Snowbanks to Grassbanks

Hypothesis:

The objectives of this proposal are to compare areas below snowbanks that have water slowing/spreading structures (treated) to areas without these structures (control) to address the following hypotheses:

1. Do cattle grazing in pastures with water slowing structures prefer the mesic areas compared to the upland sites?
2. Plant Response
   1. There are no differences in plant cover, green cover, forbs, grasses and rushes/sedges at treated areas compared to untreated control areas from 2021 and 2022.
   2. There are no differences in plant production of forbs, grasses and rushes/sedges at treated areas compared to untreated control areas in 2019 and 2020.
   3. There are no differences in soil moisture between treated areas and untreated control areas using 6 drainages from 2019 to 2022 when measured with a surface probe (Vegetronix VG-Meter)
   4. There are no differences in soil moisture between treated and untreated control areas from 4 drainages measured with hourly soil moisture probes in 24 reaches (12 treated, 12 control) from 2020 to 2022.
   5. There are no differences in soil moisture between treated areas and untreated control areas in seasonal drying rates, soil water retention and water residence time after rainfall events.

1. 3. Sage-Grouse Resource Responses:
   1. There are no differences in herbaceous vegetation preferred by sage-grouse chicks at treated areas compared to control areas.
   2. There are no differences in abundance or biomass of arthropod orders known to be sage-grouse chick food sources at treated areas compared to untreated control areas.

Materials and Methods:

In the summers of 2018 and 2019, field crews and producers installed water slowing and spreading structures.  We sampled 159 water-slowing and 3 water-spreading structures. 

The initial selection and construction of these structures was conducted by Bryant Jones, Barry McCoy, and Rulon Buhler of our producer team, Dr. Levine of University of Montana Western, Kyle Cutting of Red Rock Lakes National Wildlife Refuge (NFWF), and Jim Berkey and Sean Claffey of The Nature Conservancy.  Half of the costs associated with the construction and subsequent monitoring came from a National Fish and Wildlife Foundation (NFWF) grant.  The other half of the funds came from matching in-kind funds from the other cooperating institutions, including the Bureau of Land Management. 

These structures are within the 1.5-million-acre core sage-grouse areas of southwest Montana, in Beaverhead and Madison Counties.  This region is nestled between the High Divide and the Greater Yellowstone area.  This region is not only important to many family-run cattle grazing operations but is also vital to ecotourism.  Most of the structures are located in mountain big sagebrush (Artemisia tridentate spp. vaseyana) plant communities above 6,700 ft, with the majority located on private land, and the remaining structures on public lands. 

Experimental Design

We used identical sampling methods in treated and control reaches. Treated reaches had at least three water-slowing structures, whereas control reaches had at least three locations that could have had such structures. We centered our data collection on three by two-meter subplots in these locations (Figure 1).  We sampled the most upstream and downstream water-slowing structures (in treated reaches) or the location where structures would have been installed, as well as the most central structure/location.  In 2019 and 2020, no sampling took place within a 1.5-m buffer of each water-slowing structure (or the location where it could have occurred in control reaches) to eliminate bias associated with the disturbed ground created during construction; we placed the subplot immediately upstream of this buffer area.

During the summer of 2019, we sampled 16 reaches (8 treatment, and 8 control) in four drainages; this timing of sampling captured responses one year after treatment.  During summer 2020, we collected data in the same 16 reaches, to capture responses two years after treatment.  We also sampled 38 new reaches (18 treatment and 20 control) in three additional drainages in 2020, representing one year after treatment.  We sampled vegetation and arthropods in each reach three times (visits): once in each of the months of June, July, and August.  We chose this sampling schedule to capture the gradient of senescence throughout the summer.  During the summers of 2020 through 2022, we added one additional vegetation sampling visit during the month of September. 

Vegetation cover and production tend to increase in wet meadows, as well as trend towards increased cover of graminoids and forbs.  We expected to see denser canopy coverage and increased production in treated sites when compared to untreated sites.

Figure 1. Visual representation of a reach, our experimental unit, and sampling areas including the sampling area (black hashed line), vegetation frame (gray square), and soil moisture readings (black circles) southwestern Montana.

Cattle Observations

Cattle spent very little time in drainages that contained treatments and controls, and only 10 of the 340 observations made were of cow/calf pairs in treated drainages. Observations from the 2021 field season showed that cattle location could not be identified by treatment or controls since the scale was so small. However, we counted cattle in drainages near study sites to estimate how much time cattle spend in mesic areas vs. uplands. This was done by observing cattle in peak grazing times from a vantage point from 0600 to 1200 and 1700 to 2100 hr. During these times of day a researcher would count the number of cow/calf pairs in mesic and upland areas every 15 minutes. We obtained 110 hours of observations (n=440) during the 2021 and 2022 grazing season on the four sites that were grazed. Cattle were present in Keystone Tributary, Clover Tributary, Little Basin Seven Tributary and Little Basin 24 Tributary during the months of July and August.

 Vegetation

We used a canopy coverage approach to quantify plant communities.  We estimated total cover of vegetation and non-vegetation (e.g., woody debris, bare ground) within a 0.5 x 0.5-m frame randomly placed within each subplot, resulting in three estimates per reach (0-100%; Figure 1).  We then identified each individual plant to the lowest taxonomic level feasible and estimated percent coverage of each functional group (e.g., forbs, grasses, litter); the total of these estimates did not exceed 100%.  In 2021 and 2022, we added an additional three frame estimates resulting in six estimates per reach. In 2021 and 2022, we also clipped the half meter squared quadrats to estimate total vegetation production (g/m²).

Vegetation and Soil Moisture Data Analysis

To characterize the efficacy of the restoration structures, we analyzed variation in soil moisture, as well as vegetation cover (by functional group) and species composition. We averaged data collected within each reach. We modeled each response variable as a function of treatment and visit (month) using general linear mixed models; we included a random intercept for drainage to account for repeated sampling and inherent variation among drainages.  To account for variation within each drainage, we included four additional covariates: relative distance, gully slope, bank aspect, and the width-depth ratio for each reach. We repeated this process for years 1, 2, 3 and 4. Relative distance was the distance between the sampled reach and the most upstream reach in the drainage, to account for spatial trends.  Gully slope was the average incline of sampled areas within each reach, to adjust for lower water infiltration and drier conditions with steeper slopes.  To accommodate variation in how quickly soils dry after a period of moisture recharge, we averaged the bank aspect (river left) of sampled areas within each reach.  Finally, we computed the width-depth ratio, dividing the width of the gully by the depth, averaged for all sampled areas in the reach, to account for changes in shape leading to changes in soil saturation. Larger values of this ratio denote wider and shallower gullies, which tend to be wetter than narrower and deeper gullies.

Soil Moisture

We characterized soil moisture by measuring volumetric water content at a depth of 10 cm (Vegetronix VG-Meter-200, Vegetronix, Inc., Riverton, Utah).  In each subplot, we sampled soil moisture along a 10-m transect, which was perpendicular to and centered on the gully, two meters above the structure. We collected measurements at one-meter increments, with one measurement taken in the center of the gully (11 total measurements per subplot). Gully widths varied and we only included soil moisture measurements that fell within the bank edges of the gully in analyses.  In the summers of 2020 through 2022, we collected hourly soil moisture at 12 treated and 12 control reaches evenly divided between 4 drainages using soil moisture probes (HOBO IOHS moisture sensors) placed at 0m, 1m and 3m from the centerline of the gully. Probes were buried at 15cm. A Davis tipping bucket was attached to record total rainfall.

Arthropod Sampling

To ensure we captured both ground- and vegetation-dwelling arthropods, we used a combination of vacuum and pitfall sampling to characterize arthropod communities in treatment and control reaches in 2019 to 2020. We combined samples from each of the methods collected within a reach during the same visit. 

              Vacuum sampling - To capture flying and vegetation-dwelling arthropods, we used a vacuum/blower (Stihl SH56C, Stihl, Inc., Waiblingen, Germany) with window screen attached to the end. One vacuum sample was collected at a random location within each sampling plot, for a total of three for each experimental unit.  At the selected location, we placed a 0.25-m­^2­ plastic barrel with an 800-micron screen covering on top, to prevent arthropods from escaping.  We ran the vacuum for 30 seconds within that barrel at each sampling location. To reduce bias from disturbing vegetation, we ensured vacuum sampling was the first task completed after arriving at the experimental unit. This method was not conducted in 2021 or 2022.

              Pitfall sampling - We used pitfall trapping to capture ground-dwelling arthropods.  Ideally, pitfall traps should have a diameter between 6.5 and 15 cm, to allow for relatively efficient sampling without drastically increasing effort.  With this in mind, we used 9.5 x 12-cm plastic cups (Solo Cup Company, Lake Forest, Illinois) dug deep enough to be flush with the ground.  Additionally, we removed vegetation and debris from the area immediately surrounding the pitfall trap to allow for free movement of ground-dwelling arthropods.  We partially filled each pitfall trap with propylene glycol (LowTox Antifreeze/Coolant, Prestone Products Corporation, Lake Forest, Illinois) to kill and preserve trapped insects.  We placed one pitfall trap within each sampling unit, for a total of three for each experimental unit, for one 24-hour period.  For the first visit (June), we randomly selected a location within each sampling plot.  For the second (July) and third (August) visits, we reused the pitfall hole dug during the first visit to reduce soil disturbance. 

Arthropod Processing

After collection, arthropods were cleaned of debris and stored in 90% ethanol until further processing.  We identified each collected arthropod to order.  We measured all collected arthropods from frons to the tip of the abdomen to the nearest millimeter, then converted arthropod length to estimated biomass using taxon-specific length/mass regression equations. We computed an estimated biomass by order for each reach, combining data from the three sampled locations during one visit. In 2021 and 2022, we dried arthropods as weighed specimens to obtain dry biomass. We focused on known important arthropod foods (see Analysis).

Sage-Grouse Analysis

We used information from the literature to characterize important arthropod and plant foods for sage-grouse.  Although we aimed to find diet information from nearby areas, we used references from other locations as needed. Sage-grouse chicks overwhelmingly consume arthropods belonging to the orders of Hymenoptera (mainly Formicidae), Coleoptera, Orthoptera, and larval Lepidoptera.  We used the combined estimated biomass (in µg) for important arthropod foods in each reach for each visit as the response variable for sage-grouse chicks. In Montana, Idaho, and Oregon, chicks primarily consume forbs belonging to the families Brassicaceae, Fabaceae, Polemoniaceae, Asteraceae, and Lilaceae in areas dominated by big sagebrush (Artemisia tridentata).  We compared canopy coverage of known important genera within these families as the response variable in models for sage-grouse chicks. We modeled each response variable as a function of treatment (treated or control) and visit (month) using general linear mixed models: we included a random intercept for drainage to account for repeated sampling and inherent variation among drainages.  To account for variation within each drainage, we included four additional covariates: relative distance, gully slope, bank aspect, and the width-depth ratio for each reach.