy Research Institute Milkfat Flavour Forum. Summary of proceedings 3–4 March 1992. New Zealand Dairy Research Institute, Palmerston North, New Zealand, pp. 50–54. Kennelly, J.J., Glimm, D.R., 1998. The biological potential to alter the composition of milk. Canadian Journal of Animal Science 78 (Suppl.), 23–56. Kolver, E.S., Bryant, A.M., 1992. Changing milk composition by feeding. In: Proceedings of the 44th Ruakura FarmersÕ Conference, pp. 61–71. Larsen, W., 1997. Speech to Ruakura farmersÕ conference. In: Proceedings of the 49th Ruakura FarmersÕ Conference, pp. 1–9. LÕHuillier, P., 1999. Genetic modification of livestock for the production of therapeutics and designer foods. Proceedings of the New Zealand Grassland Association 61, 109–115. LIC, 2000. Dairy Statistics 1999–2000. Livestock Improvement Corporation Ltd., Hamilton, New Zealand. Mackle, T.R., Bryant, A.M., 1995. WhatÕs in milk? In: Proceedings of the 47th Ruakura FarmersÕ Conference, pp. 63–70. Mackle, T.R., Bryant, A.M., 1996. Feeding, milk composition and product quality – an overview. Proceedings of the New Zealand Grassland Association 57, 167–172. Newman, S.-A., Knight, T.W., OÕNeill, A.F., Death, A.F., Ridland, M., 1994. Effect of breed on plasma carotene concentration in New Zealand dairy heifers. Proceedings of the New Zealand Society of Animal Production 54, 119–120. Ng-Kwai-Hang, K.F., 1998. Genetic polymorphism of milk proteins: Relationships with production traits, milk composition and technological properties. Canadian Journal of Animal Science 78 (Suppl.), 131–147. Nickerson, S.C., 1995. Milk production: factors affecting milk composition. In: Harding, F. (Ed.), Milk Quality. Chapman and Hall, Glasgow, Great Britain, pp. 3–24. OÕBrien, B., 2000. Composition and processability of bovine milk from herds with different calving patterns. In: Agnew, R.E., Agnew, K.W., Fearon, A.M., (Eds.), Milk Composition. Occasional Publication No. 25. British Society of Animal Science, Edinburgh, pp. 77–86. Palisade Corporation, 1998. Guide to Evolver: The Genetic Algorithm Solver for Microsoft Excel. Palisade Corporation, New York, USA. Rogers, E.M., 1983. Diffusion of Innovations, third ed. The Free Press, MacMillan Publishing Ltd., New York, USA. Ryan, L., 1996. Current research into future milk quality. Proceedings of the Massey University Dairyfarming Annual 48, 191–194.
A.E. Dooley et al. / Agricultural Systems 85 (2005) 82–97
97
Scrimgeour, F., 1999. Future directions for the dairy industry. Proceedings of the Massey University Dairyfarming Annual 51, 9–16. Sutton, J.D., Morant, S.V., 1989. A review of the potential of nutrition to modify milk fat and protein. Livestock Production Science 23, 219–237. Thomson, N.A., 1993. Dairy farm production/manufacturing of dairy products – lets come closer together. In: New Zealand Dairy Research Institute Milkfat Flavour Forum. Summary of proceedings 3–4 March 1992, New Zealand Dairy Research Institute, Palmerston North, New Zealand, pp. 17–23. Van Vleck, L.D., Pollack, E.J., Oltenacu, E.A.B., 1987. Genetics For The Animal Sciences. W.H. Freeman and Company, New York. Visscher, P.M., Van der Beek, S., Haley, C.S., 1998. Marker assisted selection. In: Clark, A.J. (Ed.), Animal Breeding: Technology for the 21st Century. Harwood Academic Publishers, Amsterdam, Netherlands, pp. 119–136. Wall, R.J., 1999. Biotechnology for the production of modified and innovative animal products: transgenic livestock bioreactors. Livestock Production Science 59, 243–255. Wells, D.N., Misica, P.M., Tervit, H.R., 1998. Future opportunities in livestock production and biomedicine from advances in animal cloning. Proceedings of the New Zealand Society of Animal Production 58, 32–35. Winkelman, A.M., Johnson, D.L., MacGibbon, A.K.H., 1999. Estimation of heritabilities and correlations associated with milk color traits. Journal of Dairy Science 82, 215–224. Winkelman, A.M., Wickham, B.W., 1996. Association between milk protein variants and production traits in New Zealand dairy cattle. Proceedings of the New Zealand Society of Animal Production 56, 24–27.