| Summary: | International audience Long-term water chemistry data create both expected and unexpected insights towards how water and solutes enter and propagate through freshwater landscapes in the Anthropocene. Understanding these hydrochemical conditions is critical to protecting and managing aquatic ecosystems and ensuring human water security. However, predictive modeling and management efforts are hindered by the hydrochemical variability in catchment headwaters, where the majority of carbon and nutrients enter stream networks. While continuous monitoring of many small catchments is prohibitively expensive, particularly in developing countries where water-quality is degrading fastest, what if we could pinpoint where and when to implement conservation efforts with long-term, periodic sampling of headwater catchments? This question led us to use long-term data in an unexpected way. We developed a simple empirical framework informed by landscape ecology and catchment hydrology to quantify spatiotemporal variability in water chemistry across scales. We tested this framework with data from two regions: highly disturbed agroecosystems in western France where 56 headwater catchments were sampled periodically over 12 years, and relatively pristine tundra landscapes in Northern Alaska where we sampled 120 headwater catchments periodically over three years. In both locations, spatial variability of dissolved carbon, nitrogen, phosphorus, and major ions collapsed moving downstream, with variance thresholds occurring between 8 and 68 km2 for most solutes, corresponding to the size of sink and source patches in the landscape. Unexpectedly, temporal variability of dissolved carbon, nutrients, and major ions was preserved moving downstream and spatial patterns of water chemistry were stable on annual to decadal timescales. These long-term observations demonstrate that while hydrochemistry cannot be extrapolated among subcatchments, periodic sampling of headwaters provides valuable information about solute sources and network ...
|
|---|