The world production of sheep and goat milk was 28.8 million tons in 2014 (http://www.fao.org/faostat/en/#data) representing 3.6% of global milk production. Regulatory legislation including specific hygiene rules for food of animal origin, raw milk included, have been implemented by international organisms in developed countries ( i.e. , EC, 2004 ), some of which have published guides on milk hygiene basically centered in cow dairy farms. Increasing awareness of public health and food safety issues in recent years has led to a greater interest in milk hygiene in small ruminant livestock in order to minimize risks and to guarantee an optimal consumption quality.

Bulk tank milk (BTM) can be contaminated by bacteria (spoiling and pathogens), bacterial and fungal toxins, veterinary drugs, cleaning and disinfectant agents, macroscopic sediment, and others, many of which are important to the farmers, cheese manufacturers and consumers because they are major factors, along with somatic cells, in determining safety and hygienic quality of the final product. Some of these variables along with fat and protein contents are the target of different legal limits or payment-by-quality schemes proposed by different countries, with obvious repercussions on the marketing of sheep and goat milk ( Pirisi et al. , 2007 ).

Bacteria contamination proceeds from different sources such as flora and pathogens present in beds, milking system and facilities, storage and transport, feeding, wash water, udder, or mastitic milk. Some of these bacteria are resistant to pasteurization, or are able to grow at refrigeration temperature, or indicate fecal contamination or mastitis, or fermenting lactic acid to butyric acid, CO ₂ and H ₂ causing the late-blowing defect of ripened cheeses. Despite this, only total bacterial count of BTM (BTMTBC, really bulk tank aerobic mesophilic count) has been the target of various legal limits for small ruminant milk in the European Union (EU). Other bacterial groups, such as thermoduric bacteria, psychrotrophic bacteria, coliforms, gram-positive catalase-negative cocci (GPCNC), staphylococci, or lactate-fermenting Clostridium spp. spores would be indicators of interest for milk hygiene, udder health, safety, and marketing quality ( Arias et al. , 2013 ; de Garnica et al. , 2013a ). Mammary pathogens ( i.e. , Staphylococcus spp., Mycoplasmas spp., etc.) are also important organism causing mastitis and profitability leaks in the flocks, and their knowledge is essential to program specific strategies for mastitis control in each herd. Mycoplasma agalactiae in sheep and goats and other Mycoplasma spp. ( i.e. , M. mycoides subsp. capri, M. capricolum subsp. capricolum , and M. putrefaciens ) specifically in goats, are the etiological agents of contagious agalactia (CA). This chronic disease is considered endemic in many countries and has a great health and economic impact because it is associated with the typical triad of mastitis, arthritis and keratoconjunctivitis, along with pneumonia and sporadic abortion. Therefore the implementation of population control programs based on diagnostic and eradication of Mycoplasma spp. is very important in endemic areas ( Bergonier et al. , 1997 ; Gómez-Martín et al. , 2013 ). Bulk tank somatic cell count (BTMSCC) is the first and principal tool used by technicians and farmers to evaluate udder health in flocks and to monitor the mastitis control programs and the efficiency of production processes; nevertheless, the EU has yet to regulate values in ewe and goat milk used for dairy products sold in its region. Antibiotic residue (AR) occurrence is a result of treating dairy sheep and goats with antibiotics and not withholding milk, but little attention has been paid to investigate the factors influencing its variation ( i.e. , mastitis, antibiotic therapy, milking system, or milk production) under field conditions in these species ( Gonzalo et al. , 2010 ). Other milk contaminants as aflatoxins (AF) are secondary metabolites produced by Aspergillus spp. that contaminate plants and foods. Stored animal feed (silage, grains and cake) are at higher risk of AF (B1 and B2) contamination. The B1 AF is carcinogenic to humans ( IARC, 2002 ) and after its consumption by lactating animal it is biotransformed by liver into M1 AF, which is excreted into the milk. Nevertheless, the information on occurrence of AF is very scarce in small ruminants.

In addition, the accurate determination of sheep and goat milk hygiene variables is important because of economic, health, and management implications. The specific physico-chemical (dry matter contents, acidity, osmolality), synthetic (apocrine secretion, natural inhibitors), and hygienic (somatic cells, bacterial content, pathogens) characteristics of small ruminant milk compared to cow milk, as well as the preservation (azidiol: AZ, bronopol: BR, or unpreservation) and storage (refrigeration or freezing) strategies, makes necessary an analytical validation for the milk of these species in order to guarantee the global accuracy of analytical conditions and methodological procedures validated in cow milk.

As a whole, this review aims to present available knowledge about the variability of main variables defining milk hygiene in these species, to understand the relationship between these variables and to establish guidelines for monitoring purposes beginning on the premises that milk quality starts with management quality (good farming practices) in the flocks. Overall, this information would be useful firstly to milk laboratories with the objective to adapt analytical methods and strategies to sheep and goat milk, and secondly to farmers, farm organizations, milk regulatory organisms, and milk industry interested in developing milk quality standards for these species, as well as the most effective on-farm management and quality assurance programs with the objective of insuring higher quality standards and optimal milk quality and final products.

Udder health is a prerequisite to milk hygiene. Mastitis, particularly subclinical and chronic, is the most prevalent group of diseases of importance to milk hygiene in dairy livestock ( Heeschen, 1987 ). Previous studies have confirmed that specific bacteriological examination of milk and milk SCC are reliable methods for detecting subclinical mastitis, and an inverse relationship between SCC and milk yield has been proved in meat ( Fthenakis & Jones, 1990a ) and dairy sheep ( Gonzalo et al. , 1994 , 2002 ; Leitner et al. , 2004a ) and goats ( Baudry et al. , 1997 ; Leitner et al. , 2004b , Pleguezuelos et al. , 2015 ).

The Fossomatic (Foss Electric, Hillerød, Denmark) and Somacount (Bentley Instruments, Chaska, MN, USA) are the most widely SCC methods used in milk-testing and interprofessional laboratories, and the performance of Fossomatic has been standardized for sheep ( Gonzalo et al. , 1993 , 2003 , 2004 ; Martinez et al. , 2003 ) and goat (Sanchez et al. , 2005; Raynal-Ljutovac et al. , 2007 ) milk. Basically, Fossomatic is a DNA-specific counter based on the principle of optical fluorescence. The dye penetrates the cell and forms a fluorescent complex with the nuclear DNA. The sample is exposed to blue light, which excites the dyed cells, making these emit red light. These red light pulses are magnified, counted by a photo multiplier detector, and multiplied by the defined working factor to compute the number of somatic cells per milliliter. In cytometry on disk, the detector counts the number of cells passing by the microscope on a rotating disk. In flow cytometry, the sample is pumped through a flow cell of very small diameter that allows only one somatic cell to pass at a time. Thus, SCC by flow cytometry is slightly higher than that obtained by disk cytometry and direct microscopy ( Gonzalo et al. , 2004 ) because cells aggregated in cell clusters are forced to pass in a single row through a very small capillary tube before being counted. Automation of this process means that large numbers of samples can be analyzed per hour in milk laboratories.

In sheep milk, the comparison of the Fossomatic method vs microscopic reference method for the different analytical conditions (Table 1) evidenced that preserved milk analyzed by flow cytometry gave optimal repeatability (s _r %=1.9 to 2.6%) and their logSCC means (5.62 to 5.67) were similar to reference value (5.63). The BR was the optimal preservative for the Fossomatic method in sheep and goat milk both in refrigeration and freezing storage ( Gonzalo et al. , 2004 ; Sánchez et al. , 2005 ). The smallest SCC always corresponded to AZ-preserved milk ( Gonzalo et al. , 2003 , 2004 ; and Sánchez et al. , 2005 ); this SCC decrease was particularly dramatic in the defrosted milk. Analytical temperature (40ºC or 60ºC) did not contribute significantly to SCC variation in sheep ( Gonzalo et al. , 2004 ) or goat ( Sierra et al. , 2006 ) milks.

Fossomatic method requires, however, cumbersome and expensive equipment, so less expensive alternative methods and portable devices for on-farm or milk plant uses would be of great interest in mastitis control or quality strategies. In this sense, DeLaval cell counter (DCC, DeLaval Int. AB, Tumba, Sweden) is a portable, battery-operated optical cell counter that gives the measuring results in less than 1 min (about 40 s/sample). A cassette, which contains small amounts of fluorescent stain and Triton X-100 is used to collect milk before cell counting. A piston carries the sample toward a counting window that is exposed to a light source. Cell nuclei give fluorescent signals recorded in an image that is used to determine the SCC in milk. Similarly to cow milk, DCC showed suitable overall accuracy in goat milk compared to reference method ( Berry & Broughan, 2007 ); nevertheless in ovine raw milk, which has a higher dry matter than goat and cow milk, a soak time ≥ 1 min is recommended because a latent time is needed, after milk loading in the DCC cassette, for cellular membrane lysis and nuclei straining in this species (Table 1). Standard deviation of repeatability was, however, higher in DCC method than in Fossomatic method ( Gonzalo et al. , 2006b , 2008 ).

At half-udder level, ewe and goat can elicit very higher SCC associated to intramammary infections (IMI) than cattle, due probably to a higher rusticity of small ruminants. Thus, SCC values > 10,000 × 10 ³ cells/mL in case of subclinical mastitis are not infrequent in ewe and goat milk in case, for example, of mammary infections by E. coli , Str. agalactiae , Staph. aureus , or Aspergillus spp. pathogens. Cytokines played a major role in leucocyte recruitment to inflamed mammary tissue ( Persson et al. , 1996 ; Persson-Waller et al. , 1997 ) and bacterial pathogens isolated in mammary infections explained a high percentage of SCC total variance in sheep ( Gonzalo et al. , 2002 ). In addition, other infectious agents responsible for interstitial mastitis ( i.e. , arthritis encephalitis virus) were also associated with a significant increase of SCC ( Nord & Adnoy, 1997 ; Sánchez et al. , 2001 ) in goat udder halves. Regarding non-infectious factors influencing SCC variation, an important difference can be established between both species. Thus, in sheep, non-infectious factors (primary purpose, breed, flock, year, parity, lactation stage, etc.) contributed also significantly to SCC variation, but had smaller effects on SCC than bacterial infection factors ( Gonzalo et al. , 2002 ; Paape et al. , 2007 ; Kern et al. , 2013 ). On the contrary, in goats, the non-infectious factors had higher importance on SCC variation, so the SCC discrimination value was lesser in this species ( Luengo et al. , 2004 ; Raynal-Ljutovac et al. , 2007 ; Koop et al. , 2010b ).

Some studies ( Ariznabarreta et al. , 2002 ; Gonzalo et al. , 2002 ) allowed different infective situations to be differentiated according to SCC and isolate percentage. Isolates of main mammary pathogens and their associated SCC are shown in Table 2 for subclinical mastitis in small ruminants. Overall, goats presented higher prevalence of coagulase-positive staphylococci (22.4 vs 12.4%), gram-negative pathogens (6.3 vs 1.5%), and Mycoplasma spp. (3.1 vs 1.7%) than sheep. These three groups of organisms also elicited higher SCC than other mammary pathogens in both dairy species. Coagulase-negative staphylococci (CNS) were the most prevalent bacterial group, which elicited a large variability in SCC (4.94 to 6.27 log cells/mL in sheep, and 5.48 to 6.48 log cells/mL in goat) (Table 2). These results were consistent with the evidence that some CNS species, such as S. simulans , S. epidermidis , S. haemolyticus and S. chromogenes , in goat and sheep, or S. caprae in sheep, induced high SCC ( Deinhofer, 1993 ; Sánchez et al. , 1999 ; Pengov, 2001 ; Ariznabarreta et al. , 2002 ) and even caused clinical mastitis ( Fthenakis & Jones, 1990b ) in a way similar to the major pathogens. CNS along with other pathogens were also responsible of chronic infections associated to mammary nodular and indurative fibrous lesions in which pathogens remain for long periods. For example, in a recent study carried out in 878 ewes distributed in 10 flocks and with test-day SCC records > 2 × 10 ⁶ cells/mL, CNS were responsible of 74.9% of mammary infections, which were associated to a high prevalence (37.8%) of nodular mammary indurations evidenced by udder palpation after milking ( Linage et al. , 2017 ). In spite of these facts, studies that focus on pathogenicity mechanisms of Staphylococcus spp. are centered mainly in coagulase-positive species ( Moroni et al. , 2005 ; Linage et al. , 2012 ).

SCC test-day recording is a useful tool to monitoring subclinical mastitis and enable relevant information for effective mastitis control strategies in dairy flocks (chronic mastitis, culling, selective dry therapy, etc.). Spanu et al. (2011) evidenced that half-udder milk samples obtained from ewes with 3 or more test-day SCC records ≥400 × 10 ³ cells/mL in the previous lactation were 5.6–7.5 times more likely to be microbiologically positive for mastitis pathogens as compared to milk samples obtained from ewes with test-day SCC records below that threshold. Therefore, identification of chronic mastitis based on test-day recording is possible and this practice should be included within mastitis control programs. In addition, test-day SCC can be used to estimate the magnitude of milk yield and economic losses. Several investigations put into question, however, the relevance of SCC as a valid tool for estimating the magnitude of milk yield losses in dairy goats ( De los Campos et al. , 2006 ; Koop et al. , 2010b ). According to these authors, the negative correlations between SCC and milk yield might be confounded with the stage of lactation effect because the inverse-lactation curve between both variables ( Rota et al. , 1993 ). Therefore, a dilution effect seems to play an important role in the negative association between SCC and milk yield. Nevertheless, a more exhaustive study of the relationships between milk yield and contents and SCC in a large goat population and within stage of lactation, in order to avoid the dilution or concentration effect, evidenced important economic losses (9.5% to 31.6%) as SCC level increased within each month of lactation, from mathematical models corrected for systematic factors influencing test-day recording ( Barrón-Bravo et al. , 2013 ; Pleguezuelos et al. , 2015 ). Finally, some studies suggest that a reduction in SCC can be achieved by selection in compatibility with improving milk production and udder type and teat traits in dairy sheep ( El-Saied et al. , 1998 ; De la Fuente et al. , 2011 ) and goats ( Rupp et al. , 2011 ).

In recent years, the discovery in the milk of a beneficial or probiotic flora with different antimicrobial properties against mammary pathogens ( i.e. , mechanisms of competitive exclusion, production of H ₂ O ₂ and bacteriocins, aggregation properties, etc.) has provided grounds to consider the mastitis as a dysbiosis of a milk microbial ecosystem in woman and cow milks ( Espeche et al. , 2012 ; Fernández et al. , 2014 ). In this context, further investigations will be necessary to know the bacterial ecosystem in small ruminant milk and the importance of probiotic flora in the mastitis establishment.

At bulk tank level, BTMSCC is the principal tool used by technicians and farmers to evaluate and monitoring udder health in flocks. The main sources of variation of this variable have been identified in dairy sheep (Table 3). Special reference should be made to the flock factor, which explained a high percentage of BTMSCC total variance. Important differences in hygiene and management practices would bring about considerable differences in the prevalence of mammary infection and BTMSCC. In a total of 309 dairy sheep flocks of Castilla y León region (Spain), yield-weighted annual BTMSCC among flocks ranged from 250 × 10 ³ cells/mL and 2,500 × 10 ³ cells/mL. In this sense, recording of this variable can be used for monitoring flock udder health and as a criterion for milk payment schemes; nevertheless interpretation criteria are different from dairy cattle, due to increased SCC response to mammary infection of small ruminants. Although EU has yet to regulate values in ewe and goat milk, a practical approach of flock mammary health according to BTMSCC level can be established in both species: a) Good or acceptable flock mammary health with ewe/goat IMI prevalence < 25%: BTMSCC < 750 × 10 ³ cells/mL; b) Deficient flock mammary health with ewe/goat IMI prevalence > 50%: BTMSCC > 1,500 × 10 ³ cells/mL; and c) Intermediate situation with ewe/goat IMI prevalence between 25% and 50%: BTMSCC between 750 × 10 ³ cells/mL and 1,500 × 10 ³ cells/mL ( Gonzalo et al. , 2000 ; Silanikove et al. , 2014 ). Other authors consider, however, in the case of goats, that non-infectious factors ( i.e. , parity and stage of lactation) need to be considered when establishing legal SCC limits ( Paape et al. , 2007 ). Month within herd was also a relevant factor of variation in BTMSCC that explained the variation in IMI prevalence throughout lactation in each flock, although a BTMSCC dilution effect was also possible in more productive months ( i.e. , spring) which elicited lesser BTMSCC. Breed differences in BTMSCC were also evidenced associated with the milk yield level; greatest BTMSCC corresponded to more productive breeds (Table 3). In this sense, greater resistance to mastitis was reported by Dario & Bufano (1991) in less-productive breeds compared with more selected breeds, in which the IMI incidence would be probably greater. For milking type, machine milking elicited a lower BTMSCC than hand milking (Table 3); the inferior hygiene conditions of hand milking compared with machine milking makes it more difficult for farmers to reach milk quality standards as measured by BTMSCC. Within machine milking, BTMSCC was greater in bucket system than in parlor system; differences between dead-ended and looped milk lines were not statistically significant (Table 3). Moreover, high pulsation speeds and low milking vacuum levels were associated with small BTMSCC ( Gonzalo et al. , 2005 ). Clinical outbreaks of CA increased BTMSCC both in dairy sheep (Table 3) and goats ( Corrales et al. , 2004 ). This chronic disease is a serious limitation to improve milk hygiene in dairy flocks, therefore the implementation of programs for its prevention and eradication should be prerequisites for optimizing BTMSCC and mastitis control strategies in endemic areas. Dry therapy was other important source of variation of BTMSCC; when dry therapy was implemented BTMSCC was less than when this practice was not used, as dry therapy was efficient in reducing IMI prevalence, BTMSCC and BTMTBC, improving ewe milk hygiene ( Linage &Gonzalo, 2008 ; Gonzalo et al. , 2010 ). Systematic dry therapy was also an efficient method for cure of subclinical mastitis and SCC control in goats, although only references at udder-half level are disposables in scientific literature ( Poutrel et al. , 1997 ; Bastan et al. , 2015 ).

The evidence of high SCC (SCC test-day and BTMSCC records) in Mediterranean countries ( Gonzalo et al. , 2005 , 2010 ; Clément et al. , 2015 ), together with the important economic losses associated to subclinical mastitis in small ruminant flocks ( Fthenakis & Jones, 1990a ; Gonzalo et al. , 2002 ; Pleguezuelos et al. , 2015 ), highlight the need for implementing mastitis control programs within flocks and cooperatives, in order to improve the milk hygiene and to increase economic return for producers. Technicians and farmers must always keep in mind that mammary health of the flock is a necessary and previous condition to implementation of milk yield breeding schemes. In addition, quality assurance programs or certification schemes for dairy flocks are important tools for improving BTM quality. Because a positive relationship has been stated between BTM quality and dairy farm audits ( Velthuis & Asseldonk, 2011 ; de Garnica & Gonzalo, 2013 ), small ruminant cooperatives have currently implemented farm risk audits for monitoring mammary health in the flocks. Each farm audit record includes binary checklist items (valuated as 0: approved or 1: rejected) distributed over a number of categories or main critical areas. Each flock and each category within flock accumulates a number of risk points, which should progressively be reduced over time. Moreover, recording of BTM quality can be used to preselect high-risk flocks to be audited more frequently.

Specifications for count of different bacterial groups ( i.e. , BTMTBC, thermoduric, phychrotrophic, coliforms, GPCNC) follow American Public Health Association recommendations ( White et al. , 1992 ; de Garnica et al. , 2011 , 2013a ); lactate-fermenting Clostridium spp. spore are enumerated by the most probable number technique in Bryant and Burkey broth specific media under anaerobiosis ( Arias et al. , 2013 ) or by optimized membrane filtration technique ( Reindl et al. , 2014 ); and Staphylococcus spp. and other mammary pathogens can be counted after culture in specific media according with de Koop et al. (2010a) , de Garnica et al. (2011 , 2013a ), and Linage et al. (2012) .

Storage and preservation strategies (unpreserved or AZ-preserved milk) significantly contributed to changes in microbiological variables in BTM (TBC, thermoduric, psychrotrophic, coliforms, E. coli , Strep. agalactiae , and Staph. aureus ). Unpreserved milk stored at 4ºC increased the concentration of TBC, psychrotrophic and coliform bacteria over time. The initial concentration of thermoduric, and pathogens ( E. coli, Strep. agalactiae and Staph. aureus ) remained, however, invariable because probably of inhibition competitive effect of the main bacterial groups. Storage at 4ºC of AZ-preserved milk was a suitable method to maintain the initial concentrations for all bacterial groups and species, over 96 h. Freezing at -20ºC could be appropriate for analysis of mesophilic, thermoduric and psychrotrophic bacterial groups, but significantly decreased the viability of coliforms and pathogenic species in ovine milk ( de Garnica et al. , 2011 ). Similarly to ovine milk, AZ-preserved samples stored at 10ºC or 4ºC were appropriate for goat BTMTBC by Bactoscan flow cytometry method for up to 24 h and 11 d postcollection, respectively. Nevertheless, refrigerated unpreserved samples showed high goat BTMTBC at 24 h postcollection ( Sierra et al. , 2009c ).

BTMTBC is a practical variable related to hygiene of milking and facilities (Table 3). The goal BTMTBC of different legal limits or payment-by-quality schemes has been proposed by different countries. For small ruminants, EU determines the limits for this variable in 500 × 10 ³ cfu/mL in the case of non-pasteurized milk, and 1,500 × 10 ³ cfu/mL when milk is to be pasteurized ( EC, 2004 ). Both limits are, however, extremely high, indicatives of very deficient hygiene practices, and lead to a drastic reduction of milk shelf time and a poor quality of dairy products. In this sense, it is entirely possible to maintain annual average values of BTMTBC below 150 × 10 ³ cfu/mL even in hand milked flocks if milking hygiene is appropriate and refrigeration tank is working properly ( de Garnica & Gonzalo, 2013 ). Only very extensive livestock systems, in which the milking is carried out in areas where electrical power is not available, would justify those limits. Therefore, a reconsideration of hygiene policy about milk BTMTBC thresholds would be desirable in small ruminants, as high BTMTBC limits should be the exception rather than the rule.

The correlation between BTMTBC and BTMSCC was moderate ( i.e. , r=0.24 to r=0.42) ( Gonzalo et al. , 2005 ; Koop et al. , 2009 ; de Garnica et al. , 2013a ) suggesting that IMI is probably an important factor driving this association in small ruminants. In this sense, programs to improve udder health ( i.e. , dry therapy) may have a significant effect on both BTMSCC and BTMTBC (Table 3; Gonzalo et al. , 2010 ; de Garnica et al. , 2013a ). Other bacterial groups, such as thermoduric, psychrotrophic, and coliform were significantly related to BTMTBC, whereas GPCNC and staphylococci counts were correlated with both BTMTBC and BTMSCC variables ( Koop et al. , 2010a ; de Garnica et al. , 2013a ). Highest counts were for psychrotroph and coliform groups in dairy sheep flocks (Table 4), and a medium-high correlation ( i.e. , r=0.51) was found between both variables (de Garnica et al. , 2013a ), indicating that poor cleaning practices in the flocks tend to select for less-resistant organisms, such as gram-negative rods. Season was an important effect associated with the variation of BTM prevalence for specific bacterial groups and pathogens ( i.e. , S taph. aureus , E. coli ). Psychrotrophic and coliform bacterial groups ( E. coli included) and Staph. aureus were highest in winter in connection with more dirty beds and udders due to the wetter weather (ambient contamination) and with beginning of milking season (contagious-pathogen infections) ( de Garnica et al. , 2013a , b ). Staph. aureus and E. coli organisms showed also higher occurrences in hand milked flocks, and Staph. aureus was also higher in flocks in which dry therapy was not implemented.

Psychrotrophic flora of milk becomes predominant in milk plants ( i.e. , silo milk) exceeding 7.00 log cfu/mL from 72 h postcollection ( de Garnica et al. , 2011 ). Because of this high concentration of psychrotrophic flora and its proven relationship with a high incidence of lipolytic and proteolytic activities of Pseudomonas spp. and Acinetobacter spp. on milk and cheese components ( Hantsis-Zacharov & Halpern, 2007 ; Capodifoglio et al. , 2016 ), it is advisable to collect and process the milk ( e.g. , delivery, heat treatment) in the shortest time possible after milking to prevent significant spoilage of milk and dairy products. This bacterial group increased with increasing total bacterial load in BTM ( de Garnica et al. , 2013a ), so it can be considered as ubiquitous in all small ruminant herds. Thermodurics are a contaminant group of milk that contains thermophilic spore-forming bacteria which can survive pasteurization during dairy-product processing causing dairy-product spoilage in the post-processing. The concentration of this group in milk is, however, generally low (Table 4). The diversity of GPCNC has been recently investigated by 16S rDNA sequencing in a total of 192 isolates from ewe BTM. This molecular approach enabled the identification of 23 species of GPCNC of sanitary and technological importance within the Enterococcus, Streptococcus, Lactococcus, Aerococcus and Trichococcus genera. Some of these species are potentially pathogenic nosocomial organisms ( i.e. , Enterococcus faecalis ) which have developed mechanisms of antimicrobial resistance ( i.e. , vancomycin resistance) ( de Garnica et al. , 2013d , 2014 ), so control strategies are needed in the food chain to optimize food safety in dairy products. Regarding Clostridium spp. spore counts in BTM, a value of 10 ³ spores/L can be considered an acceptable threshold for the incentive payment systems ( Pirisi et al. , 2007 ). The main risk factors are related to housing conditions, feeding characteristics and milking parlour hygiene. Indeed, several studies in ewe and goat BTM showed that the main on-farm management risk factors associated to an increase of spore counts > 10 ³ spores/L were farm-made total mixed ration, the silages and wet brewer’s grains used for feeding, and the presence of dust in the milking parlour ( Arias et al. , 2013 , 2016 ; Reindl et al. , 2014 ). Mean values found in scientific literature for ewe BTM ranged from 2.82-4.16 log spores/L ( Scintu et al. , 2004 ; Garde et al. , 2011 ; Arias et al. , 2013 ), whereas maximum spore counts of 2.48 log spore/L and 3.70 log spore/L were described in goat BTM from suppliers without and with silage feeding, respectively ( Reindl et al. , 2014 ).

On the whole, specific (thermoduric, psychrotrophic, coliforms, GPCNC, and Clostridium spp. spore) and total (SCC and TBC) BTM counts were negatively related with good-practice audit rating of the flocks ( de Garnica & Gonzalo, 2013 ); therefore, the development of infrastructures for the improvement of BTM hygiene and flock management based on both management assurance and analytical surveillance programs, is feasible. Along with BTMTBC, some countries ( i.e. , France) have included BTM counts of coliforms and Clostridium spp. spore as criteria for milk quality payment system through a combined index including these three variables ( Pirisi et al. , 2007 ). In addition, the information relative to prevalence results for specific pathogens ( i.e. , Staph. aureus , E. coli , or Enterococcus faecalis ) or spoilage organisms ( i.e. , Pseudomonas spp., Clostridium spp.) isolated from BTM, could be incorporated to traceability systems with the objective of implementing preventive programs ( i.e. , mastitis control, flock hygiene, feeding, or milk pasteurization).

The use of screening test is a reliable and rapid option to detect antimicrobials or chemical contaminants in milk, although in case of a conflict involving differences of criteria it will be necessary to corroborate the results with official methods ( i.e. , HPLC) for positive samples ( Kang et al. , 2016 ).

Water addition is an infringement that can be considered as a hygiene problem if water is unsafe or has an unknown origin. The addition of water affects the freezing point (FP) of milk so it is used in the industry as a criterion for economic sanction. Although sheep and goat FP is influenced by many factors ( i.e. , total non-fat solids, breed, stage of lactation, season, and pasteurization) ( Antunac et al. , 2001 ; Park et al. , 2007 ; Mayer & Fiechter, 2012 ; Janstová et al. , 2013 ), both analytical methodology and preservation strategies have major effects on inter- and intra-laboratory quality testing, as well as for routine sampling strategy. Indeed, FP obtained by Milkoscan FT 6000 has a high repeatability and reliable accuracy compared to thermistor cryoscope method, which is the reference method according to IDF (2002) . Thus, Sánchez et al. (2007) suggested that Milkoscan method can be used for screening purposes in BTM samples under different analytical conditions. Moreover, within each method, the milk FP must be interpreted by taking into account the type and concentration of preservative used, as the effect of the preservative could lead to errors regarding to added water percentage. The best analytical condition for FP determination using Milkoscan FT 6000 involved the preservation with BR 0.2 g/L, while the increase in the concentration of sodium azide in the AZ formula contributed to an important reduction in the FP recorded ( Sánchez et al. , 2007 ). Limits of -540 mºC in goat milk, and -560 mºC in ewe milk have been proposed for the detection of added water, indicating a good ability of the instrumental method for FP screening in milk of both species. Besides the intentional addition of water into milk, the water of milking machine final rise, which remains in pipelines of milking installation, is the main cause of water presence in milk. Therefore, an adequate slope of milk pipelines toward the final unit, as well as a proper water drainage before milking are essential to prevent this problem.

Sediment in milk is associated generally with poor milking hygiene procedures and environmental conditions, and filtering methodologies have been developed to quantity the sediment level in milk. Milk sediments can get stuck in very small capillary tube of flow cytometers and cause failures or malfunctions of automatic analytical devices in milk laboratories. To avoid this fact, many laboratories do not analyze BTM with macroscopic sediments or from flocks with deficient hygiene conditions, and this creates problems between industry and farmers. An appropriate education and training in cleanliness practices ( i.e , animals, environment, milk filters maintenance or renewal, milk tank cleaning, etc.) is needed in such cases.

Other microbiological ( i.e. , verotoxigenic E. coli, Listeria monocytogenes, Salmonella spp.), chemical ( i.e. , aflatoxins, pesticides) or prions ( i.e. , scrapie) milk contaminants are possible and must be prevented and suppressed. In this sense, milk testing and interprofessional laboratories should include appropriate analytical methods for traceability and control; nevertheless, cost-effective screening and analytical systems have not yet been implemented for a great majority of contaminants. In the case of AF, screening commercial ELISA (Enzyme-Linked ImmunoSorbent Assay) kits are available for milk M1 AF quick detection. The Íscreen Afla M1 test kit (Tecna s.r.l., Trieste, Italy), the Aflatoxin M1 test kit (Euro-Diagnostica B.V, Arnhem, The Netherlands), the Transia Plate Aflatoxin M1 (Raisio Diagnostics SAS, Lyon, France), or the Ridascreen Aflatoxin M1 30/15 test kit (R-Biopharm AG, Darmstadt, Germany) could be reliable options to obtain acceptable results for screening M1 AF in small ruminant milk, the recovery percentages being between 78% and 130% for a range of M1 AF in milk between 25 and 100 ng/kg ( Rubio et al. , 2009a ). The combined use of refrigeration and preservation (AZ or BR) could slightly interfere the recovery of M1 AF ( Rubio et al. , 2009b ), while freezing of unpreserved milk did not modify the accuracy of ELISA tests for this AF ( Rubio et al. , 2008 ). Receptor-binding assays ( i.e. , Charm MRL Afla M1; Charm Sciences Inc., Lawrence, MA, USA) are also used in ovine and caprine milk plants to detect M1 AF at EU-MRL concentration (50 ng/kg) in 15 min. Results published on occurrence of AF in large small ruminant populations are scarce; for example, in a study to test M1 AF involving 407 BTM samples from farms and 82 silo milk and curd samples from cheese factories, collected from southeast Spain during 2 yr, 99.3% of the bulk tank samples and 98.8% of the silo milk and curd samples from cheese factories had M1 AF levels below EU-MRL for milk ( Rubio et al. , 2011 ).

In small ruminants, milk hygiene has serious economic and sanitary consequences for farmers, dairy industry and consumers due to its relationship with milk payment-schemes, production loses, discarded milk and environmental pollution, profitability and operational cost, milk lifetime, manufacturing properties and marketing, food safety and other effects on final product. Milk quality is the product of a global concept of animal production, which implies the need of using specific infrastructures for its surveillance. Analytical validation for the milk of these species is needed in order to guarantee the global accuracy for different milk variables and to interpret correctly the analytical results. Suitable control strategies based on quality assurance and analytical surveillance programs and periodical audits within herds would be the best guarantee to prevent high SCC, pathogens and indicator bacterial groups, occurrence of antimicrobial residues, aflatoxins, and other contaminants in BTM, as well as to increase the economic returns for producers and processors. Further investigations will be necessary to know in deep the balance established between beneficial and pathogenic bacteria within the milk microbial ecosystem, as well as their possible application to development of non-antibiotic mastitis control strategies in small ruminants. In addition, new cost-effective screening and analytical systems should be developed to detect a higher number of contaminants in small ruminant milk. As a whole, a better understanding of milk hygiene variables will allow producers to adopt production practices that will result in a safer and higher-quality milk that meets goal quality standards for industry and consumer acceptance.

The authors thank the collaboration of staff and farmers of Consortium for Ovine Promotion (CPO), Villalpando (Zamora, Spain).