Advertisement

From cow to cheese: Novel phenotypes related to the sensory profile of model cheeses from individual cows

Open ArchivePublished:April 18, 2018DOI:https://doi.org/10.3168/jds.2017-14342

      ABSTRACT

      Milk samples were taken once from a total of 1,224 Brown Swiss cows from 83 herds, and 1,500 mL of raw full-fat milk from each cow was processed according to a laboratory-scale model-cheese-making procedure. A sensory panel was assembled and the members trained to evaluate the sensory profile of individual model cheeses. The protocol scorecard was composed of 7 main sensory descriptors related to smell intensity, flavor intensity, taste (salt and sour), and texture (elasticity, firmness, and moisture), and 40 sensory attributes describing smell and flavor profiles. Sensory data were analyzed using a mixed model that included random effects of herd, animal, and panelist, as well as fixed effects of dairy system, days in milk, parity, and order of cheese presentation, and covariates for cheese weight and fat:protein ratio. The sensory profile was not much affected by the dairy farming systems included in the trial, but it was affected by farm within dairy system: cheeses from traditional dairy farms had a greater wood/humus attribute of both smell and flavor than those from modern farm. Of the modern farms, cheeses from those using total mixed rations including silages had a more intense smell of sour milk and a firmer, less moist texture than those using total mixed rations without silages. Moreover, for all the sensory traits, we found less variance related to herd and animals than that related to the panelists and the residuals. Stage of lactation was found to be the most important, whereas parity was not relevant. In particular, cheese smell intensity (and some related attributes) exhibited a quadratic trend with lower values in mid-lactation, whereas flavor and salt descriptors were more intense in the last period of lactation.

      Key words

      INTRODUCTION

      The perception of cheese sensory characteristics influences consumers' choices and food-related behaviors (
      • Drake M.A.
      Invited review: Sensory analysis of dairy foods.
      ). To produce a cheese with a suitable flavor and texture profile, the dairy industry has to monitor all the outcomes of the entire process, which involves many factors concerning the herd, dairy plant, and distribution network. A key issue to consider in monitoring and assessing a cheese sensory profile is the interaction between milk quality and the type of cheese to be produced. Reliable and sufficiently comprehensive milk composition and quality traits are essential to classify and assess products for their nutritional, organoleptic, microbiological, and technological characteristics (
      • Clark S.
      • Costello M.
      • Drake M.A.
      • Bodyfelt F.W.
      The Sensory Evaluation of Dairy Products.
      ). These are particularly crucial for the Protected Designation of Origin (PDO) dairy chain in light of strict regulations and restrictions governing milk treatments and modifications carried out before and during cheesemaking (
      • Verdier-Metz I.
      • Coulon J.B.
      • Pradel P.
      • Viallon C.
      • Albouy H.
      • Berdagué J.L.
      Effect of botanical composition of hay and casein genetic variants on the chemical and sensory characteristics of ripened Saint-Nectaire type cheese.
      ;
      • Bittante G.
      • Cologna N.
      • Cecchinato A.
      • De Marchi M.
      • Penasa M.
      • Tiezzi F.
      • Endrizzi I.
      • Gasperi F.
      Monitoring of sensory attributes used in the quality payment system of Trentingrana cheese.
      ;
      • Ojeda M.
      • Etaio I.
      • Fernández Gil M.P.
      • Albisu M.
      • Salmerón J.
      • Pérez Elortondo F.J.
      Sensory quality control of cheese: Going beyond the absence of defects.
      ). In these regulated production systems, all factors influencing milk yield and composition at the farm level, such as individual animal characteristics and dairy herd system, have to be strictly controlled and have particular significance for the sensory quality of dairy products (
      • Bertoni G.
      • Calamari L.
      • Maianti M.G.
      • Battistotti B.
      Milk for Protected Denomination of Origin (PDO) cheeses: I. The main required features.
      ).
      Some researchers have explored the effects of bovine milk characteristics on cheese sensory profiles by investigating the influence of lactating animals and environmental factors. These studies have looked at the effects of breed (
      • Martin B.
      • Pomies D.
      • Pradel P.
      • Verdier-Metz I.
      • Remond B.
      Yield and sensory properties of cheese made with milk from Holstein or Montbeliarde cows milked twice or once daily.
      ), lactation stage (
      • Kefford B.
      • Christian M.P.
      • Sutherland B.J.
      • Mayes J.J.
      • Grainger C.
      Seasonal influences on Cheddar cheese manufacture: Influence of diet quality and stage of lactation.
      ;
      • Coulon J.B.
      • Verdier I.
      • Pradel P.
      • Almena M.
      Effect of lactation stage on the cheesemaking properties of milk and the quality of Saint-Nectaire-type cheese.
      ), animal health status (
      • Auldist M.J.
      • Coats S.
      • Sutherland B.J.
      • Mayes J.J.
      • McDowell G.H.
      • Rogers G.L.
      Effect of somatic cell count and stage of lactation on raw milk composition and the yield and quality of Cheddar cheese.
      ), feeding regimen (
      • Verdier-Metz I.
      • Coulon J.B.
      • Pradel P.
      • Viallon C.
      • Berdagué J.L.
      Effect of forage conservation (hay or silage) and cow breed on the coagulation properties of milks and on the characteristics of ripened cheeses.
      ;
      • Carpino S.
      • Mallia S.
      • La Terra S.
      • Melilli C.
      • Licitra G.
      • Acree T.E.
      • Barbano D.M.
      • Van Soest P.J.
      Composition and aroma compounds of Ragusano cheese: Native pasture and total mixed rations.
      ), and dairy farm system (
      • Agabriel C.
      • Martin B.
      • Sibra C.
      • Bonnefoy J.C.
      • Montel M.C.
      • Didienne R.
      • Hulin S.
      Effect of dairy production systems on the sensory characteristics of Cantal cheeses: A plant-scale study.
      ) on cheese sensory profiles. Most have attempted to eliminate noise factors as much as possible through experimental design, which has involved, depending on the objectives of the trial, (1) selecting a small number of animals (based on their genetic background and their physiological status); (2) selecting animals from a few farms or just one farm; and (3) producing a small number of cheeses from bulk milk, often with a standardized fat:protein ratio. These studies show frequent low variability and have found few relationships between the factors tested and cheese quality.
      When animal factors or herd characteristics have to be investigated, a large number of cheeses produced at the individual cow level are required. For example, in previous research, we devised a model cheese manufacturing procedure with high repeatability and reproducibility of cheese-making traits, which we used to process more than 1,000 individual Brown Swiss milk samples collected from 85 herds (
      • Cipolat-Gotet C.
      • Cecchinato A.
      • De Marchi M.
      • Bittante G.
      Factors affecting variation of different measures of cheese yield and milk nutrient recovery from an individual model cheese-manufacturing process.
      ). This allowed us to simultaneously quantify the effects of farming systems (with different feeding regimens), individual farms, and cow's parity and stage of lactation on individual cheese-making efficiency and cheese yield. This procedure also allowed us to analyze the release of volatile organic compounds characterizing cheese flavor in the ripened cheeses (
      • Bergamaschi M.
      • Biasioli F.
      • Cappellin L.
      • Cecchinato A.
      • Cipolat-Gotet C.
      • Cornu A.
      • Gasperi F.
      • Martin B.
      • Bittante G.
      Proton transfer reaction time-of-flight mass spectrometry: A high-throughput and innovative method to study the influence of dairy system and cow characteristics on the volatile compound fingerprint of cheeses.
      ), and to estimate the genetic parameters of these traits; that is, the link between the cow's genetics and a given organoleptic property of the cheese produced from its milk after 2 mo of ripening (
      • Bergamaschi M.
      • Cecchinato A.
      • Biasioli F.
      • Gasperi F.
      • Martin B.
      • Bittante G.
      From cow to cheese: Genetic parameters of the flavor fingerprint of cheese investigated by direct injection mass spectrometry (PTR-ToF-MS).
      ).
      No previous study has looked at variability in the sensory profile of ripened cheeses made from individual milk samples from a large set of cows from many different farms. The aim of the present study, therefore, was to take a holistic approach to investigating the effect of several sources of variation affecting novel phenotypes related to the sensory profile of ripened individual model cheeses. In particular, we studied the effects of feeding regimens and management systems together with the effects of herd within dairy system, and the parity and stage of lactation of individual cows.

      MATERIALS AND METHODS

      Herd Selection

      The present study is part of the “Cowability-Cowplus” project described previously (
      • Cipolat-Gotet C.
      • Cecchinato A.
      • De Marchi M.
      • Bittante G.
      Factors affecting variation of different measures of cheese yield and milk nutrient recovery from an individual model cheese-manufacturing process.
      ;
      • Bittante G.
      • Cipolat-Gotet C.
      • Malchiodi F.
      • Sturaro E.
      • Tagliapietra F.
      • Schiavon S.
      • Cecchinato A.
      Effect of dairy farming system, herd, season, parity, and days in milk on modeling of the coagulation, curd firming, and syneresis of bovine milk.
      ). A total of 83 dairy herds located in Trento province (northeastern Italy) were selected from 610 farms representing different environments and dairy farming systems in this mountain area. The farms were sampled once during a calendar year, taking into account the distribution of herds over 4 different dairy systems: 1 traditional system of small farms with old barns and tied lactating cows milked at the stall and fed mainly on farm meadow hay (representing on average 61% of DMI) and a commercial compound feed (on average 18% of DMI); and 3 modern dairy systems, in all of which the cows were kept loose and milked in a milking parlor. The first modern system used a traditional feeding regimen without a TMR (no TMR), consisting mainly of farm hay (54% of DMI) and a commercial compound feed (30% of DMI). The second modern system (TMR-s) was similar to the first but used TMR including hay (22% of DMI), maize silage (in total 19% of DMI), imported alfalfa hay (16% of DMI), and concentrates consisting of either a commercial compound feed (15% of DMI) or, more often, a mix of cereals (especially maize grain; on average 23% of DMI), protein feed (often soybean meal), sometimes dry beet pulp, and supplements. The third modern system (TMR-w) used TMR but with silage replaced by a greater quantity of hay (39% of DMI) and concentrates, and with water added to the mixing wagon to increase the TMR moisture content to about 50%.

      Milk Sampling and Analysis

      Milk samples were taken once from a total of 1,224 Brown Swiss cows (generally 15 cows per dairy farm, with different parities and at different lactation stages). Briefly, during the evening milking, a 2-L milk sample was collected from each cow, taken to the Cheese-Making Laboratory of the Department of Agronomy, Food, Natural Resources, Animals and Environment (DAFNAE; University of Padova, Legnaro, Italy), and processed and analyzed within 20 h of collection. The dairy cows sampled (mean DIM = 179, mean number of parities = 2.54) produced, on average, 24.3 kg/d of milk containing 3.75% protein, 2.88% casein, 4.38% fat, 4.77% lactose, and 13.89% DM, and with an SCS of 2.98 (data not shown).

      Individual Model Cheese Procedure and Analysis

      Each raw full-fat milk sample (1,500 mL) was processed according to the protocol described by
      • Cipolat-Gotet C.
      • Cecchinato A.
      • De Marchi M.
      • Bittante G.
      Factors affecting variation of different measures of cheese yield and milk nutrient recovery from an individual model cheese-manufacturing process.
      , taking all precautions to ensure maximum reproducibility from one cheese-making session to another. Briefly, each sample was heated (35°C), inoculated with starter cultures (using an industrial freeze-dried formulation of thermophilic lactic bacteria; Delvo-Tec TS-10A DSL; DSM Food Specialties, Delft, the Netherlands) and mixed with rennet solution [Hansen standard 160 with 80 ± 5% chymosin and 20 ± 5% pepsin; 160 international milk clotting units (IMCU)/mL; Pacovis Amrein AG, Bern, Switzerland; final concentration of 51.2 IMCU/L of milk]. The resulting curd was cut, drained, shaped in wheels, pressed, salted (saturated solution, 20% NaCl), and weighed. Then, cheeses were left to ripen at 15°C and 85% relative humidity for the first month, and then at 12°C and the same relative humidity in the second month. At the end of ripening, each wheel was weighed and cheese yield was recorded as the ratio of the weight (g) of the ripened cheese to the weight of the processed milk (g). After removing the cheese rind, the chemical components (fat, protein, and DM) were measured with a FoodScan analyzer (Foss, Hillerød, Denmark). Cheese acidity, expressed as pH, was measured 3 times per sample (and averaged before data analysis) using a Crison Basic 25 pH meter with a 50 54 TC combined electrode (Crison Instruments SA, Barcelona, Spain). The color of all the cheese samples was determined with a Minolta colorimeter (CM-508c, D65 illuminant and 10° observer; Konica-Minolta Sensing Inc., Ramsey, NJ; 3 consecutive readings, averaged before data analysis) and expressed in terms of L* (lightness, ranging from black = 0 to white = 100), a* (positive values indicating red, negative green), b* (positive values indicating yellow, negative blue). Cheese hardness, expressed as maximum shear force (N/cm2), was measured on a cylindrical cheese sample (surface cross-sectional area of 1 cm2; 3 repeated measures per sample) using a TA-HDi Texture Analyzer (Stable Micro Systems, London, UK) with a Warner-Bratzler shear attachment (10-N load cell, crosshead speed of 2 mm/s). Descriptive statistics of the chemical composition and physical traits of individual model cheeses are shown in Table 1.
      Table 1Descriptive statistics of cheese yield, chemical composition, shear force, and color traits of individual model cheeses at 2 mo of ripening
      P5 = 5th percentile; P95 = 95th percentile.
      TraitNMeanSDP5P95
      CY60D,
      CY60D = cheese yield (%) at 60 d of ripening.
      %
      1,2248.731.126.9910.61
      Cheese quality traits
       Fat, %1,07238.044.0631.7045.18
       Fat in DM, %1,06447.624.2440.8754.41
       Protein, %1,08026.834.0120.0132.85
       Fat:protein ratio1,0681.450.350.982.15
       DM, %1,07780.064.6371.8486.91
       pH1,2165.170.174.875.45
       Shear force, N/cm21,20335.5216.9813.0868.08
      Cheese color
      L* = lightness, a* = redness/greenness, b* = yellowness/blueness.
       L*1,19858.996.1049.8169.33
       a*1,200−2.070.53−2.91−1.14
       b*1,2007.632.833.4712.64
      1 P5 = 5th percentile; P95 = 95th percentile.
      2 CY60D = cheese yield (%) at 60 d of ripening.
      3 L* = lightness, a* = redness/greenness, b* = yellowness/blueness.

      Quantitative Descriptive Profile Analysis

      A sensory evaluation panel made up of 14 technicians from the department (DAFNAE; University of Padova) were selected and trained in cheese evaluation under the direction of a panel leader on 8 d over 4 wk (20 h total). The reference standard and protocol scorecard were in accordance with
      • Bérodier F.
      • Lavanchy P.
      • Zannoni M.
      • Casals J.
      • Herrero L.
      • Adamo C.
      A guide to the sensory evaluation of smell, aroma and taste of hard and semi-hard cheeses.
      for classification of smell, flavor, and taste terms, and in accordance with
      • Lavanchy P.
      • Bérodier F.
      • Zannoni M.
      • Noel Y.
      • Adamo C.
      • Squella J.
      • Herrero L.
      Sensory evaluation of the texture of hard and semi-hard cheeses.
      for texture terms. After training, a maximum of 15 cheeses from cows of the same herd were evaluated on each day of sensory analysis (83 d, 1 or 2 d/wk) by 5 or 6 members of the panel selected using a sequential monadic design. Each sensory testing session lasted about 90 min and was held mid-morning under normal light conditions. Cheese samples (2 triangular prism-shaped pieces per sample, length ∼35 mm, width ∼15 mm, thickness ∼20 mm) were presented on a petri dish, and water was supplied to wash the mouth between tests.
      Briefly, panelists had to assess cheese samples according to 7 main sensory descriptors: 1 smell descriptor (intensity), 1 flavor descriptor (intensity), 2 taste descriptors (intensity of salt and sour tastes), and 3 texture descriptors (elasticity, firmness, and moisture). These traits were ranked on a 13-point discontinuous scale (from 1 to 7 including half points, where 1 represented the minimum and 7 the maximum perception of the sensory descriptor). Another level of sensory description was introduced for smell and flavor. After assessing overall intensity, panelists had to evaluate the intensity of 4 families of sensory attributes, each composed of several detailed attributes, on a 4-point discontinuous scale (1 = absence, 2 = low, 3 = medium, 4 = high). A summary of all the sensory descriptors, families, and detailed attributes included in the protocol scorecard is presented in Supplemental Table S1 (https://doi.org/10.3168/jds.2017-14342).

      Statistical Analysis

      The panelists completed 6,612 sensory scorecards (each listing 47 cheese sensory traits) to assess 1,224 model cheeses (5.4 scorecards per cheese, on average). All smell and flavor families and detailed attributes exhibited a non-Gaussian right-skewed distribution and were ln-transformed, which gave an almost-Gaussian distribution for all these traits. In the sub-data sets of each panelist (14 in total) a detailed attribute (or family) was considered missing when it was not used or was used with a frequency of ≤1% of the model cheeses evaluated by the panelist.
      The covariates of weight and squared weight of the fresh cheeses and of the fat:protein ratios of milk for all sensory data (main descriptors, families, and detailed attributes) were included in the statistical model. The rationale of including these covariates was to correct for sensory differences due to different milk compositions and differences in the initial weights of the model cheeses. Indeed, the milk processed was not standardized for fat:protein ratio before cheese-making, and the heights of the model cheese wheels differed according to the different cheese yields (each model cheese was made from 1,500 mL of milk). We used the following linear mixed model:
      yijklmnopqrs = μ + dairy systemi + herdj(dairy system)i + DIMk + parityl + OPm + cheesewn + cheesew2o + F:P ratiop + cheese/animalq(DIM × parity × OP)klm + panelistr + eijklmnopqrs,


      where yijklmnopqrs is the observed trait (sensory traits); μ is the overall mean; dairy systemi is the fixed effect of the ith dairy system (i = 1 to 4); herdj(dairy system)i is the random effect of the jth herd (j = 1 to 83) within the ith dairy system; DIMk (interval from calving to milk sampling) is the kth 60-d class of days in milk (k = 1 to 6; class 1: ≤60 d, class 2: 61–120 d, class 3: 121–180 d; class 4: 181–240 d; class 5: 241–300 d; class 6: >300 d); parityl is the fixed effect of the lth parity (l = 1 to 5 or more lactations); OPm is the fixed effect of mth order of cheese presentation (OP) to each panelist (m = 1 to 5; class 1: 1st cheese; class 2: 2nd cheese; class 3: 3rd to 5th cheeses; class 4: 6th to 10th cheeses; class 5: 11th to 15th cheeses); cheesewn is the linear covariate of cheese weight at 0 d of ripening; cheesew2o is the linear covariate of quadratic cheese weight at 0 d of ripening; F:P ratiop is the linear covariate of the milk fat:protein ratio; cheese/animalq(DIM × parity × OP)klm is the random effect of the qth cheese/animal (n = 1 to 1,224) within the kth DIM, lth parity, and mth OP; panelistr is the random effect of the rth panelist (n = 1 to 14); and eijklmnopqrs is the residual random error term ∼N (0, σ2).
      Significance of dairy system was tested on the error line of herd within dairy system; DIM class, parity, and OP were tested on the error line of animal/cheese within DIM class, parity, and OP; and panelist was tested on the residual error.
      Orthogonal contrasts were used to compare dairy systems: (1) traditional versus modern dairy systems; (2) within modern systems, those using TMR versus no TMR; and (3) within farms using TMR, those including silages (TMR-s) versus those adding water to moisten the ration (TMR-w). The orthogonal contrasts among different parities were (1) first versus second and subsequent lactations; (2) second versus third and subsequent lactations; and (3) third versus fourth and subsequent lactations. For DIM, we tested the linear, quadratic, and cubic trends of the least squares means (LSM) of the classes. Methodological effects (OP, covariates, and panelists) were beyond the objective of this study and are not presented or discussed.
      Finally, Pearson product-moment correlations were carried out to establish the relationships among the sensory descriptors. Before this, sensory variables for each panelist were standardized, scaled to unit variance, and zero centered; that is, each variable was forced to a mean of 0 and a variance of 1 (
      • Næs T.
      Handling individual differences between assessors in sensory profiling.
      ).

      RESULTS

      Sensory Profile of Individual Model Cheeses

      Smell and flavor intensities and salt taste descriptors exhibited similar means and variances, 3.07 ± 0.82, 3.33 ± 0.87, and 3.24 ± 0.68, respectively, whereas skewness and kurtosis values were <1.0, demonstrating their closeness to a normal distribution (data not shown). The means of sour taste and the 3 texture descriptors were ∼1 point away from the central value of the chart (2.51, 2.49, 4.95, and 2.80, respectively) and their standard deviations (SD) were greater than those of the other traits (0.83 to 1.09). Nevertheless, the distribution of these sensory descriptors was also considered normal, as the kurtosis and skewness indices were similar to those of the aforementioned traits (data not shown).
      Pearson product-moment correlations among sensory traits are given in Table 2 (raw data above the diagonal and model residuals below). Raw data indicated that smell and flavor intensities were moderately positively related (0.45). The correlations between flavor and the taste descriptors (salt and sour) were positive but lower (0.32 and 0.38, respectively), and smell was correlated only with the salt attribute (0.23). Coefficients (and significance) of correlations among the residuals were similar.
      Table 2Pearson product–moment correlations between sensory descriptors of cheese (above diagonal) and correlations between residuals (below diagonal)
      SmellFlavorSaltSourElasticityFirmnessMoisture
      Smell0.45
      P < 0.001.
      0.23
      P < 0.001.
      0.0040.22
      P < 0.001.
      −0.06
      P < 0.001.
      0.06
      P < 0.001.
      Flavor0.46
      P < 0.001.
      0.32
      P < 0.001.
      0.38
      P < 0.01
      0.29
      P < 0.001.
      −0.29
      P < 0.001.
      0.22
      P < 0.001.
      Salt0.17
      P < 0.001.
      0.34
      P < 0.001.
      0.26
      P < 0.001.
      −0.05
      P < 0.001.
      0.09
      P < 0.001.
      0.14
      P < 0.001.
      Sour0.10
      P < 0.001.
      0.25
      P < 0.001.
      0.18
      P < 0.001.
      0.08
      P < 0.001.
      −0.11
      P < 0.001.
      0.11
      P < 0.001.
      Elasticity0.04
      P < 0.001.
      0.05
      P < 0.001.
      −0.0050.02−0.64
      P < 0.001.
      0.57
      P < 0.001.
      Firmness0.04
      P < 0.01
      0.03
      P < 0.05
      0.010.004−0.50
      P < 0.001.
      −0.33
      P < 0.001.
      Moisture0.04
      P < 0.01
      0.10
      P < 0.001.
      0.06
      P < 0.001.
      0.07
      P < 0.001.
      0.38
      P < 0.001.
      −0.35
      P < 0.001.
      * P < 0.05
      ** P < 0.01
      *** P < 0.001.
      The 2 taste descriptors were not highly associated with each other (0.26), and the coefficient of correlation was lower for the residuals (0.18). We found that the Pearson and residual correlations among the texture descriptors had greater values: as expected, firmness was negatively correlated with elasticity and moisture descriptors, whereas elasticity exhibited a strong, positive association with moisture. Smell, flavor, and taste terms were only slightly or not at all correlated with all the texture descriptors, and, of these, flavor intensity had, on average, the highest coefficients of correlation (0.29 with elasticity, −0.29 with firmness, and 0.22 with moisture).

      Factors Affecting Variations in Model Cheese Sensory Profile

      Table 3, Table 4 summarize the importance of the fixed effects included in the linear model that explain the variability in, respectively, the main sensory descriptors and the smell and flavor attributes of the model cheeses. The hierarchical linear model included the effects of 4 random factors, illustrated in Figure 1: individual herd within dairy system, individual cow/model cheese (animal) within herd, sensory panelist (panelist) within cow/model cheese, and the residual. As clearly shown (Figure 1), panelist and residual variability had the same degree of relevance, although there were some differences among traits, with scores ranging from ±0.54 to ±0.72 (the values refer to both panelist and residual variance on a scale of 1 to 7), the only exception being the greater effect of individual panelist on the evaluation of sour taste (scores of ±1.02) although this was not confirmed by individual attributes. The effects of both herd and animal were much smaller than those of panelist and residual, with scores ranging from ±0.10 (animal variance for flavor intensity on a scale of 1 to 7) to ±0.44 (herd variance for cheese elasticity on a scale of 1 to 7).
      Table 3Effect (F-value and significance) of dairy system, DIM, and parity for sensory descriptors evaluated on individual model cheeses after 2 mo of ripening
      Fixed effect of dairy system is tested on herd variance; the other fixed effects are tested on animal variance.
      DescriptorDairy systemDIMParity
      Smell2.05.9
      P < 0.001.
      2.1
      Flavor0.93.2
      P < 0.01
      1.3
      Salt0.13.0
      P < 0.05
      3.9
      P < 0.01
      Sour0.22.3
      P < 0.05
      0.7
      Elasticity1.22.01.6
      Firmness2.01.11.2
      Moisture2.21.42.3
      1 Fixed effect of dairy system is tested on herd variance; the other fixed effects are tested on animal variance.
      * P < 0.05
      ** P < 0.01
      *** P < 0.001.
      Table 4Effect (F-value and significance) of dairy system, DIM, and parity for individual smell and flavor attributes evaluated on model cheeses after 2 mo of ripening
      Fixed effect of dairy system is tested on herd variance; the other fixed effects are tested on animal variance.
      AttributeSmell attributesFlavor attributes
      Dairy systemDIMParityDairy systemDIMParity
      Milky family0.51.12.01.52.00.7
       Raw milk1.41.51.81.32.6
      P < 0.05
      1.2
       Cooked milk1.02.2
      P < 0.05
      0.61.43.8
      P < 0.01
      0.0
       Sour milk3.4
      P < 0.05
      5.2
      P < 0.001.
      2.21.42.41.4
       Seasoned cheese1.61.62.7
      P < 0.05
      2.20.63.9
      P < 0.01
      Vegetable family1.52.20.52.30.50.5
       Hay0.21.20.81.91.12.4
      P < 0.05
       Fermented hay1.46.3
      P < 0.001.
      1.01.81.50.8
       Boiled vegetables0.42.9
      P < 0.05
      2.6
      P < 0.05
      0.62.10.5
       Garlic/onion1.60.71.02.21.50.5
       Wood/humus5.0
      P < 0.01
      0.80.82.9
      P < 0.05
      0.90.4
      Animal family1.14.6
      P < 0.001.
      0.50.42.4
      P < 0.05
      0.5
       Barn/cattle shed1.23.9
      P < 0.01
      0.51.03.8
      P < 0.01
      0.3
       Beef broth1.83.1
      P < 0.05
      3.4
      P < 0.01
      0.90.50.1
       Animal manure1.02.10.82.11.11.1
      Chemical family0.63.6
      P < 0.01
      0.51.31.90.1
       Propionic0.21.10.20.20.60.2
       Ammonic0.85.1
      P < 0.001.
      0.01.32.01.2
       Sulfurous0.60.63.0
      P < 0.05
      2.02.00.7
       Silage/VFA0.53.4
      P < 0.01
      1.10.40.81.7
      1 Fixed effect of dairy system is tested on herd variance; the other fixed effects are tested on animal variance.
      * P < 0.05
      ** P < 0.01
      *** P < 0.001.
      Figure thumbnail gr1
      Figure 1Root mean square errors of the random effects of herd, animal/cheese within herd, panelist within animal/cheese, and the residual for the main sensory descriptors and the individual smell and flavor attributes. Color version available online.
      The fixed effects considered here included dairy system and individual cow factors (DIM and parity). The LSM and the corresponding standard errors (SE) of dairy farming system and DIM for the main sensory descriptors are presented in Figure 2, Figure 3, respectively.
      Figure thumbnail gr2
      Figure 2Least squares means (±SE) of smell (a), flavor (b), salt (c), sour (d), elasticity (e), firmness (f), and moisture (g) sensory traits among dairy farming systems. Traditional = traditional system with tied animals; No TMR = modern dairy system with traditional feeding based on hay and compound feed; TMR-s = modern dairy system with TMR including silage; TMR-w = modern dairy system with silage-free TMR (water added for moisture). Letters (a, b) indicate that the contrast between TMR-s and TMR-w was significant at P < 0.05.
      Figure thumbnail gr3
      Figure 3Least squares means (±SE) of smell (a), flavor (b), salt (c), sour (d), elasticity (e), firmness (f), and moisture (g) sensory traits over DIM.
      The main sensory profile descriptors were not affected by the dairy systems tested in this trial, also because their significance was tested on the error line of herd within dairy system. However, the individual orthogonal contrasts revealed that the model cheeses obtained from milk produced on modern farms using total mixed rations exhibited different texture characteristics according to whether or not silages were included and were firmer and less moist when included (TMR-s = 5.15 and 2.62, TMR-w = 4.77 and 2.81 for firmness and moisture descriptors, respectively; Figure 2).
      For the fixed effects related to individual cows, DIM affected smell, flavor, and salt and sour intensities of the cheeses, but texture descriptors were unaffected. In particular, DIM exhibited a quadratic pattern for smell, flavor, and sour intensity, the first trait having lower values in cheeses produced from milk obtained at classes 3 and 4, respectively (Figure 3a), whereas the other 2 traits had lower values in the second and third classes of DIM for the last 2 traits (Figure 3b and d, respectively). We obtained different results for salt taste, which exhibited an almost linear increase during lactation (Figure 3c). Among the 3 texture descriptors, only elasticity exhibited a linear decreasing pattern over DIM (Figure 3e); the other 2 texture descriptors showed no clear trend. The effect of cow's age, expressed as the number of parities, was not important for any of the sensory descriptors except for salt taste (Table 4), which had lower values in model cheeses obtained from first and second calvers than in those obtained from older cows (data not shown).
      The results for families and individual attributes of smell and flavor (Table 4) were similar to the results for smell and flavor intensity descriptors. In particular, residual variability was always greater than variability among individual panelists, which was always greater than that among individual herds within dairy system and individual cows within herd. In the latter case, it was often not possible to estimate the variance. The fixed effect of dairy farm system was significant in the case of the wood/humus attribute for both smell and flavor, which was more intense for the traditional dairy system than the modern farms. Of the smell attributes, contrasts showed that cheeses from farms using silage (TMR-s) had a higher perception of sour milk than those from TMR-w farms.
      Of the fixed factors related to animals, DIM affected several smell attributes of the model cheeses, whereas flavor was less affected. Like the smell intensity descriptor, animal and chemical families exhibited a quadratic trend, with lower values in the middle of lactation, as did the individual attributes of sour milk, fermented hay, barn/cattle-shed, and ammonia, but not cooked milk, which exhibited the opposite pattern. Boiled vegetables and beef broth intensities tended to decrease linearly during lactation. As with the 7 main sensory descriptors, parity was not very important in cheese sensory perception for smell and flavor attributes and—even though it was significant for seasoned cheese, boiled vegetables, beef broth, and sulfurous smell attributes, and seasoned cheese and hay flavor attributes, respectively—we observed no clear trend among cheeses from cows at different parities.

      DISCUSSION

      Relevance of Dairy Farming System to the Sensory Profile of Cheese

      With regard to relationships between herd system and cheese sensory traits, studies have been carried out on forage type (
      • Verdier I.
      • Coulon J.B.
      • Pradel P.
      • Berdagué J.L.
      Effect of forage type and cow breed on the characteristics of matured Saint-Nectaire cheeses.
      ;
      • Martin B.
      • Verdier-Metz I.
      • Buchin S.
      • Hurtaud C.
      • Coulon J.-B.
      How do the nature of forages and pasture diversity influence the sensory quality of dairy livestock products?.
      ), forage conservation (
      • Verdier-Metz I.
      • Coulon J.B.
      • Pradel P.
      • Viallon C.
      • Berdagué J.L.
      Effect of forage conservation (hay or silage) and cow breed on the coagulation properties of milks and on the characteristics of ripened cheeses.
      ), and level of intensification in dairy farming (
      • Agabriel C.
      • Martin B.
      • Sibra C.
      • Bonnefoy J.C.
      • Montel M.C.
      • Didienne R.
      • Hulin S.
      Effect of dairy production systems on the sensory characteristics of Cantal cheeses: A plant-scale study.
      ) and in milking system (
      • Martin B.
      • Pomies D.
      • Pradel P.
      • Verdier-Metz I.
      • Remond B.
      Yield and sensory properties of cheese made with milk from Holstein or Montbeliarde cows milked twice or once daily.
      ). Important differences were found in the sensory properties of cheese when indoor feeding regimens were compared with the use of pasture (
      • Coppa M.
      • Verdier-Metz I.
      • Ferlay A.
      • Pradel P.
      • Didienne R.
      • Farruggia A.
      • Montel M.C.
      • Martin B.
      Effect of different grazing systems on upland pastures compared with hay diet on cheese sensory properties evaluated at different ripening times.
      ;
      • Bovolenta S.
      • Romanzin A.
      • Corazzin M.
      • Spanghero M.
      • Aprea E.
      • Gasperi F.
      • Piasentier E.
      Volatile compounds and sensory properties of Montasio cheese made from the milk of Simmental cows grazing on alpine pastures.
      ). Grazing is not practiced in the mountain area of the present study (with the exception of summer transhumance to highland pastures), and silages are rarely used because the majority of milk is destined for the production of long-ripened, hard PDO cheeses (e.g., Trentingrana, Vezzena, Spressa, Puzzone di Moena) without the use of lysozyme to control late blowing of cheese (
      • Bittante G.
      • Cologna N.
      • Cecchinato A.
      • De Marchi M.
      • Penasa M.
      • Tiezzi F.
      • Endrizzi I.
      • Gasperi F.
      Monitoring of sensory attributes used in the quality payment system of Trentingrana cheese.
      ). The absence of grazing among dairy farming systems could explain the small number of sensory descriptors affected by this factor. However, most of the aforementioned studies found no differences or only partial differences between treatments, probably as a consequence of the small number of cheese-makings and low variability in the milks sampled.
      The present study was designed to contemporaneously evaluate the effects of factors relating to animal and type of farming on the sensory characteristics of many individual model cheeses. However, the manufacture of model cheeses cannot mimic strict industry conditions or the production of some specific commercial cheese types. In addition, in developing the sensory panel test, we simplified the protocol scorecard in light of the need for a compromise between detailed descriptions of sensory characteristics and a simple, rapid method that could be applied to small samples of a large number of model cheeses. Despite the large number of model cheeses evaluated, the relatively small number of sensory descriptors significantly affected by dairy system was due to the large variability in the effect of individual farms within dairy system (which was assumed as the error line for testing dairy systems) and the moderate repeatability of sensory description.
      In a previous investigation on the volatile compounds of the same individual model cheeses as in the present study, we found that 16 out of 55 volatile organic compounds analyzed by solid-phase microextraction (SPME)-GC-MS (
      • Bergamaschi M.
      • Aprea E.
      • Betta E.
      • Biasioli F.
      • Cipolat-Gotet C.
      • Cecchinato A.
      • Bittante G.
      • Gasperi F.
      Effects of the dairy system, herd and individual cow characteristics on the volatile organic compound profile of ripened model cheeses.
      ) and 55 out of 240 spectrometric peaks detected in cheese headspace by proton transfer reaction-time of flight (PTR-ToF)-MS (
      • Bergamaschi M.
      • Biasioli F.
      • Cappellin L.
      • Cecchinato A.
      • Cipolat-Gotet C.
      • Cornu A.
      • Gasperi F.
      • Martin B.
      • Bittante G.
      Proton transfer reaction time-of-flight mass spectrometry: A high-throughput and innovative method to study the influence of dairy system and cow characteristics on the volatile compound fingerprint of cheeses.
      ) were affected by dairy system. In particular, there was more acetic acid in the smell of cheeses produced from herds without TMR, and there were differences in the volatile compound profiles related to esters, acids, and terpenes. However, the results from the present study regarding smell and flavor intensity (and individual attributes) show that these differences were barely perceived by the panelists (Figure 1).
      Aside from the effect of dairy farming system on the fat and protein contents (both were higher in the modern farms compared with traditional farms) of the milk used to produce the model cheeses analyzed here, it is worth noting that dairy system affected the detailed fatty acid (FA) profile, and coagulation, curd firming, and syneresis properties. In fact, the proportion of de novo SFA was higher on the modern farms than on the traditional farms, whereas the opposite was true for vaccenic acid, CLA, and branched FA (
      • Mele M.
      • Macciotta N.P.P.
      • Cecchinato A.
      • Conte G.
      • Schiavon S.
      • Bittante G.
      Multivariate factor analysis of detailed milk fatty acid profile: Effects of dairy system, feeding, herd, parity, and stage of lactation.
      ). Within the modern farms, those using TMR had lower proportions of vaccenic acid, CLA, and linolenic and vaccelenic acids. Linolenic acid, in particular, was lower in TMR-w farms. Given that different FA have different levels of susceptibility to lipolysis and oxidation, the ripened cheeses would, of course, be expected to have different sensory properties. The technological properties of milk were also affected by dairy system because, when compared with traditional farms, milk produced on modern farms exhibited delayed coagulation and a lower syneresis rate, and consequently delayed attainment of maximum curd firmness, as shown by the curd firming modeling (
      • Bittante G.
      • Cipolat-Gotet C.
      • Malchiodi F.
      • Sturaro E.
      • Tagliapietra F.
      • Schiavon S.
      • Cecchinato A.
      Effect of dairy farming system, herd, season, parity, and days in milk on modeling of the coagulation, curd firming, and syneresis of bovine milk.
      ). Among modern dairy systems, those using TMR produced milk with a higher rate of curd firming and earlier attainment of maximum curd firmness. The different degrees of susceptibility to coagulation and curd firming, and then to proteolysis, could also be responsible for the different evolution of the cheese sensory profile during ripening.
      In our study, cheeses produced on farms using silage were associated with a higher perception of firmness and a lower perception of moisture (Figure 2d and e); the influence of the milk FA profile on cheese texture could explain the greater hardness of cheese associated with maize silage feeding systems (
      • Martin B.
      • Verdier-Metz I.
      • Buchin S.
      • Hurtaud C.
      • Coulon J.-B.
      How do the nature of forages and pasture diversity influence the sensory quality of dairy livestock products?.
      ). As reported by
      • Bugaud C.
      • Buchin S.
      • Noël Y.
      • Teissier L.
      • Pochet S.
      • Martin B.
      • Chamba J.-F.
      Relationships between Abondance cheese texture, its composition and that of milks produced by cows grazing different types of pastures.
      , the different proportion of SFA:UFA can affect cheese texture. In contrast to our study, several authors reported that the inclusion of maize silage in the cows' feed, compared with other herd feeding practices, resulted in cheeses with a lower flavor intensity and a less intense taste (
      • Verdier I.
      • Coulon J.B.
      • Pradel P.
      • Berdagué J.L.
      Effect of forage type and cow breed on the characteristics of matured Saint-Nectaire cheeses.
      ;
      • Stefanon B.
      • Procida G.
      Effects of including silage in the diet on volatile compound profiles in Montasio cheese and their modification during ripening.
      ;
      • Martin B.
      • Verdier-Metz I.
      • Buchin S.
      • Hurtaud C.
      • Coulon J.-B.
      How do the nature of forages and pasture diversity influence the sensory quality of dairy livestock products?.
      ).

      Relevance of Individual Herds on the Sensory Profile of Cheese

      With respect to the effect of individual herds or dates within dairy farming system, we could not compare our results with those of previous studies because, to the best of our knowledge, this is the first attempt to use a panel test to investigate the sensory characteristics of seasoned cheese made from milk from many herds. We found relevant herd variability for all sensory descriptors and attributes. Of the sensory descriptors, the root mean square of the effect of individual herd/date varied from almost 30% (flavor intensity) to almost 70% (cheese elasticity) of the root mean square error. In the case of smell and flavor attributes, the relative variation in individual herd/date and residual error varied from ∼15 to ∼30%, which confirms the results previously found on volatile organic compounds on the same model cheeses (
      • Bergamaschi M.
      • Aprea E.
      • Betta E.
      • Biasioli F.
      • Cipolat-Gotet C.
      • Cecchinato A.
      • Bittante G.
      • Gasperi F.
      Effects of the dairy system, herd and individual cow characteristics on the volatile organic compound profile of ripened model cheeses.
      ) and on the detailed FA profile of the milk used for their processing (
      • Pegolo S.
      • Cecchinato A.
      • Casellas J.
      • Conte G.
      • Mele M.
      • Schiavon S.
      • Bittante G.
      Genetic and environmental relationships of detailed milk fatty acids profile determined by as chromatography in Brown Swiss cows.
      ), but not on the technological properties of milk samples, characterized by a much lower herd variability (
      • Bittante G.
      • Cipolat-Gotet C.
      • Malchiodi F.
      • Sturaro E.
      • Tagliapietra F.
      • Schiavon S.
      • Cecchinato A.
      Effect of dairy farming system, herd, season, parity, and days in milk on modeling of the coagulation, curd firming, and syneresis of bovine milk.
      ). Thus, milk obtained from different herds, even though it is processed in a standardized way, confers sensory profiles to cheese that specific to each farm and this variability depends only in part on the dairy system.

      Relevance of Parity and Lactation Stage to the Sensory Profile of Cheese

      Generally, lactation stage is one of the most important animal-related sources of variation in cheese quality because of the changes in the quantity and quality of milk produced each day. The present study confirms the effect of lactation stage as being very important (Figure 3). It is worth noting that we included in the statistical model the linear and quadratic effects of the weight of the fresh cheese wheel and of the milk fat:protein ratio. In some ways, this is effectively an a posteriori standardization of milk that corrects for the effect on the sensory profile of the milk fat plus protein content (as reflected by cheese yield and weight of the wheels) and the ratio between them. It is clear, then, that the effect of stage of lactation found in this work was not due to variations in the fat and protein contents of milk during lactation, but rather to other characteristics differentiating the milk produced from calving to dry-off.
      In a previous study that compared cows at 2 stages of lactation (early vs. late),
      • Auldist M.J.
      • Coats S.
      • Sutherland B.J.
      • Mayes J.J.
      • McDowell G.H.
      • Rogers G.L.
      Effect of somatic cell count and stage of lactation on raw milk composition and the yield and quality of Cheddar cheese.
      reported lower protein and fat contents and a higher moisture content in Cheddar cheese when the milk came from late-lactating cows, even though all the milk samples were standardized for fat:protein ratio before processing. Those authors also reported a higher flavor score for cheese produced at the beginning of lactation. Another study (
      • Coulon J.B.
      • Delacroix-Buchet A.
      • Martin B.
      • Pirisi A.
      Relationships between ruminant management and sensory characteristics of cheeses: A review.
      ) found that cheese had a higher moisture content at the end of lactation, which was related to faster proteolysis and lipolysis. In a study on the effect of DIM on the quality of Saint-Nectaire cheese,
      • Coulon J.B.
      • Verdier I.
      • Pradel P.
      • Almena M.
      Effect of lactation stage on the cheesemaking properties of milk and the quality of Saint-Nectaire-type cheese.
      divided the entire lactation period into 4 sub-periods and found a higher value of cheese fat in the sub-period of 145 d after calving. They obtained results similar to ours for cheese salt and sour descriptors, and a more intense and persistent taste at the end of lactation. We did not include any descriptors related to pleasantness of taste (or of flavor and smell) on our protocol scorecard, although
      • Coulon J.B.
      • Delacroix-Buchet A.
      • Martin B.
      • Pirisi A.
      Relationships between ruminant management and sensory characteristics of cheeses: A review.
      noted that cheeses from late-lactation milk had a less pleasant smell. In our study, the quadratic trend observed for smell (Figure 3a) indicated a lesser smell intensity in cheeses produced in the middle of lactation, a result that is related to the panelists' lower perception of some individual notes (sour-lactic, fermented hay, animal group, and ammonia) in the smell of cheeses produced by cows in the middle of lactation (data not shown). The results of
      • Coulon J.B.
      • Verdier I.
      • Pradel P.
      • Almena M.
      Effect of lactation stage on the cheesemaking properties of milk and the quality of Saint-Nectaire-type cheese.
      for smell intensity are in contrast to ours, but their analysis model did not include factors related to milk fat and protein concentrations and proportions. Using the same model cheeses as the present study,
      • Bergamaschi M.
      • Biasioli F.
      • Cappellin L.
      • Cecchinato A.
      • Cipolat-Gotet C.
      • Cornu A.
      • Gasperi F.
      • Martin B.
      • Bittante G.
      Proton transfer reaction time-of-flight mass spectrometry: A high-throughput and innovative method to study the influence of dairy system and cow characteristics on the volatile compound fingerprint of cheeses.
      reported a quadratic pattern for 54 peaks related to volatile organic compounds of the cheeses analyzed by PTR-ToF-MS.
      Of the texture descriptors, we found that only cheese elasticity had a tendency to decrease linearly during lactation (Figure 3e). However, in the absence of fresh curd weight in the statistical model, this effect was highly significant for all the cheese texture descriptors (data not shown).
      As expected, parity did not have any relevant influence on sensory characteristics. The results confirmed what Bergamaschi and colleagues found using solid-phase microextraction-GC-MS (
      • Bergamaschi M.
      • Aprea E.
      • Betta E.
      • Biasioli F.
      • Cipolat-Gotet C.
      • Cecchinato A.
      • Bittante G.
      • Gasperi F.
      Effects of the dairy system, herd and individual cow characteristics on the volatile organic compound profile of ripened model cheeses.
      ) and PTR-ToF-MS (
      • Bergamaschi M.
      • Biasioli F.
      • Cappellin L.
      • Cecchinato A.
      • Cipolat-Gotet C.
      • Cornu A.
      • Gasperi F.
      • Martin B.
      • Bittante G.
      Proton transfer reaction time-of-flight mass spectrometry: A high-throughput and innovative method to study the influence of dairy system and cow characteristics on the volatile compound fingerprint of cheeses.
      ): they observed that the age of cows influenced only a few of the volatile compounds analyzed in individual model cheeses: an alcohol (octan-1-ol), some fatty acids (butanoic acid, heptanoic acid, octanoic acid), and an ester (ethyl hexanoate). Methanethiol, known for its sulfurous note in cheese aroma, also increased with greater parities (
      • Curioni P.M.G.
      • Bosset J.O.
      Key odorants in various cheese types as determined by gas chromatography-olfactometry.
      ;
      • Bellesia F.
      • Pinetti A.
      • Pagnoni U.
      • Rinaldi R.
      • Zucchi C.
      • Caglioti L.
      • Palyi G.
      Volatile components of Grana Parmigiano-Reggiano type hard cheese.
      ). In our study, panelists reported few differences among the cheeses made with milk from cows of different parities and they were not relevant.

      Importance of Individual Cow to the Sensory Profile of Cheese

      Variability in cheese sensory traits attributable to individual cows was tested here for the first time. The results showed that the effect of animal (corrected for dairy system, individual herd, parity, and DIM) was relevant compared with the residual variability, with standard deviations ranging from 0.10 (flavor intensity) to 0.38 (cheese firmness), equivalent to about 20% and almost 70%, respectively, of their residual standard deviations, as shown in Figure 1. Consequently, animal repeatability varied from about 3 to 30%, and although it was lower than that of milk yield and composition and milk technological traits (
      • Stocco G.
      • Cipolat-Gotet C.
      • Bobbo T.
      • Cecchinato A.
      • Bittante G.
      Breed of cow and herd productivity affect milk composition and modeling of coagulation, curd firming and syneresis.
      ,
      • Stocco G.
      • Cipolat-Gotet C.
      • Gasparotto V.
      • Cecchinato A.
      • Bittante G.
      Breed of cow and herd productivity affect milk nutrients recovery in curd, and cheese yield, efficiency and daily production.
      ), it clearly showed that individual animal is a relevant source of variation in the sensory profile of ripened cheese.
      Whether genetics is an important part of this effect of animal is unknown but it cannot be excluded, because
      • Bergamaschi M.
      • Cecchinato A.
      • Biasioli F.
      • Gasperi F.
      • Martin B.
      • Bittante G.
      From cow to cheese: Genetic parameters of the flavor fingerprint of cheese investigated by direct injection mass spectrometry (PTR-ToF-MS).
      recently reported that the concentration of volatile organic compounds characterizing cheese flavor in these same model cheeses is heritable, and that, for some of these substances, the heritability value is similar to those of milk yield and milk fat content. It would be interesting to investigate whether sensory attributes of ripened cheese also have a genetic and genomic basis, as has been found for other quality traits obtained with the same model cheeses (
      • Dadousis C.
      • Biffani S.
      • Cipolat-Gotet C.
      • Nicolazzi E.L.
      • Rossoni A.
      • Santus E.
      • Bittante G.
      • Cecchinato A.
      Genome-wide association of coagulation properties, curd firmness modeling, protein percentage, and acidity in milk from Brown Swiss cows.
      ,
      • Dadousis C.
      • Biffani S.
      • Cipolat-Gotet C.
      • Nicolazzi E.L.
      • Rosa G.J.M.
      • Gianola D.
      • Rossoni A.
      • Santus E.
      • Bittante G.
      • Cecchinato A.
      Genome-wide association study for cheese yield and curd nutrient recovery in dairy cows.
      ), with the aim of more firmly establishing the biological and genetic pathways connecting the sensory attributes of cheese with the dairy cow's biological and productive mechanisms (
      • Dadousis C.
      • Pegolo S.
      • Rosa G.J.M.
      • Gianola D.
      • Bittante G.
      • Cecchinato A.
      Pathway-based genome-wide association analysis of milk coagulation properties, curd firmness, cheese yield, and curd nutrient recovery in dairy cattle.
      ) and providing a basis for future genetic improvement.

      CONCLUSIONS

      Novel phenotypes related to the sensory profile of ripened cheese were evaluated for the first time at the individual dairy farm level and, within farm, at the individual cow level. The effect of different dairy farming systems was moderate compared with the variability observed among the different herds within dairy system. This last source of variation merits further research in order to understand the environmental and nutritional factors related to cheese sensory profile. At the animal level, lactation stage (more than parity) proved to be an important source of variation in the cheese sensory profile, with different traits exhibiting different patterns among DIM. After correcting for parity and lactation stage, we found that individual cows were also an important source of variation, opening speculation on a possible contribution of animal genetics. The use of model cheeses made with milk from individual cows provided new knowledge concerning the influence of herd practices and animal characteristics on the cow-to-cheese sensory profile. This information could be useful in future studies focused on estimating the genetic parameters of these new phenotypes to develop a support tool for enhancing technological milk quality at the dairy cow population level.

      ACKNOWLEDGMENTS

      The authors thank the Autonomous Province of Trento for funding the project and Massimo De Marchi (DAFNAE, University of Padova, Italy) for collaborating in the organization of panel testing.

      REFERENCES

        • Agabriel C.
        • Martin B.
        • Sibra C.
        • Bonnefoy J.C.
        • Montel M.C.
        • Didienne R.
        • Hulin S.
        Effect of dairy production systems on the sensory characteristics of Cantal cheeses: A plant-scale study.
        Anim. Res. 2004; 53: 221-234
        • Auldist M.J.
        • Coats S.
        • Sutherland B.J.
        • Mayes J.J.
        • McDowell G.H.
        • Rogers G.L.
        Effect of somatic cell count and stage of lactation on raw milk composition and the yield and quality of Cheddar cheese.
        J. Dairy Res. 1996; 63 (8861348): 269-280
        • Bellesia F.
        • Pinetti A.
        • Pagnoni U.
        • Rinaldi R.
        • Zucchi C.
        • Caglioti L.
        • Palyi G.
        Volatile components of Grana Parmigiano-Reggiano type hard cheese.
        Food Chem. 2003; 83: 55-61
        • Bergamaschi M.
        • Aprea E.
        • Betta E.
        • Biasioli F.
        • Cipolat-Gotet C.
        • Cecchinato A.
        • Bittante G.
        • Gasperi F.
        Effects of the dairy system, herd and individual cow characteristics on the volatile organic compound profile of ripened model cheeses.
        J. Dairy Sci. 2015; 98 (25682146): 2183-2196
        • Bergamaschi M.
        • Biasioli F.
        • Cappellin L.
        • Cecchinato A.
        • Cipolat-Gotet C.
        • Cornu A.
        • Gasperi F.
        • Martin B.
        • Bittante G.
        Proton transfer reaction time-of-flight mass spectrometry: A high-throughput and innovative method to study the influence of dairy system and cow characteristics on the volatile compound fingerprint of cheeses.
        J. Dairy Sci. 2015; 98 (26476950): 8414-8427
        • Bergamaschi M.
        • Cecchinato A.
        • Biasioli F.
        • Gasperi F.
        • Martin B.
        • Bittante G.
        From cow to cheese: Genetic parameters of the flavor fingerprint of cheese investigated by direct injection mass spectrometry (PTR-ToF-MS).
        Genet. Sel. Evol. 2016; 48 (27852216): 89
        • Bérodier F.
        • Lavanchy P.
        • Zannoni M.
        • Casals J.
        • Herrero L.
        • Adamo C.
        A guide to the sensory evaluation of smell, aroma and taste of hard and semi-hard cheeses.
        Lebensm. Wiss. Technol. 1997; 30: 653-664
        • Bertoni G.
        • Calamari L.
        • Maianti M.G.
        • Battistotti B.
        Milk for Protected Denomination of Origin (PDO) cheeses: I. The main required features.
        in: Hocquette J.F. Gigli S. Indicators of Milk and Beef Quality. Wageningen Academic Publishers, Wageningen, the Netherlands2005: 217-228 (EAAP publication 112.)
        • Bittante G.
        • Cipolat-Gotet C.
        • Malchiodi F.
        • Sturaro E.
        • Tagliapietra F.
        • Schiavon S.
        • Cecchinato A.
        Effect of dairy farming system, herd, season, parity, and days in milk on modeling of the coagulation, curd firming, and syneresis of bovine milk.
        J. Dairy Sci. 2015; 98 (25682135): 2759-2774
        • Bittante G.
        • Cologna N.
        • Cecchinato A.
        • De Marchi M.
        • Penasa M.
        • Tiezzi F.
        • Endrizzi I.
        • Gasperi F.
        Monitoring of sensory attributes used in the quality payment system of Trentingrana cheese.
        J. Dairy Sci. 2011; 94 (22032395): 5699-5709
        • Bovolenta S.
        • Romanzin A.
        • Corazzin M.
        • Spanghero M.
        • Aprea E.
        • Gasperi F.
        • Piasentier E.
        Volatile compounds and sensory properties of Montasio cheese made from the milk of Simmental cows grazing on alpine pastures.
        J. Dairy Sci. 2014; 97 (25282410): 7373-7385
        • Bugaud C.
        • Buchin S.
        • Noël Y.
        • Teissier L.
        • Pochet S.
        • Martin B.
        • Chamba J.-F.
        Relationships between Abondance cheese texture, its composition and that of milks produced by cows grazing different types of pastures.
        Lait. 2001; 81: 593-607
        • Carpino S.
        • Mallia S.
        • La Terra S.
        • Melilli C.
        • Licitra G.
        • Acree T.E.
        • Barbano D.M.
        • Van Soest P.J.
        Composition and aroma compounds of Ragusano cheese: Native pasture and total mixed rations.
        J. Dairy Sci. 2004; 87 (15259216): 816-830
        • Cipolat-Gotet C.
        • Cecchinato A.
        • De Marchi M.
        • Bittante G.
        Factors affecting variation of different measures of cheese yield and milk nutrient recovery from an individual model cheese-manufacturing process.
        J. Dairy Sci. 2013; 96 (24094531): 7952-7965
        • Clark S.
        • Costello M.
        • Drake M.A.
        • Bodyfelt F.W.
        The Sensory Evaluation of Dairy Products.
        2nd ed. Springer, US New York, NY2009
        • Coppa M.
        • Verdier-Metz I.
        • Ferlay A.
        • Pradel P.
        • Didienne R.
        • Farruggia A.
        • Montel M.C.
        • Martin B.
        Effect of different grazing systems on upland pastures compared with hay diet on cheese sensory properties evaluated at different ripening times.
        Int. Dairy J. 2011; 21: 815-822
        • Coulon J.B.
        • Verdier I.
        • Pradel P.
        • Almena M.
        Effect of lactation stage on the cheesemaking properties of milk and the quality of Saint-Nectaire-type cheese.
        J. Dairy Res. 1998; 65 (9627848): 295-305
        • Coulon J.B.
        • Delacroix-Buchet A.
        • Martin B.
        • Pirisi A.
        Relationships between ruminant management and sensory characteristics of cheeses: A review.
        Lait. 2004; 84: 221-241
        • Curioni P.M.G.
        • Bosset J.O.
        Key odorants in various cheese types as determined by gas chromatography-olfactometry.
        Int. Dairy J. 2002; 12: 959-984
        • Dadousis C.
        • Biffani S.
        • Cipolat-Gotet C.
        • Nicolazzi E.L.
        • Rosa G.J.M.
        • Gianola D.
        • Rossoni A.
        • Santus E.
        • Bittante G.
        • Cecchinato A.
        Genome-wide association study for cheese yield and curd nutrient recovery in dairy cows.
        J. Dairy Sci. 2017; 100 (27889122): 1259-1271
        • Dadousis C.
        • Biffani S.
        • Cipolat-Gotet C.
        • Nicolazzi E.L.
        • Rossoni A.
        • Santus E.
        • Bittante G.
        • Cecchinato A.
        Genome-wide association of coagulation properties, curd firmness modeling, protein percentage, and acidity in milk from Brown Swiss cows.
        J. Dairy Sci. 2016; 99 (26947304): 3654-3666
        • Dadousis C.
        • Pegolo S.
        • Rosa G.J.M.
        • Gianola D.
        • Bittante G.
        • Cecchinato A.
        Pathway-based genome-wide association analysis of milk coagulation properties, curd firmness, cheese yield, and curd nutrient recovery in dairy cattle.
        J. Dairy Sci. 2017; 100 (27988128): 1223-1231
        • Drake M.A.
        Invited review: Sensory analysis of dairy foods.
        J. Dairy Sci. 2007; 90 (17954731): 4925-4937
        • Kefford B.
        • Christian M.P.
        • Sutherland B.J.
        • Mayes J.J.
        • Grainger C.
        Seasonal influences on Cheddar cheese manufacture: Influence of diet quality and stage of lactation.
        J. Dairy Res. 1995; 62 (7593833): 529-537
        • Lavanchy P.
        • Bérodier F.
        • Zannoni M.
        • Noel Y.
        • Adamo C.
        • Squella J.
        • Herrero L.
        Sensory evaluation of the texture of hard and semi-hard cheeses.
        Lebensm. Wiss. Technol. 1993; 26: 59-68
        • Martin B.
        • Pomies D.
        • Pradel P.
        • Verdier-Metz I.
        • Remond B.
        Yield and sensory properties of cheese made with milk from Holstein or Montbeliarde cows milked twice or once daily.
        J. Dairy Sci. 2009; 92 (19762788): 4730-4737
        • Martin B.
        • Verdier-Metz I.
        • Buchin S.
        • Hurtaud C.
        • Coulon J.-B.
        How do the nature of forages and pasture diversity influence the sensory quality of dairy livestock products?.
        Anim. Sci. 2005; 81: 205-212
        • Mele M.
        • Macciotta N.P.P.
        • Cecchinato A.
        • Conte G.
        • Schiavon S.
        • Bittante G.
        Multivariate factor analysis of detailed milk fatty acid profile: Effects of dairy system, feeding, herd, parity, and stage of lactation.
        J. Dairy Sci. 2016; 99 (27665132): 9820-9833
        • Næs T.
        Handling individual differences between assessors in sensory profiling.
        Food Qual. Prefer. 1990; 2: 187-199
        • Ojeda M.
        • Etaio I.
        • Fernández Gil M.P.
        • Albisu M.
        • Salmerón J.
        • Pérez Elortondo F.J.
        Sensory quality control of cheese: Going beyond the absence of defects.
        Food Control. 2015; 51: 371-380
        • Pegolo S.
        • Cecchinato A.
        • Casellas J.
        • Conte G.
        • Mele M.
        • Schiavon S.
        • Bittante G.
        Genetic and environmental relationships of detailed milk fatty acids profile determined by as chromatography in Brown Swiss cows.
        J. Dairy Sci. 2016; 99 (26709183): 1315-1330
        • Stefanon B.
        • Procida G.
        Effects of including silage in the diet on volatile compound profiles in Montasio cheese and their modification during ripening.
        J. Dairy Res. 2004; 71 (15068068): 58-65
        • Stocco G.
        • Cipolat-Gotet C.
        • Bobbo T.
        • Cecchinato A.
        • Bittante G.
        Breed of cow and herd productivity affect milk composition and modeling of coagulation, curd firming and syneresis.
        J. Dairy Sci. 2017; 100 (27837976): 129-145
        • Stocco G.
        • Cipolat-Gotet C.
        • Gasparotto V.
        • Cecchinato A.
        • Bittante G.
        Breed of cow and herd productivity affect milk nutrients recovery in curd, and cheese yield, efficiency and daily production.
        Animal. 2018; 12: 434-444
        • Verdier I.
        • Coulon J.B.
        • Pradel P.
        • Berdagué J.L.
        Effect of forage type and cow breed on the characteristics of matured Saint-Nectaire cheeses.
        Lait. 1995; 75: 523-533
        • Verdier-Metz I.
        • Coulon J.B.
        • Pradel P.
        • Viallon C.
        • Albouy H.
        • Berdagué J.L.
        Effect of botanical composition of hay and casein genetic variants on the chemical and sensory characteristics of ripened Saint-Nectaire type cheese.
        Lait. 2000; 80: 360-370
        • Verdier-Metz I.
        • Coulon J.B.
        • Pradel P.
        • Viallon C.
        • Berdagué J.L.
        Effect of forage conservation (hay or silage) and cow breed on the coagulation properties of milks and on the characteristics of ripened cheeses.
        J. Dairy Res. 1998; 65 (9513052): 9-21