Study site and climatic data
This study was conducted at a research facility at the South Coast Research and Extension Center in Irvine, Orange County, California (33°41′20.16″N, 117°43′24.26″W, 123 m above mean sea level, Mediterranean climate). The simulated residences consisted of model houses complete with driveways, curb and gutter, and both front and backyards designed and managed to simulate typical versus recommended practices for southern California (Photo 1). The experiment was designed and operated by the University of California Cooperative Extension of Orange County, and reflects their local expertise in assessing both common and best management practices for local landscapes (http://ceorange.ucdavis.edu/). The goals of the study were to facilitate complete instrumentation of water and N budgets and surface runoff from both pervious and impervious surfaces of Orange County residential parcels, using methods that are usually too intrusive for studies of actual residences.
Photo 1Photos of the three simulated residences in 2010 (a Typical, b Low Input, and c Low Impact). Lawns are planted in both front and back yards
This study focused on the lawn portions of the landscapes, though each contained other vegetation types and either pervious or impervious surfaces surrounding the lawns, as the study incorporated complete, parcel-scale landscape designs. Hence, the sizes of the lawns varied by landscape. The lawns in the Typical landscape (129 m2) and the Low Input landscape (43 m2) were both established in September 2006 by placing pallets of sods on a layer (5–7.5 cm) of biosolids (compost, the Biosolid Program of Orange County Sanitation District, or OCSD) on top of local soil. The typical lawn consisted of tall fescue (Schedonorus phoenix, a cool-season, C3 species formerly known as Festuca arundinacea) and was established from Marathon II sod (Southland Sod Farms, Oxnard, CA, USA) with a 6.3–7.5 cm root/soil base (sod soil N%: 0.06 ± 0.002 %),. The Low Input lawn consisted of seashore Paspalum (Paspalum vaginatum, a C4 species) established from Sea Spray sod (West Coast Turf, Winchester, CA, USA) with a 5 cm root/soil base (sod soil N%:0.11 ± 0.01 %). The Low Impact lawn (54 m2) was established in January 2007 with plugs of the cool-season, C3, California native sedge Carex (Euro American Propagators, Bonsall, CA, USA) grown in a layer of 1.3–2.5 cm OCSD biosolids on local soil. The total N% of OSCD biosolids was 1.4 % (Soil Control lab, 2007), while the local soil originally had an organic matter content of 0.42 ± 0.17 %, and a texture of 79.2 ± 2.9 % sand, 9.4 ± 1.3 % silt, and 11.4 ± 1.7 % clay (UC Davis Analytical Laboratory, 2005).
The lawn in the Typical landscape was irrigated on a regular schedule (3 days per week, 12 min per time) with an automated timer (Rain Bird 4 Station ESP Modular Series Controller, Rain Bird, Azusa, CA, USA, Hartin and Harivandi 2001) and fertilized at 192 kg N ha−1 year−1 in four equal applications, as is recommended for this lawn type (Henry et al. 2002). The fertilizer was Scotts ® Turf Builder® 29-2-4 (Scotts Miracle-Gro, Marysville, OH, USA) in 2010 and Schultz® super 21-0-0 (United Industries Cooperation, St. Louis, MO, USA) in 2011, both of which are mainly urea/coated urea. During our study, there were 6 fertilization events on March 29, May 24, August 16, and September 21 in 2010 and March 8 and May 16 in 2011. The Low Input lawn was irrigated by an automated system utilizing soil moisture sensors (Watermark Electronic Module, Irrometer, Riverside, CA, USA) which triggers at low soil moisture levels (~25 kPa at 25 cm depth) and fertilized with the same fertilizer and schedule as Typical, but at only half the rate, consistent with recommendations (25 kg N ha−1 application−1, Henry et al. 2002). However, there was a one-time application of doubled rate at 48 kg N ha−1 on August 16, 2010 to maintain visual quality, resulting in a total rate of 123 kg N ha−1 year−1 in 2010, and the first fertilization in 2011 was on March 29 instead of March 8. The lawn in the Low Impact landscape was irrigated by a weather-based Hunter ET System connected to a Hunter ICC irrigation controller (Hunter, San Marcos, CA, USA) that adjusts water application according to crop evapotranspiration (ET) rates calculated with onsite weather observations in a modified version of the Penman equation (Pruitt and Doorenbos 1977; Snyder and Pruitt 1985) and crop-specific coefficients by the California Irrigation Management Information System (CIMIS, www.cimis.water.ca.gov). The Low Impact Lawn was not fertilized before 2010, but was fertilized on March 29, 2010 and March 8, 2011 at 48 kg N ha−1 year−1 with Vigoro® Ornamental 8-4-8 Plus Minors (Vigoro Corp., Chicago, MI, USA). Following each fertilizer application, the fertilized lawns were irrigated for about 2 min. The Typical lawn and Low Input lawn were mowed regularly during the growing season once a week or every other week to a height of 7.6 and 2.5 cm, respectively, while the Low Impact lawn was only mowed once a year to the height of 2.5 cm. The clippings from all lawns were directly captured by mower and removed from the lawns.
Three time-domain reflectometers (CS616, Campbell Scientific, Inc., Logan, UT, USA) were installed in each lawn in 2008 to measure soil moisture. The installation depth was 15 cm for Typical, 20 cm for Low Input, and 25 cm for Low Impact, to correspond to rooting depth. The sensors were logged (CR10X, Campbell Scientific, Inc., Logan, UT, USA) every 30 s and averaged every 30 min. Climatic data, such as air temperature, soil temperature, and precipitation, were obtained from a CIMIS network station which was located at the site (Irvine station, #75, http://www.cimis.water.ca.gov).
N inputs from atmospheric deposition and irrigation water
The total N inputs (Ntotal ± SEt) were calculated as
$$ N_{total} = N_{fert} + N_{atm} + N_{irr} $$
(1a)
$$ SE_{t} = \sqrt {SE_{atm}^{2} + SE_{irr}^{2} } \, $$
(1b)
where Nfert, Natm, and Nirr are N inputs from fertilizer application, atmospheric N deposition, and irrigation water, respectively. Natm and Nirr were estimated as
$$ N_{ atm} = N_{dry} + N_{wet} $$
(2)
$$ N_{irr} = IN_{irr} * V_{irr} $$
(3)
where Ndry and Nwet are atmospheric dry and wet deposition, and INirr and Virr are the average IN concentrations of irrigation water and irrigation volume, respectively.
Nwet during the rainy winter seasons was measured with ion exchange resin (IER) collectors based on the design by Fenn and Poth (2004), which can last 3–12 months as needed. Five IER collectors were installed in the open area around the study site on December 12, 2010 and collected on July 21, 2011. Briefly, an IER column was attached to the bottom of a dark-colored funnel with 20-cm diameter and 10-cm vertical side walls, and was mounted at 2 m above the ground. The IER was Amberlite IRN 150 Mixed Bed IER (Amberlite™ IRN 77 + Amberlite™ IRN78), with polyester fibers in both ends to prevent particles from entering. After removal, the IER columns were sent to the Forest Fire Laboratory, US Forest Service Pacific Southwest Station (Riverside, CA. USA), where the IER was extracted with 2 M potassium chloride (KCl) and analyzed for inorganic N (NH4
+–N and NO3
−–N) concentrations colorimetrically with an TRAACS 800 Autoanalyzer (Tarrytown, NY, USA, Fenn et al. 2006). Although we used a coarse mesh screen to prevent material from entering the funnel, litter contamination was still severe, leaving only one usable sample for NH4
+–N; the NO3
−–N data were not affected. Ndry, which is usually the largest component of N deposition in California and could be 10 times higher than Nwet (Bytnerowicz and Fenn 1996; Fenn et al. 2003), was not directly measured in our study. According to Fenn et al. (2010), the atmospheric dry deposition in central Orange County ranged from 9 to 15 kg N ha−1 year−1, so we applied a value of 12 kg N ha−1 year−1 for our N budgets. Our method may tend to slightly over-estimate the atmospheric N deposition as the IER results might overlap with part of the dry deposition.
Virr was monitored on a weekly basis using five randomly placed cups, with 5 ml mineral oil added to each to prevent evaporation from June 23 2010 to May 29 2011. Irrigation water was sampled directly from the irrigation valve onsite in January, March, June, October, and December 2010 and stored at 4 °C until analyzed colorimetrically for NH4
+–N and NO3
−–N concentrations using the phenol-hypochlorite method by Weatherburn (1967) and the vanadium method by Doane and Horwath (2003). INirr was calculated as the average of the IN concentrations of the five samples.
N output in plant clippings
The biomass N in plant clippings at 30-day (Nclip_30 ± SE
i
) or annual (Nclip_365 ± SEp) scales was calculated as
$$ N_{clip\_30} = biomass * biomass \;\% N $$
(4a)
$$ N_{clip\_365} = \sum\limits_{i = 1}^{n} {(biomass_{i} * biomass\;\% N_{i} )} $$
(4b)
$$ SEp \, = \sqrt {\sum\limits_{i = 1}^{n} {SE_{i}^{2} } } \, $$
(4c)
where n is the number of mowing events in 2010 for each lawn. Equations are only shown for errors (SE) that were propagated from other errors, i.e., SEp, but not for those directly calculated from samples, i.e., SE
i
.
There were 36 and 32 mowing events for Typical and Low Input, respectively, while Low Impact was only mowed once a year. All plant clippings were collected, oven dried at 70 °C for 1 week (extended drying time due to the large quantities of clippings), and weighed. Clippings were sub-sampled at intervals throughout for analysis of N% from Typical (n = 17) and Low Input (n = 13). Three random portions of plant clippings were weighed (~10 g) from each subsample, homogenized, and ground to fine power for analysis on an elemental analyzer (EA, Fisons NA1500 NC, San Carlos, CA, USA) coupled to an Isotope Ratio Mass Spectrometer (Delta Plus IRMS, Thermofinnigan, San Jose, CA, USA). N output in plant clippings within 30 days following each fertilization event in the Typical and Low Input lawns was calculated as the product of the clipping biomass and N% during that time period. At the annual scale, clipping biomass N for each of the three lawns was estimated as the sum of biomass N from all mowing events. We did not measure the aboveground biomass that remained after mowing (verdure), since it was always left at similar height and not likely to vary very much in N stocks (Engelsjord et al. 2004; Frank et al. 2006).
N outputs in gaseous and aqueous forms
Gaseous N fluxes at 30 day (Ngas_30 ± SE
g30) and annual (Ngas_365 ± SE
g365) scales were estimated as
$$ N_{gas\_30} = \sum\limits_{i = 1}^{k} {a_{i} } + \sum\limits_{j = 1}^{30 - k} b $$
(5a)
$$ SE_{g30} = \sqrt {\sum\limits_{i = 1}^{k} {SE_{i}^{2} } + \left( {\left( {30 - k} \right) * SE_{b} } \right)^{2} } $$
(5b)
$$ N_{gas\_365} \, = \, \sum\limits_{i = 1}^{m} {N_{gas\_30i} } + \sum\limits_{j = 1}^{30 * n} b $$
(5c)
$$ SE_{g365} = \sqrt {\sum\limits_{i = 1}^{m} {SE_{g30i}^{2} } + (30 * n * SE_{b} )^{2} } $$
(5d)
where a
i
± SE
i
represents the average daily fluxes of gaseous N following fertilization and b ± SE
b
is the background flux estimated based on pre-fertilization sampling; k, m, and n represent the number of sampling days following fertilization, the number of months with fertilization, and the number of months without fertilization (m = 4, n = 8 for Typical and Low Input, m = 1, n = 11 for Low Impact), respectively.
We measured NH3 volatilization following fertilization with four open containers, each with 40 ml 2 % (v:v) sulfuric acid (H2SO4), randomly placed on each lawn and tightly covered with PVC chambers (diameter = 25 cm) for 24 h, after which the solution was collected and stored at 4 °C for IN analysis. Daily NH3 volatilization from the lawn was calculated as the increase in NH4
+–N in the solution per unit chamber area per day (Schlesinger and Peterjohn 1991). This procedure was repeated in each lawn following every fertilization event for up to 7 days, except that Low Impact was only measured following its single fertilization event in 2011. Pre-fertilization sampling was conducted 1 day before each of the 6 fertilization events in every lawn. Background NH3 fluxes from each lawn were estimated as the average of all 6 pre-fertilization sampling events.
N2O fluxes were measured using static PVC chambers (n = 4, diameter = 25 cm, Townsend-Small and Czimczik 2010; Townsend-Small et al. 2011) randomly placed on each lawn the day before fertilization and on a daily basis following each of the 6 fertilization events (early morning) for 5-7 days. Air samples were removed from the chamber using 30 ml nylon syringes at 0, 7, 14, 21, and 28 min, injected into air-tight, pre-evacuated headspace vials, and analyzed within 24 h on a gas chromatograph (GC) fitted with an electron capture detector (GC-2014 Nitrous Oxide Analyzer, Shimadzu Scientific Instruments). N2O fluxes were calculated from the slope of the fitted regression line of N2O concentrations versus time in each chamber. Fluxes were considered undetectable, i.e., zero, if R2 < 0.9. Volumetric soil water content (0–5 cm) was measured beneath each chamber immediately following sampling using a portable soil moisture meter (TH2O, Dynamax, Inc, Houston, TX, USA). NO fluxes were not directly measured; however, we estimated a range based on the two extremes of the reported NO/N2O ratios from field measurements or model simulation of lawns or fields fertilized with urea (0.005–14.7, Gu et al. 2009; Hall et al. 2008; Parton et al. 2001; Stehfest and Bouwman 2006; Williams et al. 1998). As a comparison, NO fluxes were also estimated as 1–3 % of nitrification (Baumgartner and Conrad 1992; Hutchinson and Davidson 1993) based on the net nitrification rates of each lawn.
To measure denitrification rates, four soil cores (0–10 cm), one from under each of the four PVC chambers, were extracted following the removal of chambers after each N2O sampling event and transferred to the lab immediately for denitrification analysis using the acetylene-blocking technique (Drury et al. 2008). An additional soil core was taken from the lawns in March and May 2011 in order to increase the sample size (n = 5 in total). To minimize damage to the lawns, cores were only 1.5 cm in diameter. Acetylene gas was used to block denitrification pathways and convert all denitrification products to N2O. Chambers were constructed from 8 oz mason jars with two syringe stopcocks installed in the lid. Each soil core was placed in a chamber and flushed with N2 for 5 min. At time 0, 15, 30, 45, and 60 min, air samples were removed with a 30 ml nylon syringe, and acetylene gas was injected. N2O concentrations in the air samples were analyzed and calculated in the same way as in situ air samples from the lawns. After the acetylene-blocking test, the soil cores were sieved at 2 mm, oven dried at 110 °C for 48 h, and weighed. The denitrification rates of the soil cores were estimated from the N2O fluxes over time per unit of dry soil. Measurements were conducted on a daily basis up to 7 days. Pre-fertilization sampling was conducted 1 day before each of the 6 fertilization events in every lawn. Background denitrification rates in each lawn were estimated as the average of all 6 pre-fertilization sampling events.
To assess the leaching IN concentrations, three lysimeters with ceramic tips (Irrometer® Soil Solution Access Tubes, Riverside, CA, USA) were installed in summer 2010 at 40 cm, below the rooting zone (0–10 cm). Soil solution samples were extracted every day following fertilization for up to 5 days in March 2011 from Low Impact and following fertilization in the other two lawns in August 2010, September 2010, March 2011, and May 2011. Pre-fertilization sampling was conducted in every lawn before each fertilization event. Each lysimeter was evacuated with an air pump for 60 s and left overnight in the soil to allow soil solution to flow into the tube through the ceramic tip; soil solution samples were extracted with a 60 ml nylon syringe on the second day. All samples were filtered with 0.2 μm microfilters, stored at 4 °C until analyzed colorimetrically for NH4
+–N and NO3
−–N using the phenol-hypochlorite method by Weatherburn (1967) and the vanadium method by Doane and Horwath (2003). Due to varying soil moisture conditions, we were not always able to collect all three samples. Drainage was estimated with the simplified equation:
$$ D = I + P - R - ET_{0} $$
(6)
where D, I, P, and R represents drainage, irrigation, precipitation, and runoff, respectively. Daily precipitation and ET0 rates were obtained from the CIMIS network station nearby (Irvine station, #75, http://www.cimis.water.ca.gov). Runoff was measured by collecting all surface runoff in 0.3 × 0.3 m concrete vaults downslope of each landscape. Electronic water sensing sump pumps (Water Ace, Ashland, OH, USA) transported runoff through an oscillating piston type water meter pulse flow meter (C700, Elster AMCO Water, Langley, Canada) which was logged daily (CR1000, Campbell Scientific, Inc., Logan, UT, USA).
The leaching inorganic N fluxes at 30 day (Nleach_30 ± SE
l30) and annual (Nleach_365 ± SE
l365) scales were estimated as
$$ N_{leach\_30} = IN_{fert} * D_{fert} $$
(7a)
$$ SE_{l30} = \left| {N_{leach\_30} } \right| * \sqrt {(SE_{fert} /IN_{fert} )^{2} + (SE_{df} /D)^{2} } $$
(7b)
$$ N_{leach\_365} = \sum\limits_{i = 1}^{m} {N_{leach\_30i} } + IN_{bkgd} * D_{bkgd} $$
(7c)
$$ SE_{l365} = \sqrt {\sum\limits_{i = 1}^{m} {SE_{130i}^{2} } + SE_{lb}^{2} } $$
(7d)
$$ SE_{lb} = \left| {IN_{bkgd} * D_{bkgd} } \right| * \sqrt {(SE_{bkgd} /IN_{bkgd} )^{2} + (SE_{db} /D_{bkgd} )^{2} } $$
(7e)
where INfert ± SEfert/Dfert ± SEdf and INbkgd ± SEbkgd/Dbkgd ± SEdb represent the mean ± SE of inorganic N (IN) concentrations/drainage in soil leaching solution following fertilization and at background levels, and SElb represents the error (SE) of the total leaching IN in the months without fertilization in 2010; m represent the number of months with fertilization (m = 4 for Typical and Low Input, m = 1 for Low Impact).
Residual N fluxes and potential changes in N stocks (soil and roots)
Residual N fluxes (NR ± SEr), or the difference between N outputs and inputs, were calculated for each lawn at both 30-day and annual scales as
$$ N_{R} = N_{total} {-}N_{clip} {-}N_{{{\text{NH}}_{ 3} }} {-}N_{{{\text{N}}_{ 2} {\text{ + N}}_{ 2} {\text{O}}}} {-}N_{leach} $$
(8a)
$$ SE_{R} = \sqrt {SE_{t}^{2} + SE_{p}^{2} + SE_{{{\text{NH}}_{ 3} }}^{2} + SE_{{{\text{N}}_{ 2} {\text{ + N}}_{ 2} {\text{O}}}}^{2} + SE_{l}^{2} } $$
(8b)
where (N
clip
± SEp) represents the N output in plant clippings calculated from Eqs. (4a)–(4c) \( \left( {N_{{{\text{NH}}_{3} }} \pm {\text{SE}}_{{{\text{NH}}_{3} }} } \right) \) and \( \left( {N_{{{\text{N}}_{2} + {\text{N}}_{2} {\text{O}}}} \pm {\text{SE}}_{{{\text{N}}_{2} + {\text{N}}_{2} {\text{O}}}} } \right) \) represent the fluxes of NH3 and N2 + N2O, calculated from Eqs. (5a)–(5d), and (N
leach
± SEl) represent N leaching losses, calculated from Eqs. (7a)–(7e). We did not include estimated NO fluxes in the residual calculation, as these were not directly measured; in addition, our estimates indicated that NO fluxes likely represent a very small portion of the N inputs.
Soil N immobilization/mineralization potentials (0–10 cm) at the 30 day scale (Nsoil_30 ± SE) were estimated by 30 day lab incubation (Wang and Zhu 2012; Zhu and Carreiro 1999; Zhu and Wang 2011) of soil samples with (NFS) or without (NNFS) N addition. Briefly, to determine the net N mineralization and net nitrification potentials of the lawn soils, a set of pre-fertilization soil cores were taken from the Typical and Low Input lawns in March 2010 and from the Low Impact lawn in March 2011 at five random locations each, 3 cores per location. The 3 cores from the same location were combined into one sample, sieved, and homogenized. Soil net N potentials were determined by incubating subsets (~10 g each, n = 5) of sieved soils at room temperature and field moisture levels for 30 days. The incubated soils, as well as a subset of soils before incubation, were extracted with 1.5 M KCl and measured for IN concentrations (NH4
+–N and NO3
−–N). The net potentials of soil N mineralization (or immobilization) and nitrification (μg N g−1 dry soil day−1) were calculated as:
$$ {\text{Net N mineralization }}\left( {\text{or immobilization}} \right) \, = \frac{{\left( {{\text{IN}}_{\text{final}} - {\text{IN}}_{\text{initial}} } \right)}}{{{\text{T}}_{\text{incu}} }} $$
(9)
$$ {\text{Net nitritication }} = \frac{{ ( {\text{NO}}_{ 3}^{ - } {\text{N}}_{\text{final}} - {\text{NO}}_{ 3}^{ - } {\text{N}}_{\text{initial}} )}}{{{\text{T}}_{\text{incu}} }} $$
(10)
where INinitial and INfinal were the soil IN concentrations before and after incubation, NO3
−Ninitial and NO3
−Nfinal were the NO3
−N concentrations before and after incubation, and Tincu was the incubation time. Two sets of soils were incubated, one as “Control”, and the other as “Plus-N”, to which high-concentration ammonium chloride (NH4Cl) solution was added at a rate equivalent to the fertilization rates they receive in the field. The annual change in soil N stocks (Nsoil_365 ± SE
s
) due to soil N immobilization was estimated as
$$ N_{soil\_365} = \, m * N_{FS} + \, n * N_{NFS} $$
(11a)
$$ SE_{s365} = \sqrt {m * SE_{FS}^{2} + n * SE_{NFS}^{2} } \, $$
(11b)
where m and n are the number of months with or without fertilizer application in 2010 (m = 4, n = 8 for Typical and Low Input, m = 1, n = 11 for Low Impact).
We also assessed soil N retention with other methods for comparison. We made direct measurements of soil N% and root biomass N in 2010 and 2011: five soil cores each were collected from each lawn before fertilization in March, 2010 and 2011 to assess the annual differences. Soils were sieved to 2 mm, homogenized, oven dried for 48 h at 110 °C, weighed, ground to fine powder, and analyzed for total C% and N% on the elemental analyzer. We assumed no significant variation in soil bulk density during our study period, and averaged the bulk density of soil cores in each lawn, which were 1.2 ± 0.03, 0.81 ± 0.04, and 1.2 ± 0.1 Mg m−1 for Typical, Low Input, and Low Impact, respectively. In order to assess the variation in root N content between 2010 and 2011, roots (≥2 mm) were separated from the soil cores described above, oven dried for 48 h at 70 °C, weighed, ground to fine powder, and analyzed for total C% and N%. The total N mass in soil or roots was calculated as the product of density and N%. To account for possible variation in root N at the 30 day scale, another 5 soil cores were sampled from each lawn 30 days after fertilization in March 2011 and the roots were separated and analyzed for total C% and N%.
Data analysis and residual calculation
N budgets were constructed for Typical and Low Input at a 30-day scale following each of the 6 fertilization events from January 2010 to June 2011 and for all lawns at an annual scale for 2010. Repeated measures ANOVA was used to test the effects of lawn type, time (day), and their interactions (lawn type*day) on the daily averages of soil volumetric water content, while an F test was used to compare the variances of the daily averages. Nested Analysis of Variance (ANOVA) was used to compare IN concentrations of lysimeter samples below the rooting zones (40 cm) and denitrification rates, with time (day) as the nominal factor nested in lawn types. Multivariate regression analyses were used to evaluate the relationship between denitrification rates and potential driving variables (soil water content and NO3
−–N concentrations), and the relationship between plant productivity and environmental factors (air temperature, soil water content, and plant N%, as an indicator of N availability). We used ANOVA tests to compare the soil total C%, N%, and net N potentials among the three lawns, and Student’s t tests (α = 0.05) to compare these variables in 2010 and 2011, as well as root densities and root N% of roots at the 30-day and annual scales. All analyses were performed with SAS software v9.1.3 (SAS, Cary, NC, USA).