Nutrient Loading From Residential Turfgrass

Nutrient loading has been blamed for algal blooms and the decline in seagrass in the Indian River Lagoon (Gao 2009). Popular assumptions about the role of residential fertilizer in nutrient loading to the lagoon have become the source of policies and regulatory actions. In 2010, the Florida Department of Environmental Protection (FDEP) developed the Model Ordinance for Florida-Friendly Fertilizer Use on Urban Landscapes to assist local governments in managing nutrient loading. The model ordinance defines the required best management practices for fertilizer application and allows each local government to restrict application during certain times of the year. Based on the model ordinance, five counties around the lagoon have issued weather-based restrictions on fertilizer application. Two counties ban summer fertilizer application and three counties require the use of slow-release nitrogen during certain times of the year. In addition, the University of Florida’s Institute of Food and Agricultural Sciences (IFAS) has published recommendations for fertilizer application rates and schedules for Florida homeowners.

To better understand the relationship between residential fertilizer application in the watershed and nutrient loading in the lagoon, we conducted a literature review and developed a fertilizer loading model, (here referred to as the FLM). We identified best management practices as suggested by the research and characterized model inputs such as fertilizer application rates, fertilizer uptake rates, and management actions. Our model, based on FDEP and St. Johns River Water Management District (SJRWMD) watershed models, simulates how fertilizer application and best management practices impact nutrient loading from residential areas, focusing on loadings from turfgrass.

Fertilizer-Specific Literature
We developed a bibliography of 37 scholarly reports and articles on nutrient loading in the Indian River Lagoon; however, we did not find any articles that explicitly quantified the impact of urban fertilizer input to the lagoon. For the purposes of this review, we looked at literature that examined relative amounts of nutrient leaching from varying amounts of irrigation or rainfall, programmatic applications of slow-release (aka controlled-release) and fast-release (aka soluble) fertilizers and the effect of coatings on slow-release fertilizers.

Several studies document a correlation between nutrient loadings and excessive irrigation or rainfall (Trenholm and Sartain 2010, Carey et al. 2012, and Erickson et al. 2010). Carey et al. (2012) references a study that reports phosphorus runoff rates based on simulated rainfall events. In that study, greater phosphorus losses were shown to occur from Bermuda grass four hours post-fertilizer application than 24 hours post-fertilizer application, indicating that greater nutrient runoff can be expected when fertilizer applications occur closer to rainfall events.

Much of the literature recommends incorporating slow-release fertilizers into application programs. Trenholm and Sartain (2010) references a number of studies that indicate that slow-release fertilizers result in less nitrogen loss than quick-release fertilizers. Carey et al. (2012) and Saha et al. (2007) report similar information, and suggest that regulating fertilizer content may prove to be an effective strategy for reducing nutrient leaching. Hochmuth et al. (2009) recommends using controlled-release fertilizers with no more than 0.5 pound per 1,000 square feet soluble nitrogen during the summer months, rather than implementing summer fertilizer bans. Chopra et al. (2011) finds that the appropriate use of slow-release fertilizers can significantly reduce nutrient loading, and in a case study in the paper finds that the Florida Department of Transportation reduced nutrient loading by 67% when it moved from a fast-release fertilizer program (10-10-10) to a slow-release fertilizer program (sulfur coated 16-0-8).

Other researchers note that plants absorb more nutrients in the warm summer months when growth rates are highest (Sartain 2013 and Erickson et al. 2010). While incorporation of slow-release fertilizer has been shown to reduce nutrient loading, some researchers have shown that slow-release fertilizers may remain on lawns for several months (Erickson et al. 1999, Carey et al. 2012, and Saha et al. 2007). Summer fertilizer bans, intended to reduce nutrient loss during the rainy season, limit nutrients during the active growing season, consequently resulting in less-healthy turfgrass and greater nutrient loss through leaching and erosion (Hochmuth et al. 2009, Trenholm and Sartain 2010, and Carey et al. 2013). Furthermore, Hochmuth et al. (2009) suggests that a summer “blackout” period may encourage residents to fertilize prior to and after the ban when turf growth rates and uptake rates are lowest.

The timing of the nitrogen release from coated fertilizers often depends on the thickness and type of the coating. Shaviv (2001) indicates that nutrient release curves are often characterized by an initial burst followed by a tailing effect, a pattern that is very different from the nutrient uptake pattern in plants. The typical tailing effect suggests the potential for greater nutrient loss if releases continue in the dormant season when plant uptake is low. Broschat (2005) studies release patterns of two organic polymer-coated, controlled-release fertilizers (Osmocote and Nutricote). After seven months of watering container plants, about 10% of total nitrogen was retained in Osmocote fertilizer and about 25% of total nitrogen was retained in Nutricote fertilizer.

Existing Water-Quality Models
In recent years, several watershed models that simulate the sources and nutrient loading to the lagoon have been developed. The models have a wide range of complexity and produce annual, monthly, or hourly inputs. The existing watershed models that were used to develop these loading estimates have been developed by FDEP, SJRWMD, and Brevard County. Table 1 includes estimated nutrient loading information that has been developed and published using these models (Gao 2009, US EPA 2007, Applied Technology and Management Inc. and Janicki Environmental Inc. 2012, and Adkins et al. 2004). The assumptions and formulations to predict hydrology and pollutant loading vary in each model. In addition, some models account for loading sources such as groundwater while others do not. A brief summary of each model is provided in Table 2, and brief summary of their use is presented below.

• Pollutant Load Screening Model (PLSM) and Hydrologic Simulation Program–Fortran (HSPF) 2003: In 2003, draft nutrient and dissolved oxygen (DO) total maximum daily loads (TMDLs) for the Indian River Lagoon and Banana River Lagoon were proposed by US EPA. The TMDLs were finalized in 2007 and were based on modeled nutrient loading estimates calculated by models initially developed by the SJRWMD, including the PLSM and the HSPF (USEPA 2007).
• PLSM and HSPF 2009: In 2009, a nutrient and DO TMDL report was released by the FDEP that addresses impairments for all main stem segments, not just those listed on the 1998 303(d) list. TMDLs were based on PLSM and HSPF loading estimates and consist of load estimates for sources not incorporated in the 2007 report, including direct atmospheric deposition and wasteload allocations for new point-source dischargers (Gao 2009).
• HSPF 2012: In 2012, FDEP released TMDLs for nutrients and biochemical oxygen demand (BOD) for the tributaries draining to the Lagoon (Gao and Rhew 2012). The PLSM and HSPF models established by SJRWMD were not designed to simulate BOD loadings.
• SWIL 2012: The SWIL model, a geospatial model that improves on the relatively simple PLSM model, was developed by a team of consultants on behalf of Brevard County to calculate monthly total nitrogen and phosphorus surface water and baseflow loads to the Lagoon (England and Listopad 2013).

Nutrient loading information has been developed and published using these models, and estimates of total nutrient loads to the lagoon, as well as the percent contribution of each nutrient source, vary widely (Figure 1). Table 1 includes estimated loading from the literature.

Development and Results of the Fertilizer Loading Model
None of the existing models explicitly accounts for residential fertilizer application. Without this accounting, it is not possible to quantify the amount of nutrient loading resulting from residential fertilizer application or assess the impacts of fertilizer best management practices and blackout periods. We developed the FLM to quantify these loadings.

We used the 2009 FDEP HSPF model (Gao 2009) as the basis of the model. We used one of the model’s -meteorological files, which specifies evapotranspiration potential and rainfall. We used the same data sets or assumptions as the FDEP and SJRWMD models to simulate groundwater inputs.

Model Development
We configured the FLM as a 100-acre residential neighborhood and followed HSPF in simulating nutrient inputs and transformations differently for pervious and impervious land surfaces. We randomly selected two medium-density residential neighborhoods in the lagoon watershed and quantified the percentage of lawn, roof, driveway, sidewalk, and road using ArcGIS and satellite imagery. For this hypothetical neighborhood, the overall percent imperviousness was 34%, which is typical of a medium-density residential development comprising quarter-acre to third-acre lots (USDA 1986).

Few monitoring studies have been implemented that target the individual surfaces in a development (e.g., roof and lawn) instead of comingling all sources. We used a spreadsheet-based calculation tool developed by North Carolina State University and used by the North Carolina Division of Water Resources (NC DWR) to estimate nutrient loading from residential neighborhoods.

Impervious Surfaces. We included particulate nitrogen and phosphorus atmospheric deposition rates based on aerial dry deposition rates and rainfall concentrations reported for the lagoon (Applied Technology and Management Inc. and Janicki Environmental Inc. 2012). The buildup of dry particles on impervious surfaces from the atmosphere accumulates asymptotically until deposition and resuspension reach equilibrium; 90% of the steady-state accumulation occurs within 10 to 12 days (Walker 1990, Haith et al. 1992, and USEPA 2006). The nutrient runoff from each impervious surface type was calibrated using the NC DWR data set by increasing the buildup/washoff relationships (Table 3).

We assumed that runoff from the roof traveled across the lawn after leaving the gutter system and a portion of the pollutant load from the roof area is trapped in the lawn due to infiltration, plant uptake, etc. Based on removal rates presented by Yu et al. (1993) for grass filter strips, the model was configured to remove 84% of total suspended solids, 20% of nitrate plus nitrite, and 40% total phosphorus from rooftop runoff.

For the model scenarios where fertilizer was inadvertently applied and allowed to remain on impervious surfaces, the model assumes that broadcast spreaders applied fertilizer to the entire sidewalk, half of the driveway, and 10% of the road surfaces.

Pervious Surfaces. The FLM simulates two sources of nutrient loading to pervious surfaces: background nutrient loading and fertilizer-based nutrient loading. The background nutrient loading accounted for the conditions present prior to fertilizer application, including soil chemistry and net effects of atmospheric deposition and volatilization. The simulated background event mean concentrations (EMCs) were adjusted such that nutrient concentrations from lawns in the baseline scenario are reasonably close to those reported for lawns in Table 3.

Groundwater. Nutrient loads associated with groundwater inputs are set to constant annual values based on the medians of the values assumed in the FDEP and SJRWMD models: 1.1 mg/L for total nitrogen and 0.06 mg/L for total phosphorus.

Types of Fertilizer. The FLM simulates using slow-release and fast-release fertilizer, and a baseline combination fertilizer program based on IFAS recommendations (Sartain 2013). (Table 4 summarizes the amounts, types, and application schedule recommended by IFAS.)

Slow-Release Fertilizer. Slow-release fertilizers used in our calculations are covered by a coating that inhibits the release of nutrients into the environment. Release rates from slow-release fertilizers tend to increase with increased ambient temperatures. For example, when the temperature is 20°C, approximately 70% of the nutrients are released, and when the temperature is 27°C, approximately 75% of the nutrients are released (Adams et al. 2013). To simulate the controlled release of nutrients from these fertilizers, we spread out the mass of one fertilizer application evenly over a period of time that varies depending on air temperature.

Fast-Release Fertilizer. Nutrients in fast-release, water-soluble fertilizers are readily available for plant uptake upon application. The nitrogen in water-soluble fertilizers quickly dissolves and provides a quick “green-up” (Yingling 2002). To simulate the availability of the fast-release fertilizer, we calculated that the fertilizer mass was applied to the lawn in a single day and was therefore available immediately to the turfgrass.

Turfgrass Nutrient Uptake. The HSPF model does not directly simulate nutrient uptake and growth of turfgrass. Therefore, for the FLM, we use seasonally variable, first-order decay rates to simulate the transfer of nutrients from the fertilizer into the turfgrass. Plant uptake rates varied seasonally, with higher growth occurring during warmer months (Figure 2). We incorporate seasonal rates of turfgrass nutrient uptake based on laboratory experiments with Bermuda grass done by Wherley et al. (2009). For these analyses, the uptake by the turfgrass is a permanent loss, corresponding to removing grass clippings from the lawn after cutting.

Model Scenarios
The FLM was developed to quantify the impacts of various fertilizer management strategies. We did not alter nutrient inputs from other sources such as groundwater. The total fertilizer mass applied annually for all scenarios is 6.7 lb.-N/1,000 ft2/yr and 0.22 lb-P/1,000 ft2/yr.

Baseline Scenario. This is a likely fertilizer application scenario and provides the basis for comparing loading resulting from other fertilizer management strategies. The baseline scenario assumes:

• fertilizer application rates, frequency, and composition (both fast and slow release) are based on IFAS recommendations;
• fertilizer is not applied on rainy days;
• irrigation occurs every other day (0.25 inch) even on days with rain (this assumes that residents use automatic systems on timers); and
• fertilizer and irrigation water are applied to lawn surfaces only.

Additional Modeling Scenarios. Six modeling scenarios (Table 5) test the impacts of altered fertilization schedules, composition, weather considerations, and best management practices. An additional scenario addresses irrigation management.

Model Results
In our baseline scenario results we list loads from fertilizer, groundwater, and atmospheric deposition to demonstrate the relative loading from a residential development. For the other modeling scenarios, non-fertilizer loads (e.g., background, atmospheric deposition) are grouped into a category called “other loads” (and are not discussed in detail here). In the interest of space, results are discussed only for nitrogen because many areas of Florida do not require phosphorus application. However, phosphorus results are presented in the figures to show how those compare with nitrogen.

Baseline Results
Monthly Average Loads. Monthly average total nitrogen from all contributing sources under the baseline scenario is 122 pounds per month. “Other loads” comprise the majority of the loading from the development, and these loads track closely to average monthly flow (i.e., when flows increase, loads increase, Figure 3).

The baseline scenario assumes five total applications that use fast-, slow-, and a combination of fast- and slow-release fertilizer at IFAS rates (Table 4). Loading from fertilizer is lowest from March through November when turfgrass uptake is highest. Loading is highest in December, January, and February when the temperatures in central Florida are cooler and the grass does not consume as much nitrogen.

Validation of the Annual Baseline Loads. To validate the results, we compared relative loadings from each category to those published for the Indian River Lagoon watershed (Table 1). In the baseline model, fertilizer comprises 20% of the total nitrogen load relative to other sources of nutrient loading from a residential development (Figure 4). Slow-release components contribute more than the fast-release fertilizer. We added these fertilizer-based loads to those from pervious background to estimate the contribution from runoff, which is 28% based on the FLM. This estimate is within the reported ranges for the lagoon watershed (Gao 2009, Applied Technology and Management Inc. and Janicki Environmental Inc. 2012, Sigua and Tweedale 2003).

We estimate that groundwater contributes approximately 60% of the total nitrogen load, which is in close agreement with the percentages reported by Applied Technology and Management Inc. and Janicki Environmental Inc. (2012). The third-largest contributor of nitrogen in our estimation, after groundwater and fertilizer, is atmospheric deposition. This is also within the range predicted in other studies.

Comparison of Model Scenarios–Total Nitrogen. Figure 5 shows the change in total nitrogen annual average loads for each modeling scenario relative to baseline scenario (1,468 pounds per year of total nitrogen). Three modeling scenarios result in reduced total nitrogen loads relative to the baseline scenario. The Apply Fast Release 4/yr and the No Irrigation on Rain Days scenarios each decreased the loads by approximately 200 pounds per year. The Apply Fast Release 4/yr is a hypothetical scenario developed to demonstrate how fertilizer composition affects nutrient loading. Because the annual mass is not altered, this scenario would likely damage the turf. In practice, homeowners who apply only fast-release fertilizer would likely apply less or apply smaller applications more frequently than four times per year, and the loads would actually be lower than those shown. The No Irrigation on Rain Days scenario has the greatest reduction in total nitrogen loads with an overall reduction of 16% when averaged over the modeling years. The Apply on Rain Days scenario does not have a significant impact on simulated annual nitrogen loads when averaged over the modeling period. Under the baseline scenario, fertilizer was not applied on days with rain, but assessment of rain on subsequent days was not considered. Further analysis would be needed to quantify the effect of not applying fertilizer when rain falls with 48 hours, 96 hours, etc.

The remaining modeling scenarios result in increased nitrogen loads relative to baseline. Both scenarios where only slow-release fertilizer is applied result in increased loading, but Apply Slow Release 4/yr causes more loading (over 310 pounds per year) than Apply Slow Release 2/yr (about 25 pounds per year). This is likely because the slow-release fertilizer applied four times per year results in nitrogen application and release during the cooler months when turf uptake is lower.

The worst-case scenarios for nitrogen loading are Apply Baseline to Impervious Surfaces and Apply Fast Release to Impervious Surfaces. These scenarios export an additional 1,140 pounds per year and 1,350 pounds per year, respectively, nearly doubling the nitrogen load.

Summary of Key Findings
There is a significant amount of research focused on the fate of fertilizer under various application and environmental conditions, and much of the literature provides best practice recommendations for residential fertilization. However, the lack of watershed models that account specifically for nutrient loading from urban fertilizer has meant that it has been difficult to understand the relative contribution of residential fertilizer and to hone in on the most effective and efficient ways to reduce loading from fertilizer. In some cases the results of the FLM are consistent with the existing body of literature, and in other cases there are discrepancies that may warrant further analysis.

For example, the FLM and much of the literature are in agreement on the following aspects of fertilizer application:

• Summer bans on fertilizer application may cause increased nutrient loading. Plant uptake rates are very high during the summer months, and when fertilizer is applied at recommended rates with best management practices, nutrient export during the summer months is relatively low.
• High levels of slow-release fertilizers should not be used before periods of dormancy.
• Fertilizer inadvertently applied to impervious surfaces can increase nutrient loading.
• When IFAS recommendations are followed, nitrogen loads from fertilizer comprise 20% of the loading from runoff in residential areas. Relative to the sources of nitrogen other than residential (e.g., agriculture) (Figure 1), lawn fertilizer contributes approximately 6% of the total load to the lagoon.
• Excessive irrigation, particularly on days with rain, increases nutrient loading. Irrigation management may play a greater role in nutrient transport than weather-based application restrictions because of the continuous nature of irrigation systems.

When we examined our model against other recommendations, we found some areas of conflict:

• In areas where grass goes dormant, applying fertilizer in February and October, as recommended by IFAS, may result in nutrient loading from lawns during the cooler months.
• The optional summer bans based on the Model Ordinance for Florida-Friendly Fertilizer Use on Urban Landscapes may result in increased nutrient loading to the lagoon. Under the bans, the nutritional requirements for healthy turf may not be supplied when needed, poor-quality lawns may not be retaining nutrients and soil as efficiently as high-quality lawns, and homeowners may tend to apply fertilizer when the nutrient uptake rates of the lawn are lower.

Additional study is needed to understand the optimal proportions, rates, and application times for fast- and slow-release fertilizer that will reduce nutrient loading and support a healthy lawn system.

The results from the FLM show that detailed predictive models can provide a quantitative estimate of nutrient loadings from residential areas. Model results indicate residential fertilizers may not be as significant a contributor to nutrients in the lagoon as previously suspected. Additional local, detailed sampling of water quality from residential sources (e.g., roofs, driveways, sidewalks, roads, and lawns) would provide a better calibration of the model and confirmation of the model’s results.

Acknowledgements
The authors would like to thank Xueqing Gao with the Florida Department of Environmental Protection for graciously providing the most recent HSPF model and input files. Funding for this research has been provided by ScottsMiracle-Gro Company. However, the conclusions and recommendations herein are the authors’ only and do not necessarily represent the opinions of ScottsMiracle-Gro or Cardno ENTRIX.

References
Adams, C., J. Frantz, et al. 2013. “Macro and Micronutrient Release Characteristics of Three Polymer Coated Fertilizers: Theory and Measurements.” Journal of Plant Nutrition and Soil Science 176(1): 76–88.

Adkins, M., M. Mao, et al. 2004. “Watershed Model Development for the Indian River Lagoon Basin: Providing Simulated Runoff and Pollution Load to the Indian River Lagoon Pollution Load Reduction Model.” Report to St. Johns River Water Management District, Tech. Memo. 50.

Applied Technology and Management Inc. and Janicki Environmental Inc. 2012. “Nutrient Loading Estimates: Atmospheric Deposition and Point Sources Task 1.B: Indian River Lagoon Total Maximum Daily Load Revision.” Report to Brevard County, FL.

Broschat, T. K. 2005. “Rates of Ammonium-Nitrogen, Nitrate-Nitrogen, Phosphorus, and Potassium From Two Controlled-Release Fertilizers Under Different Substrate Environments.” HortTechnology 15(2): 332–35.

Carey, R. O., G. J. Hochmuth, et al. 2012. “A Review of Turfgrass Fertilizer Management Practices: Implications for Urban Water Quality.” HortTechnology 22(3): 280–91.

Carey, R. O., G. J. Hochmuth, et al. 2013. “Evaluating Nutrient Impacts in Urban Watersheds: Challenges and Research Opportunities.” Environmental Pollution 173:138–49.

Chopra, M., M. Wanielista, et al. 2011. “Evaluation of Pollution Levels Due to the Use of Consumer Fertilizers Under Florida Conditions.” Report to Florida Department of Transportation.

England, G. and C. Listopad. 2013. “Refining the Indian River Lagoon TMDL Nutrient Loading Model.” Stormwater, July/August 2013.

Erickson, J. E., D. M. Park, et al. 2010. “Effects of Sod Type, Irrigation, and Fertilization on Nitrate-Nitrogen and Orthophosphate-Phosphorus Leaching From Newly Established St. Augustinegrass Sod.” Crop Science 50(3): 1030–36.

Gao, X. 2009. “Nutrient and Dissolved Oxygen TMDLs for the Indian River Lagoon and Banana River Lagoon.” Report to the Florida Department of Environmental Protection, Tallahassee, FL.

Gao, X., and K. Rhew. 2012. “Dissolved Oxygen and Nutrient TMDLs for Eleven Tributary Segments of the Indian River Lagoon.” Report to the Florida Department of Environmental Protection, Tallahassee, FL.

Haith, D. A., R. Mandel, and R. S. Wu. 1992. GWLF Generalized Watershed Loading Functions. Version 2.0. User’s Manual.

Hochmuth, G., T. Nell, et al. 2009. “Unintended Consequences Associated with Certain Urban Fertilizer Ordinances.” University of Florida Cooperative Extension Service SL283.

North Carolina State University. “Jordan/Falls Lake Stormwater Nutrient Load Accounting Tool” Version 2.0, http://portal.ncdenr.org/web/jordanlake/implementation-guidance-archive. Accessed June 2, 2014.

Saha, S. K., L. E. Trenholm, et al. 2007. “Effect of Fertilizer Source on Nitrate Leaching and St. Augustinegrass Turfgrass Quality.” HortScience 42(6): 1478–81.

Sartain, J. B. 2013. General Recommendations for Fertilization of Turfgrasses on Florida Soils. University of Florida Cooperative Extension Service, Institute of Food and Agriculture Sciences, EDIS (originally published 1991).

Shaviv, A. 2001. “Advances in Controlled-Release Fertilizers.” Advances in Agronomy 71: 1–49.

Sigua, G. C., and W. A. Tweedale. 2003. “Watershed Scale Assessment of Nitrogen and Phosphorus Loadings in the Indian River Lagoon Basin, Florida.” Journal of Environmental Management 67(4): 363–72.

Trenholm, L. E., and J. B. Sartain. 2010. “Turf Nutrient Leaching and Best Management Practices in Florida.” HortTechnology 20(1): 107–10.

US Department of Agriculture (USDA). 1986. Urban Hydrology for Small Watersheds Technical Release 55. June 1986.

US Environmental Protection Agency. 2006. NPS Quality (PQUAL, IQUAL) Processes, Parameters and Calibration. Lecture 11. www.epa.gov/waterscience/ftp/basins/training/b4lec11.pdf. Accessed June 2, 2014.

US Environmental Protection Agency. 2007. Total Maximum Daily Loads for the Northern and Central Indian River Lagoon and Banana River Lagoon, Florida: Nutrients and Dissolved Oxygen.

Walker, W.W. 1990. P8 Urban Catchment Model—Program Documentation. DOS Version 1.1, prepared for IEP Inc. and Narragansett Bay Project, 73 pp., 5 App, October 1990.

Wherley, B. G., W. Shi, et al. 2009. “Fate of N-Nitrate Applied to a Bermudagrass System: Assimilation Profiles in Different Seasons.” Crop Science 49(6): 2291–2301.

Yingling, J. 2002. “What You Need to Know About Fertilizing and Watering Your Lawn and Landscape to Protect Florida’s Springs.” Nutrient Remediation Working Group.

Yu, S., S. Barnes, and V. Gerde. 1993. “Testing of Best Management Practices for Controlling Highway Runoff.” FHWA/VA 93-R16. Virginia Transportation Research Council, Charlottesville, VA.