Nutritional quality modulates trait variability

Table 4 The linear and non-linear effects of food C/N ratios on the nine different traits of the oribatid mite Archegozetes longisetosus

	Linear effect	Nonlinear effect
	C/N	C/N x C/N
Chemistry
Amount (log transformation)
Estimate^a ± SE	0.125 ± 0.023	−0.007 ± 0.001
t₂₈₆	5.546	−5.242
p	< 0.0001	< 0.0001
Composition (1/4-power transformation)
pseudoF₂₈₆^b	354.390	118.350
r²	0.469	0.157
p	< 0.0001	< 0.0001
Regeneration (arcsine square root transformation)
Estimate^a ± SE	0.229 ± 0.057	−0.012 ± 0.003
t₁₀₀	4.031	−3.641
p	0.0001	0.0004
Life history
Developmental time (no transformation)
Estimate^a ± SE	−0.081 ± 0.035	0.005 ± 0.002
t₁₇₂	−2.321	2.455
p	0.0215	0.0151
Survival (arcsine square root transformation)
Estimate^a ± SE	0.130 ± 0.039	− 0.006 ± 0.002
t₂₁₆	3.357	−2.693
p	0.0009	0.0077
Offspring (square root transformation)
Estimate^a ± SE	0.144 ± 0.029	−0.007 ± 0.002
t₂₁₆	4.906	−4.140
p	< 0.0001	< 0.0001
Output (square root transformation)
Estimate^a ± SE	0.203 ± 0.033	− 0.010 ± 0.002
t₂₀₈	6.191	−5.530
p	< 0.0001	< 0.0001
Morphology
Body mass (log transformation)
Estimate^a ± SE	0.135 ± 0.021	−0.008 ± 0.001
t₂₀₈	6.543	−7.082
p	< 0.0001	< 0.0001
Morphometry (1/4-power transformation)
pseudoF₂₀₁^b	0.960	2.781
r²	0.005	0.013
p	0.3457	0.0603

^aThe estimate of the linear effect describes the slope of the relationship between the C/N ratio and the response variable (i.e. normalized trait value), while the estimate of the non-linear effect is the curvature of the quadratic relationship (i.e. normalized trait value ~ C/N x C/N). ^b For the multivariate traits we used permutational multivariate analysis of variance (PERMANOVA), instead of a normal linear model

ISSN: 1742-9994