Effect of breed on physicochemical and sensory characteristics of fresh, semihard and hard goat’s milk cheeses

ABSTRACT

The aim of this work was to study the physico-chemical parameters, texture, colour and sensorial characteristics of experimental cheeses from three equivalent groups of 15 multiparous goats from each of the three Canarian breeds, Majorera, Palmera and Tinerfeña, during ripening. A great influence on the milk basic composition was observed when comparing breed effect. The Palmera breed had the best milk quality characteristics with the highest percentage of protein (5.16%), fat (4.62%), and dry matter (14.75%). Tinerfeña breed had the lowest value of protein and non-fat solids whereas Majorera breed showed the minimum score for fat percentage. The main effects of goat breed on cheese basic composition were found in protein (P<0.01), pH (P<0.01) and Aw (P<0.05). Cheese sensorial texture was only partially affected by breed, three of nine attributes. Breed influenced all taste parameters, with higher scores in 30 and 60 days’ stages, although odour and flavour intensity was similar. Cheese from Majorera breed was the richest in aromatic descriptors showing four specific sensations, two of them, goat milk and butter, belonging to the lactic family and the other, rancid and cellar, related to miscellaneous descriptors. Finally, consumers were not able to differentiate cheeses made with milk from various breeds.

KEYWORDS:

• Goat milk and cheese
• breed
• chemical and sensory properties

Introduction

The preservation of local breeds is a strategical instrument to conserve traditional products with specific sensorial and nutritional characteristics. Autochthonous breeds, through natural selection, are adapted to their particular environments and they are also more resistant to diseases, being able to produce in harsher conditions. Therefore, local breeds are an essential tool for animal biodiversity conservation (Ciotola et al. 2009; Di Gregorio et al. 2017), and can be contemplated as traditional resources because of their role in local social life and rural culture (De Marchi et al. 2007).

The production of different Canarian cheeses, made with autochthonous breed milk, involves an important economic income and promotes the protection of indigenous genetic resources, characterized by a strong and specific relationship with the territory.

The physicochemical composition, microbiological and sensory properties of cheese depend on many factors which influence the quality of cheeses through the composition and technological behaviour of the milk (Fresno and Álvarez 2011). Breed is the main genetic aspect affecting milk quality and, consequently, milk coagulation properties and cheese characteristics. Among them, cheese yield, physicochemical characteristics, and sensorial properties are affected by several genetic factors (Coulon et al. 2004). The milk quality used for cheese manufacture is essential for certain typical products; some of them protected by the European label Protected Designation of Origin (PDO) and Protected Geographic Indication (PGI), where the connection between final product and milk origin is very strong, as it happens, for example in different cow cheeses (De Marchi et al. 2008). Breed and some technological practices can be important tools for the definition and differentiation of these traditional and labelled cheeses and can even contribute to the production of new types of cheeses (Fresno and Álvarez 2011).

Goat milk has some special and unique aroma and flavour characteristics as well as nutritional and health values (Hayaloglu et al. 2013). These properties can be influenced by different factors including breed, genetic, physiology, feed, environment, and production technology (Raynal-Ljutovac et al. 2008). Goat milk production has historical practice in southern Europe with an intensive specialization for milk production; Spain is one of the main producers of goat milk in the EU. Particularly, Canary Islands have a long tradition for raising native goats, and foreign breeds, with higher production yields, have not been allowed to exploit, to favour local breeds whose dairy production has highly distinguishable characteristics (Torres et al. 2013).

In Canary Island there are three strongly differentiated dairy goat breeds: Majorera, Palmera and Tinerfeña. All of them are adapted to different islands and environments where they have lived isolated for long time. Research studies have demonstrated significant differences on morphologic (Capote et al. 1998), genetic (Martínez et al. 2006) and productive (Torres et al. 2013) traits. Furthermore, there are three traditional cheeses awarded with a Protected Denomination of Origin (PDO), two of them made exclusively from goats’ milk and the other made from ewe milk mixture with cow and goat milk (Fresno and Álvarez 2007).

This experimental trial is part of a RTA (Agrifood resources and technologies) project that aims the valorization of traditional goat cheeses linked to an autochthonous breed establishing the milk origin traceability from molecular markers. The objective of this experiment was to determine the effect of three Canarian goat breeds on the physical–chemical and sensorial characteristics of cheese.

Materials and methods

Experimental design and procedure

Equivalent groups of 15 multiparous goats from each of the three Canarian breeds were used (Majorera, Palmera and Tinerfeña). These animals had kidded in the traditional season (December), and the trial began in March (midlactation period). All goats were kept and fed under identical conditions in El Pico farm (Canary Agronomic Research Institute, ICIA). A standard diet for lactating goats was used for both groups, composed by mixed concentrate, lucerne pellets, corn grain and rye-grass hay (1153, 578, 494, 445 g kg⁻¹ respectively). The chemical composition, on a DM (87.01%) basis, was 93.00% OM, 11.29% CP, 25.56% NDF, 17.67% ADF, 3.54% ADL, 5.97 NE (MJ kg⁻¹).

The nutritional requirements of the animals and the nutritive value of their diet were calculated in accordance with the nutritional standards set out in the INRATION 4.0 software (INRA 2007).

Milk samples

The goats were milked once daily (at 07:00 h) in a double 12-stall parallel milking parlour low-line milk pipeline (Alfa Laval Iberia SA, Madrid, Spain) equipped with recording jars (4 L ± 5%). Milking was performed at a vacuum pressure of 42 kPa, a pulsation rate of 90 pulses/min, and a pulsation ratio of 60/40, in accordance to Torres et al. (2013).

Cheese samples

Milk produced by each of the three experimental groups was processed to make cheese. In the same week 36 cheeses were processed in an experimental cheese factory placed in the Animal Production Unit: 12 goat cheeses from each experimental group (Majorera, Palmera and Tinerfeña breed). Cheeses were made following the guidelines described by Álvarez et al. (2018). The cheeses were made with raw milk on the same day as milking. After heating to 30 ± 1°C, standard animal rennet (commercial rennin powder, Marshall rennet power 50% quimosin and 50% pepsin) was added, following the manufacturer’s instructions, to obtain a 35 min clotting time. After coagulation, curds were cut to obtain grains the size of millet. Press process was the same for all cheeses: 4.9 kPa for 5 h. Subsequently, salting was achieved by rubbing dry salt onto the surface of the cheeses. 12 goat cheeses from each Majorera, Palmera and Tinerfeña breed were used for this study. The cheeses were stored in a ripening chamber at 10–12°C and 85–86% relative humidity. Four cheeses were analyzed from each group after 2 (fresh), 30 (semihard), and 60 (hard) days of ripening. Half of each cheese was used for the physicochemical analysis and the other half was used for the sensory analysis. Cheese samples were coded with a letter and with a number. For each ripening time, cheeses were sent to the laboratory in refrigerated boxes and a basic chemical analysis was performed immediately. For the sensory analysis, all samples were wrapped in aluminium sheets, stored under refrigerated conditions, and placed at room temperature (20 ± 1°C) for 2 h before testing.

Physicochemical analysis

Milk

The protein, fat, lactose and dry matter contents were measured in a representative sample taken from the vat of each cheesemaking using a MilkoScan Star (Foss Electric, Slangerupgad, Denmark). The pH value was determined using a pH metre Crison PH 25+ (Barcelona, Spain) at room 20°C temperature.

Cheese

Basic chemical composition analysis was performed in triplicate using a near infrared spectroscopy (Instalab 600, Foss Electric, Slangerupgad, Denmark). Cheese pH was measured at 20°C by introducing a penetrometric electrode into the cheese. The pH value was determined at room temperature (20°C) using a pH metre Crison PH 25+ (Barcelona, Spain). Water activity was measured by a water activity metre Pawkit (Decagon Devices, Washington U.S.A.).

Texture analysis

The textural analysis was performed in a texturometer TA-Xt2i (Stable Micro Systems, Surrey, UK), with a load cell of 5 kg, by carrying out a texture profile analysis (TPA). Prior to TPA, a 0.5-cm layer was removed from the upper surface of the cheese to obtain a regular surface for probe compression, analyses were performed at 20 ± 2°C under lubricated conditions to eliminate frictional effects. 6 representative samples of each cheese were cut into cylinders with a 40-mm diameter and a 60-mm height and used for a compression test. This test was performed using a 50-mm diameter cylindrical flat probe; a compression of 75% and a crosshead speed of 2 mm/s. From the force vs. time texturograms, 6 parameters were obtained for compression: hardness (N), fracturability (N), adhesiveness (N.s), cohesiveness, elasticity and gumminess. The interpretation of these texture parameters was made in accordance with methods advocated by Bara-Herczegh et al. (2002).

Colour measurement

Internal colour was recorded using a portable MINOLTA spectrocolourimeter (Minolta CR-400, Osaka, Japan). The L* (lightness), Croma and Hue Angle colour measurements were determined according to the CIELCH colour space, were L* corresponds to light/dark chromaticity (changing from 0% dark to 100% light), colour intensity was recorded using the Croma value and Hue angle was used as a measure of colour tone. The instrument was calibrated with a white tile prior to measuring. Each colour test was performed on ten replicates inside the cheese sample.

Sensory analysis

Sensory analyses were carried out at 2, 30 and 60 d of ripening. Samples, coded with random 3-digit codes, were presented in a balanced way to avoid the effect of the presentation order. Cheeses were served without any identification of the origin of the milk used (Majorera, Tinerfeña, Palmera). The sensorial methodology used has been described previously (Álvarez et al. 2007; Fresno and Álvarez 2012); odour and flavour attributes were in accordance with those described by Beródier et al. (1996) and texture followed the guidelines published by Lavanchy et al. (1999). This methodology has been adapted to goat cheeses by Fresno and Álvarez (2007).

Finally, a preference test was carried out to determine the acceptability of the experimental cheeses. Preference tests were done according to the Spanish Association for Standardisation and Certification standard UNE 87-005-92, as described by Fresno and Álvarez (2007) with 3 panels comprising 144, 112 and 83 individuals for each ripening stage respectively (2, 30 and 60 days); consumers scored each cheese from 1 to 5 points, with 1 rated as ‘very bad’ and 5 as ‘very good’. In addition, an experimental triangular test was conducted to ascertain the ability of no-expert judges to differentiate cheese manufactured with milk from different goat breeds.

Statistical analysis

The software package SPSS version 15.0 (SPSS Inc., Chicago, IL, U.S.A.) was used for statistical processing of the results. The significance level was set at a probability of 5% (P ≤ 0.05). All the variables of milk and cheeses chemical composition were tested for normal distribution using the Kolmogorov–Smirnov test. Likewise, Levene test was applied to verify the homogeneity of the variances. A general linear model (GLM) was used to establish statistical differences between the physicochemical parameter values and sensory analysis scores according to the type of breed, ripening time, and the interaction between these 2 factors. Post hoc multiple analyses by Tukey’s test were used for the ripening time factor. The classification of cheeses samples with different breeds using sensory characteristics were achieved by principal component analysis (PCA) and lineal discriminant analysis (LDA).

Results and discussion

Physicochemical characteristics

Milk. Table 1 shows the mean values of the chemical composition of goat’s milk from three different Canarian breeds. In general, it can be observed that, gross composition was within the values reported for goat milk (Park 2007). It is well accepted and has been demonstrated by various authors that there may be appreciable differences in the chemical composition of milk from different goat breeds (Raynal-Ljutovac et al. 2008; Trancoso et al. 2010).

Table 1. Effects of breed (Majorera, Palmera, Tinerfeña) on milk physicochemical composition.

	Breed
	Majorera	Palmera	Tinerfeña
Fat, %	4.13±0.03^c	5.16±0.02^a	4.30±0.02^b
CP, %	3.75±0.01^b	4.62±0.02^a	3.61±0.01^c
Lactose, %	4.74±0.02^a	4.16±0.01^b	4.69±0.01^a
Dry matter, %	13.18±0.02^b	14.75±0.05^a	13.22±0.02^b
Non-fat solids, %	9.14±0.02^b	9.52±0.03^a	8.95±0.01^c
pH	6.61±0.07	6.65±0.09	6.67±0.05^c

^a-d Within a row, means marked with different superscripts differ significantly (P<0.001).

In the present case there were consistent differences (P<0.001) among breeds for all the parameters considered except pH. This observation agreed with Park (2007) who referred wide variation in protein and fat contents within species, with specific influence by breed, stage of lactation, feeding, climate, parity, season, and udder health status. Although physicochemical characteristics were influenced by breed, higher percentages of fat or protein in milk do not always lead to cheeses with a superior level of these chemical compounds (Fresno and Álvarez 2011).

As was expected, the Palmera breed had the best milk quality characteristics with the highest percentage of protein (5.16%), fat (4.62%), and dry matter (14.75%). These values are quite lower to those found in a previous experiment with Palmera breed where goats fed different diets (Álvarez et al. 2018). Fat content is one of the most important components of fluid goat milk, and lipids are involved in cheese yield, firmness, colour and flavour of goat dairy products. Various fatty acids are also potentially involved as positive or negative predisposing factors in the health of human consumers (Ribeiro and Ribeiro 2010).

Tinerfeña breed had the lowest value of protein and non-fat solids whereas Majorera breed showed the minimum score for fat percentage. Fat and protein content were similar to those reported by Álvarez et al. (2007) for this same breed. Additionally, cheese making characteristics are affected by goat milk physicochemical properties, including pH, calcium and other mineral concentrations in milk, which cause differences in coagulation parameters.

Cheese. The composition of the three breed cheeses during ripening is shown in Table 2. No significant (P > 0.05) differences were found for total solids and fat content according to type of breed. The main effects of goat breed were found in protein (P < 0.01), pH (P < 0.01) and Aw (P < 0.05). Many of the variations in cheese characteristics are due to the differences in milk composition and when milk is no standardized, as in this experiment, the effect of breed in cheese chemical composition is more evident. Cheeses made from Majorera and Palmera breed showed a higher protein content at 2 and 60 days of ripening while Majorera Aw was lower at 30 and 60 days of maturation. Linking milk and cheese composition, the variations detected in gross composition of milk could affect protein cheese content. Furthermore, the lowest cheese pH was detected in Palmero cheese in all stages reaching values close to 5 after 30 days of maturation. The same research group (Fresno et al. 2001) found significant differences between Majorera and Tinerfeña goat’s milk; these variations only affected fresh cheeses, although sensory properties showed many differences between fresh, semihard and hard cheeses.

Table 2. Effects of breed (Majorera, Palmera, Tinerfeña) and ripening (2, 30, 60 days) on physical and chemical cheese composition.

	Breed^1	Ripening time
		2d	30d	60d
Total solids (TS, %)	M	46.93±1.15^C	64.63±2.03^B	73.32±1.01^A
	P	48.42±0.29^C	67.09±2.49^B	71.30±1.01^A
	T	48.59±2.26^B	66.93±2.56^A	71.65±0.59^A
Fat (% of TS)	M	42.71±1.57	45.18±1.82	43.89±1.35
	P	42.54±1.03	43.63±2.32	43.46±0.67
	T	44.06±2.51	44.05±2.28	43.48±0.55
Protein (% of TS)	M	41.57±1.66^a	41.88±2.27	43.15±0.71^a
	P	41.99±1.40^a	41.96±1.70	43.66±0.45^a
	T	36.96±1.15^bB	40.42±2.96^AB	41.73±0.81^bA
pH	M	6.27±0.07^aA	5.04±0.03^bB	5.10±0.10^B
	P	5.88±0.07^bA	4.99±0.04^bB	5.09±0.09^B
	T	6.35±0.05^aA	5.34±0.06^aB	4.91±0.10^C
Aw	M	0.96±0.01^A	0.90±0.01^bB	0.86±0.01^bC
	P	0.95±0.01^A	0.93±0.01^aB	0.90±0.01^aC
	T	0.96±0.01^A	0.92±0.01^aB	0.88±0.01^aC
Cheese yield (l/kg)	M	6.52±0.32^bA	11.00±0.80^bB	11.79±1.10^bB
	P	6.23±0.26^aA	9.62±1.38^aB	10.89±1.29^aC
	T	6.22±0.33^aA	11.35±1.62^bB	11.89±1.08^bB

^A-C Within a row, means marked with different superscripts differ significantly (P<0.05).

^a-c Within a column, means marked with different superscripts differ significantly (P<0.05).

¹M= Majorera breed; P= Palmera breed; T= Tinerfeña breed.

Analyzing TS content and the ripening time, TS increased significantly as a consequence of the loss of water. This increment is especially marked at the beginning of the conservation period. The moisture of most cheeses is established by the speed and extension of syneresis and compression of the structure of casein. After clotting, processes such as moulding, pressing and salting join a decline of pH and produce a significant loss of humidity by eliminating the whey. The fat percentage did not vary significantly with the maturation development, contrasting to other cheeses’ research made with Canarian goat's milk (Álvarez et al. 2007; Sánchez-Macías et al. 2011), cow's milk (Coppa et al. 2011) and a mixture of goat and cow's milk (Diezhandino et al. 2015). Paying attention to TS and fat in relation to breeds, no differences were detected neither fresh, semihard nor hard cheeses. These results agree with those obtained by Soryal et al. (2005) with Nubian and Alpine goats for soft cheese in U.S.A.

Cheese yield, formulated as the percentage ratio between the cheeses produced and the milk processed, is of great economic importance, being one of the main final production targets of many cheese farmers (Stocco et al. 2018). This parameter firstly depends on the fat and protein content of milk, especially casein, and on the technological properties of milk (Law and Tamine 2010). Palmera goat breed showed the best cheese yield results for semihard and hard cheeses, displaying similar ratio to Tinerfeña goats for fresh cheeses. These results assume that for every 100 litres of milk obtained, Palmera breed can produce 14% more semihard cheese than Majorera breed and 18% than Tinerfeña breed, which was expected because the Palmera milk had 23% more protein than the Majorera milk and 28% than the Tinerfeña milk. The milk protein quality and its cheese-making properties are strong influenced by the α_s1-Cn gene and its allelic variability (Martin and Grosclaude 1993). Canarian breeds have a particular allelic distribution, specifically Palmera breed show a clear predominance of high alleles (A + B), that can contribute to a better cheese yield (Jordana et al. 1996).

Instrumental texture and colour

Mean values obtained for texture attributes of TPA during ripening for the three breeds evaluated are presented in Table 3. According to certain authors, the texture of a cheese is as important as its flavour (Tunick 2000) and it is closely related to the specific proteins in the cheese (Park 2007). The quantity and distribution of the caseins, as well as the manufacturing steps, determine the structure of the cheese matrix (Van Hekken et al. 2004).

Table 3. Effects of breed (Majorera, Palmera, Tinerfeña) and ripening (2, 30, 60 days) on textural and colour characteristics.

	Breed^1	Ripening time
		2d	30d	60d
Fracturability	M	33.53±2.81^aC	58.03±5.53^bB	233.44±6.15^aA
	P	28.86±1.25^abC	51.74±8.88^bB	75.69±8.86^bA
	T	24.68±2.49^bC	106.03±13.74^aB	258.05±22.50^aA
Hardness	M	48.74±11.49^B	123.14±18.73^bB	512.90±80.33^aA
	P	35.54±5.09^C	111.36±28.25^bB	198.36±31.68^bA
	T	37.92±8.64^C	211.79±26.97^aB	655.63±49.29^aA
Cohesiveness	M	0.19±0.07	0.12±0.03	0.13±0.01
	P	0.12±0.01	0.13±0.04	0.12±0.03
	T	0.20±0.06	0.13±0.02	0.15±0.02
Adhesiveness	M	7.4±1.82^B	13.95±3.56^B	100.00±2.83^A
	P	15.32±5.56^B	21.59±6.65^AB	65.67±9.60^A
	T	14.36±4.32^B	29.15±3.65^AB	113.54±9.96^A
Elasticity	M	93.10±4.38^aA	99.23±2.01^aA	52.71±0.43^bB
	P	46.04±0.15^bB	101.86±3.22^aA	81.36±14.85^aA
	T	51.09±3.33^bC	91.16±0.54^bA	63.19±1.18^abB
Gumminess	M	914.45±462.86^aB	1531.13±508.61^B	3458.91±708.26^bA
	P	193.18±29.94^b	1536.17±928.23	2013.55±1063.19^b
	T	389.19±148.17^abC	2538.60±196.96^B	6260.05±1288.29^aA
L*	M	91.26±0.45^A	82.77±0.56^bB	79.67±1.27^C
	P	90.74±0.61^A	83.79±0.69^bB	79.86±0.71^C
	T	90.77±0.04^A	85.41±0.46^aB	77.92±1.63^C
a*	M	−0.98±0.20^B	−1.35±0.26^AB	−2.01±0.36^aA
	P	−1.43±0.13^B	−2.14±0.46^B	−3.92±0.16^bA
	T	−1.07±0.27^B	−1.60±0.21^AB	−2.26±0.47^aA
b*	M	8.32±0.82^b	9.45±0.84	8.36±0.51^b
	P	10.49±0.31^aB	11.17±1.23^B	13.90±0.02^aA
	T	8.85±1.05^abB	11.16±0.64^A	8.99±0.89^bB
Croma	M	8.38±0.83^b	9.54±0.86	8.61±0.56^b
	P	10.59±0.32^aB	11.37±1.30^B	14.44±0.03^aA
	T	8.92±1.08^abB	11.27±0.66^A	9.28±0.89^bB
Hue angle	M	96.65±0.77^B	98.07±0.92^bB	103.47±1.91^A
	P	97.79±0.48^C	100.76±1.08^aB	105.73±0.62^A
	T	96.82±1.03^B	98.17±0.64^bB	104.14±2.76^A

^A-C Within a row, means marked with different superscripts differ significantly (P<0.05).

^a-c Within a column, means marked with different superscripts differ significantly (P<0.05).

¹M= Majorera breed; P= Palmera breed; T= Tinerfeña breed.

Breed factor significantly affected (P < 0.05) fracturability, hardness, elasticity and gumminess. Majorera and Tinerfeña cheeses were more fracturable at 30 and 60 aging days while hardness was not affected by breed in fresh cheeses (2 days). Tinerfeña cheese was the hardest at semihard and hard ripening stages. In other similar studies, with ricotta soft whey cheese, (Pizzillo et al. 2005) where four Italian goat breeds were analyzed (Girgentana, Siriana, Maltese and Local), this factor had a limited influence in textural and colour properties.

As was observed previously in other Canarian goat cheeses (Álvarez et al. 2018) the ripening period revealed significant changes for most of the textural parameters analyzed except cohesiveness. Regardless of the breed analyzed, fracturability, hardness, adhesiveness and gumminess increased from 0 to 60 days of ripening while elasticity behaviour was inconsistent between breeds. The increase in hardness is related to decreasing moisture which acts as a plasticizer in the protein matrix, thereby making it less elastic and more susceptible to fracture upon compression (Fox et al. 2000). Table 3 also shows the CIELab and CIELch colour coordinates for the interior of the cheeses evaluated. The values for lightness (L*) vary from 91.26 for Majorera cheese at 2 days’ stage to 77.92 for Tinerfeña cheese at 60 days of ripening, thus indicating a gradual darkening. Lightness was only affected by breed at 30 days of maturation where the Tinerfeña cheese was the brightest. Moreover, the values for greenness/redness (a*) vary from −0.98 for Majorero fresh cheese (0 days) to −3.92 for Palmera hard cheese (60 days) while the values for yellowness (b*) were higher in Palmero cheese in these two same stages.

Ripening time affected all colour parameters in mostly cheeses by breed, only b* and chroma values in Majorera cheese were not significant. These values are in the same range as those reported by Fresno and Álvarez (2012) in cheeses made with Majorera milk. In the development of the aging process, the interior colour loses whiteness tending to acquire an ivory colouration, as can be noted by the values of the colour parameters a* and b*.

Sensorial analysis

Table 4 resumes the textural sensory properties evaluated for the cheeses at the study, considering the three different breeds and the three ripening stages analysed.

Table 4. Effects of breed (Majorera, Palmera, Tinerfeña) and ripening (2, 30, 60 days) on sensorial texture characteristics.

	Breed^1	Ripening time
		2d	30d	60d
Roughness	M	1.43±0.53^C	3.43±0.31^B	5.43±0.45^A
	P	1.71±0.27^C	3.57±0.24^B	5.20±0.35^A
	T	1.71±0.27^C	3.57±0.24^B	5.50±0.25^A
Superficial moisture	M	6.57±0.54^aA	4.29±0.23^aB	2.00±0.29^C
	P	5.79±0.27^bA	3.89±0.13^bB	1.90±0.25^C
	T	5.79±0.27^bA	3.89±0.13^bB	2.05±0.26^C
Elasticity	M	5.00±0.82^A	3.21±0.42^B	1.43±0.45^C
	P	4.50±0.41^A	2.96±0.22^B	1.43±0.45^C
	T	4.50±0.41^A	2.96±0.22^B	1.43±0.45^C
Firmness	M	2.71±0.49^bC	4.11±0.41^B	5.50±0.65^A
	P	3.57±0.45^aC	4.54±0.44^B	5.50±0.65^A
	T	3.57±0.45^aC	4.32±0.28^B	5.07±0.73^A
Friability	M	2.07±0.45^C	3.86±0.28^B	5.64±0.63^A
	P	2.29±0.57^C	4.07±0.49^B	5.86±0.48^A
	T	2.29±0.57^C	3.96±0.34^B	5.64±0.63^A
Adhesivity	M	1.21±0.39^C	1.96±0.49^B	2.71±0.70^A
	P	1.21±0.39^C	1.96±0.49^B	2.71±0.70^A
	T	1.21±0.39^C	1.96±0.49^B	2.71±0.70^A
Solubility	M	3.43±0.54^B	4.25±0.38^AB	5.07±0.84^A
	P	3.00±0.29^B	4.04±0.42^AB	5.07±0.84^A
	T	3.00±0.29^B	4.29±0.23^AB	5.07±0.84^A
Mouth moisture	M	4.71±0.49^aA	3.82±0.24^aB	2.93±0.19^bC
	P	3.36±0.38^b	3.14±0.24^b	2.93±0.19^b
	T	3.36±0.38^b	3.68±0.47^a	4.00±0.58^a
Granulosity	M	3.71±0.49^B	4.25±0.52^AB	4.79±0.69^A
	P	4.00±0.29^B	4.39±0.43^AB	4.79±0.69^A
	T	4.00±0.29	4.00±0.38	4.00±0.58

^A-C Within a row, means marked with different superscripts differ significantly (P<0.05).

^a-c Within a column, means marked with different superscripts differ significantly (P<0.05).

¹M= Majorera breed; P= Palmera breed; T= Tinerfeña breed.

Cheese sensorial texture may be defined as a composite sensory attribute resulting from a combination of physical properties and perceived by the senses of sight, touch, and hearing (Pinho et al. 2004).

Breed type had a minor influence on textural properties, affecting only three of nine parameters considered (superficial and mouth moisture and firmness). The cheeses made with Majorera milk had the highest superficial and mouth moisture values in the two first ripening stages while Palmera cheese showed highest values for mouth moisture at 60 days of maturation with differences of 1.07 in the score. Moreover, firmness was only affected by breed at day 2 of manufacture. Nevertheless, in previous experiments Álvarez et al. (2008) found higher breed effect when comparing Palmera and Majorera texture sensory characteristics showing significant differences (P<0.001) in six of eight parameters.

The cheese texture is, at first, controlled by the pH and by the ratio of casein and moisture. The reduction of the pH induces the demineralization of the micelles which, successively, has a considerable impact on the association between the casein protein network (Guiné et al. 2015).

Regarding the effect of maturation, in general, the values for these texture parameters tended to increase gradually during the 60 days of ripening at rates that depended on the moisture loss. Roughness, firmness, friability, adhesivity, solubility and granulosity increased while superficial moisture and elasticity decreased their scores till the end of maturation. Mouth moisture was only affected by ripening time at Majorera cheese.

Taste, odour and flavour characteristics of experimental cheeses made with three milk goat breeds at differing ripening stages are showed in Table 5. Breed influenced all taste parameters, with higher intensity in 30 and 60 days’ stages. Palmera cheese was more acid and pungent but less astringent at these two periods. On the other hand, odour and flavour intensity was similar between cheeses regarding of the breed milk used. Although with no significant effect, Palmera values were higher than Majorera and Tinerfeña scores. Other breed experiment with Canarian goats showed differences (P < 0.001) in odour and flavour intensity between Majorera and Palmera cheeses, showing higher intensity in Majorera ripened (semihard and hard) cheese but with favourable scores for Palmera cheeses in fresh ones (Álvarez et al. 2008).

Table 5. Effects of breed (Majorera, Palmera, Tinerfeña) and ripening (2, 30, 60 days) on taste, odour and flavour intensity.

	Breed^1	Ripening time
		2d	30d	60d
Acidity	M	0±0	0±0^b	0±0^b
	P	0±0	1.57±0.35^a	3.14±0.69^a
	T	0±0	0±0^b	0±0^b
Sweetness	M	1.71±0.81^aA	0±0^B	0±0^B
	P	1.71±0.81^aA	0±0^B	0±0^B
	T	0±0^b	0±0	0±0
Pungent	M	0±0^C	0.80±0.27^bB	1.57±0.54^bA
	P	0±0^C	1.32±0.24^aB	2.64±0.48^aA
	T	0±0^C	0.80±0.27^bB	1.57±0.54^bA
Astringent	M	0±0^bC	0.79±0.27^aB	1.57±0.54^aA
	P	0.86±0.38^aA	0±0^bB	0±0^bB
	T	0±0^bC	0.79±0.27^aB	1.57±0.54^aA
Odour intensity	M	1.43±0.73^C	3.21±0.62^B	5.00±0.65^A
	P	1.43±0.73^C	3.46±0.44^B	5.50±0.60^A
	T	1.43±0.73^C	3.21±0.62^B	5.00±0.65^A
Flavour intensity	M	1.71±0.57^C	3.68±0.40^B	5.64±0.48^A
	P	2.36±0.48^C	4.07±0.43^B	5.79±0.57^A
	T	1.71±0.57^C	3.68±0.40^B	5.64±0.48^A

^A-C Within a row, means marked with different superscripts differ significantly (P<0.05).

^a-c Within a column, means marked with different superscripts differ significantly (P<0.05).

¹M= Majorera breed; P= Palmera breed; T= Tinerfeña breed.

Ripening had a major impact on these sensorial characteristics, where acidity was the only unaffected parameter. Pungent, astringency and odour and aroma intensity increased till 60 days of ripening while sweetness was only detected at primary stage in Majorera and Palmera cheeses.

Flavour of cheese is determined by its taste and aroma and results from the correct balance and concentration of numerous sapid and aromatic compounds perceived during cheese consumption (Albenzio and Santillo 2011). With regard to the odour and flavour specific descriptors (Table 6), the different milk used in the cheese manufacture (breed) significantly affected (P < 0.05) 8 of 10 parameters (cellar and whey odour and all aroma descriptors). Sensory properties of goat cheeses are an important factor for consumer acceptability and marketability of the products. Most sensory and textural attributes of cheeses increase during ripening (Park and Drake 2005). When cheese quality is analyzed at market through surveys of consumer perceptions, selection, freshness and flavour were listed as the primary reasons for purchasing (Teng et al. 2004).

Table 6. Effects of breed (Majorera, Palmera, Tinerfeña) and ripening (2, 30, 60 days) on odour and flavour specific descriptors.

	Breed^1	Ripening time
		2d	30d	60d
Odour
Goat milk	M	2.86±0.38^aB	2.54±0.37^B	5.07±0.73^A
	P	2.64±0.48^aB	2.54±0.37^B	5.07±0.73^A
	T	0±0^bC	2.54±0.37^B	5.07±0.73^A
Whey	M	0±0	0±0^b	0±0^b
	P	0±0^C	1.86±0.28^aB	3.71±0.57^aA
	T	0±0	0±0^b	0±0^b
Rancid	M	0±0^C	1.07±0.31^B	2.14±0.63^A
	P	0±0^C	1.07±0.31^B	2.14±0.63^A
	T	0±0^C	1.07±0.31^B	2.14±0.63^A
Flavour
Goat milk	M	3.14±2.44^aB	2.50±0.38^aB	5±0.76^aA
	P	0±0^b	0±0^b	0±0^b
	T	0±0^b	0±0^b	0±0^b
Whey	M	0±0	0±0^b	0±0^b
	P	0±0^C	2.18±0.24^aB	4.36±0.48^aA
	T	0±0	0±0^b	0±0^b
Butter	M	0±0^C	1.29±0.30^aB	2.57±0.61^aA
	P	0±0	0±0^b	0±0^b
	T	0±0	0±0^b	0±0^b
Rancid	M	0±0^C	0.82±0.24^aB	1.64±0.48^aA
	P	0±0	0±0^b	0±0^b
	T	0±0	0±0^b	0±0^b
Cellar	M	0±0^C	0.93±0.24^aB	1.86±0.48^aA
	P	0±0	0±0^b	0±0^b
	T	0±0	0±0^b	0±0^b
Pungent in nose	M	0±0	0±0^b	0±0^b
	P	0±0^C	2.39±0.35^aB	4.79±0.69^aA
	T	0±0	0±0^b	0±0^b

^A-C Within a row, means marked with different superscripts differ significantly (P<0.05).

^a-c Within a column, means marked with different superscripts differ significantly (P<0.05).

¹M= Majorera breed; P= Palmera breed; T= Tinerfeña breed.

Regarding the influence of the ripening time in the specific sensorial descriptors, a general increase is observed in all the sensations perceived. In 2 days’ cheeses only goat milk descriptor was detected in Majorera and Palmera odour and in Majorera aroma. Moreover, goat milk and rancid odour is perceived in the three breeds analysed while whey descriptor only in Palmero cheeses. In an Italian exhaustive study, sensorial characteristics of ricotta cheeses were affected by breed; cheeses made with Siriana goats’ milk had higher score for ‘goat taste’ and additionally a greater granulosity (Pizzillo et al. 2005). It is been known for a long time that characteristic ‘goat’ flavour has been linked to animal breed (Ronningen 1965) and that cheeses made with local goat milk showed a stronger taste than others made with foreign breed milk as Saanen (Skjevdal 1979).

Majorera cheese is the richest in aromatic descriptors showing four specific sensations; two of them, goat milk and butter, belonging to the lactic family and rancid and cellar, related to miscellaneous descriptors (Fresno and Álvarez 2007).

It is interesting to note that typical odour and flavour descriptors in Canarian cheeses from vegetal and fruity families did not appear at all in our experimental cheeses, perhaps because no high quality forage was used in the common diet. It has been confirmed in previous studies, when supplying local forages as tedera and tagasaste in Palmera goats (Álvarez et al. 2018), and vinagrera, saltbush and local barley hay in Majorera goats (Álvarez et al. 2007), a significative increase in specific hay, fruity, and dried fruit descriptors. In contrast, in standard diets where wheat straw was used as forage, less complex descriptors appeared, being goaty, lactic acid and rennet the most important. In our study, where fescue hay was supplied, goat milk and whey were detected as main odour and flavour descriptors. Cheeses made with milk from goats fed local forages were also more appreciated by the expert panel and by a consumer test. Furthermore, less acidity and trigeminal sensation and more solubility, taste persistence, and odour and flavour intensity was determined (Álvarez et al. 2007).

Consumer test results (data non shown) showed highly valued cheeses in the three breeds and ripening stages. Majorera (4.01), Palmera (4.03) and Tinerfeña cheeses (4.06) were valued above 4 points (good level). No significant differences were observed in the cheeses assessment according to the breed or the ripening period. On the other hand, triangular trials with consumers revealed a great difficulty in differentiating the breed origin of cheeses. Although in semihard cheeses there were about 50% successes, in fresh and hard cheeses this percentage decreased significantly (38% and 22% respectively).

Factor analysis, using principal component analysis (PCA) was applied to 15 sensorial variables including textural parameters, taste descriptors and odour and aroma intensity. Sensory data was subjected to PCA, to ascertain which variables contributed most to the total variance, obtaining a more simplified view of the relationship between the sensorial parameters analysed.

Two factors were chosen (75.83% of the total variance) because their eigenvalues were higher than 1, and therefore, they explain more variance than the original variables (Table 7). A Varimax rotation was carried out to minimize the number of variables that influence each factor and then facilitate the interpretation of the results.

Table 7. Factor matrix obtained after a Varimax rotation.

% cumulative variance	Factor
	1	2
Superficial moisture	−.939	−.253
Elasticity	−.939	−.229
Flavour intensity	.927	.253
Roughness	.925	.251
Friability	.916	.256
Odour intensity	.911	.258
Solubility	.836	.059
Adhesivity	.829	.130
Firmness	.825	.280
Pungent	.786	.503
Astringency	.656	−.604
Sweetness	−.566	−.317
Acidity	.194	.911
Mouth moisture	−.219	−.589
Granulosity	.389	.431

The analysis showed that about 64.35% of the total variation is explained by the first principal component (PC1), 75.83% by the first two principal components (PC1 and PC2). The first factor that explains the higher percentage of variance is positively correlated with odour and flavour intensity and textural attributes and negatively correlated with superficial moisture and elasticity. PC2 is characterized by pungent trigeminal sensation and granulosity and in a negative extend with astringency and mouth moisture.

Representing the score plots for all the cheese samples on the first and second factor (Figures 1 and 2), it was observed that cheeses made with milk from different breeds were well differentiated. A discriminant model was built with the purpose of separating the cheeses according to the breed. The discriminant functions were obtained from a strategy of selection of variables by steps. Thus, the variables are included or excluded according to the Wilks’ lambda. Majorera and Palmera are very well classified; while in Palmera breed only two cheeses were grouped in different groups (Table 8).

Figure 1. Principal component analysis of the sensory data. Plot of the cheese samples.

Figure 2. Principal component analysis of the sensory data. Correlation plot.

Table 8. Results of the stepwise lineal discriminant analysis (LDA) of all cheese samples according to type of breed, numbers and (%).

	Breed	Predicted group			Total
		1	2	3
Initial Group	Majorera	21(100)	0	0	100.0
	Tinerfeña	0	19 (90.5)	2(9.5)	100.0
	Palmera	0	0	21(100)	100.0

96.8% samples well classified.

Conclusions

The results indicate that physicochemical and sensory properties of cheese vary according to goat breed (Majorera, Palmera and Tinerfeña). For PDO cheeses like Majorero and Palmero linked to a specific autochthonous breeds, experimental breed studies are very important for demonstrating such influence. In addition, this could prevent the use of milk from other breeds in protected cheeses. The effect on milk composition was high, whereas cheese basic composition and texture and colour parameters were less affected. Expert judges had found differences in sensorial texture, odour, flavour and taste while consumers were not able to differentiate between breeds. Majorera cheeses were more appreciated by the expert panel. This result can be due to a less bitterness and sweeter sensation with higher odour and flavour intensity and the presence of goat, butter and hay descriptors.

Acknowledgements

This study was supported financially by the RTA2014-00047-00-00 Project with FEDER Funds.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This study was supported financially by the RTA2014-00047-00-00 Project with FEDER Funds.