*2.8. Statistical Analysis*

Loading and unloading duration of the two genetic types were compared using T-test. The incidence of pigs showing the different behaviours at loading, unloading and lairage were calculated and the data were processed by Fisher Exact Test procedure of SAS v. 9.3 (SAS Institute, Cary, NC, USA).

Data from blood parameters were transformed to meet assumptions of homogeneity of variance and normality of residuals. Concentrations of CK, lactate, albumin, albumin/globulin ratio, creatinine, and AST were log10 transformed. A square root transformation was used to normalize cortisol and ALT results, while an inverse transformation was used to normalize glucose and K+ results. All transformed estimates were back-transformed for presentation to their original scale. Blood parameters were analysed using the mixed model (PROC MIXED of SAS) including the genetic type (two levels), sampling time (two levels: T0 and T1) and their interaction as fixed effects and subject within the day of slaughter as random effect. Sex as a fixed effect and hot carcass weight as a covariate were initially included in the model but they never reached statistical significance (*p* > 0.05) and were removed. Differences between means were tested by the Tukey-Kramer test (*p* < 0.05).

Data of pH, colour, cooking loss and shear force were analysed using PROC MIXED of SAS for repeated measures. The same model for blood parameters was used replacing sampling time factor with measuring time (five levels for pH, four levels for colour, three levels for cooking loss, and shear force). Sex and hot carcass weight did not reach the significant level (*p* > 0.05) and were removed from the model. The data of drip loss was analysed using the same model without measuring time.

In order to highlight possible differences between the two genetic types in blood parameters responses to short distance transport, the variation between blood parameters at exsanguination (T1) and basal blood parameters (T0) has been used to perform an unsupervised multivariate principal component analysis (PCA). All the new variables resulting from the difference between T1 and T0 blood parameters were normally distributed except for cortisol difference, which was root squared

transformed in order to meet normal distribution criteria. Furthermore, a PCA has also been used to test the presence of differences in meat quality traits between the two studied genetic types. Unsupervised PCAs have been performed using *ropls* package in the R environment version 3.4.4 [28]. The data were mean centred and unit- variance scaled. The results of multivariate models were plotted on both scores and loadings plot. The combined use of univariate and multivariate analyses was employed in order to test if the results obtained with the multivariate analysis (PCA) were in agreemen<sup>t</sup> with what could be observed with the mixed model.

PROC GLIMMIX was used to analyse the effects of genetic type on skin damage scores recorded on each quarter separately as well as on the whole carcass. Because these data approximated a Poisson distribution, the GLIMMIX procedure's POISSON option was used. The differences in least squares means (L.S.M.) were evaluated using Tukey–Kramer's test.

## **3. Results and Discussion**

Both genetic types did not carry the recessive allele (c.1843T) of the *RYR1* gene [27] and the dominant allele (c.599A or p.200Q) of the *PRKAG3* gene [26] that influence performance and meat quality traits [29,30].

Few research studies have been focused on the effect of transport on local breeds [16,17] although a grea<sup>t</sup> deal of literature exists on conventional commercial pigs and their crossbreeds [4,31,32]. The present study reports for the first time the effects of short distance transport on blood parameters and meat quality traits of Apulo-Calabrese. The results obtained showed different physiological response and meat quality attributes in both genetic types after the transport procedure.

#### *3.1. Behavioural Recordings and Carcass Bruises*

Behavioural occurrences on both genetic types were collected at loading, unloading and lairage. The two genetic types showed no differences in the behavioural occurrences during unloading and lairage, while at loading Apulo-Calabrese pigs showed significantly lower percentages (*p* < 0.05) of reluctance to move and vocalization (Table S1). During lairage, the posture was recorded after 25 min close to the end of the resting time (30 min) as planned routinely by the abattoir. For the duration of time spent in lairage no pigs were observed sitting, 94% of the pigs were lying down and only 6% of the pigs were observed standing. The genetic type did not show significant effect (*p* > 0.05) on skin damage score (whole carcass: 1.14 ± 0.34 and 1.12 ± 0.32 for Apulo-Calabrese and crossbreeds, respectively).
