New concepts have emerged in theoretical ecology with the intent to quantify complexities in ecological change that are unaccounted for in state-and-transition models and to provide applied ecologists with statistical early warning metrics able to predict and prevent state transitions. With its rich history of furthering ecological theory and its robust and broad-scale monitoring frameworks, the rangeland discipline is poised to empirically assess these newly proposed ideas while also serving as early adopters of novel statistical metrics that provide advanced warning of a pending shift to an alternative ecological regime. We review multivariate early warning and regime shift detection metrics, identify situations where various metrics will be most useful for rangeland science, and then highlight known shortcomings. Our review of a suite of multivariate-based regime shift/early warning indicators provides a broad range of metrics applicable to a wide variety of data types or contexts, from situations where a great deal is known about the key system drivers and a regime shift is hypothesized a priori, to situations where the key drivers and the possibility of a regime shift are both unknown. These metrics can be used to answer ecological state-and-transition questions, inform policymakers, and provide quantitative decision-making tools for managers.

Long-term vegetation dynamics across public rangelands in the western United States are not well understood because of the lack of large-scale, readily available historic datasets. The Bureau of Land Management's Soil-Vegetation Inventory Method (SVIM) programwas implemented between 1977 and 1983 across 14 western states, but the data have not been easily accessible. We introduce the SVIM vegetation cover dataset in a georeferenced, digital format; summarize howthe datawere collected; and discuss potential limitations and biases. We demonstrate how SVIM data can be comparedwith contemporarymonitoring datasets to quantify changes in vegetation associated with wildfire and the abundance of exotic invasive species. Specifically, we compare SVIM vegetation cover data with cover data collected by BLM's Assessment, Inventory, and Monitoring (AIM) program (2011– 2016) in a focal area in the northern Great Basin. We address issues associated with analyzing and interpreting data from these distinct programs, including differences in survey methods and potential biases introduced by spatial and temporal variation in sampling. We compared SVIM and AIM survey methods at 44 plots and found that percent cover estimates had high correspondence for all measured functional groups. Comparisons between historic SVIM data and recent AIM data documented significant declines in the occupancy and cover of native shrubs and native perennial forbs, and a significant increase in exotic annual forbs. Wildfire was a driver of change for some functional groups, with greater change occurring in AIM plots that burned between the two time periods compared with those that did not. Our results are consistent with previous studies showing that many native shrub-dominated plant communities in the Great Basin have been replaced by exotic annuals. Our study demonstrates that SVIM data will be an important resource for researchers interested in quantifying vegetation change through time across public rangelands in the western United States.

Restoring western US rangelands from a site dominated by invasive annuals, such as cheatgrass and medusahead, to a diverse, healthy, perennial plant−dominated ecosystem can be difficult with native grasses. This study describes the establishment and trends in persistence (plant/m²) of native grass cultivars and germplasm compared with typically used crested and Siberian wheatgrasses at four locations in Idaho (one),Wyoming (one), and Utah (two) that range in mean average annual precipitation (MAP) from 290 to 415 mm. Sites were cultivated and fallowed 1 yr before planting using two glyphosate applications to control weeds. We monitored seedling establishment of 10 perennial cool-season grass species and plant persistence over 5 yr. Precipitation during the seeding year varied with the Utah sites locations reviving below MAP (4% and 14%), while the Wyoming and Idaho sites received above MAP at 8% and 26%, respectively. Across these four sites, native grass seedling establishment of bottlebrush squirreltail (29±0.08 [standard error] seedling/m²), bluebunch (28±0.05), slender (30 ± 0.05), and Snake River wheatgrasses (28 ± 0.08) was similar to “Vavilov II” Siberian wheatgrass (36 ± 3.20). By yr 5, western, Snake River, and thickspike wheatgrasses were the only native grasses to have plant densities similar to Vavilov II (37 ± 0.29) Siberian and “Hycrest II” (36 ± 0.29) crested wheatgrasses. On sites receiving between 290 and 415 mm MAP, our data suggest that native grasses are able to establish but in general lack the ability to persist except for western, Snake River, and thickspike wheatgrasses, which had plant densities similar to crested and Siberian wheatgrasses after 5 yr.

Invasive species control requires understanding the mechanisms behind their establishment and their interactions with other species. One potential ecosystem alteration influencing the establishment and spread of invasive species is anthropogenic nitrogen enrichment, from sources like introduced or invasive nitrogen (N)-fixing legumes, which can alter competition between native, non-native, and invasive plants. Kentucky bluegrass (Poa pratensis) and N-fixing yellow sweet clover (Melilotus officinalis) are exotic to the Great Plains and are currently invading and degrading native rangelands by altering ecosystem processes and displacing native plants. Therefore, we investigated how N enrichment from yellow sweet clover affects the aboveground biomass production of Kentucky bluegrass and western wheatgrass (Pascopyrum smithii), a native cool-season grass, the ranges of which overlap in the northern Great Plains. In a controlled greenhouse environment, we conditioned experimental pots by growing yellow sweet clover and terminating each plant after 8 wk. Conditioned soils contained ≈340% more plant-available N than untreated soils 2 wk after yellow sweet clover death. We then grew Kentucky bluegrass and western wheatgrass transplant seedlings in interspecific and intraspecific pairs in pots conditioned either with or without yellow sweet clover for 12 wk. Aboveground biomass production of both Kentucky bluegrass and western wheatgrass grown in interspecific and intraspecific pairs increased in conditioned soils. However, when grown together in conditioned pots, the increase in Kentucky bluegrass biomass relative to untreated pots (520%) was double that of the increase in western wheatgrass biomass (260%). Our results reveal that Kentucky bluegrass can use increased soil N to produce proportionally more aboveground biomass than western wheatgrass, a native grass competitor. Thus, our results suggest yellow sweet clover and other sources of N enrichment may facilitate the invasion of Kentucky bluegrass.

Big sagebrush (Artemisia tridentata Nutt.) plant communities often require management to reduce shrub density and rehabilitate understory vegetation. We studied vegetation responses to a two-way chain harrow treatment and broadcast seeding of 12 herbaceous species at eight Wyoming big sagebrush (A. tridentata Nutt. subsp. wyomingensis Beetle & Young) sites. These sites differed in land-use history; five were cultivated for dryland wheat production during the 1950−1980s and then seeded with introduced forage grasses (C-S), while three had not been exposed to this land-use legacy (non C-S). Our objective was to evaluate whether the C-S legacy influences the magnitude of vegetation change following contemporary treatment. Before treatment, C-S sites had lower sagebrush cover, higher dead sagebrush cover, and higher broom snakeweed (Gutierrezia sarothrae [Pursh] Britton & Rusby) cover than adjacent non C-S sites. Plant community change 3 years after treatment, determined with multivariate ordination analysis of species composition, varied between site histories, and response to treatment was most strongly correlated with reductions in sagebrush cover, increases in perennial grasses, and increases in 10 other herbaceous species—including some undesirable species and four that were seeded in 2010. Five years after treatment, mature sagebrush cover remained reduced for both land-use histories, yet density of sagebrush seedlings and broom snakeweed increased in C-S sites during the second and third years after treatment. In addition, perennial forb cover increased for C-S sites, while perennial grass biomass increased for non C-S sites. Our results emphasize that broad variability in plant community responses to sagebrush reduction and seeding is possible within the same ecological site classification and that legacy effects due to the combination of past cultivation and seeding should be considered when planning restoration projects, including the consideration that seeding may not always be necessary on C-S sites.

Seeding native plants into degraded grasslands presents major challenges. Often, seeded species fail to establish and areas become/remain dominated by unwanted plants. We combined herbicides and seeding in former coal mining fields dominated by exotic winter annual grasses (downy brome [Bromus tectorum L.] and Japanese brome [Bromus arvensis L.], hereafter “annual bromes”). The main interest was restoring Wyoming big sagebrush (Artemisia tridentata spp. wyomingensis [Beetle & A. Young] S.L. Welsh, hereafter “big sage”), a very difficult species to restore to North American grasslands. We tested the nonselective herbicide glyphosate and the grass-specific herbicide quizalofop. The summer following herbicide applications and seeding, annual brome cover in controls 22% (CI₉₅% 13%, 36%) was significantly greater (P < 0.03) than in glyphosate 11% (CI₉₅% 5%, 25%) and quizalofop 16% (CI₉₅% 7%, 35%) treatments. At Decker mine, glyphosate increased seeded big sage density (P < 0.04) from 0.76 (CI₉₅% 0.27, 2.11) to 3.05 (CI₉₅% 1.42, 6.56) plants ⋅ m^-2 the second summer after seeding. Corresponding increases for Spring Creek mine were from 0.11 (CI₉₅% 0.03, 0.43) to 0.43 (CI₉₅% 0.13, 1.40) plants ⋅ m^-2 (P < 0.04). These results were consistent across two experiments initiated in different years. In addition to big sage, our study's seed mixes contained native grasses and forbs, and herbicide treatments tended to promote establishment of these plant groups. In annual brome-dominated areas of the northern Great Plains, conditions amenable to big sage seedling establishment do not appear entirely uncommon, and herbicides can increase establishment.

Fire plays a central role in influencing ecosystem patterns and processes. However, documentation of fire seasonality and plant community response is limited in semiarid grasslands. We evaluated aboveground biomass, cover, and frequency response to summer, fall, and spring fires and no fire on silty and clayey sites in semiarid, C₃-dominated grassland. The magnitude of change in biomass between years was greater than any differences among fire treatments. Still, differences existed among seasons of fire. Summer fire reduced non-native annual forb frequency (3% vs. 10% ± 2%) and Hesperostipa comata, reduced native annual forbs the first year, increased Poa secunda and bare ground, and increased Vulpia octoflora the second year. Fall fire increased grass biomass (1224 vs. 1058 ± 56 kg·ha^-1), but fall fire effects were generally similar to those of summer fire. Spring fire effects tended to be intermediate between no fire and summer and fall fire with the exception that spring fire was most detrimental to H. comata the first growing season and did not increase bare ground. All seasons of fire reduced litter, forb biomass, and frequency of Bromus japonicus and Artemisia spp., and they reduced H. comata, V. octoflora, and native annual forbs the first year, but increased basal cover of C₃ perennial grasses (2.2% vs. 0.6% ± 0.4%). Fire during any season increased dominance of native species compared with no fire (6.6% vs. 2.0%± 1.0% basal cover) and maintained productivity. Seasonal timing of fire manipulated species composition, but increased C₃ perennial grass cover and native species dominance with fire during any season indicated that using fire was more important than the season in which it occurred. In addition, fire effects on the vegetation components tended to be counter to previously observed effects of grazing, suggesting fire and grazing may be complementary.

Statistically defensible information on vegetation conditions is needed to guide rangelandmanagement decisions following disturbances such as wildfire, often for heterogeneous pastures. Here we evaluate sampling effort needed to achieve a robust statistical threshold using >2 000 plots sampled on the 2015 Soda Fire that burned across 75 pastures and 113 000 ha in Idaho and Oregon. We predicted that the number of plots required to generate a threshold of standard error/mean ≤ 0.2 (TSR, threshold sampling requirement) for plant cover within pasture units would vary between sampling methods (rapid ocular versus grid-point intercept) and among plot sizes (1, 6, or 531 m²), as well as relative to topography, elevation, pasture size, spatial complexity of soils, vegetation treatments (herbicide or seeding), and dominance by exotic annual or perennial grasses. Sampling was adequate for determining exotic annual and perennial grass cover in about half of the pastures. A tradeoff in number versus size of plots sampledwas apparent,whereby TSRwas attainable with less area searched using smaller plot sizes (1 compared with 531 m²) in spite of less variability between larger plots. TSR for both grass types decreased as their dominance increased (0.5–1.5 plots per % cover increment). TSR decreased for perennial grass but increased for exotic annual grass with higher elevations. TSR increased with standard deviation of elevation for perennial grass sampled with grid-point intercept. Sampling effort could be more reliably predicted from landscape variables for the grid-point compared with the ocular sampling method. These findings suggest that adjusting the number and size of sample plots within a pasture or burn area using easily determined landscape variables could increase monitoring efficiency and effectiveness.

Policymakers and managers are promoting Rangeland Fire Protection Associations (RFPA) as one way to better incorporate private citizens as active participants who contribute to fire suppression efforts on public rangelands. While the RFPA programis growing in popularity, little is known about the way that RFPAs establish and operate. This is especially true in mosaic management scenarios characterized by fragmented land ownerships and a variety of land or fire management entities responsible for wildfire suppression. Our goal was to investigate how an RFPA forms and functions in a management scenario characterized by: 1) proximity to exurban residential development; 2) agreements with multiple local, state, and federal wildfire suppression entities; and 3) a geographically disperse protection district. We conducted in-depth interviews with RFPA members, land or fire management professionals, emergency managers, and local interest groups who interact with the Black Canyon RFPA (BCRFPA) in southwestern Idaho. We found that the BCRFPA leveraged the insights, documents and support of existing RFPAs during their establishment, but ultimately had to adapt the RFPA idea to specific elements of their local context. Members of nearby rural fire districts were initially apprehensive about the formation of the BCRFPA due to concerns about resource competition (e.g., funding and large equipment). RFPA members with professional firefighting experience helped alleviate those tensions by explaining how the RFPA would integrate into existing wildfire management networks. The BCRFPA provided local knowledge about road conditions, water resources, and fuel conditions and initial attack to fill in gaps in landscape-level wildfire protection. However, the proximity of residential areas to the BCRFPA protection district made decisions about fire suppression more complex by introducing trade-offs between residential and rangeland resource protections. Ultimately, our results indicate that RFPAs can help rangeland human populations better adapt to wildfire risk, but that social fragmentation may challenge RFPA functioning.

A ranch-level model using state-and-transitions models for three ecological sites is used to determine the tradeoffs of providing various ecosystem services. The hypothetical ranch is located in northern Colorado and is based upon area average ecological site characteristics and livestock production practices. Management decisions include stocking rate and brush control. The model includes exogenous factors such as precipitation and fire. The model solves for optimal decisions over an infinite planning horizon using stochastic dynamic programming. Results showthat a ranch cannot provide all ecosystem services in tandemat their highest level, implying that land managers must decide which ecosystems service theywant to provide. Also, it is much cheaper in terms of foregone profitability for a ranch to continue to provide a specific ecosystem service that is already provided by the ranch rather than try to transition the ranch to a new ecological state in order to provide a service currently not provided.

Mechanical cutting and mastication of juniper trees aims to restore grassland habitat by reducing the density of encroaching woody species. However, the associated soil disturbance may also create conduits for invasive species, a risk that must be mitigated by land managers. We characterized herbaceous communities in treated and adjacent untreated areas in a piñon-juniper (Pinus edulis and Juniper monosperma) woodland in northern Arizona 2.5 years after treatment. Untreated plots had 4× the herbaceous cover (82%) than treated plots (21%). Within treated plots, native species cover (19%)was 10× higher than invasive species cover (2%). Furthermore, treated plots exhibited greater plant community variability and diversity than untreated plots, driven by an increase in the diversity of native grasses and non-native forbs. No new recruits were Arizona listed noxious weeds, indicating that, at least in the short term, mastication is not producing invasive species hot spots in this piñon-juniper woodland.

Diet selection and performance of Criollo Chaqueño and Brahman × Criollo Chaqueño were studied in three seasons in the dry forests of the Bolivian Chaco using direct observations. During the dry season (DS) and the rainy season (RS), the Criollo cattle diet consisted of a higher proportion (P < 0.05) of woody plants compared with the crossbreds, while the crossbreds included a higher proportion of grasses. Leaf litter was selected in the DS by Criollos but almost not by crossbreds (P < 0.05; 23% and <1% of diet selection, respectively). Season (P < 0.001) had an overall effect on body weight change with a loss of weight in the DS. However, between the DS and the RS the bodyweight of the Criollos increased (P < 0.05),whereas it did not differ (P > 0.05) for the crossbreds. Other than for the Criollos, the body weight of the crossbreds decreased (P < 0.05) from the DS to the transition period. Especially during the DS, Criollos made more use of forage resources from shrubs and trees and thus, seem to be better adapted to year-round forest grazing compared with the crossbreds.

Wild harvesting is an efficient option for supplying seed to be used for restoring seminatural grasslands. Several methods are currently used to implement wild harvesting, but few controlled experiments have investigated its efficiency regarding seed amount and number of species collected. A harvesting trial was conducted in a species-rich, low-productivity grassland of the calcareous Italian Alps (1 030 m above sea level [a.s.l.). Three mechanical methods were tested in three replications using a completely randomized block design: green hay (GH), dry hay (DH), and seed stripping (SS) harvesting. The number of fertile shoots, mature seeds, and species collected was recorded and compared with the standing seed yield (SSY). GH, DH, and SS harvested approximately 84%, 70%, and 29% of SSY, respectively. Forbs were harvested more efficiently than grasses in all methods but in most cases at very low seed amounts per m² due to their low seed density for SSY. No significant difference among methods was found for the number of species collected as mature seeds, but SS, implemented on larger plots, tended to collect more forb species. Comparing the results with those of other experiments demonstrated that the relatively cool temperature of the seed maturation period at the mountain site favored stronger seed retention and therefore increased the GH and DH efficiency but decreased the SS efficiency. In cool mountain areas, wild harvesting from forb-rich grasslands should bemore successful by SS implemented on wide areas and several times over the vegetative season. In less species-rich grassland, GH and DH can efficiently collect high seed amounts of the fewer species present, even if implemented over smaller areas.

Exclusion of large grazers from rangelands that evolved with significant grazing pressure can alter natural processes and may have legacy effects by changing magnitude or direction of community responses to subsequent disturbance. Three moderately grazed pastures were paired with 12-ha areas with 15 yr of livestock exclusion. Six treatments were assigned to each in a 2 × 3 factorial arrangement of fire (fall fire or no fire) and grazing utilization (0%, 50%, or 75% biomass removal) to determine grazing history effects on rangeland response to subsequent disturbance. Livestock exclusion increased C₃ perennial grass (1 232 vs. 980±50 kg ⋅ ha^-1) and forbs (173 vs. 62± 19 kg ⋅ ha^-1) and reduced C₄ perennial grass (36 vs. 180± 25 kg ⋅ ha^-1) with no effect on total current year biomass. Diversity was greater in pastures than exclosures (H' = 1.5400 vs. 1.3823 ± 0.0431). Every biomass, cover, and diversity measure, except subshrub biomass, was affected by fire, grazing utilization, or both. Contrary to expectations, grazing history only interacted with fire effects for old standing dead material and interactions with grazing utilization were limited to old dead, bare ground, richness and dominance. Fire by grazing utilization interaction was limited to bare ground. Fire reduced annual grass (64 vs. 137±29 kg ⋅ ha^-1), forbs (84 vs. 133 ± 29 kg ⋅ ha^-1), and diversity (H'=1.3260 vs. 1.5005± 0.0537) with no difference in total current-year biomass (1 557 vs. 1 594±66 kg ⋅ ha^-1). Grazing to 75% utilization reduced total current-year biomass (1 467 vs. 1 656±66 kg ⋅ ha^-1) and dominance (0.4824 vs. 0.5584 ± 0.0279). Grazing history affected starting points for most variables, but changes caused by grazing utilization or fire were similar between pastures and exclosures, indicating management decisions can be made based on independent knowledge of grazing or fire effects.

Rotational grazing is sometimes promoted for grassland bird conservation, but the benefits to wildlife have not been comprehensively documented. We examined effects of twice-over rotational grazing on nesting success of grassland songbirds in southwestern Manitoba, Canada in comparison to season-long grazing. We monitored nesting attempts and collected structural vegetation data in 2011 (during a 1/300 flood event), and 2012 (average water levels), for five species of obligate grassland bird species (n = 110) and one shrub-nesting species (n = 41).Nesting analyses were conducted using logistic exposure models. Nesting success was 2.4 to 4 times lower in twice-over grazed pastures compared with season-long grazed pastures, perhaps because of the increased cattle density during the short grazing periods of the twice-over system. Nests protected by shrubs from grazing activities of cattle did not show this pattern. The grazing system did not have an effect on vegetation structure. This suggests that twice-over rotational grazing does not benefit grassland songbirds in northern mixed-grass prairies, and that caution must be taken before implementing this grazing system in areas intended to promote biodiversity conservation.

Despite a shift from yr-round bison grazing throughout the Great Plains before European settlement to extensive seasonal cattle grazing, little is known about ungulate grazing impacts on grassland streams. In this study we 1) determine whether grazing management is a significant driver of grassland streammorphology within the Flint Hills Ecoregion (Kansas, United States); 2) determine if yr-round bison grazing (the precolonial condition) and seasonal cattle grazing (the currently dominant grazing practice in the region) result in distinct stream morphology; and 3) determine if the introduction of cattle into ungrazed watersheds produces significant changes to channel morphology. We use a replicated watershed-scale study design and survey 17 streams across four grazing treatments (ungrazed, long-term bison grazed [yr-round], long-termcattle grazed [seasonal], and newly cattle grazed [seasonal]). Baseline geomorphic surveys were completed in 2010 following consistent grazing management since 1992, and resurveys were completed in 2011 and 2013 to determine short-term grazing impacts. Under the conditions of the experiment, we did not detect significant differences (P > 0.10) in channel morphology or streambed substrate size among grazing treatments following nearly 2 decades of consistent grazing management. Cattle introduction into ungrazed watersheds resulted in modest (P < 0.05) stream widening (0.19 m, 3.9%) following two grazing seasons. Bison grazed watersheds also experienced modest (P < 0.05) stream widening (0.20 m, 5.1%) during the resurvey period. Stream widening from 2010 to 2013 within newly cattle-grazed and long-term bison-grazed treatments indicates that cattle and bison are capable of producing moderate alterations to grassland stream morphology over short time periods. However, longer time periods containing more diverse hydrologic conditions may be necessary to generate larger geomorphic changes between surveys. Although we detected modest changes to streammorphology in response to grazing over short time periods, overall, stream morphology does not vary among grazing treatments in the study area.

Livestock production is important for local food security and as a source of income in sub-Saharan Africa. The human population of the region is expected to double by 2050, and at the same time climate change is predicted to negatively affect grazing resources vital to livestock. Therefore, it is essential to model the potential grazing output of sub-Saharan Africa in both present and future climatic conditions. Standard tools to simulate plant productivity are dynamic vegetation models (DVMs). However, as they typically allocate carbon to plant growth at an annual time step, they have a limited capability to simulate grazing. Here, we present a novel implementation of daily carbon allocation for grasses into the DVM Lund-Potsdam-Jena General Ecosystem Simulator (LPJGUESS) and apply this to study the grazing potential for the Kordofan region in Sudan. The results show a latitudinal split in grazing resources, where the northern parts of Kordofan are unexploited and southern parts are overused. Overall, we found that the modeled grazing potential of Kordofan is 16% higher than the livestock usage reported in the Food and Agricultural Organization of the United Nations, indicating a mitigation potential in the form of a spatial relocation of the herds.

Plant selection pattern and performance of lactating cattle and camels were compared on semiarid savanna rangelands in Kenya in the rainy season (RS) and a transition period (TP) between the RS and the main dry season. It was further evaluated whether supplementation with rumen-degradable protein (RDP) had an effect on these parameters. In both seasons, two cattle types (local ‘Pokot’ cattle and Guernsey × Boran crossbreds) and camels were used, with six females per treatment group (supplemented and nonsupplemented) each (n = 72 animals in total). The experimental periods consisted of 8–10 d of adaptation and 36–40 d of data and sample collection. The diet selected by the cattle types was similar and consisted almost exclusively of grasses. The camel diet consisted mainly of herbs and shrubs with higher contributions of woody plants in the TP than in the RS. Forage from woody plants overall made up a higher proportion of the diet, which was also reflected by a longer browsing time (overall and in the TP) of the supplemented camels compared with the nonsupplemented camels. This result indicates that supplementation of browsers like camels with RDP can be used to increase the intake of forage from woody plants rich in plant secondary compounds, which could be an effective measure for managing rangeland to limit bush encroachment. Overall, no seasonal differences in milk yield were found for the camels and Pokot cattle, but crossbreds had a lower yield in the TP compared with the RS. Overall, the cattle had higher milk fat content than the camels while the camels had slightly higher protein content. Supplementation had no effect on milk yield and composition. The results of diet selection and performance (milk yield) reflect the advantage of camels in arid rangelands.

We investigated the effectiveness of rotational and permanent grazing exclosure periods for improving topsoil quality in three commercial farms devoted to cattle breeding in sodic grassland (halophytic steppe) soils of the Flooding Pampa of Argentina. We compared two plots under continuous grazing (C1-C2) with two plots under more than 8 yr of rotational grazing management (R1-R2) and two adjacent plots under permanent grazing exclosure for more than 8 (E1) and 4 (E2) yr. Periodic and permanent grazing exclosure periods caused significant (P < 0.05) and progressive increases in topsoil organic carbon content and organic carbon stock (0–20 cm; from 24 to 61 Mg ha^-1) as follows: (C1 = C2) < (R1 = R2 = E2) < E1 plots. Topsoil physical properties (bulk density, structural instability, and bearing capacity) and salinity were higher (P < 0.05) in C1 and C2 than in the other plots, while infiltration rate was higher in the oldest exclosure (E1) than in the other plots. Topsoil pH decreased from C1-C2 plots (9.5–9.9) to R1-R2 plots (7.3–8.2) to E1-E2 plots (6.5–7.5), while SAR was highest in C1-C2 and lowest in E1 plots. We propose a conceptual model leading to soil recovery in this halophytic steppe community, triggered by organic carbon accumulation induced by grazing management. Short-time grazing exclusion periods (i.e., rotational grazing) are a plausible and low-cost management option to be recommended to the farmers in this highly restrictive environment.

Predator-livestock interactions are a major concern for both agriculture and conservation globally. Using retrospective survey data from 274 ranches in Wyoming, United States, we used information theory to model how ranch attributes and large carnivores influenced the timing, duration, and severity of livestock predation. We then used constrained ordination to understand how 1) landscape, weather, and animal features influence predation and 2) how livestock behavior and nonlethal loss relate to ranch attributes and large carnivores. Timing, duration, and severity of livestock predation were generally not explained by ranch size or number of counties but were explained by livestock type, livestock parturition (either timing or duration), and documented large carnivore loss. Addition of the large carnivore loss variable to global models always improved Akaike information criterion scores. Rangelands characterized as rough, forested, shrubby, or a public grazing allotment reportedly increased predation risk, in part, due to large carnivore exposure. Approximately two-thirds of participants noticed livestock nervousness if a predator was nearby, half of participants noted changes in livestock distribution patterns, and a quarter of participants noted a reduction in livestock grazing time. Nonlethal losses such as lower weight gains, lower conception rates, lower birth rates, and delayed birth season were reported by 27%, 19%, 12%, and 11% of participants, respectively. Ordination revealed separation between behavioral changes and nonlethal losses, attributed to large carnivore exposure. Parturition relative to livestock type was also strongly correlated to timing and duration of predation for cattle-only operations but not for operations with sheep. The predictive cattle predation-parturition model suggests that for each additional month of calving, producers should anticipate 21 additional d of predation. Understanding predator-livestock interactions relative to ranch and rangeland features, parturition, large carnivore exposure, and losses that extend beyond mortalities can assist in developing novel strategies to mitigate lethal and nonlethal losses.