Salting our freshwater lakes

Due to landscape position, lake ecosystems are influenced by surrounding terrestrial processes, and their generally long water residence times can contribute to the accumulation of external inputs and pollutants (1). Therefore, although lakes cover only 3% of the continental land surface (2), long-term trends in lakes are often early warning indicators of significant local, regional, or global changes (3). One such early warning indicator is change in lake chloride concentrations. Naturally occurring in freshwaters at low concentrations, chloride is a highly soluble and conservative ion that has also been shown to be a reliable proxy for chloride-based road salts (typically sodium chloride) (4, 5). Although chloride concentrations in freshwaters can vary cyclically due to climatic processes, such as extended periods of drought (6), elevated chloride concentrations in lakes often result from agricultural, industrial, and transportation practices (7). Elevated chloride concentrations can have adverse effects on water quality and aquatic ecosystems (8–11), including both immediate and long-term alterations to community structure, diversity, and productivity (12–14).

Salt application for de-icing roadways has been recognized as a major source of chloride to groundwater (15–17), streams and rivers (5, 10, 18, 19), and lakes (7, 9, 20, 21, 22) across north temperate climates in North America and Europe. In the United States, road salting became a standard practice in the 1940s, and road salt sales over the subsequent 50 y increased from 0.15 to over 18 million metric tons per year (4). In Canada, despite its addition to the List of Toxic Substances (23) and the implementation of the Code of Practice for the Environmental Management of Road Salts in 1999, an average of 5 million metric tons of road salt per year was applied to roadways between 1995 and 2001 (23, 24). Following application, road salt quickly dissolves and is transported into rivers and lakes through leaching and runoff (5, 25). A few studies have characterized the negative short term or localized impacts of elevated road salt concentrations in freshwaters (5, 15, 25), but there have been no large-scale analyses of chloride trends in freshwater lakes.

Here, we investigate trends in lake chloride concentration, using a dataset of long-term chloride concentrations in lakes and reservoirs in North America. We identify regions of high salinization, where aquatic ecosystems may be at risk, and contrast the role of climate versus the anthropogenic practice of road salting in driving chloride variability. Lakes included in the dataset were required to have at least 10 y of chloride data, a mean chloride concentration ≤1 g L⁻¹ (to exclude brackish lakes), and a surface area ≥4 ha. The median length of an individual time series was 21 y. The dataset included lake morphometric characteristics, climate statistics on temperature and precipitation, and atmospheric sea salt deposition. As a proxy for road salt application, land cover metrics were calculated, including road density (26) [length of road in a given area (km km⁻²)] and percent impervious land cover (25) within a 100- to 1500-m buffer surrounding each lake. Road density and impervious land cover represent the best proxies for road salt application, given that variability in road salt application, both spatially and on a year-to-year basis, prevents application rates from being calculated at spatial and temporal scales relevant to lakes.

Lakes in this dataset were not randomly sampled and thus do not necessarily represent the distribution of lakes within each state or province. To limit sampling bias in this dataset, we focused our analyses on a geographic area with dense sampling coverage: a North American lakes region (NALR), which includes Connecticut, Maine, Massachusetts, Michigan, Minnesota, New Hampshire, New York, Ontario, Rhode Island, Vermont, and Wisconsin (Fig. 1). We excluded North Dakota lakes from this grouping, as many are part of the Devil’s Lake watershed, an endorheic (closed-basin) system where water levels have risen ∼10 m since 1992, and therefore, the hydrology is vastly different from exorheic (open) lakes (27). Likewise, Manitoba lakes were excluded, as many were enlarged or drained during hydroelectric construction along the Churchill and Nelson Rivers (28). Of the 371 North American lakes in our dataset, 284 were in the NALR (Fig. 1). Mean chloride concentrations in lakes across the NALR ranged from 0.18 to 240.8 mg L⁻¹, with a median value of 6.0 mg L⁻¹.

Chloride time series for each lake differed in the frequency, duration, and depth of sampling. We pooled all depth samples for analyses, based on observations that chloride concentrations track similar trends throughout the water column of most lakes and that previous studies of long-term chloride trends have shown similarity with depth (7, 29). To reduce autocorrelation due to seasonality, we reduced all time series to annual averages. To enable comparison of chloride trends across lakes, a linear model was fit to the annual data, where chloride (standardized to a distribution with mean = 0 and SD = 1) was a function of time. Lakes were classified by simple linear regression models into three possible long-term trends: decreasing (n = 42, slope < 0, P < = 0.01), stationary (n = 204, slope = 0, P > 0.01), or increasing (n = 125, slope > 0, P < = 0.01). Of the 125 lakes with a positive trend in chloride, 99 were in the NALR (Fig. 1).

To investigate both linear trends and time-series patterns over a comparable period, any site in the NALR that had at least biennial data from 1985 to 2010 was included in a subset of long-term continuous (LTC) data. Clustering the 56 LTC lakes into three groupings using a hierarchical clustering analysis revealed three characteristic trends in chloride concentrations: neutral/decreasing (cluster 1, n = 16), oscillating (cluster 2, n = 4), and increasing (cluster 3, n = 36) (Fig. 2A). Cluster 1 was a geographical mix of lakes with both decreasing and neutral slope trends, cluster 2 lakes were exclusively in Maine and had neutral slope trends, and cluster 3 lakes, 21 of which were in Minnesota, had predominantly increasing slope trends (35 of 38) (Fig. 2B).

Potential drivers of increasing lake chloride were first assessed by relating slope values to lake, climate, and landscape characteristics of lakes in the NALR (Fig. 3 A–C). Due to the prevalence of zero-values in the data, it was not possible to build robust log-linear models for most of the landscape characteristics. Therefore, we used both classification/regression trees and random forests to build predictive models for the NALR data. A classification/regression tree and a random forest were created for each of three response variables: linear slope, tested as both continuous numerical and categorical (positive, zero, negative) variables, and hierarchical cluster grouping (1, 2, or 3). Categorical slope was used as a response variable to further remove any bias in our linear model application by removing magnitude. The motivation for using two approaches and three response variables was to improve the accuracy of our analytics, in much the same way as ensemble modeling.

Results of the three classification/regression trees and three random forests revealed that impervious land cover and road density surrounding each lake were the primary classification splits and the most important predictors for lake chloride trends and cluster grouping (Table 1).

The predictors used in the tree-based models were all static variables, meaning values did not vary with time. This limitation may misrepresent relationships between chloride concentrations and drivers that vary on a subannual basis (e.g., precipitation). Monthly precipitation data were obtained from the PRISM high-resolution spatial climate dataset, which covers the United States at a spatial resolution of 4 km (30). To account for the lag in chloride retention in a watershed (19), a LOESS curve was fit to mean monthly precipitation (mm/d) from 1985 to 2010 at each LTC site. A correlation between precipitation and chloride concentration at each LTC lake was calculated from annual data predicted from the LOESS precipitation curve and the generalized additive model (GAM) of chloride concentration. There was a strong negative correlation (r² = 0.71–0.87) between precipitation and chloride concentration for the four Maine lakes that group into cluster 2 (oscillating pattern). These four lakes are all less than 0.25 km² and receive ∼1.25 m of precipitation per year and have no impervious land cover within 500 m. Without knowledge of the groundwater hydrology of these lakes, it may be that precipitation controls the chloride balance, with heavy rains and large snowfalls diluting the chloride concentrations. A strong relationship between precipitation and chloride is not evident for lakes that group into cluster 1 or 3 (median r² = 0.12, range = 0–0.61).

Of our NALR sites, 44% of freshwater lakes have undergone long-term salinization. Positive chloride trends were present in lakes with as little as 1% impervious coverage. This finding is consistent with studies of US streams that found increased chloride concentrations associated with any urban land cover (31) or roads (32, 33) and substantiates findings of ecological community thresholds associated with low levels of catchment urbanization (34). Across the NALR, lakes with mean chloride concentrations >1 mg L⁻¹ (mean value of the time series) were more likely to be associated with positive trends in chloride (Fig. 3D). This suggests that high chloride concentrations in this region may be an indicator and warning sign of recent salinization.

If impervious land cover surrounding a lake is a robust predictor of water quality, it is important to understand the probability of its occurrence across all lakes within a region or country. Using national hydrography and land cover datasets for the continental United States, we found that the median percent impervious land cover within 500 m of all lakes greater than 4 ha is 0.31% (n = 149,350; Fig. 4). Of these US lakes, 28% had greater than 1% impervious land cover in a 500-m buffer zone. The density of roads and other impervious surfaces surrounding lakes in US regions where road salt is applied should therefore be of high concern. In the NALR, 70% (94 out of 134) of lakes with > 1% impervious land cover in the 500-m buffer had increasing chloride trends. If this result is extrapolated to all lakes in the US NALR (CT, MA, ME, MI, MN, NH, NY, RI, VT, and WI), ∼7,770 lakes may be experiencing elevated chloride concentrations, likely due to road salt runoff. This is calculated as 70% of the 11,104 out of 38,603 lakes in the US NALR greater than 4 ha that have >1% impervious land cover within 500 m. We note that data from Wisconsin and Minnesota are heavily biased toward urban lakes, whereas data from Maine, New York, and Vermont are heavily biased toward lakes in remote areas. This dataset (Fig. 4) includes lakes from all environments and should be representative of the Midwest and Northeast US region as a whole.

In North America, specifically in the Midwest and Northeast, local salt application leaves freshwater lakes vulnerable to salinization. Of the 284 lakes in the NALR, 26 already have a chloride concentration above 100 mg L⁻¹ at their last sampling date. The median impervious land cover within a 500-m buffer surrounding these 26 lakes is 24.8%, compared with the US mean 0.31%. If a linear relationship between time and chloride concentration is extrapolated, 47 lakes are on track to reach 100 mg L⁻¹ by the year 2050, and 14 are expected to surpass the EPA’s aquatic life criterion concentration of 230 mg L⁻¹ by 2050 (Fig. 4B). This is also the concentration at which a deterioration in drinking water taste is perceptible.

Elevated chloride concentrations in lakes can alter the composition and function of phytoplankton, zooplankton, macroinvertebrate, and fish communities (10–12, 35). As a consequence of salinization, aquatic species richness and abundance may decline, which could result in trophic cascades and altered water quality and ecosystem structure and function (36). In extreme cases, salinization can generate density gradients within the lake water column that prevent vertical mixing. Permanent stratification can result in anoxia and internal nutrient and metal resuspension, which decreases lake habitability and water quality (37). All of these ecosystem alterations can significantly affect lake water quality, which has millions of dollars in economic value (38, 39).

Our estimate that 7,770 lakes in the US NALR may be at risk for elevated chloride concentrations is likely an underestimate, as it does not consider regions of heavy road salt application where no long-term lake data were available, such as Québec or the Maritime provinces of Canada. Many states and municipalities are aware of the importance of shoreline management for maintaining healthy lakes; however, many shoreline zoning regulations are only enforced within 300 m or less of a lake (e.g., Wisconsin and Minnesota regulate 300 m, whereas Vermont and Maine only regulate 76 m). Because impervious surfaces and road density within at least 500 m of a lake are associated with increased chloride in areas that apply road salt, best management practices should recognize that lakeshore management extends well beyond the lake perimeter. Further, many jurisdictions lack consistent long-term monitoring programs, which provide data for predictive models and can be used to raise awareness and inform policy and management decisions used to curtail the threat of lake salinization. Clearly, keeping lakes “fresh” is critically important for protecting the ecosystem services freshwater lakes provide, such as drinking water sources, commercial fisheries, tourism, recreation, irrigation, and aquatic habitat.