Table 1. Output of two separate principal components analyses (PCA; N = 155 individuals) whose resulting factor scores (PC1) were used for statistical analyses (see Fig 3).

First PCA—metabolic PC score
Tissue variable	PC1 loading	Communality (h2)
Eigenvalue	3.86
Variance explained	77%
Plasma lactate	0.91	0.83
Plasma osmolality	0.89	0.79
Muscle lactate	0.94	0.88
Muscle ATP	-0.85	0.73
Muscle PCr	-0.80	0.63
Second PCA—plasma ion PC score
	PC1 loading	Communality (h2)
Eigenvalue	2.23
Variance explained	74%
Plasma Cl-	0.91	0.83
Plasma K+	-0.78	0.61
Plasma Na+	0.89	0.79

The first PCA, whose resulting factor scores are referred to as metabolic PC (principal component) scores, resulted after an initial PCA with all ten original physiological metrics, from which variables were successively removed because of having either a) a low Kaiser-Mayer-Olkin (KMO) measure of sampling adequacy (see Field et al. 2012), or b) not having a loading ≥ |0.6| (shown in bold) for any factor which also had other ≥ |0.6| loadings (i.ef., not agreeing strongly with other variables within a factor). PC loadings represent correlation coefficients (r) between the original variable and the new synthetic (e.g., PC1) variable. The second PCA (bottom) whose resulting factor scores are referred to as plasma ion PC scores, was initially run using the five remaining variables, and was simplified after the same iterative procedure used to refine the first PCA.