An official website of the United States government
Abstract River metabolism and, thus, carbon cycling are governed by gross primary production and ecosystem respiration. Traditionally river metabolism is derived from diel dissolved oxygen concentrations, which cannot resolve diel changes in ecosystem respiration. Here, we compare river metabolism derived from oxygen concentrations with estimates from stable oxygen isotope signatures (δ18O2) from 14 sites in rivers across three biomes using Bayesian inverse modeling. We find isotopically derived ecosystem respiration was greater in the day than night for all rivers (maximum change of 113 g O2 m−2 d−1, minimum of 1 g O2 m−2 d−1). Temperature (20 °C) normalized rates of ecosystem respiration and gross primary production were 1.1 to 87 and 1.5 to 22-fold higher when derived from oxygen isotope data compared to concentration data. Through accounting for diel variation in ecosystem respiration, our isotopically-derived rates suggest that ecosystem respiration and microbial carbon cycling in rivers is more rapid than predicted by traditional methods.
more »
« less
Fitting metabolic models to dissolved oxygen data: The estuarine Bayesian single‐station estimation method
Abstract Continuous measurements of dissolved oxygen (DO) are useful for quantifying ecosystem metabolism, which is critical for understanding estuarine biogeochemistry and ecology, but current methods applied to these data may lead to estimates that are physically impossible and poorly constrained errors. Here, we present a new approach for estimating estuarine metabolism: Estuarine BAyesian Single‐station Estimation (EBASE). EBASE applies a Bayesian framework to a simple process‐based model and DO observations, allowing the estimation of critical model parameters, specifically light efficiency and respiration, as informed by a set of prior distributions. EBASE improves upon the stream‐based model from which it was derived by accommodating missing DO data and allowing the user to set the time period over which parameters are estimated. We demonstrate that EBASE can recover known metabolic parameters from a synthetic time series, even in the presence of noise (e.g., due to tidal advection) and when prior distributions are uninformed. Optimization periods of 7 and 30 d are more preferable than 1 d. A comparison with the more‐conventional method of Odum reveals the ability of EBASE to avoid unphysical results (such as negative photosynthesis and respiration) and improves when the DO data are detided. EBASE is available using open‐source software (R) and can be readily applied to multiple years of long‐term monitoring data that are available in many estuaries. Overall, EBASE provides an accessible method to parameterize a simple metabolic model appropriate for estuarine systems and will provide additional understanding of processes that influence ecosystem status and condition.
more »
« less
- Award ID(s):
- 1924559
- PAR ID:
- 10501785
- Publisher / Repository:
- Wiley Blackwell (John Wiley & Sons)
- Date Published:
- Journal Name:
- Limnology and Oceanography: Methods
- Volume:
- 22
- Issue:
- 8
- ISSN:
- 1541-5856
- Format(s):
- Medium: X Size: p. 590-607
- Size(s):
- p. 590-607
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
ABSTRACT Intermittent streams are characterized by significant periods of low to no flow, yet are also frequently subjected to flashy, high floods. Floods alter ecosystem function and result in variable successional patterns across the stream network. Yet, the timing of restored function after floods in intermittent stream networks is relatively unexplored. We measured recovery of stream ecosystem function using rates of gross primary production (GPP), ecosystem respiration (ER), net ecosystem production (NEP), and the primary production to respiration ratio (P/R) across eight locations in the Kings Creek drainage basin with differing preflood conditions (previously dry [intermittent] or flowing [perennial]) over a 30‐d period following a 2‐yr return interval flood. We found that all metabolic rates (GPP, ER, NEP, P/R) varied primarily by time (days since flood) and antecedent flow, but not spatial network position (i.e., drainage area). Intermittent sites exhibited high rates of ER (0.17–3.33 g dissolved oxygen [DO] m−2d−1) following rewetting compared to perennial sites (0.03–1.17 g DO m−2d−1), while GPP, NEP, and P/R were slower to recover and varied less between sites of differing preflood conditions. Metabolic rates were not strongly influenced by other environmental conditions. A large proportion of variation was explained by the random effect of location. Our results suggest that metabolism is temporally asynchronous and highly heterogenous across intermittent watersheds and that antecedent hydrology (drying prior to rewetting) stimulates heterotrophic activity, likely dependent on terrestrially derived organic matter and nutrient subsidies.more » « less
-
Abstract The temporal structures of gross primary production (GPP) and ecosystem respiration (ER) vary across time scales in response to complex interactions among dynamic drivers (e.g., flow, light, temperature, organic matter supply). To explore emergent patterns of river metabolic variation, we applied frequency‐domain analysis to multiyear records of metabolism across 87 US rivers. We observed a dominant annual periodicity in metabolic variation and universal fractal scaling (i.e., power spectral density inversely correlated with frequency) at subannual frequencies, suggesting these are foundational temporal structures of river metabolic regimes. Frequency‐domain patterns of river metabolism aligned best with drivers related to energy inputs: benthic light for GPP and GPP for ER. Simple river metabolism models captured frequency‐domain patterns when parameterized with appropriate energy inputs but neglecting temperature controls. These results imply that temporal variation of energy supply imprints directly on metabolic signals and that frequency‐domain patterns provide benchmark properties to predict river metabolic regimes.more » « less
-
Abstract Ecosystem metabolism of freshwater ecosystems has been studied for several decades, with theory and synthesis largely derived from temperate streams and rivers in North America and Europe. Advances in sensor technology and modeling have opened a wider range of streams to be included to test theories beyond temperate streams. In this paper, we review and synthesize ecosystem metabolism data from tropical streams and rivers to determine to what extent the constraints of metabolism measured in temperate streams are similar in tropical streams. We compiled 202 measurements of gross primary productivity (GPP) and ecosystem respiration (ER) from 83 tropical streams spanning 22.2°S to 18.4°N. Overall, tropical streams were heterotrophic (ER > GPP), with GPP ranging from 0.01 to 11.7 g O2m−2d−1and ER ranging from −0.2 to −42.1 g O2m−2d−1, similar on average to rates reviewed from temperate streams, but with higher maximum ER in tropical streams. Gross primary productivity increased with watershed area; a result also observed in temperate streams. ER decreased with elevated phosphorus and higher annual rainfall. We constructed a structural equation model that explained greater variation of ER (74%) than GPP (26%), and reflects similar drivers, such as land‐use and watershed area, as in temperate streams. We conclude that tropical stream ecosystem metabolism has similar drivers as temperate streams, and a warmer and wetter climate and human use of tropical lands will influence metabolic rates in streams.more » « less
-
Machine learning models often perform poorly under subpopulation shifts in the data distribution. Developing methods that allow machine learning models to better generalize to such shifts is crucial for safe deployment in real-world settings. In this paper, we develop a family of group-aware prior (GAP) distributions over neural network parameters that explicitly favor models that generalize well under subpopulation shifts. We design a simple group-aware prior that only requires access to a small set of data with group information and demonstrate that training with this prior yields state-of-the-art performance -- even when only retraining the final layer of a previously trained non-robust model. Group aware-priors are conceptually simple, complementary to existing approaches, such as attribute pseudo labeling and data reweighting, and open up promising new avenues for harnessing Bayesian inference to enable robustness to subpopulation shifts.more » « less
