Overall, SOD was greater in sediments with more fauna (H1), consistent with previous studies (Aller 1988; Pelegri and Blackburn 1995; Waldbusser et al. 2004; Webb and Eyre 2004). In samples with low faunal biomass, SOD was relatively low throughout the diel cycle, which is not unexpected in sandy sediments low in organic content (Hargrave 1972; Burdige 2006). Our measurements of SOD and the observed increase with faunal presence also agree well with chamber measurements from other studies in similar habitats. Banta et al. (1999) measured basal SOD rates of 30 mmol m− 2 d− 1 in lab microcosms with unpopulated sandy sediments, and the addition of the polychaete Hediste diversicolor (formerly Nereis diversicolor) inflated SOD by 80–90% (Banta et al. 1999). Likewise, Webb and Eyre found that burrowing thalassinid shrimp in in situ benthic chambers increased SOD by 80% compared to unoccupied sediments (Webb and Eyre 2004).
We found significant variability in SOD on two distinct temporal scales; diel variability driven by the presence of fauna and sub-diel variability on the scale of minutes to hours. The significant effects of DO and faunal biomass on SOD indicates that changing DO did not itself affect SOD, but rather, where fauna were present, changing DO drove shifts in their activity and behavior that affected SOD. This is consistent with our hypothesized response (H2B), as well as the observations of Wenzhofer and Glud (2004), who found that the majority of nightly DO uptake could be attributed to faunal effects. Their variability in SOD appeared to be driven by light availability driving circadian behaviors. Our measurements were made in shaded chambers, and although the duration of our deployment was short enough that light-driven circadian behaviors cannot be eliminated as a potential driver, our data do directly link change in SOD to DO variability. We hypothesized that fauna may flatten the relationship between DO and SOD (H2B), however in the regression of the DO vs SOD slopes against faunal biomass increasing sediment faunal presence essentially inverted the relationship of DO and SOD throughout the diel cycle, surpassing even our hypothesized response. The regression of faunal biomass and the SOD/DO slope crosses zero which indicates that there exists some tipping point where increasing faunal presence begins to drive higher nightly SOD. Our data demonstrate that the presence of fauna changes SOD patterns considerably throughout the diel cycle, and in ways that could not be predicted from measurements at any single time point or even from daytime measurements alone.
Though we were unable to directly measure animal activity within the chambers during the deployments, the increase in SOD in more populated sediments at night was likely driven by fauna responding to low DO by increasing bioirrigation. Nereid worms irrigate their burrows (Wells and Dales 1951), and several common species are known to shift from oxygen conforming to regulatory behaviors at low DO concentrations (Kristensen 1983). However, the two samples with highest nightly SOD values (Samples 3 and 4) were dominated by orbiniids and had relatively little nereid biomass. Orbiniids are not known to construct and irrigate burrows like some nereids do (Jumars et al. 2015), so we would not have predicted that they would be driving the elevated SOD measurements. They are, however, active burrowers (Francoeur and Dorgan, 2014), and have many dorsal external gills along their posteriors (Rouse et al., 2022); irrigation during active burrowing may contribute more to bioirrigation than we expected. A more likely explanation is that these samples may have contained individuals of the tube-dwelling Onuphid polychaete Diopatra cuprea that evaded collection by retreating deep in their tubes when the sediment cores were extracted. D. cuprea tubes create subsurface habitat and refuge for other burrowing animals (Woodin 1978), and their irrigation activity is well documented (Mangum et al. 1968; Dales et al. 1970). Both the higher faunal abundance and the greater SOD measured in those samples may have been due to the presence of D. cuprea.
The short time scale of irrigation activity also explains the sub-diel variability in our measurements. In most of our samples, SOD varied substantially measurement-to-measurement, i.e., on ~ 20 min time scales, and samples with greater faunal presence exhibited greater high-frequency variability. Infauna typically irrigate intermittently in cycles of activity and rest which vary in intensity, rate and duration and be altered by DO availability (Wells and Dales 1951; Mangum et al. 1968; Kristensen 1983; Kristensen 1989; Kristensen 2001; Volkenborn et al. 2010; Volkenborn et al. 2012; Camillini et al. 2019). The irrigation pattern depends on the taxon, however most irrigating infauna cycle through behaviors on a scale of minutes or tens of minutes. On the time scale of our SOD incubations (~ 20 min) fauna intermittently irrigating likely contributed to the measurement-to-measurement variability. In sediments with large and complex faunal communities, intermittent irrigation would drive swings in SOD on very short timescales and depending on the resident fauna. Even accounting for diel variation by measuring SOD at maximum and minimum DO and interpolating measurements between them would not accurately describe SOD dynamics throughout the entire cycle and would still yield unreliable estimates of net SOD over time. This may be compounded by variation in benthic primary production, which was purposefully excluded in this study.
It is common practice in sediment research to extrapolate single SOD measurements through time and space. SOD measurements taken in laboratory core incubations and with “batch” style benthic chambers are questionable in their representativeness, and the faunal effects that can substantially increase both the magnitude and the variability in SOD add important context to high-frequency eddy correlation SOD measurements. Measurement of SOD on small temporal and spatial scales has only recently become methodologically feasible but is proving potentially highly significant, and variability can now be captured but has yet to be fully explained. Our observed high spatial and temporal SOD variability, and their apparent interdependence, indicate that increasing resolution of SOD measurement in both dimensions may be not only beneficial but necessary to accurately characterize metabolic processes in these systems.