Exceptions to the Heterotrophic Rule: Prevalence and Drivers of Autotrophy in Streams and Rivers

Streams and river ecosystems contribute to global carbon cycle via collecting, storing, transforming and producing organic matter. The general paradigm is that most stream ecosystems are heterotrophic, respiring more carbon than they produce, nonetheless, many streams are autotrophic for some time, and a few may autotrophic on an annual scale. Periodic autotrophy is associated with high light, low disturbance, and low watershed carbon inputs, but limited examples of annual-scale autotrophy make it difficult to estimate how frequently net autotrophy occurs and what drives this net storage or export of carbon. Here, we use a spatially and temporally extensive dataset of 236 rivers across the continental USA with 921 years of daily metabolism data to assess the scale of annual and seasonal autotrophy and estimate its strongest covariates. Only 6% of rivers in this dataset were consistently autotrophic on an annual timescale, but 67% of rivers experienced at least one autotrophic event lasting longer than 7 days. These periodic autotrophic events frequently began in spring and summer. By comparing the 37 annually autotrophic rivers identified in our dataset to examples in the literature, we found that anthropogenically driven autotrophy arising from agricultural and urban land use is underrepresented in the study of riverine metabolism. We used quantile regression models to test hypothesized drivers of autotrophy and a sparse lasso technique to broadly explore a range of 86 watershed and in-stream covariates. Our models support previous findings that high light, long periods between disturbances, and low terrestrial production covary with more autotrophy. Additionally, through sparse regression, we found that elevation and river width may be useful synthesis variables that explain the distribution of autotrophic rivers.