NOTE: The notation (su3) means that the data is used in the document analyzing the sustainability of the productivity of the world's food, fiber and water supply systems. (See elsewhere in this website.)

SECTION (2-A) - Grazing Basics - [A1] General, [A2] Aridity Issues, [A3] Short-duration- and Rest-Rotation Grazing, [A4]~ Input/ Output, [A5]~Livestock Mortality, [A6] Range Vegetation Changes, [A7]~Range Soil Changes, [A8] Desertification, [A9]~Riparian Habitats, [A10]~Suitability/ Need for Grazing, [A11] Forested Grasslands, [A12] Microbiotic Crusts, [A13] Holistic Range Management, [A14] Grazing Animal Differences,

A general discussion of over-grazing basics is given in p. 778-787 of Ref. (56D1).

The world's 3 billion cattle, sheep, goats and camels convert lignocelluloses (a main product of photosynthesis that is indigestible to humans) into high-quality protein (89P2). Comments: As noted below, 20 lb. of forage produces one lb. of beef containing about 0.132 lb. of animal protein. Comments: Livestock populations tend to parallel human populations so the "3 million" figure is probably closer to 4-5 million around 2005.

The world's rangelands are the source of nearly 25% of the world's meat (98H2). Comments: Other major sources of meat are fisheries, grain supplied by croplands and pasturelands (Chapter (4-H).).

In the late 1970s, the world's livestock consumed 8.71x10¹⁵ kcal or 1.74 Gt./ year of usable dry organic matter, 75% from pasture (+rangeland??), 17% from grain, and 8.5% from other agricultural products (86V2). Another estimate gives 0.5 Gt. grain (dry wt.) + 0.15 Gt. other agricultural products. If the fraction of livestock food derived from grain is slightly less than 17%, overall consumption of dry organic matter by livestock would be 2.8 Gt./ year (86V2). Pimentel estimates that 3.2 Gt./ year of forage and grain are fed to livestock in developed countries, and 1.8 Gt. in developing countries, for a total of 5.0 Gt./ year (86V2). The quantity of animal products humans consume: 0.15 Gt./ year (apparently excluding hides), so if domestic animals consume 2.2 Gt. of dry plant material/ year, the efficiency of conversion of plant material to human food via animals is 0.15/2.2 = 6.88% (86V2).

Of the estimate 2.2 Gt./ year of NPP eaten by livestock, 1.5 Gt./ year comes from natural and derived grazing lands - about half from each (Wheeler et al, 1981, in (86V2))

Cattle are passive grazers, grazing repeatedly on the same site, rarely venturing far from water. Bison are adapted to the climate, vegetation and topography of the Great Plains and Rocky Mountains. They are twice as efficient as cattle when feeding on native grasses, highly mobile, and venture far from water. They don't require supplemental feeds, e.g. corn and alfalfa (p. 56 of (94O1)).

Although heavy grazing is more profitable than conservative grazing for a few years, in the long term (5-10 years), it generally gives a lower rate of return and increases financial risk (Ref. 18, 29, 42 of (93H1)). Under heavy grazing, livestock are forced to select a diet lower in nutritional quality; they consume less forage, they eat more poisonous plants, and they spend more energy in foraging and other daily activities. These things reduce productivity (93H1).

Maps of Africa, Asia, Australia, North America, and South America showing the location of extremely arid, arid, and semi-arid regions are found in Ref. (70D1). These maps can be measured planimetrically to produce the results shown in the table below. (All areas are in units of millions of km², and include only ice-free land.) Precipitation rates on semi-arid land, arid land, and hyper-arid land are be converted to run-offs in the review of Irrigated Land Degradation.

A National Academy of Science study found that 1/3-2/3 of all rainfall in the Sahel comes from soil- and plant-moisture evaporation (Ref. 15 of (85B1)). Comments: rather than from the oceans. Hence loss of vegetation from over-grazing and loss of water-holding capacity from soil erosion produces less evaporation and hence less rainfall. Less rainfall produces less carrying capacity that, for a constant- or growing herd size, produces more over-grazing and soil erosion - a positive feedback loop.

Much of the western US public grazing lands receive less than 10 inches of rain a year (99R1). Comments: This means it is arid land. It is doubtful that any grazing occurs on hyper-arid land.

Livestock reduce the rate at which water penetrates the soil surface by reducing vegetative and litter cover and by compacting the soil. Hence livestock grazing is associated with decreased water storage and increased runoff. Lower soil moisture (increased water stress) reduces plant productivity and vegetative cover, creating a positive feedback loop that further degrades both plant communities and soil structure. Increased water stress increases tree mortality and fire frequency (97B2).

Ref. (87G1) discusses the Sahel's climate from a historical viewpoint. The history of rainfall in the sub-Sahara is discussed in Ref. (85K1). (The author claims that albedo effects can't explain the 15-year draught (the worst since 1820-40) because "the draught is simply too big to be man-made") (85K1).

The albedo-change hypothesis of Charney et al is challenged, arguing that albedo changes have been much smaller than has been assumed (87G1). Ref. 16 and 17 of Ref. (79S1) give arguments for believing that a decline in surface density of vegetation alters the surface albedo and changes the surface-water budget, thereby inducing changes in the large-scale circulation that feeds back in a positive (unstable) sense by causing further declines in vegetation through reduced precipitation.

Ref. (76K1) discusses the application of the albedo-plant-cover-rainfall theory to the Near-East (Sinai/Negev). Otterman's rebuttal follows the article.

Desertification of the Sahelian region of West Africa may increase regional albedo by as much as 4% (Refs. 38 and 39 of (90S1)). Such changes in radiation balance are likely to affect regional climate, and potentially lead to further decreases in regional rainfall (Ref. 40 of (90S1)). The Sahel issue in relation to both rainfall and soil nutrients are discussed in much technical detail in Ref. (83B1).

Decreases in plant cover causes a decrease in rainfall that may, in turn, decrease plant cover (a positive feedback loop) (74O1), (75C1).

The rate of transpiration of soil moisture in semi-arid grassland is greater than in shrub-land or bare ground (Refs. 20 and 35 of (90S1)). As a result, there is greater cooling of grassland soil by the loss of latent heat. As grassland is replaced by shrubland and bare soil, soil-surface temperature and air temperature increase, even though the albedo of exposed desert soil is greater (Refs. 36 and 37 of (90S1)). Hot, dry soil retards accumulation of organic N in the soil (90S1). Comments: The above statement is also in "Range Vegetation statements" (below).

(NOTE: Also see [13] Holistic Range Management)
A study of short-duration grazing concluded (89B1):

• It results in a decline in individual animal performance;
• It doesn't improve diet-quality for grazing animals;
• Dramatically increased stocking is not feasible;
• Stocking rate is still the primary factor in grazing success or failure;
• It does not improve range condition at the same or higher stocking rates, compared with continuous grazing;
• It does not increase grass- or forbs standing crop;
• It produces no positive influence on germination or establishment of seeded- or native plants;
• It results in soil compaction,
• It doesn't work, i.e. stocking rate is the major factor affecting the degradation of range land resources

Ref. (79F1) claims rest-rotation grazing is not yet tested by research investigations (p. 106 of (79F1)). Rest-Rotation grazing is described and defended in (70H1) and (82R1).

• Studies that examine ecological impacts of winter-only grazing or rest-rotation grazing vs. no grazing (a list as of 9/00)
• Edgar F. Kleiner, K. T. Harper, "Soil Properties in Relation to Cryptogamic Ground-cover in Canyonlands National Park", Journal of Range Management, 30(3) May 1977) pp. 202-5.
• Dee Galt, Greg Mendez, Jerry Holechek, James Joseph, "Heavy Winter Grazing Reduces Forage Production: An Observation", Rangelands, 21(4) (8/99) pp.18-21. (The comparison is to year-long grazing.)

Eckert and Spencer (87E2) concluded that season of use has not been found to compensate for heavy grazing at any time of the year (97B3).

Ortega et al (97O1) concluded "stocking rates a have a more significant impact on phytomass [plant biomass] than grazing systems" (97B3).

Sauer (78S1) found that standing dead litter is beneficial to bluebunch wheatgrass, which declines with overuse in the winter (97B3).

Taylor et al (97T1) concluded, "rotational stocking was not able to sustain initial species composition at any stocking rate tested" (97B3).

Clary and Webster (89C2) concluded: "recent information on grazing uplands suggests that although conventional grazing systems have great intuitive appeal, they are less effective at maintaining ecological quality and livestock production than previously thought" (97B3).

Hart et al (89H1), (93H2) found that proper stocking rates were more important than grazing systems in improving rangeland vegetation (97B3).

Mueggler (75M1) found that Idaho fescue and bluebunch wheatgrass required 3 and 6 years of rest, respectively, to recover from grazing, a longer period than any grazing system permits (97B3).

In a review of grazing systems, Herbel (74H2) concluded that grazing systems did not result in an improvement in range condition and that stocking rate is likely the overriding factor in determining whether a grazing system works (97B3).

Pieper and Heitschmidt (88P1) concluded that "stocking rate is and always will be the major factor affecting degradation of rangeland resources" (97B3).

Plant protein yield from US pasture and rangeland is 5.34 tonnes/ km²/ year (80P1).

Energy issues related to production of plant- and animal protein are discussed in Ref. (80P1).

In the US, 37 million tons/ year (33.6 million tonnes/ year) of plant protein are fed to livestock to produce 5.4 million tons (4.91 million t.) of animal protein for human consumption. The plant protein source is 14.8 million tons (13.45 million tonnes) from grains, 20.2 million tons (18.36 million tonnes) from forage, and 2 million tons (1.8 million tonnes) from miscellaneous plant- and animal by-products (80P1).

FAO (1983) estimated that 0.5 Gt. of grain + 0.15 Gt. (dry weight) of other agricultural products, were fed to livestock yearly (globally) in the early 1980s (86V1).

Pimentel et al (1980) estimated that 3.2 Gt./ year of forage and grain are fed to livestock in developed countries, and 1.8 Gt./ year in "developing" countries (86V1).

Wheeler et al (1980) estimated that, in the late 1970s, the world's livestock consumed 8.7x1015 kcal, or 1.74 Gt. of usable dry organic matter/ year (0.79 Gt. of carbon) - 75% from pasture, 17% from grain, and 8.5% from other agricultural products (86V1).

It takes 7 kg. of grain to produce 1 kg. of beef. The conversion is 4:1 for pork, and 2:1 for chicken. This is perhaps why beef production has stagnated in recent years, and pork- and chicken production has surged ahead (98H2). Comments: The Soils Degradation review document has a more complete table that includes the ratio of live weights to dressed weights for various meat animals.

38% of the world's grain production is fed to animals (Gary Gardner, Chapter 5 of State of the World 1996).

Over 20 lb. of herbage are needed to produce 1.lb. beef (p. 33 of (91J1)). 7.5 lb. of grain, fed to cattle, produce 1.lb. beef (85B2).

The average efficiency of conversion of plant materials into human food by livestock is 6.8%, i.e. (animal products humans consume as food) = 0.068 x (plant biomass livestock consume) (86V1).

About 40-50% of the weight of an average 800-lb. cow becomes beef (p. 351 of (91J1)).

An average cow consumes 700-800 lb. of vegetation/ month. An average US range steer eats 12,000 lb. of range plant material + 2850 lb. of feedlot food by slaughter time (86J1), (91J1). Sheep eat 20% as much; goats eat 75% as much as sheep (91J1). Comments: This is probably US-, not global, data.

Livestock mortality from causes other than predators is 10%/ year (e.g. toxic plants, hunters, poachers, disease, road-kill) (p. 347 of (91J1)).

(Juniper) The current expansion of western juniper differs significantly from prior expansions (1600, 850, and 200-400 years ago) in that this time, expansion is occurring under increasingly xeric (water-deprived) conditions. Also densities are higher (94M1).

(Juniper) Prior to introduction of livestock into the Northwest in the late 1800s, western juniper was confined by recurrent fires and competitive interactions with herbaceous species to rock outcrops, shallow soils on fractured bedrock and deep pumiced soils (78D2), (78M3), (93B2).

(Juniper) Ref. (96B2) concludes that most of the available evidence suggests that western juniper has few, if any, negative effects on water infiltration or soil erosion.

(Juniper) Ref. (96B2) contends that there is little scientific evidence supporting the belief by ranchers, rangeland managers and range scientists in the Pacific Northwest that western juniper reduces water infiltration, dries up springs and streams, increases erosion, reduces biodiversity and reduces the quality and quantity of forage for livestock and wild species.

(Juniper) Ref. (79F1) argues than pinon-juniper manipulation ("chaining") is a highly questionable practice that needs to be carefully investigated. The real issue is argued to be poor grazing practices rather than water-consumption by pinon-juniper (79F1). Programs like brush-control do little to recover over-grazed range, and can accelerate rangeland deterioration (p. 40 of (94O1)).

(Juniper) Juniper control often causes weed explosions due to nutrient enrichment from decomposing tree roots (85E2) in (97B3).

(Shrubs) Net primary productivity (NPP) is similar in native grasslands and invasive shrub communities of southern New Mexico (Ref. 26 of (90S1)). Shrubs, however, lower the economic potential of the landscape as rangeland relative to grass-cover (90S1). Comments: This statement is also in Section (2-B).

(Sagebrush) Regeneration after severe disturbances on the western grass-sagebrush region is slow. 40+ years are required for reestablishment of sagebrush communities on abandoned cropland in eastern Wyoming (81T1).

(Effects of vegetation on hydrology) The rate of transpiration of soil moisture in semi-arid grassland is greater than in shrub-land or bare ground (Refs. 20 and 35 of (90S1)). As a result, there is greater cooling of grassland soil by the loss of latent heat. As grassland is replaced by shrub-land and bare soil, soil-surface temperature and air temperature increase, even though the albedo of exposed desert soil is greater (Refs. 36 and 37 of (90S1)). Hot, dry soil retards accumulation of organic N in the soil (90S1). Comments: The above statement is also in "Aridity Issues" (above).

(Effects on soil Chemistry) Losses of soil Nitrogen (N) that result from transition from grassland to bare soil may result in an environment that favors N-fixing shrubs such as mesquite which augments local fertility (Ref. 32 of (90S1)).

(Biodiversity Changes) Reconstructed prairie hosts, at most, 10% of the 150-200 plant species typically found on virgin grassland (Stephanie Simon (Los Angeles Times) in Pittsburgh Post Gazette, 5/30/99).

(Exotic Species) Introduced weeds damage western ecosystems by:

• Increasing fire frequency; ((90B2) in (99G1))
• Reducing biodiversity; ((96R1) in (99G1))
• Reducing wildlife habitat; ((92B2) in (99G1))
• Disrupting nutrient cycling, hydrology and decomposition;(90V2 in (99G1))
• Increasing topsoil loss; ((89L1) in (99G1))
• Altering soil microclimate; ((84E1) in (99G1))

Weed control chemicals: (99G1)

• Picloram (kills non-target plant species, is persistent in the soil, and leaches into groundwater and streams),
• Dicamba (can leach into groundwater and streams and causes a variety of cancers and genetic damage),
• glyphosate (acutely toxic to many species, including beneficial insects, fishes, birds, and earthworms, is persistent in soils, and significantly impacts non-target plant species)

("Weed" species) In discussing invasive and noxious weeds in the interior Columbia Basin, Belsky (97B3) noted that:

• Most of the invasive weeds are "pioneer" species that colonize soil surface disturbances, such as those created by the trampling of soils and microbiotic crusts by livestock hooves;
• A few noxious weed species can invade relatively undisturbed sites, making it critical that effective preventative measures be used to stop invasions or reduce weed abundance in surrounding lands;
• The very low densities of noxious weeds on reserves and lands with healthy, vigorously growing bunchgrasses and intact microbiotic crusts demonstrate that healthy rangeland communities are our best, and maybe our only, tools for fighting the invasion and spread of weeds.
• "Noxious weed control on BLM- or USFS-administered lands has generally been ineffective", mainly due to limited budgets (weed control is expensive), lack of coordination, and an inability to get ahead of the weed problem; and
• "The least expensive, most effective, and the highest priority weed management technique is prevention"

(Weed eradication) Roger Sheley (94S1) (97B3) emphasized that:

• "Preventing the introduction of rangeland weeds is the most practical and cost-effective method for their management. Prevention programs include such techniques as limiting weed seed dispersal, minimizing soil disturbance, and properly managing desirable vegetation."
• "Once the infestation becomes established (in riparian areas), eradication is unlikely";
• "Minimizing soil disturbance by vehicles, and livestock is central to preventing weed establishment"

("weed" species) Most grass lands and savannas are located in semi-arid areas where heavy grazing destroys the ability of plants to resist drought, and leads to eventual loss of palatable species in favor of "weed" species (Ref. 3 of (81C2)).

("Weed" species) Noxious weed control on BLM- or USFS-administered lands has "generally been ineffective" (97B3).

(Weed eradication) The overall consensus of weed professionals (e.g. the individual authors of reports in Sheley (94S1)) is that the best method for preventing weed invasions is by:

• Limiting weed seed dispersal,
• Minimizing soil disturbances,
• Managing native grasses so they remain vigorous and competitive,
• Establishing competitive grasses where they have been lost (97B3).

(Grazing effects) Trlica et al. (77T1) found that most species required more than two years of rest after defoliation during most seasons of the year (97B3).

(Grazing tolerance) Unlike grasses of the Great Plains, which evolved under thousands of years of intense grazing by American Bison, bunchgrasses of the Intermountain West were only lightly grazed and trampled. Consequently, these species evolved little tolerance of intense grazing and trampling, causing them to be highly sensitive to actions of introduced cattle and sheep. As a result, within 20-30 years of the beginning of intense livestock production, many western rangelands were severely damaged and completely denuded (72Y1), (91H1) in (99G1)).

(Exotic species) Exotic species are the second main cause, following loss of habitat, of species extinction in the US (94F1), (94F2) in (99G1)).

(Juniper Eradication) Spraying at the level necessary to kill most of the weeds also kills native plants and many soil organisms necessary for nutrient cycling (conclusions of several speakers at a conference on Rangeland Restoration, Washington State Univ., 3/24-25/97) (97B3).

Ref. (70L1) cites several studies of soil erosion rates in arid lands, and gives rates in the range of 0.5-1.5 mm. soil/ year (730-2190 tonnes/ km²/ year). On the border of the desert zone, destruction of natural vegetation leads to intense wind erosion (several cm/ year on sandy soil) (3 references are cited in support of this.) (1 cm. of topsoil weighs 14,600 tonnes/ km²) (70L1). Comments: Arid land soils are low in organic matter and hence are erosion-prone.

Studies by the USDA-SCS found that range land with good palatable grasses and good vegetation-density lose topsoil at under 0.5 ton/ acre/ year (under 112 tonnes/ km²/ year). Rangelands with a mixture of weeds and grasses, and with medium vegetation-density lose topsoil at under 2.5 tons/ acre/ year (under 560 tonnes/ km²/ year). Rangeland where annual weeds make up most vegetation, and which have low vegetation-density, lose topsoil at 7.5 tons/ acre/ year (1680 tonnes/ km²/ year) (p. 89 of (71R1)).

On prairie with good sod on a 10% slope, no topsoil loss was measured after a 2.5" rain. Over-grazed land experienced a 0.1 ton/ acre soil loss, while bare soil experienced 3.4 tons/ acre soil loss for the same 2.5" rainfall (p. 89 of (71R1)).

Trampling by livestock compacts soil, reducing water infiltration rates ((78M1), Ref. 25 of (90S1)). Increased runoff increases erosion and transport of water, N, and other plant nutrients. The net effect is reduced availability of soil moisture and nutrients in the landscape and increased heterogeneity of their horizontal distribution (78M1), (90S1). Increased grazing intensity increases water-runoff (Refs. 2, 21, 27, 28, 30, 31, 39, 46, 51, 54, 56 of (78M1)). The primary causes are soil compaction and the resultant reduction in infiltration rates, as well as cover-depletion (78M1).

As grassland soils erode in arid regions, CaCO₃ is exposed at the surface. This promotes volatilization of NH3 (90S1). Ammonia volatilization, denitrification of soil organic matter and wind erosion are the main contributors to losses of N from desert soils. All these processes are likely to increase when grasslands are converted to desert shrub-lands (90S1). On semi-arid soils, bursts of denitrification presumably occur during moist conditions that prevail after rain storms (Ref. 57 of (90S1)). In semi-arid grasslands, competition with grasses for available nitrate and rapid plant-uptake of soil moisture probably limit denitrification. When shrubs replace grass, and greater over-land flow transports NO₃ to basin depressions, greater levels of denitrification are expected (90S1).

Eroded rangelands are particularly difficult to restore because fertilization usually is not economical in low-rainfall areas (Ref. 86 of (81N1)).

Major symptoms of desertification (81S1):

• Declining groundwater tables;
• Unnaturally high soil erosion;
• Salinization of topsoil and water;
• Loss of native vegetation,
• Reduction of surface water

Climate has not changed significantly in the past 2000 years (78D1). Comments: The basic contention being made here is that desertification is not due to long-term decreases in rainfall.

See the desertification map of the world in Ref. (78D1).

When net long-term desertification of productive grasslands occurs, a relatively uniform distribution of water, N, and other soil resources is replaced by an increase in their spatial and temporal heterogeneity. This leads to invasion of grassland by shrubs. In these new plant communities, soil resources are concentrated under shrubs, while wind and water remove material from inter-shrub spaces and transform soil materials to new positions on the landscape (Ref. 10 of (90S1)).

Worldwide, more than 23,000 square miles/ year (59,600 km²/ year) become desert (Seattle-PI-Com Desertification, 11/23/99).

Cattle spend a disproportionate fraction of their time in riparian habitats (See 6 references listed in Ref. (94F2)). Comments: Bison do not do this.

In the Great Basin of southeastern Oregon, more than 75% of all wildlife species are dependent on, or use, riparian habitats (90C1).

In southeastern Wyoming more than 75% of all wildlife species depend on riparian habitats (90C1).

In Arizona and New Mexico, 80% of all vertebrates depend on riparian areas for at least half of their life cycles: more than half of these are totally dependent on riparian areas (90C1).

Riparian areas provide habitat for more species of birds than all other western US rangeland vegetation types combined. More than half of all bird species in the southwestern US are completely dependent upon riparian areas (90C1).

(Hydrologic Effects) Proper functioning riparian areas store water, reduce flooding and provide late-season flows (99W1).

(Biodiversity) As many as 80% of wildlife species in Arizona and New Mexico ((90C1) in (99B1)) and in southeastern Oregon ((79T1) in (99B1)) are dependent on riparian habitats.

(Biodiversity) Though riparian areas make up only 1% of the landscape, they shelter and feed 60-80% of species in the western US (99W1).

(Grazing Effects) Cattle spend 5-30 times as much time in riparian areas than would be predicted from surfaced area alone (82R2), (84S1) in (99B1)).

(Grazing Effects) Cattle evolved in moist woodlands in Eurasia, and are not well adapted to arid landscapes. They use more water than bison, spend more time in riparian areas, and have been bred for lack of mobility (99W1). Cattle evolved in cool, wet meadows in northern Europe and Asia (99B1). Comments: This is seen as an explanation of why cattle have a much stronger affinity for riparian areas than buffalo.

(Grazing Effects) Ref. (97H1), referring to work by Refs. (89C2), (89P3), and (94E3), state that, in riparian habitats:

• The three-pasture, rest rotation grazing system promotes shrub browsing that can exceed growth during rest years;
• Three-pasture, deferred rotation inhibits growth of woody species;
• Early rotation damages herbaceous species;
• Rotational grazing leads to declining woody species; and
• Short duration grazing damages stream banks, shrubs, and herbaceous regrowth;
• Season-long grazing, spring and fall grazing, and spring and summer grazing do not provide for successional advancement of riparian vegetation. (97B3) Ref. (97H1) contradict these conclusions by stating that winter grazing, early growing season grazing and early rotation are acceptable because they lead to increases in shrubs and herbs, and stabilization of stream banks (97B3).

(Grazing Effects) Clary and Webster's 1989 summary (89C2) (see 97B3) of research on different riparian grazing systems concluded, "Experience in riparian areas has generally failed to show an advantage to any specific grazing system". They also concluded that "the level of utilization [i.e. the percentage of the current year's crop consumed] to be the most important consideration". Platts, (89P3) (see (97B3)) recommended:

• Use of riparian pastures (which strictly limits riparian grazing),
• Streamside fencing,
• Increased periods of rest, and
• Reduced intensity of use.

These are the best ways to restore degraded streams. Reduced levels of use and non-use were therefore his main recommendations for restoring streams (97B3).

(Riparian productivity) Streams make up 1% of the landscape in the 11 Western states, yet 70-80% of plants and animals depend on them for survival. Unlike elk or bison, which eat and roam, cows love to congregate in streams, especially in summer (99R1).

(Riparian Productivity) One study found that a riparian zone in eastern Oregon comprised 1.9% of the grazing allotment by area, but produced 21% of the available forage and 81% of the forage consumed by cattle ((82R2) in (99B1)).

(Riparian Productivity) Riparian zones typically represent 2-5% of a grazing allotment, but may supply up to 80% of forage used by cattle (94O1). Over-grazing of riparian areas causes compaction of soil, reduced water infiltration, and reduced vegetation cover (94O1).

(Riparian recovery) Ref. (97H1) admitted that exclusion of livestock for two or more years was necessary for initiation of recovery of degraded streams (see (97B3)). However, exclusion of livestock was not recommended in any of the alternatives, not even for non-functional streams or to improve highly sensitive riparian areas such as critical spawning areas for sensitive fish species. Instead, Hann et al. (97H1) repeatedly, and contrary to scientific evidence, stated that reduction in livestock numbers will typically have little positive effect on riparian systems. To support this, they cited a paper finding that salmon spawning was enhanced by the presence of livestock. This last paper was never published in a refereed journal and was severely criticized by top fisheries biologists who were studying the same stream segment (97B3).

(Grazing system validity) Refs. (78M2) and (84P1) (see (97B3)) found no grazing system that was compatible with healthy aquatic ecosystems.

(Grazing system validity) The only grazing system consistently found to restore and protect riparian habitats at an acceptable rate is "no grazing" and "corridor fencing" (94E3), (96O1), Joy Belsky et al. (in press) and most fisheries biologists) (97B3).

(Grazing effects on riparian habitats) Many studies on riparian habitats in the range science literature such as Refs. (94E3) and (89C2) report that spring-time grazing leads to loss of native bunchgrasses and serious damage to wet soils and stream banks; summer grazing leads to loss of woody species, loss of plant vigor, and loss of seed crops; autumn grazing leads to loss of woody species, long-term stream bank disturbance, and loss of vegetative cover needed to protect riparian soils; and winter grazing can lead to loss of woody species, compaction and disturbance of wet soils, and loss of litter needed to protect soils (97B3).

(Grazing impacts on exotic weeds) In the Intermountain West and Great Basin, exotic weeds have been able to displace native species partly because native grasses there are not adapted to frequent and close grazing (97B2). Comments: The large buffalo herds were on the Great Plains -not the Intermountain West and Great Basin.

Regarding bison and shrub steppe and Allan Savory's argument that it must be grazed by large herds of ungulates to survive, see Richard N. Mack and John N. Thompson, "Evolution in Steppe with Few Large, Hooved Mammals", The American Naturalist 119(6) (1982) pp. 757-73.

A study shows that Great Plains grasslands are a grazing-dependent ecosystem and need grazing to maintain their character, diversity, and vigor. This study involved the long-term resting of mixed grass prairie in the Coteau region of North Dakota and was conducted by Brian Martin in 1994. He studied an area that the Nature Conservancy had acquired. The Conservancy had grazed a portion of the area and allowed the remainder to rest for approximately 10 years. The same phenomenon that Mike Brand showed on sites with 30 years of rest occurred on the Conservancy land with only 10 years of rest, invasion by non-native species (mainly Kentucky Blue Grass) occurred on the ungrazed site. The grazed site retained native species. Litter or mulch made is difficult for native species to exist on the ungrazed site. (kirk.koepsel@sfsierra.sierraclub.org 8/17/99) Comments: Great Plains grasses evolved with large herds of buffalo. The Intermountain West did not (See Section (4-D)), so it is unlikely that this result applies to any grasslands other than those on the Great Plains.

Changes in species composition and forest dynamics in the dry and moist forests of the Interior Basin have been attributed to fire suppression and selective logging. However numerous studies show that these changes to Eastside forests began decades before fire suppression or selective logging was introduced. ((51R1), (83M1), (84Z2), (90S3), (96B1) in (97B3)). Comments: The authors argued that these changes were a result of grazing.

Microbiotic (cryptogamic) soil crusts are delicate symbioses of cyanobacteria, lichens and mosses. These crusts perform the major share of nitrogen fixation in desert ecosystems (78R1). The availability of nitrogen in the soil is a primary limiting factor on biomass production in deserts. In the Great Basin Desert of the US, it is secondary in importance only to the lack of moisture (78J1). Microbiotic crusts in arid ecosystems have been correlated with increased organic matter and available phosphorus (77K1), increased soil stability (72K1) (78R1) and increased soil water infiltration (72L1) (78R1). Crusts also play an important role in ecological succession because they provide favorable sites for the germination of vascular plants (84S2).

Observations of recovery of crypto-biotic crusts from trampling by livestock at three sites in Utah yielded estimates for full recovery of 45-85 years. Moss recovery was much slower than that of the lichens. At two of the three sites where mosses were found, no moss recovery at all was seen. At the third site, where some recovery was seen, full recovery of moss cover would take over 250 years at the observed rate of recovery (93B3).

Prior to domestic livestock introduction common ungulates were small (e.g., pronghorn antelope versus cow/bison, 70 vs. 500 kg.) and/or present in low numbers; their localized trampling damage could be tolerated even by communities ill equipped to cope with such disturbance. It appears that herbivorous mammals are incompatible with maintenance of steppe where cryptogams (particularly crustose lichens) occupy a significant fraction of the soil surface ((82M3): p. 764).

Living crusts of lichens, mosses, and algae and cyanobacteria blanket exposed soils in deserts, dry grasslands and shrubland around the world (99G1).

Living crusts of lichens, mosses, algae and cyanobacteria increase soil stability, enhance soil fertility, increase plant-nutrient content, often enhance water infiltration and water-holding capacity, and contribute to mycorrhizial colonization ((88H1), (96L1) in Ref. (99G1)).

Cyanobacteria in these crusts also fix atmospheric nitrogen ((93E1) in (99G1)) and may be the main source of N input into desert- and semi-desert ecosystems (93E1), (99E1) in Ref. (99G2).

There is evidence that intact microbiotic crusts reduce or prohibit weed establish by preventing weed seed germination ((42M1), (87H1), (86E1), (89M1) in Ref. (99G1)).

Ref. (97B3) notes that 100+ scientific studies demonstrate the importance of microbiotic crusts in arid ecosystems.

In steppe environments, where they are so critical to the function of the entire ecosystem, microbiotic crusts normally require 60-100 years to re-establish once they have been destroyed (Claude Treanor, Keating Highway Allotment #2108 Allotment Evaluation, 2/6/87. Contact Vale Oregon District, BLM.).

The ecosystem component perhaps most critical to healthy rangelands, and the one most disturbed by livestock, is microbiotic crust. There is a broad consensus among botanists and range scientists that these fragile, living covers of the soil surface play a vital role in arid and semiarid rangeland ecosystems ((93S1), (88H1), (94W1), (94K1) in (97B3)).

The most important role of microbiotic crusts is considered to be:

• Stabilized soil surfaces and improved soil fertility;
• Increased herbaceous productivity,
• Reduced weed invasions,
• Fixes nitrogen from the air, and
• Increased plant nutrient content (97B3).

1. Improve water infiltration into the soil as a result of hoof action.
2. Increase mineral cycling.
3. Reduce the percentage of non-grazed plants.
4. Improve livestock distribution (more uniform use of the range).
5. Increase the period when actively growing forage is available to livestock.
6. Accelerate plant succession.

Researchers at 13 locations in North America have attempted to evaluate the validity of Savory's ideas. Below is a summary of findings.

Two comprehensive reviews (87S1) (92O2) of over 50 grazing experiments that consider African experiences with short-term grazing both concluded that:

According to Vicari and Bazely (93V1), "there is little evidence that the act of grazing per se increases the fitness of grasses or any other plant species, except under highly specific circumstances".

SECTION (2-B) - Grazing Land Productivity - [B1] Net Primary Production, [B2] Other Estimates of Grassland Productivity, [B3] Grassland Burning, [B4] Wild Herbivore Productivity, [B5] Productivity Dependence on Precipitation, [B6] Analysis: Livestock Carrying Capacity of the World's Grasslands,

Net primary productivity (NPP) is similar in native grasslands and invasive shrub communities of southern New Mexico (Ref. 26 of (90S1)). Shrubs, however, lower the economic potential of the landscape as rangeland relative to grass-cover (90S1). Comments: This statement is also in Sect. (2-B).

The world's grazing lands have an average carrying capacity of 5 AU/ km² (PSAC, 1967). Production is under 500 kg. meat/ km²/ year (78B2). Comments: This number must be in error. Since the global average grazing intensity is about 60 AU/ km², it implies that the average grazing land is over-grazed by a factor of about 12. No data suggest an over-grazing factor of more than about 4.

Another estimate of pastureland productivity: 840 tonnes/ km²/ year (presumably dry organic matter) (Hingane, 1990) (91J2). Comments: "pastureland" is usually taken to be grassy areas in humid regions, not in semi-arid or arid regions where grassy areas are typically called "rangelands". The productivity of pasturelands is typically far greater that that of rangelands.

Values of Net Primary Productivity (NPP) of grasslands (See table above) comes from Whittaker and Likens (73W1) whose results for all global ecosystems are summarized below. Comments: Whittaker/ Likens data tend to come from more natural, undisturbed areas and, possibly for that reason, tend to be higher than all of the 6-8 other similar NPP studies.

A number of other estimates have been made, and their plant carbon mass results are summarized below.
Values for the global phytomass carbon pool size (Gt. C) (87E1):

• 550 [Early estimate based on production and residence time (Schroder, 1919)]
• 300 [Estimate using production and stand age (Muller, 1960)]
• 560+100 [Vegetation information on 0.5-degree grid; considers agriculture (Olson et al, 1983);]
  560 [Considers agriculture (Ajtay et al, (79A1));]
• 592 [Considers agriculture (Duvigneaud in Bolin et al, 1979)]
• 827 [Upper limit for potential natural vegetation. Based on a high production estimate of 72 Gt. Carbon/ year (land+water) (Whittaker and Likens, (73W1));]
• 638 [Natural vegetation only (Aselmann, 1985);]
• 737 [Considers agriculture (Matthews, 1984);]
• 594 [Considers deforestation (Goudriaan et al, 1984);]
• 759 [Potential natural vegetation (Esser, 1986);]
• 657 [Considers agriculture (OBM) (Esser, (87E1)).]
• 658 [Average of the above 9 studies later than 1970]

Comments: Recent studies of forest biomass show that Whittaker and Likens (73W1) significantly over-estimate global tropical-forest biomass. It seems probable then that computing grasslands carrying capacity based on Whitaker-Likens data is likely to produce a significant over-estimate of carrying capacity - perhaps by 20% or more. 

(Rainforest) After 7-10 years of beef cattle grazing, the effects of overgrazing and torrential rains turn a rainforest's nutrient-poor soils into eroded wasteland (Ref. 8 of (83N1)).

(Rainforest) Stocking on newly cleared rainforest land is typically 100 AU/ km² during Year 1, and 14-20 AU/ km² during Years 5-10 (83N1).

On some semi-desert grasslands in Cochise County AZ, as few as one grazing cow per 50 acres is considered ecologically sound (13 cows/ sq. mile) (85C1).

Natural fires in grasslands are believed to occur about every 1-3 years in humid areas ((85F1), p. 232) and every 1-20 years in dry areas ((85W1), p. 85). Today the number of natural fires is insignificant relative to the number of fires caused by humans (99L1). 5 million km² of tropical and subtropical savannas, woodlands and open forests now burn annually (Goldammer, 1995, cited in (99L1)).

Seiler and Crutzen (1980) estimate that grassland is burned at 6 million km² of savanna yearly, mostly by human-caused fires. Aboveground biomass in the herb-grass layer of the burned area is 1.8-2.9 Gt. organic matter (Divide by 2.2 to get carbon mass.). 75% is consumed in the fires (86U1).

Hav et al (1990) estimate that, globally, savannas are burned at the rate of 7.5 million km²/ year. Half of this burning is in Africa (75%/ year?) (91A1).

African savanna burned yearly = 10 million km². Savanna biomass = 100 tonnes/ km² to 2500 tonnes/ km² of dry organic matter (Menaut, 1979) (91D3).

The area under permanent pastures and other grazing grounds in India is 122,000 km². This is burned annually. Biomass burned: 198 tonnes/ km² (91J2). Comments: Burning improves grazing. Apparently the ash is fertilizer.

The concentration of large herbivores (kg/ km²) on African savanna is plotted vs. annual precipitation for three soil-nutrient-availability groups in Ref. (93F1). Both pastoral- and natural sites are plotted. Herbivore biomass per unit-area of range is plotted vs. net primary productivity of the range for natural- and agricultural systems in South America. Agricultural stocking densities are about 10 times natural stocking densities (92O1).

Dregne and Chou examined degradation-induced percentage losses of land productivity in the world's dry regions (Ref.17 of (97C1)). Using FAO data they found that 51 million km² of land fell into the "dry" category (39% of the earth's ice-free land). 85% of this land was used at rangeland (43.3 million km²), 9% as rain-fed croplands, and 3% as irrigated croplands (97C1). Comments: In the Irrigation Degradation review, an analysis finds 22.09 million km² of semi-arid land, 22.18 million km² of arid land, and 5.86 million km² of hyper-arid land. Total (hyper-arid+ arid+ semi-arid) = 50.13 million km². Since it is doubtful that any hyper-arid land is grazed, it appears that 85% of rangeland is semi-arid or arid, so (15/85)*43.3 = 7.6 million km² of rangeland is humid or semi-humid land.

Grazing more than 50% of a grass quickly curtails root growth and severely limits future yields (US Utah extension agents James Barnhill, 801-399-8208 and Dean Miner, 801-370-8469, utah@ext.usu.edu, 1997).

Average precipitation over the US and Alaska is mapped on p. 20 of (80U1). Comments: The capacity of grazing land is strongly dependent on rainfall or access to water (as illustrated by the high productivity of riparian habitats) as will be seen throughout this document.

In the US Great Basin, all riparian lands cover fewer than 2% of the land area, yet receive 50% of livestock pressure. Riparian meadows occupy 1-2% of the interior northwest, but account for 81% of forage removed by livestock (p. 95 of (91J1)).

Grazing provides 64% of feed consumed by US beef cattle, and 79% of feed consumed by US sheep (p. 275 of (80H1)). Comments: Grain allocation to livestock has been trending upward, so this figure may be quite obsolete.

An average of 13.7 acres (0.055 km²) is needed for an AUM (AU??) (18 AU/ km²) for all public lands within BLM grazing districts in the Western US. This figure ranges from 6.1 acres in Montana to 21.8 acres in Nevada (88D1).

21.8 acres/ AUM are needed on BLM lands in NV (261 acres/ AUM if grazed year-around) (89W1) (GAO estimate). Comments: This implies a capacity of 11.3 AUM/ km² or 0.95 AUM/ km² for year-around grazing (11 or 1 AU/ km²? - interpretation of information is difficult.).

Range grazing in the contiguous US in 1976: 217 million AUMs, a little over a third of biological potential (80H1).

The ultimate biological potential production from US rangelands is estimated to be 566 million AUMs under intensive management. The 1976 supply level was 213 million AUMs. If one assumes that only the range currently being grazed is available for intensive management, herbage and browse could potentially be increased from 169 million tons/ year to 220, and range grazing could be increased from 127 million tons/ year to 247 (by upgrading poor, fair, and very poor range to "good" range) (p. 295 of (80H1)).

Africa (northern): Productivity of browse plants: 150 kg DM/ km²/ mm rain/ year, of which 50% is actually consumed (80L1). Comments: This implies the absence of grass, i.e. a fairly desert-like environment.

Asia (Soviet Central): In sandy deserts, 6-7 km² of grazing land are required per 100 sheep (Capacity = 3 AU/ km²) (70P1).

Australia: Arid soils are formed on, or from, erosion products of ancient, deeply weathered land. Thus they differ from soils of many other arid areas in that they have very low fertile (p. 307 of (70P2)).

China's Loess Plateau: Grassland carrying capacity dropped from 300 sheep/ km² in 1950's to 100-140 sheep/ km² in 1970's (89F1). Comments: These numbers seem high; loess soils are poor soils (low in organic matter, highly erodible).

Iraq: Rangelands of northern Iraq can safely sustain 250,000 sheep. (The sheep population in the 1970s was 1,000,000.) (78E1).

Latin America: Range land that is natural grass land, woodlands, and savanna where tree cover is limited by drought, fire, flooding and poor soil includes the cerrado of central Brazil, the llanos of the Orinoco River basin, the chaco of Argentina and Paraguay, the matorral of Chile and Peru, and the dry regions of north-central Mexico. These rangelands require 15-50 km² to support 100 cows (Capacity = 2-6 AU/ km²) (See p. 113 of (90W1)).

Mexico: Livestock carrying capacity of Mexico's rangeland: 11-22 ha/ cow (4.5-9 cows/ km²) but the average livestock density is 17-33 cows/ km², producing serious soil erosion (Ref. 97 and 98 of (90W1)).

Scotland: Most common grazing lands have about 100 sheep/ km² (97M1). Comments: This may entail over-grazing, so carrying capacity is probably less.

US - Arizona: On average, over 240 acres are required to sustain one cow on BLM grazing lands in Arizona (99W2).

US - California's Mojave Desert: Livestock graze on 7281 mi² (18858 km²) of BLM land of mid- to upper-elevation desert. This 5% of CA produced 103,191 AUMs in 1987 (1 cow/ mi²) (0.39 cow/ km²) (p. 64 of (91J1)).

US - Wasatch Front (Utah): A typical pasture provides a little more than 670 tonnes/ km²/ year. (Some of these pastures are irrigated.) A horse requires over 1.6 tons of feed during a 160-day grazing season. A 400-lb. steer needs slightly more than one ton. Grazing more than 50% of a grass quickly curtails root growth and severely limits future yields (U. S. Utah extension agents James Barnhill, 801-399-8208 and Dean Miner, 801-370-8469, utah@ext.usu.edu, 1997).

In this analysis two methods are used to estimate the livestock carrying capacity of the world's grasslands. The first uses net primary production (NPP) data - essentially rates of photosynthesis. Grassland NPP data pertinent to grazing is tabulated below.

To compute NPP in tonnes/ km²/ year of dry organic-matter, Col.4 should be multiplied by about 2, since plant dry organic matter is 45-50% carbon by weight. To convert the results to tonnes/km²/ year of normal (non-dry) organic matter (grass), one must divide by the ratio of dry weight to normal weight for grass. This ratio for grain is 0.8 (86V1), so for grass estimate a ratio of 0.7. Grazing over 50% of a grass quickly curtails root-growth and severely limits future yields (US Utah extension agents James Barnhill, and Dean Miner, utah@ext.usu.edu, 1997). For annual grasses, assuming half of the plant mass is in the roots, this means that only about 25% of NPP is to be considered usable herbage. (The part of NPP going to non-grass soil organisms is neglected here.) Globally, livestock stocking densities are about 10 times natural stocking densities of herbivores (92O1), so about 10% of NPP should be allocated to grazing wildlife and 90% to domestic wildlife. An animal-unit (AU) requires 10,800 lb/ year (4.90 tonnes/ year) of usable herbage (p. 25 of (91J1)). Some NPP occurring on woodlands, shrub lands and savanna goes into trees and other woody (inedible) plants, not grass. To allow for this, the NPPs per unit area of woodland, shrubland and savanna are taken to be the same as that for temperate grassland. (This is optimistic because savannas are known to be poor grazing lands - Section (3-C).).

The NPP data of Whittaker and Likens (73W1) used here are known to refer to ecosystems in a non-degraded state. The eight other global NPP inventory studies done since then (Ref. (87E1)) are more realistic in this regard and give an average global phytomass carbon pool size about 20% smaller. Arid grasslands are known to be degraded more than most other ecosystems however. The weighted average productivity-loss for the world's dry rangelands has been computed to be 43% (97C1). For the US, the productivity loss has been computed by range expert August Hormay to be at least 50%. Non-dry grasslands have been degraded much less. So to take account of NPP degradation, Whittaker-Likens' NPPs are multiplied by 0.8 in computing Col. 6 of the Table below.

Burning of grasslands also needs to be accounted for. Data given later in this report suggests a 5% reduction in forage and carrying capacity as a result of grassland burning.

These conversion factors enable one to compute carrying capacities of the world's grasslands in AU. The net results are tabulated in Col. 5 and 6 of the table below.

The above livestock carrying capacity of the world's grazing lands will be compared to the world's grazing livestock population below -immediately after an alternative calculation of carrying capacity.

In Section (2-B)[5] are lists of carrying capacity of rangelands in various regions of the US. These can be summarized in the table below (Col. 2). In Col. 3 are the areas of the grassland categories as computed in Section (3-A)[5]. From this data the livestock carrying capacity of the world's grazing lands can be computed.

The discrepancy between the world's livestock carrying capacity computed here (894 million AU) and in the NPP-based analysis (1105 million AU) is probably a result of assuming that the useable NPP of a unit area of savannas and woodland/ shrubland (which are invariably semi-arid) is the same as that of temperate grasslands, a significant but unknown fraction of which are in sub-humid or humid climates and therefore significantly more productive.

The conclusion of these two analyses is that the world's grazing lands are overgrazed by 66-100%. It is doubtful that any significant portion of this overgrazing could be undone by allocating additional "surplus" grain reserves to the feeding of grazing-type livestock. Meanwhile the soils of the world's grasslands, and hence their NPP, continue to degrade.

Human Carrying Capacity of the World's Grazing Lands
It is also of interest to use the tables above to compute the human carrying capacity of the world's grasslands if they were used exclusively for grazing. To convert the results above to meat productivity, note that it takes over 20 pounds (presumably non-dried weight) of useful herbage to produce a pound of beef (p. 33 of (91J1)). About 10% of grazing animals are lost to poisonous plants, predators, hunters, poachers, disease, road-kill, etc. (US experience) (p. 347 of (91J1)). To compute protein productivity note that meat is 13.2% protein. To convert the results to human carrying-capacity note that an adequate animal-protein diet is considered to be 13.1 kg/ person/ year (US per-capita animal-protein consumption is 23.7 kg./ year). One pound of beef contains about 1120 Calories, and an adequate diet is assumed to require 2500 Calories/ day. Using these factors the human carrying capacity of the world's grasslands as grazing lands based on animal-protein- and Calorie productivity can be computed. Results are given in the table below, based on the NPP analysis above.

If US animal-protein consumption is used as the standard, the above human carrying capacity of the world's grasslands drops from 2419 million to 1337 million people.

The world's 17 million km² of tropical rainforests can be (and are) cleared and used for grazing land. However grazing lands prepared from land cleared of tropical forest lasts for only about 8 years before it must be abandoned for about 30 years to allow soil productivity to be restored. Thus the effective potential grazing land supply offered by tropical rain forests is reduced to 3.6 million km² (neglecting the fact that over the 8 "productive" years, productivity is undergoing rapid degradation to near-zero). Data in the forest degradation review permit a detailed calculation of the potential additional livestock carrying capacity represented by rainforests.