Ruminants grazing mixed-species pastures face many choices, including when and where to graze and how much herbage to consume. These choices affect not only the nutritional status of the animal, but also sward composition and nutritive value through selective defoliation. Limited research has been conducted in the area of dietary selection and preference, most of which has been limited to simple model systems often involving a choice between only two herbage species. Although these studies have provided a vital tool to allow understanding of the fundamental principles of foraging behavior, in reality, grazing ruminants are faced with more complex situations. Understanding and managing animal preferences in mixed swards and thereby altering dietary selection can result in greater primary (plant) and secondary (animal) productivity. Key issues to improve this understanding include a better linking of behavioral and nutritional studies, a better understanding of the genetic factors influencing diet selection, and the development of more explicit spatial models of foraging behavior that incorporate multiple scales of decision making. This article, as part of a set of synthesis articles, reviews the current state of knowledge and research methodologies related to diet selection of grazing domestic ruminants with particular reference to improved temperate grazing environments, including how well we understand each part of the complex decision-making process a grazing ruminant faces, the links with primary and secondary productivity, and developments in methodologies. Finally, we identify key areas where knowledge is lacking and further research is urgently required. 

Quantifying the effects of watershed improvement efforts is critical to agencies responsible for protecting water resources of the semiarid western United States. A complex water quality data set collected from 1994 to 2004 of upper Muddy Creek Basin was subjected to cluster analysis, discriminant analysis, and canonical correlation analysis to improve understanding of basin fluvial processes and to investigate whether livestock grazing best-management practices (BMPs) improved the water quality of the watershed. Hierarchical agglomerative cluster analysis grouped nine sampling sites into two clusters based on similarity of biological indices, separating the clusters into aquatic communities more and less tolerant of degraded stream conditions. Discriminant analysis yielded strong spatial and temporal distinctions, providing important data reduction by rendering seven key parameters (total dissolved solids [TDS], temperature, elevation, slope, 10-dominant taxa, percent collector-gatherers, and percent Plecoptera) for the spatial variation and four parameters (TDS, dissolved oxygen, total taxa, and community tolerance quotient) for the temporal variation. Canonical correlation analysis identified strong negative relationships among Plecoptera taxa and total taxa with TDS and turbidity in addition to strong positive associations with elevation, slope, and channel substrate weighted embeddedness value. Despite the onset of severe drought midway through the study period, overall reductions of 13% for TDS and a 30% increase in macroinvertebrate total taxa occurred across years, strongly suggesting that improvements in water quality were correlated to BMPs that stabilized stream channels and improved the condition of riparian areas. 

Past research has shown that changes in grazing-resistance traits may be associated with genetic changes in plant populations. Little is known about spatial genetic relationships within plant populations (spatial genetic structure) and any grazing effects on these relationships. Here we present observations of the fine-scale spatial genetic structure in three grass species in semiarid environments (Arizona, Mexico, and Argentina). In each environment, populations of a dominant grass species were sampled from two sites with contrasting livestock grazing histories. Plant genotypes were described with the use of amplified fragment length polymorphism markers. In Arizona, populations of sideoats grama (Bouteloua curtipendula var. caespitosa Gould and Kapadia) differed in that one has never experienced livestock grazing, whereas cattle have grazed the other. In the other two environments, populations exposed to long-term heavy grazing were examined, along with those that experienced either moderate grazing (Mexico, blue grama [Bouteloua gracilis {Willd. ex Kunt} Lag. ex Griffiths]) or extended exclusion of livestock (Argentina [Poa ligularis Nees ex Steud.]). Based on independent analysis of each population, we observed no differences in average gene diversity between populations of each species. With the use of analysis of molecular variance we found slight but significant genetic differentiation between populations with different grazing histories in Arizona and Argentina. Significant genetic structure was present in all populations and indicated an inverse relationship between spatial and genetic distance. Interestingly, this relationship was most pronounced in the cattle-free sideoats grama population, suggesting larger genetic neighborhood areas in the absence of livestock. Less distinct differences in spatial genetic structure associated with grazing history were evident in the other two species. We hypothesize that livestock grazing may lead to increased homogeneity in genetic structure at the landscape scale. Effectively examining this hypothesis presents many experimental challenges. 

Artemisia ordosica Krasch. is a semishrub native to the Ordos Plateau of Inner Mongolia, northern China, and forms a unique and dominant vegetation type in the sandland of the region. To determine the variation of productivity in A. ordosica rangeland, we investigated net primary production (NPP), fine root turnover, soil microbial C (Cmic), and soil organic carbon density (SOCd) on sand dunes differing in mobility (i.e., fixed, semifixed, and shifting sand dunes) in Mu Us sandland. We found that, on an area basis, the NPP, SOCd, Cmic, and fine root turnover rates all increased with increasing vegetation cover. However, the ratios of root NPP to total NPP (RMRN) increased with declining vegetation cover. Total NPP varied markedly among habitats and ranged from 18.3 g m-2 yr-1 for communities on the shifting sand dunes to 293.8 g m-2 yr-1 for communities on the fixed sand dunes; whereas the rates of fine root turnover varied from 0.16 yr-1 to 0.54 yr-1. Our study demonstrated that habitat change in sandland has significant impacts on ecosystem productivity by affecting many related aspects of NPP. From the perspective of biomass production, protection of the semifixed dunes from degradation should be taken as a higher priority than trying to convert shifting sand dunes to semifixed sand dunes; whereas conversion of semifixed sand dunes to fixed sand dunes would appear to be a much easier task than restoring shifting sand dunes. 

Vegetation features radiating from residential areas in response to livestock grazing were quantified for an arid steppe rangeland in the Keshiketeng Banner, Chifeng Prefecture, in northeastern Inner Mongolia in 2004 and 2006. The aim of this study was to estimate grazing impacts on the vegetation dynamics of these historical grazed ecosystems. Grazing intensities were classified as reference area (RA), light (LG), moderate (MG), and heavy (HG) according to the vegetation utilization across the study area. Rangelands were studied along a grazing gradient, where characteristics of plant communities, heights of dominant species, aboveground vertical structures, and belowground biomass were investigated. Along this grazing gradient, vegetation changed from the original dominant plant species Leymus chinensis (Trin.) Tzvel. to a semi-subshrub species Artemisia frigida Willd. when moving from the reference area (RA) to the region around the settlement. Canopy coverage, aboveground productivity, and the number of perennial species declined as one moved toward the residential area. Heights of five dominant species, except for Stipa grandis P. Smirn., declined with increased grazing intensity. Aboveground vertical structure in the RA treatment showed more resilience than the other treatments. There was no difference in root biomass in the top 1 m of soil (P > 0.05) between the RA treatment and the area immediately around settlement (HG treatment). Generally, we found that the intensity of grazing disturbance did not exceed the tolerance of the rangeland ecosystem within LG treatment. However, vegetative conditions in HG treatment became worse with increased grazing pressure. Rangelands in this arid steppe are under tremendous threat due to excessive forage utilization, which cannot be considered a sustainable practice. 

Long-term effects of two mechanical interventions, shallow plowing and harrowing, on degraded Leymus chinensis (Trin.) Tzvel. grassland were studied. Species composition and standing biomass of the grassland were monitored at peak biomass each year for 24 yr after application of these two measures, together with grassland in natural recovery and that under public grazing. Results showed a high resilience of degraded grassland, which recovered naturally after excluding grazing animals to a structure similar to the intact L. chinensis community. In comparison with natural recovery, harrowing facilitated restoration of L. chinensis population and community structure and improved grassland production. Shallow plowing accelerated recovery of L. chinensis population to a larger extent than harrowing and led to a flourish of annual species and improvement of herbage production in the years following its application. But the production improvement was unsustainable and was associated with a decrease in grassland species richness and community complexity. We conclude that the best measure for restoring degraded grassland depends on the restoration objectives and severity of grassland degradation. Harrowing is a feasible technique to assist restoration of the degraded grassland. In contrast, shallow plowing is not appropriate for ecological restoration, but may be applied for quick restoration of herbage production. 

Grasslands of northern China are of great ecological, economic, and cultural importance (Kang et al. 2007). These immense grasslands cover 400 million ha or 40% of the land area of China and stretch 4 500 km northeast-southwest (lat 28 degreesN to lat 51 degreesN). They extend from the northeastern plains adjacent to Mongolia to the southern Tibetan Plateau and consist of four major types: meadow steppes, typical steppes, desert steppes, and alpine steppes (Sun 2005; Kang et al. 2007). Inner Mongolia has 87 million ha of natural grassland, which is a significant constituent of the Eurasian Steppe—the largest contiguous biome in the world (Li 1962, 1979; Wu and Loucks 1992). From east to west, meadow steppe, typical steppe, and desert steppe zones occur in response to the decreasing moisture gradient. 

Optimal distribution of cattle on forested rangelands has long been a subject of concern specifically related to uniform and sustainable use of forage resources. Our objective was to determine if cow age influenced distribution and resource use on forested rangelands. This study was conducted from 1991 to 2001 at the US Department of Agriculture Starkey Experimental Forest and Range, northeastern Oregon, a mixed-conifer forested rangeland. We used 43 039 locations of cattle taken from 1 h prior to sunrise until 4 h after sunrise and 4 h prior to sundown until 1 h after sundown from 15 July to 30 August to evaluate cattle distribution patterns during peak foraging time. Cattle were grouped into four age classes: 2- and 3-yr-old cattle, 4- and 5-yr-old cattle, 6- and 7- yr-old cattle, and cattle > 8 yr old. All age classes preferred areas with gentler slopes (P < 0.05), westerly aspects (P < 0.05), farther from water (P < 0.05), and with greater forage production (P < 0.05) than pasture averages. Cattle older than 3 yr of age selected areas with less canopy closure (P < 0.05) than the mean value for the pasture. Young cows (< 4 yr old) selected lower elevations and steeper slopes than the oldest cows (P<0.05). In summary, cow age and correspondingly its experience directly influences distribution patterns and forage resource use of cattle at the Starkey Experimental Forest and Range. 

Our objective was to determine the short-term response of bluebunch wheatgrass and medusahead to defoliation of wheatgrass designed to stimulate regrowth through tillering. We hypothesized that defoliating bluebunch wheatgrass by 20% at the 3 to 3.5 leaf stage followed by a 50% defoliation at peak standing crop would increase its tillering and biomass production. Consequently, we expected a reduction of the density and biomass of medusahead over that of bluebunch wheatgrass defoliated 50% at peak standing crop. Treatments included four initial medusahead densities (200, 333, 444, 600 plants m-2) created by hand-pulling and three defoliation regimes factorially arranged (12 treatment combinations) in a randomized complete-block design and replicated four times at two sites. In 2006 and 2007, defoliation was accomplished by hand-clipping bluebunch wheatgrass 1) by 50% once at peak standing crop (late June); 2) by 20% at the 3 to 3.5 leaf stage, then again to 50% at peak standing crop (mid May, late June); or 3) plants were not clipped. Density was sampled in 2006 and 2007, and biomass was harvested only at Star Mountain (near Riverside, Oregon) in 2007 because Warm Springs (near Drewsey, Oregon) was burned by a wildfire before final 2007 data could be collected. In 2006, no treatments applied at either site detectably altered the number of tillers produced by bluebunch wheatgrass nor did they affect bluebunch wheatgrass density or biomass in 2007 at Star Mountain. Changes in medusahead density were not detected in 2006, but this annual invasive grass increased in density and biomass in 2007 at Star Mountain in plots receiving two defoliations. The relatively short growing period caused by summer drought and the relative intolerance of bluebunch wheatgrass to grazing make the twice-over grazing an unlikely practice for arid rangelands in the western United States. In fact, it could possibly increase the risk of annual grass invasion. 

It is hypothesized that Utah beef producers in certain locations could intensify private land use via improved forages and irrigation. Although intensification could increase ranch productivity and help compensate for any future restrictions in public grazing, is the approach profitable and sustainable in a dynamic environment? We investigated the efficacy of intensification using linear programming for three size-classes of model ranches. Model solutions maximize returns net of forage costs; outputs include brood-herd dynamics, optimal forage mixes, and net returns. The model is driven by 11-year risk scenarios combining high or low precipitation with high or low beef prices. We then consider current or no access to public grazing—a policy uncertainty. In general, results support the idea that intensification could be profitable, sustainable, and strategically useful under several sets of conditions. Modeled brood- herds expand and contract in response to precipitation. Optimal forage use is dominated by reliance on treated, improved, and irrigated forages. Critical irrigated forages include alfalfa hay and improved pasture. Profitability generally increases with operation size, but when public grazing is eliminated, herd sizes and profitability drop. Small and medium-sized operations respond to loss of public grazing by using more irrigated pasture and alfalfa hay, while larger operations use a wider variety of irrigated and nonirrigated forages. Sensitivity analysis indicates that optimal forage mixes for all operations remain stable even when input costs for fossil fuels double. Further increases in fuel costs, however, begin to reduce the contributions from irrigated pasture and alfalfa hay. Low precipitation (drought) has very large and negative effects on profitability in general. When drought combines with restricted access to public grazing, profitability of small and medium-sized operations drops further while profitability of large operations increases. Empirical research is needed to test model results and examine what the limiting assumptions reveal about real-world production constraints.