Production and profitability of dairy farms producing milk with different concentrations of unsaturated fatty acids: a simulation study

ABSTRACT

This simulation study investigated the characteristics of farms that produced milk with a high concentration of unsaturated fatty acids (UFA) under New Zealand conditions. A stochastic farm model was used to simulate a population of 1,820,000 cows and 5600 dairy farms (AVE farms). From this population, the top 17,150 cows for UFA concentration in milk fat were segregated into 50 farms (UFA farms). The financial performance of dairy farms was evaluated at a milksolids payout of NZ$5.21/kg milksolids. The simulation was replicated 1000 times and confidence intervals were estimated by bootstrapping. Milk produced by UFA farms had a higher UFA concentration (35.15 vs 30.12 g/100 g fat, P < 0.05), but lower concentrations of fat (3.55% vs 4.33%, P < 0.05) and protein (3.19% vs 3.60%, P < 0.05) than milk produced by AVE farms. The operating profit of UFA farms was NZ$1324/ha lower than that of AVE farms. A premium for UFA concentration is necessary to compensate dairy farmers that invest in herd segregation to supply milk with high UFA concentration in fat.

KEYWORDS:

• Dairy cattle segregation
• milk fat composition
• New Zealand
• profit
• stochastic farm model
• unsaturated fatty acids

Introduction

In recent years there has been a growing interest in the manipulation of milk fat composition at the farm level, prompted by the trend in consumer demands towards food products with health benefits and convenience (Özer & Kirmaci 2010). The modification of milk fat composition to increase its concentration of unsaturated fatty acids (UFA) affects the processability of milk (e.g. spreadability of butter), its potential nutritional value (e.g. cis 9 trans 11 C18:2, omega 3 fatty acids, cis polyunsaturated fatty acids) and the organoleptic characteristics of dairy products (Smet et al. 2009; Augustin et al. 2013).

The concentration of UFA in milk fat is influenced by nutrition, genetics, stage of lactation, parity and energy status of the cow (Auldist et al. 1998; Thomson & Van-der-Poel 2000). Several studies have investigated the possibility of increasing milk fat UFA concentration through dietary manipulation (Oeffner et al. 2013; Sun et al. 2013), genetic selection (Heck et al. 2012; Marchitelli et al. 2013) and the segregation of dairy cows (Bobe et al. 2003; Thomson et al. 2003b; Chen et al. 2004).

Previous studies have reported negative genetic and phenotypic correlations between milk fat UFA concentration and milk fat yield, fat percentage and protein percentage (Soyeurt et al. 2007; Schennink et al. 2008; Stoop et al. 2008; Arnould & Soyeurt 2009). Protein yield per cow could also be negatively affected in dairy cows that produce fat with a high UFA concentration. Under the current New Zealand milk payment system, which rewards fat and protein yields and penalises milk volume (LIC 2015), these losses in production could negatively affect farm profit.

Currently little is known about the economic impact of increasing the milk fat UFA concentration on-farm. The objective of this study was to investigate the physical and financial characteristics of a group of farms formed by the segregation of dairy cows that produce milk with high UFA concentration under New Zealand conditions.

Materials and methods

The bio-economic stochastic farm model developed by Silva-Villacorta et al. (2016) was used to simulate the segregation of dairy cows to form herds that produce milk fat with high UFA concentration. The model, which accounts for all the inputs and outputs of a typical New Zealand dairy farm, uses the Cholesky decomposition algorithm of (co)variance matrices to simulate the daily performance of individual cows for milk yield, the percentages of milk fat and protein, live weight and milk fat UFA concentration. This study assumed that there was a small dairy production site interested in collecting and processing milk high in UFA.

Simulation of the base population

A population of 1,820,000 cows distributed in 5600 dairy farms (each 120 ha and 325 cows) was simulated and assumed to be located in the North Island of New Zealand (AVE farms). The number of cows and dairy farms simulated corresponded to approximately 60% of cows and 63% of farms present in the North Island of New Zealand (LIC 2015). This group of dairy farms constituted the base population from which some dairy cows were segregated.

For each farm, calving was assumed to start on 20 July, with the last cow calving by 10 October. A compact spring calving was simulated so that 50% and 90% of the herd calved within 22 and 46 days from the planned start of calving, respectively. Calving dates and calving pattern were obtained from LIC (2015) and Holmes et al. (2002). In each herd, all cows were dried off by 10 May the following year.

Herd structure and replacement rate

On each farm, herds were composed of 12 age classes: calves (<2 months old), R1 heifers (<1 year old), R2 heifers (<2 years old) and 2-, 3-, 4-, 5-, 6- 7-, 8- 9- and 10-year-old cows (cows in their first to ninth lactation). On all farms, survival rates considered for each age class were 0.66 (percentage of female calves raised as replacement), 0.86, 0.86, 0.86, 0.87, 0.86, 0.81, 0.77, 0.71, 0.66, 0.64 and 0, respectively (LIC 2015).

Culling rates were obtained from Lopez-Villalobos et al. (2000). Cows were culled due to age at the end of their ninth lactation. Culling rates due to death were 4%, 3% and 1.4% for R1, R2 and cows, respectively. Culling rates due to disease were 1%, 1% and 3.4% of R1, R2 and cows, respectively. The culling rate due to poor fertility was 8% for R2 and cows. The culling rate due to unsatisfactory performance was the difference between the culling rate of each age class minus the culling rates due to age, death, disease and poor fertility for the corresponding age class.

For each herd, it was assumed that 90% of cows (artificially inseminated) and R2 heifers (naturally mated) were pregnant at the end of the mating season (October to December). It was also assumed that 50% of calves born were females, that calf mortality rate was 6%, and that male calves and female calves not used as herd replacements were sold as bobby calves for veal (Lopez-Villalobos et al. 2000).

Feed supply

On each farm, it was assumed that 11 t of pasture (ryegrass–white clover) dry matter (DM) was eaten by the herd (76% utilisation) per hectare per year (MacDonald 2011). The monthly energy concentration of pasture from June to May of the following year was 10.9, 11.0, 11.1, 11.5, 11.4, 11.3, 11.2, 10.7, 10.5, 10.3, 10.9 and 11.0 MJ ME/kg DM, respectively (Litherland & Lambert 2007). It was assumed that cows met their ME requirements at all times and deficits in feed were filled with imported supplements made of conserved pasture (10 MJ ME/kg DM). Calves were fed milk (4 L/day, 3.6 MJ ME/L) and offered enough meal (12.5 MJ ME/kg DM) to meet their ME requirements. Heifers were raised off-farm and offered only pasture.

Economic inputs/performance

Income from the sale of milk was estimated using the New Zealand A + B – C milk pricing system, which pays for milk fat and protein, and penalises milk volume. The financial performance of simulated farms was investigated at a milk price of NZ$6.58/kg milksolids (10 year average of milk price in New Zealand), determined by the following values for milk components: NZ$4.63/kg fat, NZ$9.01/kg protein and −NZ$0.03/L milk (NZAEL 2013; LIC 2015).

Stock income (from sales of male calves, culled cows and surplus female calves and heifers) was determined by the carcass weight of cattle sold and the value of beef (Table 1). Carcass yield of calves and heifers was assumed to be 50% and 53% of live weight, respectively (Lopez-Villalobos et al. 2000). Carcass yield of culled cows was estimated as indicated by McCall & Marshall (1991):

Table 1. Beef prices assumed for culled cattle (NZAEL 2012).

Type of animal and carcass weight	Value of beef (NZ$/kg carcass)
Male and female calves
<13.5 kg	0.73
13.5–18.5 kg	1.09
>18.5 kg	2.54
Cows and heifers
<145 kg	2.25
145–170 kg	2.52
170–195 kg	2.67
196–220 kg	2.74
>220 kg	2.83

Live weight of culled calves, R1 and R2 heifers was assumed to be 8%, 40% and 70% of mature live weight, respectively (Clark & MacDonald 2007).

Gross farm income was determined by the sale of milk, the sale of stock and the sale of other dairy-related products (assumed to be NZ$46/ha, DairyNZ 2016). Farm expenses were estimated as the sum of the marginal expenses (Table 2) and feed expenses incurred on each farm.

Table 2. Marginal expenses (10 year average) per cow (DairyNZ 2016).

Item	NZ$/cow
Variable costs
Wages	209
Animal health	81
Breeding, herd improvement	45
Farm dairy	20
Electricity	35
Young stock grazing	125
Runoff lease	22
Vehicles, fuel	66
Repairs, maintenance	121
Freight	19
Overhead costs
Administration	40
Insurance	19
ACC	11
Rates	32
Depreciation	134
Operating profit adjustments
Labour adjustment	153
Runoff adjustment	28
Livestock adjustment	27
Total	1187

Note: ACC, the Accident Compensation Corporation.

Feed expenses were estimated based on the feed dry matter eaten by the herd (kg DM), assuming pasture and supplement prices of NZ$0.10/kg DM and NZ$0.25/kg DM, respectively (Holmes & Matthews 2001; Clearwater & Wright 2003).

Farm operating profit was estimated as the difference of gross farm income minus farm expenses.

Model inputs: segregation of dairy cows

The number of dairy cows to segregate and the number of dairy farms needed for the production of milk with high UFA concentration in fat was determined taking into consideration the market for this type of milk and the examples of other dairy companies in New Zealand that manufacture speciality dairy products. The present study assumed that there was a small dairy production site with 50 milk suppliers interested in the manufacture of dairy products with a higher than average concentration of UFA. The market for UFA milk and the example of other dairy companies in New Zealand that manufacture speciality dairy products were considered when determining the number of UFA farms to simulate (Campina 2007; DFO 2008; DairyNews 2011; Synlait 2013).

The stocking rate (cows/ha) of UFA farms, an input of the model, was determined by running different scenarios of stocking rate to identify one that matched their feed demand per hectare to that of AVE farms. From the population of AVE farms (1,820,000 cows in 5600 dairy farms), the top 17,150 cows for milk fat UFA concentration were segregated and randomly ‘purchased’ by 50 farms (120 ha and 343 cows each) for the production of milk with high UFA concentration in fat (UFA farms). The number of cows to segregate was determined by the number of UFA farms to simulate and a stocking rate of 2.86 cows/ha (this stocking rate matched the feed demand per hectare of UFA farms to that of AVE farms). It was assumed that the high UFA concentration in milk fat of segregated dairy cows was determined by genetic factors.

The cost of segregation (NZ$220/purchased cow) was estimated assuming that each UFA farm spent an additional NZ$200/purchased cow when replacing (sell/buy) its current herd by a herd of cows producing milk fat with high UFA concentration (Dooley et al. 2005), and that cattle freight costs were NZ$20/segregated cow (DairyNZ 2016). It was also assumed that milk fat UFA concentration, estimated by infrared spectrometry, was part of routine milk testing and did not contribute additional cost to the segregation of cows.

Statistical analysis

Physical and financial performance indicators of UFA farms and AVE farms were compared by bootstrapping methodology (Henderson 2005; Schmidheiny 2012). The simulation of AVE farms (base population) and the segregation of dairy cows producing milk high in UFA to form the 50 UFA farms were repeated 1000 times, so that each trait studied had 1000 bootstrap values. A 95% (α = 0.05) confidence interval was estimated using the percentile method of bootstrapping (Henderson 2005): once bootstrap values (, , … ) of a trait (θ) were sorted from smallest to largest (, , … ), the lower and upper confidence intervals of each trait corresponded to the 25th and 975th ordered elements [θ_25th, θ_975th], respectively.

Results

Physical characteristics of UFA farms

On a per cow basis, there were no significant differences in lactation length, milk yield, live weight and supplementary feed demand between UFA farms and AVE farms (Table 3). Cows in the UFA farms produced milk with significantly higher UFA concentration in fat, but lower concentrations of fat and protein, than cows in AVE farms. As a consequence, cows in the UFA farms produced less fat, protein and milksolids than cows in AVE farms. Milk supplied by UFA farms to a dairy processor had 15.0%–18.4% (average 16.7%) more UFA in fat than milk supplied by AVE farms (Figure 1).

Figure 1. Concentration (mean and 95% confidence interval) of unsaturated fatty acids (UFA) in milk fat produced by AVE farms (– – –) and UFA farms (—) during the dairy season.

Table 3. Physical characteristics (mean and 95% confidence interval) of AVE farms and UFA farms.

Trait	AVE farms (5600 farms)		UFA farms (50 farms)
Farm size (ha)^2	120		120
Herd size (cows)^2	325		343
Stocking rate (cows/ha)^2	2.71		2.86
Per cow production
Lactation length (days)	267	[266–268]	268	[267–268]
Milk yield (L)	4,351	[4267–4425]	4385	[4357–4414]
Milk fat yield (kg)	188^a	[184–191]	156^b	[155–157]
Protein yield (kg)	156^a	[153–160]	140^b	[139–140]
Milksolids yield (kg)^1	345^a	[337–350]	296^b	[294–298]
Fat (%)	4.33^a	[4.28–4.38]	3.55^b	[3.53–3.57]
Protein (%)	3.60^a	[3.57–3.63]	3.19^b	[3.18–3.20]
g UFA/100 g of milk fat	30.12^a	[29.92–30.32]	35.15^b	[35.10–35.20]
Live weight (kg)	491	[488–495]	489	[488–491]
Feed demand (t DM)	4.58^a	[4.52–4.63]	4.34^b	[4.33–4.36]
Supplement demand (t DM)	0.52^a	[0.46–0.57]	0.49^a	[0.47–0.51]
Per hectare
Milk yield (L)	11,788^a	[11,556–11,983]	12,534^b	[12,453–12,618]
Milk fat yield (kg)	510^a	[500–521]	445^b	[442–447]
Protein yield (kg)	423^a	[414–430]	399^b	[397–402]
Milksolids yield (kg)^1	935^a	[918–948]	845^b	[839–849]
Pasture utilised (t DM)^2	11		11
Supplement demand (t DM)	1.4	[1.2–1.5]	1.4	[1.3–1.5]
Total feed demand (t DM)	12.4	[12.2–12.5]	12.4	[12.3–12.5]

Notes: DM, dry matter. Means within a row with different superscript letter are significantly different (P < 0.05).

¹Milksolids = milk fat + protein.

²Model input.

Cows in UFA farms had lower feed demand than cows in AVE farms, but the greater stocking rate (+0.15 cows/ha) on the UFA farms resulted in the UFA farms utilising an amount of feed (t DM/ha) similar to that of AVE farms (Table 3). Due to their greater stocking rate, UFA farms produced more milk per hectare (+746 L/ha/year), and could supply a dairy processor with more milk per farm (+28 to +394 L/day) during the dairy production season (Figure 2), than AVE farms. Although UFA farms had a greater stocking rate, they produced less fat (−65 kg/ha/year), protein (−24 kg/ha/year) and milksolids (−90 kg/ha/year) and supplied less milksolids (−5 to −45 kg/day) to a dairy processor, than AVE farms (Table 3, Figure 3).

Figure 2. Milk production (mean and 95% confidence interval) of AVE farms (– – –) and UFA farms (—) during the dairy season.

Figure 3. Milksolids production (mean and 95% confidence interval) of AVE farms (– – –) and UFA farms (—) during the dairy season.

Financial characteristics of UFA farms

Table 4 shows the financial performance of UFA farms and AVE farms. Milk income and gross farm income were lower for UFA farms than AVE farms, both per cow and per hectare. There were no significant differences in stock income per cow between UFA farms and AVE farms.

Table 4. Financial characteristics (mean and 95% confidence interval) of AVE farms and UFA farms.

Trait	AVE farms		UFA farms
Per cow (NZ$)
Milk income^1	2142^a	[2096–2184]	1840^b	[1830–1849]
Stock income	184	[181–188]	185	[183–187]
Other income	16.98		16.09
Gross farm income	2343^a	[2298–2386]	2041^b	[2030–2050]
Marginal expenses^2	1187		1187
Feed expenses	534^a	[519–550]	508^b	[504–513]
Segregation expenses^2	0		220
Farm expenses	1721^a	[1706–1737]	1915^b	[1911–1920]
Operating profit	622^a	[590–650]	126^b	[119–133]
Per hectare (NZ$)
Milk income	5801^a	[5677–5915]	5260^b	[5232–5287]
Stock income	499^a	[491–508]	529^b	[522–535]
Other income^2	46		46
Gross farm income	6346^a	[6222–6461]	5835^b	[5805–5862]
Marginal expenses	3215		3393
Feed expenses	1447	[1405–1491]	1453	[1441–1465]
Segregation expenses^2	0		629
Farm expenses	4662^a	[4620–4705]	5475^b	[5462–5487]
Operating profit	1684^a	[1600–1762]	360^b	[342–377]

Note: Means within a row with different superscript letter are significantly different (P < 0.05).

¹Value of milk components (NZ$6.58/kg milksolids): NZ$4.63/kg fat, NZ$9.01/kg protein and −NZ$0.03/L milk.

²Model input.

Per cow, UFA farms had significantly lower feed expenses than AVE farms, but per hectare there were no significant differences in feed expenses between UFA farms and AVE farms. Segregation costs significantly increased farm expenses in UFA farms, both per cow and per hectare. The operating profit of UFA farms, per cow and per hectare, was significantly lower than in AVE farms.

Discussion

Bio-economic farm models have been used in the past to investigate the characteristics of dairy farms and herds that produce milk with modified composition (Dooley et al. 2005; da Cunha et al. 2010). The farm model used in the present study simulated the natural variation that exists in milk production, milk composition and concentration of UFA in fat of Holstein-Friesian dairy cows and farms under New Zealand conditions. Therefore, the results of the present study could not be compared with or extrapolated to studies comprising other dairy breeds or studies in which milk fat composition was altered by dietary manipulation.

The base population in this study comprised 1,820,000 simulated dairy cows distributed in 5600 dairy farms. The number of dairy cows simulated corresponded to the whole population of Holstein-Friesian cows in New Zealand, and represented 60% of dairy cows and farms present in the North Island of New Zealand (LIC 2015). Therefore, the results from this study give an indication of the extent to which UFA concentration in milk fat could be modified if all Holstein-Friesian cows in New Zealand were included in a segregation programme to form 50 herds that produced milk with high UFA concentration.

The segregation of dairy cows has been practised in the past to alter milk fat composition (Bobe et al. 2003; Thomson et al. 2003a). In a study conducted in New Zealand, Friesian cows were segregated according to their phenotype for solid fat content (Thomson et al. 2003a). As solid fat content is correlated to the concentration of UFA in milk fat (MacGibbon 1996), the segregation of cows according to their phenotype for solid fat content resulted in herds that produced milk with high (soft fat) or low (hard fat) UFA concentration in fat (Thomson et al. 2003a). Using segregation, the composition of milk produced by a farm could be changed within a year (Dooley et al. 2005). In the present study, UFA farms produced at least 15% more UFA than AVE farms in their first year. However, these results should be interpreted with caution given that in this study it was assumed that environmental effects remain constant and that the difference in fat UFA concentration between cows was exclusively due to genetic factors. Some studies reported that cows that produced milk with a high UFA concentration in fat maintained this characteristic during the lactation and between seasons (Thomson et al. 2003b), but changes in grazing management and location could also influence milk fat UFA concentration significantly (Thomson et al. 2002; Alonso et al. 2004; Frelich et al. 2012). More studies are necessary to understand the influence of grazing management and location on milk fat composition under New Zealand conditions. This is important to ensure that milk fat UFA concentration is still favourable when cows are transferred to a new location or grazing management is changed.

As cows on UFA farms had a milk yield similar to that of AVE cows (Table 3), but with significantly higher fat UFA concentration (+5.03 g/100 g milk fat) and significantly lower percentages of fat (−0.78%) and protein (−0.41%), their yields of fat, protein and milksolids were significantly reduced (−32 kg/cow, −16 kg/cow and −49 kg/cow, respectively). The differences in milk fat UFA concentration, fat percentage and fat yield between cows on UFA farms and cows on AVE farms were consistent with the differences reported by Thomson et al. (2003a) between cows that produced milk with hard or soft milk fat: +3.48 g UFA/100 g fat, −0.94% fat percentage and −23 kg fat, respectively (average of three trials and fat yield estimated to 270 days in milk). As in the present study, Thomson et al. (2003a) did not report a significant difference in milk yield between cows that produced milk fat with high or low UFA concentration. However, in the present study about 100 cows were screened for every cow segregated into a UFA farm, while in the study by Thomson et al. (2003a) about 30 cows were screened for every cow segregated into the soft fat (high UFA) herd.

The low production per cow on UFA farms could be due to a negative relationship between fat UFA concentration and other milk production traits. Several studies have reported negative genetic and phenotypic correlations (moderate to high) between milk fat UFA concentration and fat yield, fat percentage and protein percentage (Soyeurt et al. 2007; Schennink et al. 2008; Soyeurt et al. 2008; Stoop et al. 2008). These studies also reported a low phenotypic correlation between milk fat UFA concentration and milk yield, which could be positive (Schennink et al. 2008) or negative (Soyeurt et al. 2007).

In the present study, milk supplied to a dairy processor by UFA farms had at least 15% more UFA in milk fat than milk supplied by AVE farms. Couvreur et al. (2006) reported significant changes in the sensorial properties of butter (melting score, firmness in mouth) when the difference in UFA concentration between the control and modified milk fat was 11%. The percentage difference in milk fat UFA concentration between UFA farms and AVE farms varied during the season (Figure 1). Under a premium scenario for milk fat UFA concentration, changes in milk fat UFA concentration during the season can affect the ability of UFA farms to qualify for a premium.

The lower milk income of UFA farms when compared with AVE farms (−NZ$302/cow and −NZ$541/ha) was a consequence of their lower milk fat and protein yields (per cow and per hectare), their greater milk yield per hectare, and a milk payment system that rewarded the yields of fat and protein, and penalised milk volume. As the income from the sale of milk represents about 90% of gross farm income (DairyNZ 2016), the reduced milk income of UFA farms significantly affected their gross farm income and operating profit, both per cow and per hectare.

Due to their higher stocking rates and the cost of segregation, UFA farms had significantly higher farm expenses than AVE farms. However, the cost of segregation could be different from that assumed in this study (NZ$220/segregated cow) if dairy farmers had to pay for the measurement of UFA concentration in milk fat. Also, it is possible that the value of dairy cows that produce milk with high UFA concentration in fat is different from the value assumed in this study under a milk payment system that rewards the concentration of UFA in milk. It is also possible that segregation costs were overestimated considering that cows that produce milk with high UFA concentration could be selected to be culled due to their low milksolids production. Although the present study investigated the annual establishment of UFA farms by the segregation of dairy cows only, once formed, genetic selection could be used to generate replacements that produce milk with high UFA concentration in fat. With the use of genetic selection, segregation costs could be lower given that fewer cows that produce milk fat with high UFA concentration would need to be purchased.

The effect on farm production and profit of segregating dairy cows that produce milk with high UFA concentration may vary depending on animal, environmental and financial factors. Therefore, more studies are necessary before a programme to form farms that produce milk with high UFA concentration is established. The economic performance of the UFA farms in the present study is an indication of what may happen if dairy cows were selected and segregated to form herds that produce milk with high UFA concentration. At the payout used in the present simulation (NZ$6.58/kg MS), the operating profit of UFA farms was negative, both per cow and per hectare (Table 4). The premium needed for UFA farms to break even their operating profit with that of AVE farms was NZ$3.00/kg milk fat. If the cost of segregation was removed from the analysis in the present study, the premium needed by UFA farms to have a similar operating profit (NZ$/ha) as AVE farms would be NZ$1.56/kg MS approximately (a scenario where UFA farms have a breeding programme to generate their own replacements). Dairy companies would need to investigate if premiums like the ones reported in this study were worth the effort to produce milk high in UFA.

However, the limitations of this study should also be considered, especially with regard to the practicality of segregation, the cost of segregation and the effect of environmental factors on milk UFA concentration. Also, it is important to determine how segregating cows for the production of milk high in UFA can affect the financial position of dairy farms (stock value, return on assets, return on equity).

Conclusion

The concentration of UFA in milk fat supplied to a dairy processor could be increased by segregating cows that produce milk with high UFA concentration. In the present study, high concentrations of UFA in milk fat were associated with lower yields and percentages of fat and protein, per cow and per hectare. Under the current milk payment system in New Zealand (milk fat + milk protein – milk volume), the operating profit of UFA farms was negatively affected, per cow and per hectare. This study highlights the importance of developing a milk payment system that includes a premium for concentration of UFA in fat. Such a premium would need to not only break even the operating profit of UFA farms with that of AVE farms, but also include an economic incentive to encourage dairy farmers to change their farm system.

Disclosure statement

No potential conflict of interest was reported by the authors.

ORCID

N. Lopez-Villalobos http://orcid.org/0000-0001-6611-907X

H. T. Blair http://orcid.org/0000-0003-4415-4282

Additional information

Funding

The authors acknowledge the financial contribution of the Institute of Veterinary, Animal and Biomedical Sciences (Massey University), through the Colin Holmes Dairy Scholarship, and Fonterra Co-operative Group for their support towards this study.