These comments were submitted by the Florida Stormwater Association in response to an article in our November/December 2014 issue, “Nitrogen Loading From Residential Turfgrass.” –Editor
Background
In Florida, city and county governments are under increasing pressure to develop and implement cost-effective tools to address state and federal water-quality goals for nitrogen and phosphorus in waters subject to regulation (or Waters of the United States) as expressed in total maximum daily loads (TMDL) for nutrients, and also numeric nutrient criteria. One mechanism available to local governments to help lower the amount of nutrients being discharged into receiving waters is the adoption of ordinances regulating the use of residential fertilizer. Ordinances that contain policies prohibiting the application of residential fertilizer during the summer rainy season have been adopted in many urban areas. Such “blackout” periods have been the subject of much debate—both at the local level and in the Florida legislature for the past 10 years.
The authors’ stated purpose was to better understand the relationship between residential fertilizer application in Florida’s Indian River Lagoon (IRL) watershed and nutrient loading in the lagoon, generally concluding that local “blackout” policies restricting the use of residential fertilizer during the rainy season may be unnecessary or counterproductive. However, incomplete or misrepresented citations as used in the article cause the validity of the conclusions to be suspect.
What follows are a series of concerns with the article’s research methodology and conclusions.
1. Cited Literature
The authors developed a bibliography of 37 reports and articles; however, there are numerous concerns with how the referenced information was used. For example, a publication by Carey et al. (2012) was referenced. The article says that Carey et al. references a study, but the authors never cite the actual study. A review of Carey et al. 2012 shows the correct reference is L. M. Shuman (2002), “Phosphorus and Nitrate Nitrogen in Runoff Following Fertilizer Application to Turfgrass,” Journal of Environmental Quality 31:1710–1715. The Shuman study was designed to look at the transport of nitrate and phosphate from simulated golf course fairways of “Tifway” Bermuda grass. Shuman found that nitrate concentrations in runoff were 0.5 milligrams per liter (mg/L) after the first three runoff events, and 1–1.5 mg/L 168 hours after treatment. Mass nitrate losses were highest at four and 24 hours after treatment. The authors state that greater phosphorus losses were shown to occur from Bermuda grass four hours post-application with a decrease at 24 hours post-application and thereafter, indicating that greater nutrient runoff can be expected when fertilizer applications occur closer to rain events. While the phosphorus data were one finding, the original research also showed a significant increase in nitrate concentrations seven days after the application. The conclusion of the Shuman paper states that the “results indicate that turfgrass management should include applying minimum amounts of irrigation after fertilizer application and avoiding application before intense rain or when soil is very moist.” Only citing one portion of the research is misleading.
The authors also cite Martin et al. (no date) in Figure 1 stating that lagoon sediments may account for 15 times the amount of nitrogen loading to the IRL compared to all of the other sources combined. The publication is not listed in the reference section. The author, Dr. Jonathan Martin, was contacted about the statement. He stated that they could be referencing research from 15 years ago, but that he could not find anything, and he did not believe the statement is correct.
Additional concerns about how referenced data and assumption have been used in this article will be pointed out.
2. Existing Water-Quality Models
The Pollutant Load Screening Model (PLSM) and the Hydrologic Simulation Program-FORTRAN (HSPF) 2003 and PLSM and HSPF 2009 are not considered to be accurate or representative models and are in the process of redevelopment.
Figure 1 lists the nitrogen sources as being from urban/agricultural runoff, while actually those numbers represent total surface and baseflow loads to the lagoon, which is not just from urban or agricultural land uses, but from all lands surrounding the IRL.
In Table 2, the authors state that runoff in the Spatial Watershed Iterative Loading (SWIL) Model is “time variable, land-use specific.” This is not accurate. SWIL runoff calculations are all monthly values.
3. Development and Results of the Fertilizer Loading Model
The authors do not specify what period of time was actually simulated, what meteorological files were actually used, or which data sets and assumptions were used. This information is critical.
The HSPF (Gao 2009) model is not designed to evaluate detailed surface hydrology. HSPF is not appropriate for this type of detailed hydrologic analysis. The model does not account for interaction between pervious and impervious areas, such as the roof draining to the lawn. This weakness in the model allows for assumptions that can greatly affect runoff and loading rates. For example, nutrient transport caused by roof runoff draining on to the lawns, picking up fertilizer, then running onto the road into the storm system is not accounted for. Furthermore, generic removal rates of 84% for total suspended solids (TSS), 20% total nitrogen (TN), and 40% total phosphorus (TP) were utilized. These removal rates are generous, and would likely not be acceptable for site permitting purposes.
The atmospheric deposition data were incorrectly applied to land-runoff when the data used were for atmospheric deposition over Tampa Bay proper, not the watershed. Further, the data used were for Tampa Bay not for the IRL and its contributing watershed.
The analysis does not discuss hydrologic soil properties; sandy soils will result in less surface runoff but greater groundwater nutrient loadings, as has been documented by many of the springs-related Basin Management Action Plans. In this analysis, groundwater loads are not considered to include any nutrients from fertilizer.
If analysis used sandy soils, the surface nutrient loads will be lower and the nutrients routed into the ground are excluded from the fertilizer-based portion of total loading. Infiltration does not result in complete nutrient removal as this analysis assumes.
The loads developed for the all of the Florida Department of Environmental Protection (FDEP) and St. Johns River Water Management District (SJRWMD) models represent total loads, which include the loads from fertilized lawns (based on event mean concentration [EMC] values for residential neighborhoods). Therefore, defining them as loads that are separate from lawns is not appropriate and the percent comparisons provided are misleading.
4. Model Development
Outdated impervious data were used from 1986, when data for current conditions are available. Further, there is no documentation of what medium-density residential areas were used to calculate this number (year developed, single lot or entire development).
The event mean concentration values for TN used in Table 3 (area weighted average of 1.9 mg/l) look comparable to Harvey Harper’s (Harper and Baker 2007) compilation of data for single-family residential land use in Florida (TN: 2.07 mg/l mean, 1.85 mg/l median, 1.87 mg/l log-normal mean).
Unless Harper’s compilation was limited to only areas that did not fertilize, additional (uncounted) fertilizer loading is inflating the background runoff loading values.
5. Impervious Surfaces
These are very generous removal rates for roof runoff crossing a yard. Furthermore, the roof runoff crossing the yard will actually pick up nutrients and transport downstream. It does not appear this additional load was accounted for. The reference used for the removal rates (Yu et al. 1993) was an evaluation of the treatment of roadway runoff by two treatment systems—a detention pond and a constructed swale. Yu et al. did not evaluate roof runoff across lawns nor are these removal efficiencies found in the Yu et al. results.
6. Groundwater
In many cases throughout the article, the authors confuse nutrient loadings, which largely depend on the hydrologic loadings, and nutrient concentrations.
For groundwater, the authors assumed a background loading rate of 1.1 mg/l TN, which is reported as “the median of the value assumed in the FDEP and SJRWMD models.” PLSM did not have such a value, so perhaps this was from HSPF? However, Harper’s separation of runoff and baseflow in IRL tributaries using Purdue University’s WHAT model produced a significantly different geometric mean of 0.886 mg/l TN (data table copied below), similar to runoff, since these tributary basins included areas that were fertilized, additional (uncounted by Matos et al.) fertilizer loading is inflating these background baseflow loading values.
7. Types of Fertilizer
The authors are not explicit in the units—are they the same for quick-release fertilizer?
8. Turfgrass Nutrient Uptake
The assumptions for seasonal nutrient uptake will significantly drive the results of any modeling. The authors did not provide a reference for Figure 2. The uptake rates used were from Wherley et al., which is a study on Tifway Bermuda grass receiving reuse water for irrigation. The purpose of the study was to investigate nitrate (NO3-) uptake efficiency during growth and dormancy cycles, and thus the potential to filter reuse effluent at different times of the year. Further, the authors’ analyses included the removal of grass clippings, which is not typical behavior either by residents that mow their own lawns or by professional maintenance companies, nor is the use of Tifway Bermuda grass representative of residential communities.
This was stated in the work by Wherley et al., these results were “very different” from the results of “other grass systems” and assumed uptake by turfgrass was a permanent loss (total removal of grass clippings after lawn cutting). This species of Bermuda grass is more commonly used on golf courses and athletic fields.
Also curious, although the Matos et al. Figure 3 graphs show peaks in February and October that correlate to application of fast-release fertilizer, there is no peak associated with the April application, making these results questionable.
9. Baseline Scenario
- The authors assumed application rates, frequencies, and composition based on the University of Florida Institute of Food and Agricultural Sciences (IFAS) recommendations. This is a significant assumption that is not validated.
- Fertilizer is not applied on rainy days. What about within 24 to 48 hours? This is a significant assumption given the social surveys that indicate a good number of residents polled do not know that fertilizer should not be applied when it is raining or rain is predicted.
- Irrigation occurs every other day (0.25 in.), even on days with rain (assuming residents use automatic systems on timers). There are data available from the water management districts, local utilities, and extension on irrigation practices.
- Fertilizer and irrigation water are applied to lawn surfaces only. This is a significant assumption given that overspray from irrigation, more frequently than not, reaches hardened surfaces. It is not unusual to see fertilizer applied to driveways, walkways, etc. Reuse water and its associated nutrient inputs, soil types, and slopes do not appear to be accounted for.
10. Modeling Results
There was no model calibration. This is never acceptable, particularly when so many assumptions are made.
11. Baseline Results
Given the significant differences in the manner in which the loads are calculated, this comparison is not valid. For example, ATM/Janicki Environmental nutrient loads were for the entire IRL estuary and a comparison of these estimates to those from a poorly defined 100-acre hypothetical watershed that apparently doesn’t percolate is wrong. Further, the referenced study by IFAS was very narrow in scope and did not include tests of representative summer conditions.
12. Summary of Key Findings and Response
- Statement: 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.
Response: If nutrient export during summer months occurs, even at a “relatively low rate,” how does a summer ban “increase” loading? If any nutrient export occurs, removing a nutrient source would reduce loading.
- Statement: High levels of slow-release fertilizers should not be used before periods of dormancy.
Response: The baseline condition didn’t support this statement. - Statement: 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.
Response: This estimate is significantly biased as a result of the assumptions made in the model of which there are two major ones: 1) bagging clippings; and 2) abiding by the IFAS recommendations. What percentage of fertilizer applications actually conform with IFAS recommendations? Additionally, this comparison is not accurate as the loadings the authors are comparing to also include fertilizer loads, either in the shallow groundwater or as part of the runoff from residential areas (included in residential EMCs). - Statement: 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.
Response: The baseline did not agree with this statement. - Statement: 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.
Response: The authors failed to look at Florida’s regional and subtropical conditions that affect grass dormancy: Average temperatures North Florida (Tallahassee): February 67°F, October 81°F; Central Florida (Orlando): February 73°F, October 84°F; and South Florida (Miami): February 78°F, October 86°F. In the IRL region of Florida, grass grows year-round. - Statement: The results from the FLM [the authors’ Fertilizer Loading Model] show that detailed predictive models can provide a quantitative estimate of nutrient loadings from residential areas.
Response: FLM is greatly dependent upon the assumptions used in this study. The assumptions are poorly documented and often inappropriately applied. - Statement: Model results indicate residential fertilizers may not be as significant a contributor to nutrients in the lagoon as previously suspected.
Response: Though not quantified, the net effect of the authors overestimating total IRL loading from background runoff and groundwater, together with underestimating IRL loading from fertilizer, work together to erode the concluding statement that “residential fertilizers may not be as significant a contributor to nutrients in the lagoon as previously suspected.” Even so, the authors conclude that “nitrogen loads from fertilizer comprise 20% of the loading.” Figure 4 shows this fertilizer contribution in the context of atmospheric (12%) and groundwater (60%) loading. So of the remaining 28% (100% – 12% – 60% = 28%) of IRL loading that the authors attribute to runoff when IFAS fertilizer recommendations are followed, most of it, 20 of the 28% (6% fast-acting plus 14% slow release), is attributable to fertilizer. For municipal separate storm sewer systems, which are mandated to reduce total TN loading to the IRL by 21 to 56% (depending on location), which have zero authority or means to reduce atmospheric contributions, and limited authority to reduce groundwater load contributions, the excessive contribution of fertilizer to runoff is exceptionally significant and merits the utmost attention and action.
References
Harper, H. H., and D. M. Baker. 2007. Evaluation of Current Stormwater Design Criteria within the State of Florida. Final Report submitted to the Florida Department of Environmental Protection for Agreement SO108. Environmental Research and Design Inc. Orlando, FL.
Florida Stormwater Association
719 East Park Avenue
Tallahassee, FL 32301
www.florida-stormwater.org
888-221-3124
