Management and sanitary practices in ewe dairy farms and bulk milk somatic cell count

This study was done to establish relationships between management and sanitary practices on ewe dairy farms and the quality of milk produced. For this purpose, a survey was carried out on 118 farms and a total of 121,117 animals in the Castilla-La Mancha region (Southeast Spain) in which the veterinarians of the Association of Sanitary Defence participated. Flocks varied considerably in size, ranging from 120-7,200 sheep, with an average milking period lasting 3-5 months and milk yields of between 50 and 150 litres per lactation by mechanical systems on 75.0% of farms. Sanitary practices during milking are still not commonplace; 53.3% performed post milking teat disinfection, 37.0% did the California mastitis test, 31.1% had a dry therapy programme and only 18.6% of farms kept records of animals with clinical mastitis. Nonetheless, the results revealed that 42.7% of the farms surveyed showed good quality hygienic milk (bulk milk somatic cell count, BMSCC < 600 × 10 cells mL). On most farms, the rate of animals with clinical mastitis was less than 5.0%, while that of subclinical mastitis cases was less than 5.0% on 25.6% of farms. Certain factors like mechanical milking, post milking teat disinfection, recording animals with clinical mastitis and controlling the frequency of animals with subclinical mastitis did not exceed 5.0%. Those risk factors for increased BMSCC levels were significant. Therefore, farmers are recommended to bear these factors in mind, and to instil the importance of carrying out these practices when implementing a good dairy farm practices system. Additional key words: ewe milk, good dairy farming practices, milk quality.


Introduction
In recent years, ewe dairy farming has become more important thanks to genetic selection and to better feeding conditions (Haenlein, 2007) which have led to higher milk yields and improved milk quality.Sheep milk is mainly destined for dairy products, basically pure sheep cheeses, but also mixed types (Pirisi et al., 2007).Many of these products have a distinction of quality (Protected Denomination of Origin (PDO) etc.), which involves basic food products having certain quality standards.The quality of PDO products must be guaranteed, which is a practice that starts on the farm.Currently, the production guidelines and marketing standards for hygienic quality milk within the food safety framework are set by Council Directives 92/46/EC and 94/71/EC (OJ, 1992(OJ, , 1994)).
European guideline compliance has greatly improved the quality of milk on sheep dairy farms (Gallego, 2002;Gonzalo et al., 2005;Pirisi et al., 2007) thanks to the cooperation of dairy farmers, veterinarians and laboratory technicians.Quality control systems for milk payments have been defined and applied.The hazard analysis and critical control point (HACCP) is the off icial quality management method for dairy industry milk (Cullor, 1997).On dairy farms, good dairy farming practices (GDFP) guarantee the hygienic quality of the milk, and include (EC) 852/2004 and (EC) 853/2004853/ (OJ, 2004a,b) ,b) regulations.These food safety guidelines encourage the GDFP in which different aspects, such as animal health, milk hygiene, animal feed and water, animal welfare and the environment, are integrally controlled.It is necessary to develop milk quality and safety improvement programmes for ewe dairy farms which adapt to their particular production conditions (Bencini and Pulina, 1997;Bergonier and Berthelot, 2003;Serrano et al., 2003;Gonzalo et al., 2005).
Many authors consider that somatic cell count (SCC) values are not only valuable indicators of udder health, but also indirectly reveal parameters for sani-tary control and management practices (Paape et al., 2001).In cows, the factors associated with a low bulk milk somatic cell count (BMSCC) are those related to improved environmental and hygiene applications before and during milking (Tadich et al., 2003;Jayarao et al., 2004).Currently, there is a regulatory limit set for this species by Council Directives 92/46/EC and 94/71/EC.However, despite there being no set maximum BMSCC or individual SCC values for the milk of small ruminants, there appears to be good agreement (De la Cruz et al., 1994;González-Rodríguez et al., 1995;Gonzalo et al., 2005) as to considering that BMSCC values over 500-600 × 10 3 cell mL -1 would be associated with an increased prevalence of mammary infections and with significant losses in the milk production of small ruminants.Therefore it is necessary to conduct studies which could help improve the hygiene-sanitary quality of ewe's milk by f irstly improving the BMSCC.
With this in mind, the aim of our work was to study the characteristics of ewe dairy farms in terms of management and sanitary practices during milking and their relationship with milk quality and udder health.This work is considered necessary before implementing a GDFP system to guarantee the sanitary quality of milk destined for cheese making.

Material and methods
A questionnaire was drawn up (Table 1) to obtain information about general farm management characteristics and sanitary practices during milking as regards milk hygiene.A total of 490 ewe dairy farms located in Castilla-La Mancha (Southeast Spain) participated in the survey which was conducted over one complete year.These dairy farms supplied milk for PDO Manchego Cheese production and participated in a study into the presence of antibiotic residues and the methods to detect inhibitors in ewe's milk (Yamaki et al., 2004(Yamaki et al., , 2006)).debajo del 5% de animales con mastitis clínicas, y entre las subclínicas, más de la cuarta parte de las granjas (25,6%) tienen una incidencia < 5%.Prácticas como el ordeño mecánico, el sellado postordeño, el registro de animales con mastitis clínicas y controlar que la frecuencia de animales con mastitis subclínica no supere el 5% fueron significativos factores de riesgo del incremento del recuento de células somáticas en tanque.En conclusión, cuando se pretenda implantar un programa de buenas prácticas ganaderas en un grupo de explotaciones de ovino de leche, es recomendable incorporarlas para mejorar el recuento celular y por lo tanto la calidad higiénico-sanitaria de la leche producida.
The questionnaire was completed with the help of the veterinarians in charge of the sanitary control on ewe dairy farms in Castilla-La Mancha.A total of 118 questionnaires were answered, whose results are provided in this work and which include information obtained from a total of 121,117 ewes.

Statistical analysis
A multiple logistic regression model (Hosmer and Lemeshow, 1989) was conducted to test the effects of two groups of factors: general farm management characteristics and sanitary practices during milking and their effect on BMSCC levels.The data collected from the questionnaires were entered as binary factors as follows: BMSCC levels were coded as high (0: BMSCC < 600 × 10 3 cells mL -1 ) and low (1: BMSCC > 600 × 10 3 cells mL -1 ).
Each factor group was tested separately according to models [1] and [2].After firstly analysing the logistic regression models, all the significant factors of each group were reassessed together.A forward stepwise procedure was used to test the final model [3] and was based on the significance tests done on the change in deviance (indicated D) when including, or not, factors in the model.This statistic is approximately distributed as a chi-square with the appropriate degree of freedom.Wald's statistic tested whether the coefficients (and the constant) of the model were significant.All the analyses were done with SPSS (SPSS, vers 15).
where FS: flock size, MP: milking period, MY: milk yield, DP: dry-off period, PS: doing partial suckling,  AS: doing artificial suckling, Mp: milking procedures, ChM: checking milking machine, TD: teat disinfection in post-milking, KR: keeping records of animals with clinical mastitis, DT: dry therapy, CMT: California mastitis test, RCM: ratio of animals with clinical mastitis in the flock, and RSM: ratio of animals with sub-clinical mastitis in the flock.

Results
The general farm management characteristics group revealed that flock sizes varied considerably, ranging from 120 to 7,200 animals with a mean size of 1,035 ewes.Group distribution was well balanced as 35.9% (42/117) of the farms had less than 500 animals; 35.9% of the farms had 500-1,000 animals, and 28.2% had more than 1,000 animals.
As regards milking periods (Fig. 1), a large percentage of farms (72.4%: 84/116) milked for periods of 3-5 months, while 19.8% of the surveyed farms milked less than 3 months.A close relationship was found between the milking period and the milk yield as 79.7% (75/94) of both the intermediate yield levels (50-100 and 100-150 L) were obtained for milking periods of over 3-5-months.The groups which presented the highest yield of 150-200 L (6.8%) and over 200 L (4.2%) could be linked to those farms highly specialized in dairy production.Other data related to the handling practices during milking indicated that the dry off period lasted 2 months on roughly half the farms (51.9%; 54/104), while a similar percentage of farms (57/107) undertook partial suckling.Nonetheless, 95.7% (110/115) of the farms did not carry out artificial suckling.
A forward stepwise logistic regression model was carried out according to [1].The effect of the farm management factors on the BMSCC level was statistically significant (D 1,104 = 5.643, P < 0.05), and showed that only one factor, mechanical milking, was affected by BMSCC > 600 × 10 3 cells mL -1 (the coeff icient for mechanical milking, -1.224, Wald's test = 5.158, df = 1, P = 0.023).Moreover, the negative significance noted with the BMSCC could indicate that the less frequent mechanical milking, the greater the risk of increasing BMSCC values (data not shown).
In terms of milking procedures, 74.6% (88/118) of farms did mechanical milking, whereas only 25.4% manually milked.All the farms with mechanical milk systems had records of checking the milking machine.However, such maintenance was recorded once or twice a year for 70.4% (62/88) of farms, while 15.9% of farms recorded such checks more than twice a year.
The survey into post milking teat disinfection revealed that this practice has been slowly incorporated onto ewe dairy farms given that 53.3% (63/118) of farms actually performed this.The frequency distribution of post milking tip dipping demonstrated that 31.4% of farms conducted this practice after each milking, 16.9% (20/118) did so with only mastitis cases, and 5.1% did so in the first two weeks of milking.
In this study, only 18.6% (22/118) of farms kept records of animals with clinical mastitis.Furthermore, 31.4% (37/118) of farms answered the questions about dry therapy; moreover, 43 of the 118 farms carried out the CMT, indicating that this practice was less wide in small dairy ruminants than in dairy cattle, and that it should be extended because it improves the hygienic quality of milk and helps detect those animals infected by subclinical mastitis.
Concerning clinical and subclinical mastitis, the results of the questionnaire indicated that 82.7% (81/98) of the farms claimed they had less than 5.0% of clinical mastitis cases in their flocks.However, 25.6% (22/86) of farms had less than 5.0% of cases of subclinical mastitis among their flocks.The highest rate was 5-10% (40.7%), while 22.1% of farms presented a rate of 10-15%.A low rate (11.6%) was observed for more than 15.0% of subclinical mastitis cases in flocks.The effect of the sanitary practices associated with milk hygiene on the BMSCC level was statistically significant (D 7,86 = 65.428,P < 0.001).Indeed, the significant factors with a BMSCC value of > 600 × 10 3 cells mL -1 were associated with post milking teat disinfection (P = 0.005), keeping records of animals with clinical mastitis (P = 0.024), and the rate of animals with subclinical mastitis in flocks (P = 0.004) (data not shown).
When the post milking teat disinfection factor was compared with the remaining factors by using the «after every milking» option, the results revealed that the options «did not do this» or «only for mastitis cases» were significant practices in both positive and statistical terms (P = 0.003 and P = 0.029, respectively).However, conducting this practice in the first two weeks was not significant (P = 0.361).
As regards keeping records of animals with subclinical mastitis, a negative significance was observed for a high BMSCC (coeff icient, -2.282, Wald's test = 5.903, df = 1, P = 0.015; Table 2) in relation to not keeping records.When comparing the change in the BMSCC levels to the rate of animals with subclinical mastitis above 15.0% in flocks, only the < 5% group was seen to be negatively associated with a high BMSCC (P = 0.017), while the remaining rates ranged from 5-10% (P = 0.231) to 10-15% (P = 0.861), were not significant.

Discussion
The SCC in milk is a reliable parameter to indirectly diagnose the health status of mammary glands (De la Cruz et al., 1994;Bencini and Pulina, 1997;Paape et al., 2001;Gonzalo et al., 2002;Contreras et al., 2007) and is, therefore, an effective tool to control mammary illnesses like mastitis.Factors such as breed, flock, animal's age, lactation stage, udder health and hygiene, and handling during milking, all influence SCC values (Bergonier and Berthelot, 2003;Gonzalo et al., 2005).Consequently, these values are not only necessary indicators of milk quality in sanitary terms, but also indirectly reveal parameters for sanitary control and management practices on small ruminant dairy farms (González-Rodríguez et al., 1995;Gonzalo et al., 2005).On 42.7% of the farms surveyed, the BMSCC was < 600 × 10 3 cells mL -1 , and these results coincide with other studies done previously on the Manchegan breed (Serrano et al., 2003).
In the survey, flock size distributed almost equally in each category, and this distribution could be explained by the different production systems in the study The estimation change of the BMSCC levels in relation to the reference level of: a using teat disinfection after every milking; b not keeping a record of animal with sub-clinical mastitis; c more than 15% animals in the flock with sub-clinical mastitis.
area which ranged from semi-extensive to intensive (Gallego, 2002).The majority of farms reported milking their animals for 3 to 5 months (72.4%) and performing dry off for 2 months or for more than 2 months (84.6%).These results coincide with those of other studies carried out in the La Mancha area and are associated with a reproduction cycle of 3 parturitions in two years (Gallego, 2002).Of all the farms surveyed, 19.8% stated they milked less than 3 months on account of their mixed production (meat and milk).The rest of the farms mentioned long milking periods (over 5 months), which could be due to them having animals with a high potential milk production, a reproduction rate of 1 parturition per year in many cases, or being in the breed selection scheme.With regard to the average milk yield, the majority of the farms (94/118) responded between 50 to 100 L and 100 to 150 L. These yields came close to the estimated average yield in the region of the local breed, that is, the Manchega breed (Gallego, 2002;Serrano et al., 2003).
The partial suckling method involved milking ewes and nursing lambs during the lactation period, which was done simultaneously.In this study, half the farms carried out this method since its offers a series of advantages such as increased marketable milk and the progressive weaning of lambs without affecting their subsequent growth.However, artificial suckling, as the most commonly used method on dairy goat farms, was not as extended on dairy sheep farms.
Mechanical milking was seen to be a highly extended practice among the farms surveyed, and was slightly higher than that which other authors have observed (Gallego, 2002).This could possibly be due to the expansion of new milking parlours thanks to the favourable perspectives on the price of ewe's milk in the study area in recent years.Indeed, it was one of the reasons that has contributed to improving milk quality.According to the analysis done in the stepwise logistic regression model for farm management characteristics, only mechanical milking was seen to be statistically and negatively significance with a BMSCC of > 600 × 10 3 cells mL -1 .In other words, it is somewhat difficult to keep good BMSCC levels with manual milking because of the complexity involved in sustaining sanitary conditions (Bencini and Pulina, 1997;Gonzalo et al., 2006Gonzalo et al., , 2009)).After the analysis of the stepwise logistic regression model however, we did an assessment with the significant variables of the group of hygiene practices associated with milk hygiene.This factor was seen to be non-significant (P = 0.997) in terms of high BMSCC levels (see Table 2), suggesting that performing hygiene practices was a more important factor to maintain good quality milk with desirable BMSCC levels.
With regard to checking milking machines, this factor was not associated with the BMSCC (P = 0.163).However, various authors have mentioned the importance of maintaining milking machines because if they did not work properly, they could markedly influence the sanitary condition of ewes' udders.The value obtained was low (70.4%) if we bear in mind that optimising milk machine standards and periodically checking milking equipment should be carried out if these measures are to improve the health of flocks (Gonzalo et al., 2005;Contreras et al., 2007).
One of the effective prevention methods to avoid mammary infections is post milking teat disinfection, which has mainly been applied in highly infected flocks of small ruminants (Paape et al., 2001;Begonier and Berthelot, 2003).This is a very important finding given the effectiveness of the prevention method to avoid mammary infections as this practice prevents mastitis pathogens from penetrating the teat canal which remains open for some time after milking, thus preventing exposure to possible contamination (Contreras et al., 2007).This factor was associated with a high BMSCC when assessed with only the hygiene practice variables (Model [2]) and when tested with all the significant variables (Model [3]).The risk of new bacterial infections occurring during lactation could increase during the weaning period in sheep (Gonzalo et al., 2006).This study indicated that many farms perform partial suckling, and that this period coincided with the final weaning and the beginning of the milking period when ewes were more exposed to high-risk infection.Therefore, the result of the estimation change of the BMSCC levels could be accounted for by applying this practice after each milking in order to maintain appropriate BMSCC levels.However, applying this practice only in mastitis cases affected the BMSCC levels just as not performing this practice did (Paape et al., 2001;Jayarao et al., 2004).
To improve milk quality, it is useful to administer an intramammary antibiotic during the dry off period since this treatment is able to reduce old infections, or even new ones which might appear during the dry off period (Paape et al., 2001;Gonzalo et al., 2005Gonzalo et al., , 2006Gonzalo et al., , 2009)).The affirmative response rate to antibiotic treatment during dry off periods was lower than in cattle dairy farms where this treatment is employed routinely (Jayarao et al., 2004).However, applying antibiotic treatment during the dry off period in this study was not significant in the BMSCC in model [2] (P = 0.463).
Another tool to indirectly diagnose subclinical mastitis is the CMT.Even though it was not significant with a high BMSCC (P = 0.103) in this study, this test could estimate the cellular count semi-quantitatively, and it proves most useful on farms because it is swift, simple to use and low-cost (Hueston et al., 1986;Paape et al., 2001).Furthermore, all the cited authors have mentioned its convenient use for selective antibiotic therapy during dry off periods to avoid excessive administration in healthy udders and residue contamination in milk (Paape et al., 2001).Consequently, it could be necessary to apply this practice in a higher percentage of animals in flocks to obtain more benefits.
In relation to keeping records of animals with subclinical and clinical mastitis, all the frequencies obtained were within the intervals described for small ruminants (Begonier and Berthelot, 2003;Contreras et al., 2007).In this study, the keeping of records of animals with clinical mastitis factor associated negatively with a high BMSCC by the first (P = 0.024: data not shown) and the final (P = 0.015) stepwise logistic regression models.Various authors (Schultz, 1997;Tadich et al., 2003) have reported how the effect of a few animals with an extremely high individual SCC was particularly noticeable in relation to the BMSCC levels in the flock.Therefore, it is important that farmers keep records of the animals with clinical mastitis to avoid mixing bad quality milk with good quality milk.Regarding the change in the BMSCC levels of the rate of animals with sub-clinical mastitis being over 15%, only less than 5% associated negatively with a high BMSCC (P = 0.017).According to Gonzalo et al. (2002), the BMSCC levels related to the severity of subclinical mastitis in ewes.Therefore, farmers and veterinarians should pay more attention to treating infected animals in a flock with a subclinical mastitis rate over 5% in order to improve milk quality.
In conclusion, the importance of factors like mechanical milking, post milking teat disinfection, recording animals with clinical mastitis and controlling the frequency of animals with subclinical mastitis not exceeding 5.0%, are related with increased BMSCC levels and should, therefore, be considered on ewe dairy farms.It is also necessary for ewe dairy farmers to instil the importance of carrying out these practices when implementing a good dairy farm practices system.

Figure 1 .
Figure 1.Farm distribution related to milking period and milk yield (n = 116).

Table 1 .
List of questions in the survey

Table 2 .
Results of the final logistic regression model with the factors associated with BMSCC > 600 × 10 3 cells mL -1