Long-term variations in primary production in a eutrophic sub-estuary: Contribution of short-term events

We determined the role of events in generating short-term variability in production, how they contribute to interannual variability, and their relationship to variability in the determinants of production, primarily biomass and photo-physiological parameters. We examined residuals from the seasonal and spatial mean daily rate in a 20-year time series of primary production in a eutrophic sub-estuary of Chesapeake Bay. The seasonal and spatial means of production and residuals were based on natural log-transformed measurements of daily primary production calculated from measurements of light saturation curves of photosynthesis for 20 years at 6 stations on the Rhode River, Maryland (USA). The variance of the residuals was greater than that explained by the seasonal and spatial statistical model, so that the event scale was the largest single source of variance. Residuals were classified as events if they exceeded +/-In(2), signifying a multiplier or divisor of 2 above or below the seasonal-spatial mean. Spatially, events were most frequent at the most upstream station affected by runoff from the local watershed, and temporally most frequent in spring at all stations. Principal component analysis (PCA) of monthly averaged residuals revealed 3 characteristic temporal modes of residual variance, the first of which was associated with variability in spring due to the occurrence of extremely large spring blooms or their complete absence. Interannual variation in annual production was correlated with the strength of expression of these modes. Production events were analyzed in relation to residuals in the determinants of productivity, i.e. phytoplankton chlorophyll biomass, B, the light-saturated photosynthetic maximum normalized to chlorophyll, P-max(B) the diffuse attenuation coefficient for light, and the depth integral of the dimensionless photosynthesis profile. Negatively correlated changes in B and P-max(B) were the most common mode of variation among the determinants, and this mode dampened variations in production and were common in fall months. Positively correlated changes in B and P-max(B) constituted the second most common mode of variation amongst determinants, and this mode was positively correlated with variations in production and were most common in spring months. The prevalence of the first mode in fall months modulated the impact of major named storms on primary production in this system. Published by Elsevier Ltd.