Effects of Land Use on Lake Nutrients: The Importance of Scale, Hydrologic Connectivity, and Region

Abstract

Catchment land uses, particularly agriculture and urban uses, have long been recognized as major drivers of nutrient concentrations in surface waters. However, few simple models have been developed that relate the amount of catchment land use to downstream freshwater nutrients. Nor are existing models applicable to large numbers of freshwaters across broad spatial extents such as regions or continents. This research aims to increase model performance by exploring three factors that affect the relationship between land use and downstream nutrients in freshwater: the spatial extent for measuring land use, hydrologic connectivity, and the regional differences in both the amount of nutrients and effects of land use on them. We quantified the effects of these three factors that relate land use to lake total phosphorus (TP) and total nitrogen (TN) in 346 north temperate lakes in 7 regions in Michigan, USA. We used a linear mixed modeling framework to examine the importance of spatial extent, lake hydrologic class, and region on models with individual lake nutrients as the response variable, and individual land use types as the predictor variables. Our modeling approach was chosen to avoid problems of multi-collinearity among predictor variables and a lack of independence of lakes within regions, both of which are common problems in broad-scale analyses of freshwaters. We found that all three factors influence land use-lake nutrient relationships. The strongest evidence was for the effect of lake hydrologic connectivity, followed by region, and finally, the spatial extent of land use measurements. Incorporating these three factors into relatively simple models of land use effects on lake nutrients should help to improve predictions and understanding of land use-lake nutrient interactions at broad scales.

Fig 1. The location of the study lakes by lake class.

The study area is Michigan USA and the map shows the regions used in all analyses (Ecological Drainage Units) delineated by brown lines within state boundaries delineated by grey lines. The 346 lakes are shown by lake hydrologic class (DR_ST are drainage lakes with stream inflows; DR_ST-LK are drainage lakes with stream inflows and at least one upstream lake ≥ 10 ha, see Fig 2 for details). The regions are labeled with a three-digit code as in Table 1. The inset shows the location of Michigan within North America.

Fig 2. A description of lake hydrologic classes and the spatial extents for land use measurements.

A diagram explaining the features of the 3 different lakes classes and the spatial extents tested in this study. The lake zones are shown on the isolated lake diagram, the stream zone is shown in the DR_ST diagram, and the network catchment is shown in the DR_ST-LK diagram. However, such scales are also quantified for some (but not all) of the different lake classes as depicted in the bottom half of the diagram. na is not applicable.

Fig 3. Box plots of lake and landscape characteristics by lake class.

Where relevant, the landscape features are calculated for the local catchment spatial extent (see methods for details). The lake classes are ordered from left to right in order of increasing complexity of connections to other water bodies. Outliers are not plotted for all figures. The F statistic and p value are shown for a one-way ANOVA to test the difference among the three lake classes for each variable. Statistics in bold are significant at p ≤ 0.05. Samples sizes are 164 for isolated lakes, 97 for DR_ST lakes, and 85 for DR_ST-LK lakes.

Fig 4. Model fit and effect sizes for the fixed effects in models of land use-lake total phosphorus (TP).

For the relationships between TP and individual LULC types by lake class and spatial extent, the R² (A-D) and slopes (E-H) from mixed-models with random intercepts and fixed effects of the LULC type. Individual models were fit for each lake class and spatial extent of a particular LULC type combination (i.e., for % agriculture land use, 21 individual models were fit). In plots E-H, the error bars are confidence intervals (CI’s) of the slopes from the above models. Filled symbols are for the fixed effect slopes in which the CI does not overlap zero. Note, we show lines connecting the points of each lake class for ease of comparison across lake classes only.

Fig 5. Model fit and effect sizes for the fixed effects in models of land use-lake total nitrogen (TN).

For the relationships between TN and individual LULC types by lake class and spatial extent, the R² (A-D) and slopes (E-H) from mixed-models with random intercepts and fixed effects of the LULC type. Individual models were fit for each lake class and spatial extent of a particular LULC type combination (i.e., for % agriculture land use, 21 individual models were fitted). In plots E-H, the error bars are confidence intervals (CI’s) of the slopes from the above models. Filled symbols are for the fixed effect slopes in which the CI does not overlap zero. Note, we show lines connecting the points of each lake class for ease of comparison across lake classes only.

Fig 6. Synthesis of results by spatial extent and nutrient.

The dark grey shading corresponds to evidence for a stronger relationship at one or more spatial extents than at the other non-shaded spatial extents based on both R² and the absolute value of the slope (larger positive or negative number indicating a stronger effect). Dashes indicate a lack of an ecologically relevant relationship (in which either the slope CIs overlap zero or an R² value close to zero). LULC is the land use/land cover type, Wetl. is wetland, Nut. is nutrient, Loc. is local catchment, 100m-S is the 100 m zone around inflowing streams, Net. is the network catchment. The DR_ST lake class includes drainage lakes with streams flowing in; and the DR_ST-LK lake class includes drainage lakes with streams flowing in and upstream lakes ≥ 10 ha.

Fig 7. Region-specific slopes in models of land use-lake total phosphorus (TP).

For the relationships between TP and individual LULC types by lake class, spatial extent, and region, the region-specific slopes (colored circles, without the CI’s, for clarity) from mixed-effects models with varying intercepts and varying slopes of the LULC type. The light blue vertical bars indicate whether the varying slope parameter in this model was significant using a likelihood ratio test. We also show the fixed effect slopes including the non-significant estimates from Figs 4 and 5 as black diamonds for comparison only. The region codes are as in Table 1.

Fig 8. Region-specific slopes in models of land use-lake total nitrogen (TN).

For the relationships between TN and individual LULC types by lake class, spatial extent, and region, the region-specific slopes (colored circles, without the CI’s, for clarity) from mixed-effects models with varying intercepts and varying slopes of the LULC type. The light blue vertical bars indicate whether the varying slope parameter in this model was significant using a likelihood ratio test. We also show the fixed effect slopes including the non-significant estimates from Figs 4 and 5 as black diamonds for comparison only. The region codes are as in Table 1.

Fig 9. Synthesis of results for the relationship between lake nutrients and LULC.

Categories and descriptions as for Fig 6. The dark grey shading corresponds to either spatial extent (the fixed effects), region (the random effects), or lake class (comparison across models) being important for a specific land use/cover (LULC)-nutrient combination based on visual inspection (and some statistical evaluation) of graphs and model output. The numbers at the bottom row indicate the number of the tested LULC types (4 total) in each column that were shown to be important predictors of lake nutrient concentrations. The dashes indicate no relationship.