Assessing heterogeneity of the composition of mare ’ s milk in Flanders De beoordeling van de samenstelling van paardenmelk die in Vlaanderen wordt verkocht

In this study, the effect of farm, time, season and health was evaluated on the composition of mare's milk sold in Flanders. The content of the analyzed components (i.e. fat, fatty acids, protein, lactoferrin, lysozyme, lactose and urea) differed significantly (p < 0.0001) between farms, at a given moment in time. Within each farm, large month-to-month variations for most milk components (p <0.01 to 0.0001) were observed. The variation over time between different farms was smaller. These findings indicate that the composition of the mare's milk consumer portions varies substantially between the different farms and also over time within each farm. Season, nutrition, udder health and worm burden are believed to contribute significantly to this variation.


INTRODUCTION
Little is known about the variation of the composition of mare's milk in time or space (different farms).Mare's milk is commonly used in the diet of certain Asian and Russian populations.Recently, the interest in mare's milk has increased in Western European countries due to the possible health promoting characteristics of this product (Stoyanova In deze studie werd de invloed van het bedrijf, het seizoen en de gezondheid van de dieren op de samenstelling van consumptiemelk afkomstig van merries beoordeeld.Ook de fl uctuaties in de tijd werden bekeken.Uit de resultaten blijkt dat er op een gegeven tijdstip een signifi cante variatie (p < 0,0001) bestaat in vet, vetzuren, eiwit, lactoferrine, lysozyme, lactose en ureum tussen bedrijven onderling.Op bedrijfsniveau werden er grote signifi cante fl uctuaties in de tijd vastgesteld voor de meeste melkcomponenten (p < 0,01 tot 0,0001).De fl uctuaties tussen de verschillende bedrijven waren beduidend kleiner dan die binnen de bedrijven.Er kan besloten worden dat de samenstelling van paardenmelk weinig homogeen is, zowel tussen de bedrijven op een vast tijdstip als binnen de bedrijven in de loop van het jaar.Er zijn signifi cante aanwijzingen dat seizoen, voeding, uiergezondheid en het niveau van wormbesmetting de samenstelling van deze melk kunnen beïnvloeden.et al., 1988;Jahreis et al., 1999;Sarwar et al., 2001;Benkerroum, 2008;Salamon et al., 2009).

A
The composition of mare's milk and its comparison to human and bovine milk have been reported in the literature.Mare's milk has a lower fat content, a higher lysozyme content than and a comparable lactoferrin content to human and cow's milk (Malacarne et al., 2002).The total amount of unsaturated fatty acids in mare's milk is highly comparable to human milk and higher than the amount in cow's milk.The polyunsaturated fatty acids (PUFA) linoleic acid of the omega-6 group and alpha-linolenic acid of the omega-3 group are present in considerably larger proportions in mare's milk than in human or cow's milk.Medium chain fatty acids (MCFA) consist of 8 to 12 carbons and are considered to be antibacterial (Skřivanová et al., 2009;Sprong et al., 2001).The proportion of these MCFA, especially capric acid (C10:0) and lauric acid (C12:0), is approximately two times higher in mare's milk than in human and cow's milk.Combined with the lower amount of saturated fatty acids (SFA), the presence of these MCFA could be benefi cial to human health (Malacarne et al., 2002).Additionally, the relatively high proportion of the whey protein fractions lactoferrin and lysozyme, which are considered antimicrobial active factors, might also play a role in the health claim of mare's milk.
The content of mare's milk is variable over the period of lactation (Mariani et al., 2001;Summer et al., 2004;Pikul and Wójtowski, 2008;Salamon et al. 2009).Experiments show that the content of fat and protein decreases (Mariani et al., 2001;Summer et al., 2004;Pikul and Wójtowski, 2008), and that lactose increases (Mariani et al., 2001) during lactation.Salamon et al. (2009) observed that the linoleic acid and linolenic acid concentrations decrease, while the antibacterial fatty acids capric and lauric acid increase over lactation time.Pikul and Wójtowski (2008) also found a decrease of linoleic acid, but linolenic acid increased signifi cantly during month four and fi ve of lactation.The diet fed to the mare is considered to be a major determinant of the equine milk composition.The composition of milk differs clearly between mares that are fed a diet rich in forage and mares that receive a diet rich in concentrates (Doreau et al., 1992;Hoffman et al., 1998).Most studies report that breed does not affect milk composition (Neseni, 1958;Kulisa, 1977;Doreau, 1991;Pikul and Wójtowski, 2008;Salamon et al., 2009), except for Pietrzak-Fiećko et al. (2009) who reported that the fatty acid composition of mare's milk is breed specifi c.Gibbs et al. (1982) found a signifi cant decrease of protein content with higher parity.In Murgese mares, machine-milking signifi cantly increases the fat content in milk compared to hand-milking (Caroprese et al., 2007).A study of asinine milk has revealed daily rhythmicity in lipid, protein and lactose content (Piccione et al., 2008).Fat and lactose have been found to peak during the night, while protein content peak during the day.
The aim of this study was to evaluate the variability of the composition of mare's milk sold at farms in Flanders (Belgium) and to explain part of this variability.The study consisted of two parts.In the fi rst part, the variation between and within ten different farms at a given time was evaluated.In a second study, the variation over time between and within four different farms was assessed, and several contributors to the variations were identifi ed.
To be consistently health promoting, variations in milk composition should be minimal.

Collection of milk samples
Consumer portions are a mixture of milk from all lactating mares of a single milking time or milking day.The milk is collected in a tank by machine-milking, and is cooled, stored and frozen in 250 mL consumer portions.In the fi rst study, fi ve consumer portions were bought at ten equine farms in East and West Flanders in November 2008 during a one-week period.Out of these ten farms, four were chosen for a monthly (fi rst week of the month; one consumer portion) collection during 20 months (May 2009 -December 2010; second study).At some of these four farms, there were periods of non-milking.
All samples were immediately transported to the laboratory using a cooling device, and stored at -20°C until further analysis.At the time of analysis, the frozen samples were thawed in a water bath at 25°C.

Farm management information
Information about the diet composition, the amount of concentrates given to each mare, about breed and lactation stage was obtained by verbal communication with the farmers.

Collection of feces samples
At the four farms of the longitudinal study, the eggs per gram feces (EPG) of strongyles were determined monthly during 20 months to evaluate the worm burden at farm level.Rectal feces samples were collected from maximum fi ve mares (in average 3.23) per farm.Fecal samples were mixed, and analyzed according to the McMaster technique (Ward et al., 1997).

Analysis of gross milk composition
The fat, protein, urea and lactose determinations were performed in a laboratory at the University College Ghent.The fi ve consumer portions bought in November 2008 on the ten different farms (n = 50) were analyzed at the same time.The samples (one at a time and per farm) gathered on the four farms between May 2009 -December 2010 (n = 66) were analyzed twice.Milk fat was determined by the IDF Gerber method No.105:1981(IDF, 1969).Verifi cation of the fast Gerber method was done in the fi rst study according to the Röse-Gottlieb AOAC method No. 905.02 (AOAC, 1996).The total nitrogen (N) amount in the different samples was verifi ed using the Kjeldahl method.The total protein content was calculated as total N amount x 6.38 according to AOAC method No. 991.20 (AOAC, 1996).An enzyme kit (Megazyme, catalogue No. K-URAMR) was used according to the manufacturer's instructions to determine the urea content of the samples spectrophotometrically.Another enzymatic test kit (Megazyme, catalogue No. K-LACGAR) was used to determine the lactose content of the samples.
Fatty acids were determined in the form of fatty acid methyl esters, obtained by extraction (cf.Röse-Gottlieb) and methylation.They were quantifi ed using gas chromatography (GC).This analysis includes saturated fatty acids (SFA), mono unsaturated fatty acids (MUFA), omega 3 polyunsaturated fatty acids (n-3 PUFA) omega 6 polyunsaturated fatty acids (n-6 PUFA) and some other fatty acids not included in the previous categories (Vlaeminck et al., 2005).The lysozyme and lactoferrin determinations using SDS-PAGE (Wambacq, 2005) were done in the same laboratory.All of the analyses in Lanupro were done on two consumer portions of the ten different farms and in duplicate on each of the consumer portions bought at the four selected farms.

Statistical analysis
SAS Version 9.2 learning edition 4.1 was used to explore and analyze the data.A paired t-test was used to search for signifi cant differences between the Gerber method and the Röse-Gottlieb reference method.As the different components in the fi rst study were analyzed on two (fatty acids, lysozyme and lactoferrin) or fi ve (all of the other components) samples per farm, the variance within (= due to measurement method and different consumer portions examined) and between the ten farms was evaluated using a linear mixed model with farm as random effect.The possible infl uence of the amount of provided concentrates was analyzed using a linear mixed model with farm as random effect and concentrates as continuous fi xed effect.
The analysis of heterogeneity within and between farms over time was based on a linear mixed model with farm as random effect.Additionally, season and EPG were added as covariates to the analysis, with season as categorical and EPG as continuous fi xed effect.Pearson correlation coeffi cients between the different milk components were derived.

RESULTS
The management information of the ten farms involved in this study is shown in Table 1.At all of the ten farms, roughage was provided ad libitum.The average somatic cell counts (SCC) of farm 1 and 2 were high compared to the normally low SCC in mare's milk (Dankow et al., 2006), but this was mainly due to one high SCC in both farms (farm 1: 4, 671,000 in November 2010; farm 2: 1,248,000 in November 2009).The overall average SCC of the four farms equaled 122,734 cells/ml (n = 66).
The composition of mare's milk is summarized in Table 2.The results by the quick Gerber method and Röse-Gottlieb method for fat determination did not differ signifi cantly (p < 0.05).Therefore, only the quick Gerber method results are shown.
The fatty acid composition is summarized in Table 3.
Except for lysozyme (74.8%), most of the variability in the results of the fi rst study (95.7% to 99.8%) was found to be due to differences between the farms, indicating no variation due to the analysis method or   The amount of concentrates in the diet was not signifi cantly associated with the amount of fat, protein, lysozyme, lactoferrin, urea and lactose.Linoleic acid appeared to be substantially higher (p < 0.01) in milk from farms where mares received more concentrates.Figure 1 shows the relationship between the daily amount of concentrates and n-6 and n-3 content in the milk.
Table 4 presents the individual and combined farm means, the minima and maxima of the different milk components during the 20-month-sampling period.Figure 2 shows the variation of the protein content over time.Similar patterns (not shown) were found for fat, n-3 and n-6.
The variability of different components caused by between-farm and intra-farm factors over a period of 20 months is shown in Table 5.
The statistical analysis of seasonal and EPG infl uences on milk composition revealed a signifi cant effect of season (p < 0.05) for fat, n-6 to n-3 ratio and urea.The fat content appeared to be higher in milk produced during autumn and winter, while the ratio of omega 6 to omega 3 fatty acids was higher during winter and spring.The values for urea were highest in autumn.The n-6 fatty acids were the only component on which the EPG had a signifi cant (p < 0.05) infl uence.

DISCUSSION
The aim of this study was to assess the variation of the composition of mare's milk sold at farms in Flanders (Belgium) both over space and time and to explain part of this variability.The fi rst observation was the large variations in milking procedures and management at the different farms.
The total fat content showed extreme variations between farms at a fi xed time with a very low fat content found in three of the farms (≤ 0.60%).Only Mariani et al. (2001) mentioned a similar low fat content (0.44 g/100 g at 180 days of lactation), most 0.24 (0.2-0.4) 0.32 (0.2-0.4) 0.32 (0.2-0.5) 0.36 (0.2-0.5) 0.30 (0.2-0.5) Lactose (g/100 g) 6.91 (6.5-7.5)6.97 (6.7-7.5)6.77 (5.5-7.2) 6.72 (6.4-7.0)6.87 (5.5-7.5)FA: fatty acids; SFA: saturated fatty acids; MUFA: mono unsaturated fatty acids; n-6 PUFA: omega 6 polyunsaturated fatty acids; n-3 PUFA: omega 3 polyunsaturated fatty acids likely because hand-milking causes a lower fat content (Caroprese et al., 2007).In the present study, no hand-milking was performed.Doreau et al. (1992) found a higher fat content in milk from mares that were fed with a forage-rich diet.In all of the ten farms of the present study, the mares were provided with a diet consisting of roughage ad libitum and a variable amount of concentrates.However, the amount of concentrates did not affect the fat content of the milk in this study.Over time, the fat content ranged from 0.26 to 1.91 g/100 g (Table 4).Depending on the farm, large fl uctuations in fat content were observed during the 20-month-observation period.The highest fat content of the 20-month period was found in milk of farm 2. This might be due to a diet rich in hay given to the mares at this farm, which confi rms the results of Doreau et al. (1992).Moreover, in autumn and winter, the fat content was signifi cantly higher than in the other two seasons, even though a low fat content is to be expected at the end of lactation in those periods (Mariani et al., 2001;Pikul and Wójtowski, 2008).This is most likely due to a diet with relatively more roughage (Doreau et al., 1992) given in those seasons and to the presence of mares at different lactation stages.
It was found that the mean protein content in all samples was less than the crude protein fraction mentioned by Malacarne et al. (2002).Nevertheless, this result is comparable to values of milk from mares after two months of lactation as has been reported by Mariani et al. (2001).Although in this study, the protein content showed signifi cant differences between farms, the farm with end-stage mares represented a rather high protein content per 100 g milk compared to most of the other farms.This might be explained by the fact that at the end of lactation, less milk is produced with a relatively higher protein content.The higher protein concentrations in milk from mares fed a diet rich in forage, as reported by Doreau et al. (1992), could not be proven statistically in the present study.The mean protein percentages of the four different farms during the 20-month study were relatively close: the variability between farms was only 29%, so the largest variations over time were present within the farms.The mean lysozyme content was similar to the value reported by Jauregui-Adell (1974).Malacarne et al. (2002) found a slightly lower value, and reported a higher lactoferrin amount than found in the present study.Although the average concentrations of whey proteins lysozyme and lactoferrin at the four farms were very similar, the minimum and maximum values widely varied within the farms over time (Table 4).Salimei et al. (2002) reported comparable results for the mean urea content.Large fl uctuations were observed within the farms of the present study.There was a signifi cant seasonal infl uence, with the highest amounts of urea in spring, summer and autumn, decreasing in winter.In winter, mares are housed inside, and fed a more balanced diet.Most likely, a more balanced diet limits the loss of N and the formation of urea.In general, as the dietary N intake increases, the plasma urea concentrations, urea pool size, urea entry rate and urea excretion rate also increase (Prior et al., 1974).
The mean lactose content was in agreement with values found in the literature (Mariani et al., 2001;Malacarne et al., 2002).A high lactose content (7.11 g/100 g) was found in the milk of two farms in November 2008.A minimum content of 5.47 and a maximum content up to 7.5 g/100 g were found in the second study.However, a very low milk lactose value of 5.47 g/100 g was observed in milk of farm 3 in July 2010.This low value has not been reported in the literature so far.The lactose content in the study by Mariani et al. (2001) was never lower than 6.36 g/100 g (at four days post parturition).
There are remarkable differences between the fatty acid content found in the present study (Table 3) and those reported in the literature.The mean SFA in this study was 10% lower than the results reported by Malacarne et al. (2002), but comparable to the values found by Pikul and Wójtowski (2008).Both studies reported slightly lower SCFA amounts and higher amounts of MCFA.The mean LCFA content in the present study is in agreement with the amount mentioned by Pikul and Wójtowski (2008), but lower than that reported by Malacarne et al. (2002).Unsaturated fatty acids account for more than 50% of the total fatty acid content.Experiments performed by Pikul and Wójtowski (2008) suggested approximately the same ratio of MUFA and PUFA as mentioned in Table 3, but Malacarne et al. (2002) reported less PUFA.The present study showed that the average amount of n-6 PUFA, which consists mainly of linoleic acid, was low in comparison to the values reported by Pikul and Wójtowski (2008).However, the results of Malacarne et al. (2002) are much more comparable to the result of this study.In contrast with the amount of n-6, the amount of n-3 PUFA and the results for linolenic acid were very high compared to the literature values (Malacarne et al., 2002;Pikul and Wójtowski, 2008).Pikul and Wójtowski (2008) suggest a signifi cant effect of lactation stage, and have found that the n-6 PUFA content decreases from approximately 20% of total fatty acids in the colostrum to 15.7% in milk of the fi fth month after foaling.n-3 PUFA was found to increase during lactation from approximately 5.8% in colostrum to 7.6% after fi ve months of lactation.However, at the end of lactation, the milk of the mares showed rather opposite values, as the amount of n-6 was the highest of all of the farms, and n-3 was one of the lowest values.The trend of a higher milk concentration of linoleic acid and a lower concentration of linolenic acid is in agreement with the results of Hoffman et al. (1998) and Doreau et al. (1992) who stated that there is no hydrogenation before the absorption of dietary fatty acids in the horse, which results in a correlation of the long chain fatty acids found in the milk and the composition of the provided nutrition.Simopoulos (2004) mentioned an optimal balance of n-6 to n-3 fatty acids of 1/1 to 5/1 in the human diet to obtain benefi cial health effects on different diseases.Because an increasing amount of concentrates results in increasing linoleic acid and decreasing linolenic acid in milk, a more favorable n-6/n-3 ratio could be achieved.In the present study, n-6/n-3 ranged from 0.25 to 1.99 with a mean of 0.96.Other studies reported ratios of 0.33 to 3.5; large differences seem to be present in mare's milk (Hoffman et al., 1998;Malacarne et al., 2002;Pikul and Wójtowski, 2008;Salamon et al., 2009).In the present study, it was found that over time, large ranges could be seen between minimum and maximum values for the different groups of fatty acids (Table 4).In contrast to n-6, where 58% of the variability could be explained by between-farm differences, the different values for n-3 were mostly due to within-farm factors.Milk of farms providing different amounts of concentrates seemed to contain fatty acids with various ratios of n-6 and n-3 PUFA.The lowest mean n-6 PUFA and the highest mean n-3 PUFA over time were found in milk from the farm giving only concentrates (farm 6).As the precise composition of the concentrates was unknown, the component responsible for differences seen in the milk of mares that received more concentrates than mares that were given less concentrates, was also unknown.The farm that provided the highest amount of concentrates (farm 1) showed opposite values with the highest mean n-6 fraction and the lowest mean n-3 fraction.Besides concentrates to roughage ratio, the precise diet composition and lactation stage may also infl uence the composition of fatty acids (Doreau et al., 1992 1998; Mariani et al., 2001).It has been shown that season has no signifi cant effect on neither n-6 nor n-3 content.However, in the present study, the n-6/n-3 ratio was signifi cantly infl uenced by season.In spring and winter, this ratio seems to be more favorable towards human nutrition than the ratio in the other months (Simopoulos, 2004).
Monitoring the health of lactating mares seems to be important.In the present study, the n-3 fatty acids showed a signifi cant decrease when EPG increased (p < 0.05).However, no large impact was expected of low to normal worm infestations.Somatic cell counts > 100,000 cells/ml were found, which refl ects udder disease in horses (Böhm et al., 2009), Nevertheless, the farm keepers did not mention mastitis as a problem, nor did they wonder about the potential causes.Inadequate milking hygiene and milking mechanisms could be the reasons for this fi nding.However, the reasons why mastitis and occasionally high SCC's are not an issue in horse milking farms may be the lower milk yield, more milking sessions per day and the fact that foals are kept with their mothers overnight.High SCC's have been measured and controlled several times in other, unpublished studies.Furthermore, negative infl uences of increased SCC's on some milk components were seen in the present study.Monitoring udder health on horse milking farms is therefore advisable.

CONCLUSION
There is a large between-farm variation of mare's milk composition at a fi xed point in time.Over time, there is a larger within-farm than an inter-farm variation concerning the composition of mare's milk.Differences in nutrition, worm burden and udder health cause signifi cant intra-farm variations.

Figure 1 .
Figure 1.Effect of variable amount of concentrates on linoleic (increase) and linolenic (decrease) acids in mare's milk.

Figure 2 .
Figure 2. Monthly protein percentage in mare's milk of four farms.

Table 1 . Farm information.
* Farms followed over a 20-month period (May 2009 -December 2010); EPG: eggs per gram feces; SCC: Somatic Cell Count.(n1): number of mares milked on a daily basis; (n2): number of samples that were collected and analyzed; Irregular: mares are milked depending on the commercial needs; if not milked, foals stay with their mothers to sustain milk production.

Table 5 . Variability of milk components between and within four farms over a 20-month period.
; Hoffman et al.,