1. Mitigating ecosystem service (ES) tradeoffs is a key management goal in locations where stakeholders value different and potentially conflicting ecosystem services (ESs). However, studies are not often designed to examine how local management actions address ES tradeoffs, and therefore do not provide options that can alleviate conflict.

2. In semi-arid rangelands, we examined the potential for managers to mitigate tradeoffs between livestock production and water quality. To move away from solutions that offer cattle removal as a singular management strategy, we examined how cattle presence, plus two elements of rotational grazing - the length of time cattle spend on rangeland (i.e., duration), and the season grazed (i.e., timing), affected stream Escherichia coli (E. coli concentrations). We also modeled how grazing duration and timing affected the ability to meet regulatory benchmarks for water quality throughout a grazing season.

3. Grazing duration controlled the length of time E. coli concentrations were high in streams. In short- and medium-duration systems, E. coli concentrations were high for shorter periods of time than in long-duration systems, resulting in fewer violations of national and state water quality standards.

4. Stream E. coli concentrations showed a consistent seasonal pattern, starting low in spring, peaking in summer, and declining towards fall. Thus, grazing during spring or fall, rather than in summer, reduced the number of days that E. coli levels exceeded water quality standards.

5. Our results suggest that reducing the grazing duration and shifting its timing are complementary strategies that can mitigate the tradeoffs between livestock grazing and water quality without fencing-off riparian areas or removing cattle from pastures with streams.

6. Synthesis and applications. In this study, we found grazing duration and timing can be used as tools to mitigate ES tradeoffs between cattle production and water quality in rangeland streams. Shorter grazing durations reduced the number of days E. coli levels were above regulatory limits, as did grazing that occurred either early or late in the season. These results support the idea that rotational grazing can be an effective strategy to manage water quality in semi-arid rangelands. They also highlight the need for more grazing studies that incorporate gradients of duration and timing into study designs.

We measured stream E. coli levels over three years in rangelands employing target grazing treatments. Grazing treatments included: continuous-turnout (long-duration, no rotation), deferred-rotation (medium-duration, with rotation), and time-controlled rotation (short-duration, frequent rotation). The first two of these are common in this region, the third is used less often. All were in-use across our sampling sites prior to this study. Long durations ranged from 82-138 days, medium durations from 31-81 days, and short durations from 1-30 days. Timing for continuous-turnout spanned the entire grazing season (mid-May through mid- September). For deferred-rotation, grazing began across a range of timings including mid-May, mid-June, early July, and mid-July. For time-controlled rotation, grazing timing ranged anywhere from spring through fall. All sites were grazed with beef cattle cow-calf pairs. Stocking densities were 1.03 - 1.78 pairs · ha^-1 in time-controlled areas, 0.11 – 0.28 in deferred-rotation, and 0.02 - 0.09 pairs· ha^-1 in continuous-turnout areas. These equated to stocking rates of ~0.7 pairs · ha^-1 · month^-1 in time-controlled areas, and ~0.3 pairs · ha^-1 · month^-1 in deferred-rotation and continuous-turnout areas.

We sampled every two to three weeks from May through October/early November, a timeframe that encompasses the grazing and recreation season in this area of Utah. By sampling twice per month, we captured fluctuating E. coli levels as cattle moved in and out of pastures. We collected water grab samples according to UT Division of Water Quality’s Standard Operating Procedures for collection, handling, and quantification of E. coli samples. At each site, we collected 100 mL grab samples from flowing stream channels in sterile jars, which we stored on ice. We analyzed samples within eight hours of collection using the Idexx E. coli Quanti-Tray 2000 System (Westborook, MA), adding pre-packaged Colilert reagent to jars, sealing mixture into analysis trays, and incubating samples at 35ºC for 18 - 28 hours. E. coli concentrations were identified as most probable number (MPN) of colony forming units per 100 ml via florescence under a UV light. The Quanti-Tray System can detect E. coli concentrations to a maximum of 2419.6 MPN without dilution. We did not dilute samples because this value is above all regulatory benchmark values.