We present a Bayesian statistical model of diel oxygen dynamics in aquatic ecosystems to simultaneously estimate gross primary production, ecosystem respiration, and oxygen exchange with the atmosphere (and their uncertainties) on the basis of changes in dissolved oxygen concentration, water temperature, irradiance, and, if desired, the 18 O to 16 O ratio (d 18 O-O 2 ). We test this model using simulated data with realistic measurement errors to demonstrate that it accurately estimates the model parameters and that parameter uncertainties correctly scale with error in the observations and number of data points. Application of the model to field data from two productive stream ecosystems with substantial daily dissolved oxygen variation quantified the underlying physical and biological factors that control oxygen dynamics in these ecosystems and provided empirical support for a light saturation model of the photosynthesis-irradiance relationships at the ecosystem scale. Although inclusion of d 18 O-O 2 provides a second oxygen budget, analysis of field data shows that metabolic and reaeration parameters can be accurately estimated by modeling the transient dynamics of dissolved oxygen concentration alone in relation to daily changes in water temperature and light regime. This model is particularly suited to lowgas exchange, high-productivity systems, which have thus far proved challenging to measure ecosystem metabolism accurately. The modeling framework is applicable to single-station, open-system experimental designs and provides a rigorous and generalizable framework for estimating ecosystem metabolism in aquatic ecosystems.

Hollowed, A. B., Bond, N. A., Wilderbuer, T. K., Stockhausen, W. T., A'mar, Z. T., Beamish, R. J., Overland, J. E., and Schirripa, M. J. 2009. A framework for modelling fish and shellfish responses to future climate change. – ICES Journal of Marine Science, 66: 1584–1594. A framework is outlined for a unified approach to forecasting the implications of climate change on production of marine fish. The framework involves five steps: (i) identification of mechanisms underlying the reproductive success, growth, and distribution of major fish and shellfish populations, (ii) assessment of the feasibility of downscaling implications of climate scenarios derived from Intergovernmental Panel on Climate Change (IPCC) models for regional ecosystems to select and estimate relevant environmental variables, (iii) evaluation of climate model scenarios and select IPCC models that appear to provide valid representations of forcing for the region of study, (iv) extraction of environmental variables from climate scenarios and incorporation into projection models for fish and shellfish, and (v) evaluation of the mean, variance, and trend in fish and shellfish production under a changing ecosystem. This framework was applied to forecast summer sea surface temperature in the Bering Sea from 2001 to 2050. The mean summer surface temperature was predicted to increase by 2°C by 2050. The forecasting framework was also used to estimate the effects of climate change on production of northern rock sole (Lepidopsetta polyxystra) through projected changes in cross-shelf transport of larvae in the Bering Sea. Results suggest that climate change will lead to a modest increase in the production of strong year classes of northern rock sole.

Abstract. Dynamics of dissolved oxygen in aquatic ecosystems reflect the biological, physical, and chemical processes that regulate ecosystem metabolism. Organic matter that supports ecosystem respiration (ER) is produced both by in situ photosynthesis and via loading from terrestrial ecosystems. Terrestrial-derived organic matter is relatively recalcitrant and its availability is stable at diel time scales relative to substrates produced through photosynthesis, which are more labile and often show distinct diel changes in availability. Here, we explored whether the contributions of these two sources of organic matter to ecosystem metabolism could be quantified by a process model of photosynthesis and ER fit to high-frequency observations of oxygen concentration in streams. We found that a two-stage model of respiration provided a better fit to diel oxygen dynamics in most streams than a model assuming that the substrates supporting ER were one temporally stable pool. Two-stage models estimated peak daytime respiration rates that were~2.99 higher on average than nighttime rates, but this increase was variable and ranged from 1.1 to 11.69. Estimates of gross primary production were 1.359 higher, on average, (range = 1.04-7.919) compared to estimates generated by a single-stage model. Streams draining watersheds with less than about 7% gradient exhibited oxygen dynamics that provided comparable statistical support for single-stage metabolism, likely due to the higher loading of allochthonous organic matter that swamped metabolism based on autochthonous production in these streams. Ecosystem metabolism in streams draining steeper watersheds was best characterized by two-stage metabolism, reflecting the greater importance of autochthonous contributions to labile organic carbon pools in these ecosystems.