Importance of spatial scale in structuring emergent lake vegetation across environmental gradients and scales: GIS-based approach

Geographical information systems (GIS)-based approaches support the traditional empirical assessment and monitoring measures to indicate large-scale environmental alterations caused by global change. We examined the relationship between cover patterns of emergent aquatic plants derived from the CORINE land cover database, and environmental gradients (i.e. lake, climate and land cover and soil variables) at four different buffer scales (50 m, 100 m, 300 m and 500 m) in 408 boreal lakes in Finland. Our main statistical methods were hierarchical partitioning (HP), generalized linear models (GLM) and variation partitioning (VP). Based on correlations and HP, the importance of different buffer zones varied between environmental variables in explaining the vegetation cover. Our study also demonstrated that landscape productivity variables (i.e. Normalized Difference Vegetation Index and Tasseled Cap Greenness) strongly structured the vegetation cover. These productivity variables proved to be proxies for agricultural land and nutrient leaching from terrestrial areas to water, but they were also associated with shading of riparian vegetation. In the VP, land cover and soil variable groups influenced emergent vegetation cover the most, followed by lake and, lastly, by climate variables. Our work emphasized the importance of spatial scale in structuring freshwater assemblages that should be taken into account in ecological research and bioassessment. The use of a single spatial scale may lead to flawed results on the relevance of various environmental gradients on freshwater assemblages. Moreover, our study showed that satellite-based productivity variables can be powerful proxies for the amount of nutrients eventually entering aquatic ecosystems from adjacent terrestrial areas. The GIS-based approach used here provided an ecologically sound and cost-efficient early-warning estimation of vegetation overgrowth in boreal lakes. This approach is applicable to other regions with similar land cover data sets.