*2.2. Field Treatments and Agronomic Considerations*

During the calibration period (2006–2008), all fields were managed identically with fall liquid dairy manure application (at a rate to meet 80% of annual corn N need) and chisel plow tillage to a depth of 15 cm (2.7 m wide seven-shank plow Landoll Farm Equipment, Brillion, WI, USA) the same day. Field cultivation to a depth of 7.5 cm (6.2 m wide John Deere, Moline, IL, USA) was performed in the spring before planting corn.

The same management was continued on the designated control field (M1) during the treatment period, but management was changed on the others. Field M4 (BFMT) was treated identically to M1 but with the addition of a mixed grass-legume buffer strip (9.8 m wide) along the lower side of the field above the drainageway. The buffer strip was planted on 3 October 2008 using a no-till drill (John Deere 1590, Moline, IL, USA) with 112 kg ha−<sup>1</sup> winter rye (*Secale cereale* L.), 2.2 kg ha−<sup>1</sup> alsike clover (*Trifolium hybridum* L.), 9.0 kg ha−<sup>1</sup> tall fescue (*Festuca arundinaceum*), and 3.4 kg ha−<sup>1</sup> smooth bromegrass (*Bromus inermis* L.). While the vegetative buffer was implemented as a conservation practice, it can also provide an economic return as harvestable hay forage crop for dairy animals and removes soil N and P. In our study the buffer vegetation was harvested approximately twice per year with field-scale equipment, but yield and nutrient data are incomplete. In M2, a winter rye cover crop was planted in early Oct after the silage corn harvest at 112 kg ha−<sup>1</sup> with a no-till drill (John Deere 1590, Moline, IL, USA) and manure application and chisel tillage was delayed until spring (RSMT). On field M3, liquid dairy manure was broadcast on the surface in the fall (without incorporation) and chisel tilled the following spring (FMST). While not considered a BMP because of the lack of manure incorporation, it is still a relatively common practice, so we chose to include it to assess potential effects on surface runoff water quality.

Liquid dairy manure was sampled at each application and analyzed for N, P, potassium (K), NH4 +-N, and dry matter content by the University of WI Soil and Forage Lab [40] (Table 1). Manure application rates averaged 45,880 L ha−<sup>1</sup> and ranged from 3.2% to 18% dry matter content. Nutrient application rates are presented in Table 1. Average total N and P application rates from manure were approximately 155 and 24 kg ha−<sup>1</sup> year−1, respectively. This N application met approximately 60% of crop needs (based on N availability of 50% in the first year and 10% in the second) slightly less than during calibration due to lower manure N contents in 2010 and 2011 than expected. Each year, corn was planted on all fields on the same day. Corn (*Zea Mays* L.) (2905RB; 89-day RM; YGCB RR, in 2009 and 2010; RK212GT; 81-day RM; RR, in 2011) was planted in May or late April at 87,500 seeds ha−<sup>1</sup> with 112 kg ha−<sup>1</sup> of 9-11-30-6S-1Zn starter fertilizer applied as a band (50 mm to the side of the seed row and 50 mm deep) via the planter. Additional fertilizer N was applied in June or July as needed (Table 1) based on a pre-sidedress nitrate soil test [41]. Corn was harvested for silage on the same day for all fields between mid-Sept and early Oct (Table 1). Yields were estimated by hand-harvesting above-ground corn biomass samples (all but the bottom 25 cm of stalk) from nine randomly selected, 3 m row-length sub-plots from each field when whole-plant DM content had reached approximately 350 g kg−1. Subsamples were taken and a composite plot sample was dried at 55 ◦C, ground to pass 1 mm, and analyzed for total N (Elementar Variomax CN analyzer, Ronkonkoma, NY, USA) and P content after nitric acid digestion by ICP-OES (University of WI Soil and Forage Lab) to estimate corn N and P removal. Soil samples were collected: nine 2.5-cm diameter, 20-cm deep samples were taken in each watershed with a hand sampler, every fall throughout the treatment phase, and in the spring of 2012. Plant-available P was extracted using the Bray 1 solution (0.03 *N* ammonium fluoride + 0.025 *N* hydrochloric acid; [42]). Phosphorus in extracts was determined colorimetrically (ammonium molybdate solution) using standard techniques (abbreviated as B1P).


**Table 1.** Field activities and manure composition during the 2008–2011 treatment period.

† N = nitrogen, P = phosphorus, K = potassium, NH4-N = ammonium nitrogen; ‡ RSMT = fall rye (cover crop) with spring applied manure and chisel tillage; FMST = fall applied manure with spring tillage; BFMT = fall applied manure/chisel tillage with grass buffer; †‡ 42 kg ha−<sup>1</sup> on M1 and M4; 92 kg ha−<sup>1</sup> on M2 and M3.

#### *2.3. Hydrology and Runoff Measurements*

Details on runoff instrumentation and monitoring are found in Jokela and Casler [39] and are briefly described here. Runoff was sampled and monitored at gauging stations located at the low elevation point of each field. Original flume design and monitoring procedures were based on those of the US Geological Survey with slight modifications [43]. A 60 cm fiberglass H-flume (Tracom, Inc., Alpharetta, GA, USA) was attached to pressuretreated plywood wingwalls (driven to approximately 60 cm deep and extending approximately 3 m on each side). In November 2007, 1.8 m long channels were installed between the wingwall and the flume to provide more uniform flow entering the flume and a greater distance for deposition of sediment ahead of the flume [44]. Plywood wingwalls were replaced with steel sheet pilings placed 1.2 m deep at M2 on 11 May 2012 and M3 on 24 Mar 2010 because of failure due to frost heaving. Flumes were mounted with threaded rods for leveling as needed. Shallow earthen berms directed surface runoff to each flume. During late fall through early spring, plywood enclosures were attached to the approach channel/flume and a quartz heater was used as needed to prevent freezing of sample lines. Instrumentation was housed inside a 1.8 × 2.1 × 2.0 m3 high shed (Niagara model, Yardmate Series, Royal Outdoor Products, Inc., Middleburg Heights, OH, USA) equipped with AC power for data loggers, sampling equipment, heaters, and heat tape with battery backup power.

Runoff volume was determined by the measuring stage in the H-flumes with an air bubbler/pressure transducer flow meter (ISCO Model 4230, Teledyne Isco, Inc., Lincoln, NE, USA). A bubbler PVC tube (3.175 mm i.d) was attached to the floor of the flume 40 mm back from the outlet. Staff gages were also installed in the H-flumes to allow simultaneous comparison of the stage with that from the flow meter. Time-based runoff samples were collected at intervals based on estimated event runoff quantity by an automated 24-bottle (1 L) refrigerated sampler (ISCO 6712SR, Teledyne Isco, Inc., Lincoln, NE, USA). A sampling tube (9.3 mm i.d.) was attached to the flume floor near the flume outlet and extended approximately 2 m to the automated sampler inside the enclosure (protected from freezing by heat tape and foam insulation). A CR10X datalogger (Campbell Scientific, Inc., Logan, UT, USA) was used to read and store data and control the runoff sampling collection scheme. A weather station (Campbell Scientific, Inc., Logan, UT, USA) was located 1000 m from the site and measured precipitation (tipping bucket), air temperature, humidity, wind, and solar irradiance. Real-time, two-way radio telemetry allowed remote communication with each station and the weather station. A Campbell scientific software program (PC208W) was used for real-time communication to modify sampling intervals as needed.

#### *2.4. Nutrient and Runoff Water Quality Measures*

Samples from individual autosampler bottles were combined into a flow weighted composite for each runoff event. Samples were analyzed for suspended sediment (SS; gravimetric method 3977-97B) [45], total P (TP; block digestion, method 4500 P F; [46]), and total Kjeldhal N (TKN; block digester automated colorimetric, 4500 NH3 G; [46]), a filtered (0.45 μm) subsample was analyzed for DRP (automated colorimetric method 4500 P F; [46]), nitrate + nitrite-N; abbreviated as NO3-N (automated Cd reduction; 4500 NO3 F5; [46]), and NH4 +-N (automated phenolate, 4500 NH3 G; [46]). Values for TKN and NO3-N were added to provide an estimate of total N (TN) since NO3-N is not measured in the TKN procedure.
