INTRODUCTIONEcuador produces between 5.5 to 5.8 million litersof milk on a daily basis and this production has beensteadily growing in recent years [1]. About 75% of milkis produced in the Andean region, mostly by smallassociations of farmers, far from large urban centers [2].The quality of milk determines the quality of dairyproducts. It refers to the content of microorganisms(pathogenic or not) and somatic cells, as well as thepresence of antibiotics and medicines [3]. Milk qualityis guaranteed by the health of the herd, as well as goodmanagement and milking practices (GMMP). To checkthe microbiological quality of raw milk, dairy producerscommonly use the counts of total bacteria (TBC),somatic cell counts (SCC), and the methylene bluereduction test (MBRT) [4].The presence of somatic cells in milk has beenmainly related to the increase of white cells (leukocytes)as a result of an immune system’s response to mastitis.It is a livestock disease caused by the inflammationof the udder due to the action of pathogenic microorganismssuch as Staphilococcus aureaus, Streptococcus141Pérez-Lomas M. et al. Foods and Raw Materials, 2020, vol. 8, no. 1, pp. 140–148dysgalactiae and Streptococcus agalactiae. Mastitisalters the chemical composition of milk and decreases itsyield [5, 6].However, the TBC and SCC are usually carried outin accredited laboratories located in urban areas, farfrom small farmers and their associations. They areexpensive for milk producers and, moreover, the latterhave to wait quite long for the results before they canimprove the microbiological quality of their milk.The MBRT, on the other hand, is an old but effectivemethod which has been correlated, with some success,with the total microbial load and, therefore, with themicrobiological quality of milk [7–9]I. It is a simple andfast method, although sometimes it lacks the expectedprecision.The goals of this work were to evaluate the quality ofmilk produced by a group of cattlemen’s associations inthe province of Carchi, Ecuador, and to suggest a cheap,fast, and reliable alternative methodology that wouldallow the associations to evaluate the quality of theirmilk in situ.STUDY OBJECTS AND METHODSHerd size and geographic location. The studyinvolved six cattlemen’s associations (A–F) located inthe Andean province of Carchi in Ecuador. There were11 small farmers in Association A, 27 in AssociationB, 16 in Association C, 20 in Association D, 15 inAssociation E and 19 in Association F. Their milk wassampled for eight months, from October 2016 to May2017. As a result, 709 samples were taken from a herdof 814 milking cows (34 from Association A, 235 fromAssociation B, 50 from Association C, 120 fromAssociation D, 230 from Association E, and 145 fromI ISO 4833-2:2013. Microbiology of the food chain – Horizontalmethod for the enumeration of microorganisms – Part 2: Colonycount at 30°C by the surface plating technique. Geneva: InternationalOrganization for Standarization; 2013.Association F). All the samples were analyzed for totalsolids, total protein, fat, acidity and density. The totalbacteria and somatic cells were also counted.Physicochemical and microbiological propertiesof the samples. The determinations of total solids, totalprotein, fat, acidity and density, as well as somatic cells(SC) and total bacteria counts (TBC) were performedin an accredited laboratory of the Phyto- and Zoo-Sanitary Regulation and Control Agency of Ecuador(AgroCalidad) (www.agrocalidad.gob.ec) located inTumbaco (Quito, Pichincha, Ecuador) [10]II,III,IV,V. Thesomatic cell count (SCC, SC/mL) was performed in aFoosmatic™7 (Foss, Hilleroed DK-3400, Denmark)according to the standard procedureVI. The totalbacterial count (TBC, CFU/mL) was performed in aBactoScan™FC+ (Foss, Hilleroed DK-3400, Denmark),obtaining values equivalent to those that would beobtained from a standard plate count (SPC)VII.Standard and extended methylene blue reductiontest. The standard methylene blue reduction test(MBRT) and the 24 h extended methylene blue reductionII ISO 6731:2010 [IDF 21:2010]. Milk, cream and evaporated milk:determination of total solids content (reference method). Geneva:International Organization for Standarization; 2010. 5 p.III ISO 8968-1:2014 [IDF 20-1:2014]. Milk and milk products –Determination of nitrogen content – Part 1: Kjeldahl principle andcrude protein calculation. Geneva: International Organization forStandarization; 2014. 18 p.IV ISO 1211:2010 [IDF 1:2010]. Milk – Determination of fat content– Gravimetric method (Reference method). Geneva: InternationalOrganization for Standarization; 2010. 18 p.V ISO/TS 11869:2012 [IDF/RM 150:2012]. Fermented milks –Determination of titratable acidity – Potentiometric method. Geneva:International Organization for Standarization; 2012. 7 p.VI ISO 13366-2:2006 [IDF 148-2:2006]. Milk – Enumerationof somatic cells – Part 2: Guidance on the operation of fluoroopto-electronic counters. Geneva: International Organization forStandarization; 2006. 13 p.VII ISO 4833-1:2013. Microbiology of the food chain – Horizontalmethod for the enumeration of the microorganisms – Part 1: Colonycount at 30°C by the pour plate technique. Geneva: InternationalOrganization for Standarization; 2013. 9 p.Table 1 Types, characteristics, and possible causes of clots obtained from the extended methylene blue reduction test (MBRTe)Type Classification Characteristics Causes ImageMBRTe-I Solid or liquidhomogenous clot*Homogeneous clot, with an acidicodor and taste, without cracks orfissures, white in color, withoutor with few bubblesPresence of Lactobacillus spp.or antibiotics in milkFigure 1aMBRTe-II Clumpedheterogeneous clotHeterogeneous clot with lumps,with a whitish, yellowish serum,or other abnormal colorsProduced by germs with bitter tastes andunpleasant odors; mastitic milk at the endof lactation; or milk cooled for a long timeFigure 1bMBRTe-III Gaseousheterogeneous clotHeterogeneous clot with bluishshades and numerous bubblesand gaseous groovesColiform bacteria; milk obtained and preservedin poor sanitary conditions or refrigerated fora long timeFigure 1cMBRTe-IV Clumped + gaseousheterogeneous clotHeterogeneous clot with lumpsand numerous bubbles or gaseousfurrowsCombined action of coliform bacteria and somaticcells; milking and conservation of milkwithout complying with the GMMP; masticmilk or milk refrigerated for a long timeFigure 1d* If it is liquid, check the presence of antibiotics or substances that can inhibit microbial growth (such as detergents, pesticides, etc.)142Pérez-Lomas M. et al. Foods and Raw Materials, 2020, vol. 8, no. 1, pp. 140–148test (MBRTe) were carried out in the associations’ ownlaboratories. For the quantitative MBRT, 10 mL of asample was incubated at 37°C with 0.5 mL of methyleneblue and the time (in hours) for the blue coloration todisappear was measured according to the techniquedescribed in ISO 4833-2:2013I. We used the methyleneblue reagent produced by Merck KGaA (Darmstadt,Germany). The samples were incubated at 37°C in aconventional water bath (Thermo Scientific™ TSGP10,Waltham, Massachusetts, USA).The procedure of the qualitative MBRTe was similarto that of the MBRT, but the samples were incubated for24 h. As a result, we obtained clots of the following fourtypes (Table 1, Fig. 1): a homogeneous solid or liquid clot(MBRTe-I); a heterogeneous lumpy clot (MBRTe-II);a heterogeneous gaseous clot (MBRTe-III), and a heterogeneouslumpy + gaseous clot (MBRTe-IV).The MBRT is, therefore, a quantitative test(measured in hours), while the MBRTe is a qualitativetest (one of the four possible sample types after 24 hincubation with methylene blue).Statistical Analysis. The statistical analysis wasapplied using the free statistical package R version 3.6.1(2019-07-05).RESULTS AND DISCUSSIONWe analyzed 709 samples for eight continuousvariables (SCC, TBC, MBRT, fat, protein, total solids,density and acidity) and three categorical variables:(1) eight dates (Oct-16, Nov-16, Dec-16, Jan-17,Feb-17, Mar-17, Apr-17, and May-17);(2) six associations (A, B, C, D, E, and F); and(3) four MBRTe clots (MBRTe-I, MBRTe-II, MBRTe-III,and MBRTe-IV).The Lilliefors test (a normality test based on theKolmogorov-Smirnov test) [11–13] was used to explorethe continuous variables, and none of them showed anormal distribution of the samples (P < 0.05).Recently, a similar finding has been reported in astudy conducted to determine the quality of milk (totalbacterial and somatic cell counts) among small livestockproducers where the values obtained did not follow anormal distribution [14]. This is probably due to the nonhomogeneityof the samples, the differences betweenthe producers with respect to compliance with goodpractices, as well as uncontrolled factors that fall outsidethe framework of the studies.The Kruskal-Wallis rank-sum test was performedto establish the influence of categorical variables overcontinuous variables [15, 16]. A pairwise comparisonwith the Wilcoxon nonparametric rank-sum test wasused to determine which of the associations or MBRTetypes differed from each other (P < 0.05) for eachspecific continuous variable [17]. Different lettersnear each of the magnitude values showed significantdifferences (P < 0.05).In this study, the values of acidity, total solids,MBRT, SCC and TBC differed significantly (P < 0.05)between the associations, while density, protein and fatconcentrations were not different (P > 0.05) (Fig. 2).As we can see in Fig. 2, only Association A, whichfully implemented the GMMP, showed a better qualityof fresh milk during the whole period. Association B,which began to implement the GMMP during the studyperiod, achieved high quality in the final months of thestudy. Associations C, E and F are still in the processof organizing their quality assurance system, and theirresults oscillate between regular and low quality. Finally,Association D always had contamination problems andshowed poor quality milk, so all of their work protocolsneed revising.The values of SCC, TBC, MBRT, fat, total solids,and acidity were significantly influenced by the type ofMBRTe (P < 0.05), whereas there were no differences(P > 0.05) for the protein content and density (Fig. 3).The significant difference (P < 0.05) observed inthe fat content between the MBRTe-II and MBRTe-IIIsamples (Fig. 3c) could be due to a high concentrationof somatic cells and a low concentration of totalbacteria in the MBRTe-II sample group. In fact, themethod of fat determination presupposes the additionof sulfuric acid which causes the breakdown of somaticcells incorporated into milk fat. It was also possiblethat exogenous bacteria species that contaminatedmilk, which were present in the MBRTe-III samples,exerted a greater lipolytic effect on the fat and loweredits concentration in milk, compared to the rest of theMBRTe samples.In Fig. 3c, we can observe an increase in acidity anda decrease in total solids when moving from MBRTe-I(а) (b) (c) (d)Figure 1 Four types of clots. (а) MBRTe-I (homogeneoussolid/liquid clot), (b) MBRTe-II (lumpy clot), (c) MBRTe-III(gaseous clot), and (d) MBRTe-IV (lumpy + gaseous clot)143Pérez-Lomas M. et al. Foods and Raw Materials, 2020, vol. 8, no. 1, pp. 140–148to MBRTe-IV. This trend seems to be associated witha combined increase in the microbial load and somaticcells in these groups, enhancing the presence of organicacids and therefore leading to higher acidity, and adecrease in carbon sources, such as lactose, leading tolower total solids.When comparing the magnitudes of the qualitativeMBRTe and the quantitative MBRT with the SCC andTBC values in a Kruskal-Wallis rank-sum test, we cansee that unlike the MBRT, which is only significantlyinfluenced (P < 0.05) by TBC, but not SCC (P > 0.05),the qualitative variable of MBRTe correlates significantly(P < 0.05) with both the TBC and SCC values.In the proposed MBRTe test, some of the samplesincubated with and without the presence of methyleneblue had a similar behavior and formed the same type ofclot after 24 h. This finding suggests that the presenceof the methylene blue dye does not play the same role asit does in the MBRT test. However, there is a need formore detailed experiments to corroborate the influenceor necessity of this dye in the MBRTe test. They needto use the same samples, incubate them under the sameconditions for 24 h and then observe the type of clotforming after that time.(c) (d)(а) (b)Figure 2 Average values of (a) acidity (°D), (b) total solids (g/100 mL), (c) MBRT (h), (d) SCC (SC/mL), and (e) TBC (CFU/mL)for each association during eight months of the study. The dashed red lines represent the values that delimit the thresholds of goodquality, regular quality and poor quality of milk or the minimum acceptable values by the Ecuadorian standards. Different lettersmean statistically significant differences (P < 0.05)(e)××144Pérez-Lomas M. et al. Foods and Raw Materials, 2020, vol. 8, no. 1, pp. 140–148When correlating the MBRT values with the TBC, orvice versa, we can observe similar correlations to thosepreviously reported by other authors [8, 9], althoughwith somewhat lower correlation coefficients R² (Fig. 4).Thus, the qualitative MBRTe not only would allowus to assess the microbiological quality of milk samplesthrough TBC values, but it could also detect a healthydairy herd (< 310 000 SC/mL) or the presence of mastitisin its preclinical (310 000 ≤ SCC ≤ 700 000 SC/mL) orclinical (> 700 000 SC/mL) stages, which is impossibleto do with the standard MBRT test.Fig. 5 shows the distribution of the MBRTe samplesFigure 3 Relationship between average values of (a) TBC + SCC, (b) MBRT, (c) Fat (g/100 mL), and (d) Total solids (g/100 mL) +Acidity (°D) and the MBRTe-types. Equal letters mean no significant differences (P < 0.05) according to the Wilcoxonnonparametric rank-sum test [17](c) (d)Figure 4 Correlation between (a) MBRT vs log10 (TBC) and (b) log10 (TBC) vs MBRT. The dashed red lines represent the valuesthat delimit milk quality thresholds(а) (b)(а) (b)SCClog10, TBC145Pérez-Lomas M. et al. Foods and Raw Materials, 2020, vol. 8, no. 1, pp. 140–148Figure 5 Distribution of MBRTe samples in each association (A‒F) showing the herd health and compliance with good practices(A) (B) (C)(D) (E) (F)Figure 6 Correlation between the MBRTe and TBC + SCC (CFU or SC/mL, respectively). (a) MBRTe-I: n = 265 (37% of the total);(b) MBRTe-II: n = 130 (18% of the total); (c) MBRTe-III: n = 99 (14% of the total); (d) MBRTe-IV: n = 84 (12% of the total)(а) (b)(c) (d)×× ×× ××××146Pérez-Lomas M. et al. Foods and Raw Materials, 2020, vol. 8, no. 1, pp. 140–148in each association. As we can see, the associations withthe highest proportions of MBRTe-I, compared to theother types, had the best quality milk.As we can see in Fig. 5, Association A providedbetter milk quality than the rest of the associations. Thelowest-quality milk was produced by Association D.This means that the health of its herd and the proceduresfor milking, handling, and storage of fresh milk shouldbe reviewed.When correlating the TBC and SCC values with thequalitative MBRTe, we observed that the MBRTe couldadequately predict not only the samples with a highbacterial load, but also those with a significant presenceof somatic cells. The latter might indicate preclinical orclinical mastitis in the herd (Fig. 6).As we can see in Fig. 6a, more than 94% of theMBRTe-I samples had SC values below 310 000 SC/mLand TBC values below 300 000 CFU/mL. Fig. 6b showsthat more than 94% of the MBRTe-II samples had TBCvalues below 300 000 CFU/mL, while 57% of themhad SC values between 310 000 and 700 000 SC/mL,which could indicate a preclinical condition of mastitis.Moreover, 42% of the MBRTe-II samples had SC valuesof over 700 000 SC/mL, which suggests the presenceof mastitis in at least part of the dairy herd. Of theMBRTe-III samples, 97% had SC values below310 000 SC/mL, which indicates a healthy dairy herd,without mastitis problems.However, as we can see in Fig. 6c, 32% of thosesamples showed moderate values of microbial contamination(between 300 000 and 600 000 CFU/mL)and 67% of them had high values (> 600 000 CFU/mL).These data suggest that the samples came from ahealthy dairy herd, but the GMMP were not followedproperly. Finally, all the MBRTe-IV samples (Fig. 6d)showed moderate to high values of both TBC and SCC,suggesting a dairy herd with mastitis problems and badmanagement and milking practices. Such a productcannot be recommended for direct consumption – ithas to be carefully pasteurized before being used in themanufacture of dairy products.Likewise, we analyzed a possible relationship betweenthe physicochemical and hygienic-sanitary propertiesof the samples and the month in which these sampleswere taken (from October 2016 to May 2017). For this,a Kruskal-Wallis rank sum test was applied to each ofthe measurements made to each sample and the monthof sampling. We found that the TBC, MBRT, proteinand fat contents did not depend on the months in whichthe samples were taken. However, the determinationsof density, total solids and SCC, in at least a couple ofmonths, were influenced by the month of sampling. Todetermine the significance (P < 0.05) of these differences,we performed a pairwise comparison using the Wilcoxonnonparametric rank-sum test and the Bonferronimethod (Fig. 7), as we did with the previous categoricalvariables (type of association and MBRTe) [17].The differences associated with the month in whichthe samples were analyzed could be explained by someuncontrolled factors in the experiments. These includevariations in the periods of rain, which could influencethe type and abundance of the grass consumed by thedairy herd, and changes in the management of the herd,as well as milking and storage of fresh milk. Also,Figure 7 Dependence of average values of (a) SCC (SC/mL), (b) density (g/mL), (c) total solids (g/100 mL), and (d) acidity (°D) onthe month of sampling. Equal letters mean no significant justified (P < 0.05)(а) (b)(c) (d)×147Pérez-Lomas M. et al. Foods and Raw Materials, 2020, vol. 8, no. 1, pp. 140–148possible measures taken by the associations to dealwith mastitis problems may have been reflected in theSC values, as well as the time they were taken.To sum up, we can say that the MBRTe correctly prequalifiedfresh milk and, therefore, allowed us to suggestpossible industrial uses for it and set fair market prices(Table 2).We all know of difficulties that small cattlemen’sassociations have with assessing the microbiologicalquality of milk and detecting sub-clinical mastitis inreal time to continuously improve the quality of milkdelivered to the industry and consumers. Accreditedlaboratories that perform somatic cells and totalbacterial counts, as well as the methylene blue reductiontest (MBRT), are located in provincial cities or capitals,far away from the rural areas where most of the smallfarmers’ associations are, at least in Ecuador [16]VIII.This means that the farmers’ associations usually haveto wait a few days (an average of 3 days) for the testresults. Thus, they cannot quickly identify individualproducers that affect the milk quality of the wholeassociation to take prompt corrective measures.Moreover, the cost of such analysis in Ecuador,including transportation (for a distance of ~ 50 km), isapproximately $9.56 per sample. In contrast to that,the qualitative MBRTe takes only one day and costsapproximately $0.46 per sample. In addition, it is easyto perform and its interpretation is straightforward andsimple: fresh milk is pre-qualified as good (MBRTe-I),intermediate to good (MBRTe-II), poor to intermediate(MBRTe-III), and poor (MBRTe-IV).The qualitative MBRTe would allow us not onlyto know if the association follows good practices ofmilking, handling and storage of milk, but also toexamine the health of the dairy herd, as far as mastitisis concerned. In addition, it is a cheap test since itrequires only a conventional thermostatic bath, the bluemethylene reagent, and a set of common glass tubes. Itis significantly cheaper than modern equipment for thedetection and counting of somatic cells.The above makes the MBRTe suitable for smallassociations of livestock farmers that are isolated fromcities and towns where accredited laboratories aregenerally located.VIII NTE INEN 9:2012. Norma Técnica Ecuatoriana. ServicioEcuatoriano de normalización. Leche cruda. Requisitos [EcuadorianTechnical Standard. Ecuadorian Normalization Service. Raw milk.Requirements]. Quito: INEN; 2012. 7 p.Additionally, the MBRTe can be applied not onlyto raw fresh milk collected from all the farmers in theassociation, but also from individual farmers who are itsmembers. This last feature could help identify individualcattlemen who own dairy herds with preclinical orclinical mastitis or those who do not comply with goodpractices of milking and handling of fresh milk. Bydoing so, the association can make a corrective plan toimprove the microbiological quality of raw fresh milkin the near future and establish better market prices forits producers.CONCLUSIONIn this work, we evaluated the quality of milkproduced by six dairy associations of small farmers inthe province of Carchi for eight consecutive months.We determined the hygienic and sanitary status ofmilk and dairy herd, respectively. The study found anadequate correlation between the quality of milk andthe farmer’s compliance with good practices of milking,handling and storage of fresh milk. Thus, it served toencourage some of the associations to comply with thesegood practices.We demonstrated a relationship between thequalitative MBRTe and somatic cells and total bacteriacounts. As a result, we proposed the MBRTe to thecattlemen’s associations in the Ecuadorian highlandsto pre-qualify milk collected from both the entireassociation and individual farmers. Also, the proposedmethodology can be useful for isolated ranchers, awayfrom accredited labs, to check the quality of their milkby themselves. This test can identify the presenceof sub-clinical or clinical mastitis and inadequatemanagement of milking, handling, storage andtransportation of fresh milk. The results can be used tomake appropriate improvement plans to correct thesedeficiencies and enhance the quality of milk.CONTRIBUTIONMayra Pérez-Lomas collected the data. MiltonCuaran-Guerrero, Lucía Yépez-Vásquez, HolgerPineda-Flores, Jimmy Núñez-Pérez and Rosario Espin-Valladares contributed the data and analysis tools. JoséPais-Chanfrau conceived and designed the analysis.Edmundo Recalde-Posso performed the analysis. Luis E.Trujillo-Toledo wrote the paper in cooperation with JoséM. Pais-Chanfrau.CONFLICT OF INTERESTThe authors declare that there were no conflicts ofinterest during the elaboration of this work or later,during the preparation of the manuscript.ACKNOWLEDGEMENTSWe would like to thank the engineers Luis Aldeanand Vanessa Bastidas from the Alpina S.A. Foundation(www.alpina.com.co) for the support given to thisresearch.