Integrating a Spatial Dimension Into the Relationship Between Landscape and Lake Chemistry Through the Use of a Large, Multi-Themed Database

Abstract:

Development of large-scale databases (from data collection networks or from a compilation of small datasets) has opened a new window for evaluating biogeochemical patterns on regional to continental scales. Adding a new spatial dimension to traditional relationships between landscape properties and lake chemistry will likely lead to identification of patterns that are often not considered in small-scale biogeochemical studies. Further, studies at a large spatial scale may highlight the importance of interactions between local and regional drivers. We tested these ideas using LAGOS (Lake multi-scaled GeOSpatial and temporal database), a recently compiled database that integrates lake nutrient chemistry data for ~ 5,000 lakes, lake characteristics (e.g., depth, area, watershed area) and geospatially-referenced landscape data (e.g., land use, climate, atmospheric deposition, hydrologic connectivity) over a 17-US state region. Specifically, we explored the spatial structure of different variables and identified which landscape variables were important in explaining variation in lake chemistry and stoichiometry.

Geostatistical analysis revealed a large range in spatial structure in landscape and lake variables, ranging from continental (temperature, precipitation, atmospheric deposition), to regional (water clarity, lake N and P, land use), to local scales (lake depth, lake area). Variables with similar spatial structure were most strongly related to each other. These spatial patterns were reflected in the relationship between lake chemistry and landscape drivers. Variables at multiple scales, including land-use, lake depth, and atmospheric N deposition, had similar effects on lake nitrogen and phosphorus chemistry. While landscape-scale effects on N and P were similar, local variables and internal nutrient processing influenced N and P differently and had the strongest influence on ratios of nutrients. Hence, N:P ratios were more locally structured than either element alone. These results demonstrate that spatial structure can have a strong influence in the analysis of links between limnological and landscape variables. Use of large, spatially-explicit datasets can lend additional insight into examination of drivers of aquatic nutrient chemistry on a regional to continental scale.