Plasma markers of oxidative stress are uncorrelated in a wild mammal

Oxidative stress, which results from an imbalance between the production of potentially damaging reactive oxygen species versus antioxidant defenses and repair mechanisms, has been proposed as an important mediator of life‐history trade‐offs. A plethora of biomarkers associated with oxidative stress exist, but few ecological studies have examined the relationships among different markers in organisms experiencing natural conditions or tested whether those relationships are stable across different environments and demographic groups. It is therefore not clear to what extent studies of different markers can be compared, or whether studies that focus on a single marker can draw general conclusions regarding oxidative stress. We measured widely used markers of oxidative damage (protein carbonyls and malondialdehyde) and antioxidant defense (superoxide dismutase and total antioxidant capacity) from 706 plasma samples collected over a 4‐year period in a wild population of Soay sheep on St Kilda. We quantified the correlation structure among these four markers across the entire sample set and also within separate years, age groups (lambs and adults), and sexes. We found some moderately strong correlations between some pairs of markers when data from all 4 years were pooled. However, these correlations were caused by considerable among‐year variation in mean marker values; correlation coefficients were small and not significantly different from zero after accounting for among‐year variation. Furthermore, within each year, age, and sex subgroup, the pairwise correlation coefficients among the four markers were weak, nonsignificant, and distributed around zero. In addition, principal component analysis confirmed that the four markers represented four independent axes of variation. Our results suggest that plasma markers of oxidative stress may vary dramatically among years, presumably due to environmental conditions, and that this variation can induce population‐level correlations among markers even in the absence of any correlations within contemporaneous subgroups. The absence of any consistent correlations within years or demographic subgroups implies that care must be taken when generalizing from observed relationships with oxidative stress markers, as each marker may reflect different and potentially uncoupled biochemical processes.