. 2021 Sep 14;9(1):coab076. doi: 10.1093/conphys/coab076

Table 5.

Statistical test results and associated significance for various blood physiology parameters throughout crowding trials on Atlantic mackerel

			Main effects models						Interactive models
			X₁ (‘trial’)			X₂ (‘monitoring period’)			X₃ (X₁ × X₂ interaction)
Parameter	n	Variance structure	df	F-value	P	df	F-value	P	Variance structure	df	F-value	P
Cortisol^a	143	NA	4	12.707	<0.001	4	2.555	0.041	NA	16	2.458	0.003
Lactate	143	VarIdent (trial × monitoring period)	4	85.935	<0.001	4	12.666	<0.001	VarIdent (trial × monitoring period)	16	10.948	<0.001
Haematocrit	143	None	4	4.851	0.001	4	1.752	0.142	None	16	1.393	0.157
pH	143	VarIdent (monitoring period)	4	1.740	0.145	4	8.360	<0.001	VarIdent (monitoring period)	16	2.720	0.001
Osmolality	142	VarIdent (trial)	4	6.670	<0.001	4	10.380	<0.001	VarIdent (monitoring period)	16	3.570	<0.001
Glucose	143	VarIdent (monitoring period)	4	3.049	0.019	4	14.471	<0.001	VarIdent (monitoring period)	16	5.367	<0.001
Potassium ions	143	VarIdent (trial)	4	5.430	<0.001	4	4.202	0.003	None	16	3.447	<0.001
Sodium ions	143	VarIdent (monitoring period)	4	3.040	0.0196	4	19.49	<0.001	None	16	2.870	<0.001
Chloride ions	143	VarIdent (trial)	4	13.400	<0.001	4	9.260	<0.001	VarIdent (trial)	16	4.390	<0.001

^§

^aData modelled using GLM with gamma error distribution and log link function.

Both main and interactive effects of ‘trial’ and ‘monitoring perio’ were fit using GLS models (unless otherwise indicated) using a variety of variance structures.