HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Figures Only Full Text (PDF) Free Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Provenza, F. D. Articles by Wiedmeier, R. D. Search for Related Content PubMed Articles by Provenza, F. D. Articles by Wiedmeier, R. D. Agricola Articles by Provenza, F. D. Articles by Wiedmeier, R. D. Related Collections Forage Management Ecosystem Management Grazing Management

SYMPOSIA

The Value to Herbivores of Plant Physical and Chemical Diversity in Time and Space

^a Wildland Resources, Utah State Univ., Logan, UT 84322-5230
^b Plant, Soils, and Biometeorology, Utah State Univ., Logan, UT 84322-4820
^c Animal, Dairy, and Veterinary Sciences, Utah State Univ., Logan, UT 84322-4810

^* Corresponding author (stan{at}cc.usu.edu)

	ABSTRACT

TOP ABSTRACT INTRODUCTION DESCENT OF LIVESTOCK AND... SIMPLIFYING COMPLEX SYSTEMS HEALTH BENEFITS OF PLANT... CONTRASTING VIEWS ON INTAKE,... DYNAMICS OF FORAGE SELECTION... CONCLUSIONS REFERENCES

 
Whether foraging on pastures or rangelands, herbivores encounter plant species that differ in their concentrations of nutrients. They also all contain various secondary compounds that at too high doses can be toxic, but at the appropriate dose many of these toxins may have medicinal benefits. The quantity of forage an animal consumes depends on the other forages it selects because nutrients and toxins interact. Food intake also depends on an individual's morphology and physiology, and marked variation is common, even among closely related animals, in needs for nutrients and abilities to cope with toxins. Thus, individuals can better meet their needs when offered a variety of foods that differ in nutrients and toxins than when constrained to a single food. Nonetheless, we have focused on a few species, often grown in monoculture, and we have reduced concentrations of secondary compounds with little appreciation for their roles in protecting plants against herbivores, pathogens, and competitors. In nature, where diversity of plants is the rule and not the exception, eating a variety of foods is how animals cope with, and may benefit from, secondary compounds. The potential benefits of creating mixtures of plant species whose nutrient and secondary compound profiles complement one another are obvious, though much remains to be learned about how to reconstruct agro-ecosystems with plants that complement and enhance one another structurally, functionally, and biochemically.

Abbreviations: BW, body weight • PEG polyethylene glycol • VFAs, volatile fatty acids • T, toxins • HQ, high-quality forage

	INTRODUCTION

TOP ABSTRACT INTRODUCTION DESCENT OF LIVESTOCK AND... SIMPLIFYING COMPLEX SYSTEMS HEALTH BENEFITS OF PLANT... CONTRASTING VIEWS ON INTAKE,... DYNAMICS OF FORAGE SELECTION... CONCLUSIONS REFERENCES

 
OUR ANCESTORS domesticated livestock from wild herbivores some ten thousand years ago (Diamond, 1999). While most people are aware of the various species presently extinct, we are not likely to know much about the foraging behaviors of their progenitors. Who were the predecessors of today's domesticated herbivores, and when and where did they and their diets evolve? What roles did plants and plant diversity play in the health and well-being of soils, plants, and herbivores? While cultivating food for livestock is presently a common practice for providing nutrients, if we investigate the feeding systems where cattle evolved, we may discover valuable components of these ecosystems that should be incorporated back into the diets of domesticated ruminants.

	DESCENT OF LIVESTOCK AND THE VALUE OF PLANT DIVERSITY

TOP ABSTRACT INTRODUCTION DESCENT OF LIVESTOCK AND... SIMPLIFYING COMPLEX SYSTEMS HEALTH BENEFITS OF PLANT... CONTRASTING VIEWS ON INTAKE,... DYNAMICS OF FORAGE SELECTION... CONCLUSIONS REFERENCES

 
The earliest evidence of the progenitors of domesticated goats (Capra hircus hircus), sheep (Ovis aries), and cattle (Bos taurus and Bos indicus) is found in a region stretching from the eastern edge of the Mediterranean to western Asia. Goats originally inhabited high mountains, while sheep lived in hilly regions and the foothills of mountains (Clutton-Brock, 1987). Goats and sheep are both accustomed to harsh environments, they both eat a variety of grasses, forbs, and shrubs, and they are complementary grazers, but sheep eat proportionally more grass and less shrubs than do goats (Clutton-Brock, 1987). Sheep and goats were domesticated about 10 000 years ago. The earlier domestication of ancestors of the dog (Canis familiaris), about 12 000 years ago, which hunted and therefore herded these small ruminants for its own consumption, facilitated the hunting and the eventual domestication of livestock by man (Zeuner, 1963). The transition from man the hunter, following herds of goats and sheep, to man the herder, moving his own flocks, is not difficult to imagine.

By all accounts, modern cattle descended from the aurochs (Bos primigenius), a now-extinct giant herbivore still in existence in Caesar's time. From fossil evidence, Montgomery (2004) calculates the weight of the relatively small British aurochs as over 2000 kg, while continental aurochs weighed up to 4000 kg—nearly 4.5 tons. In Gallic Wars, Caesar describes aurochs as "somewhat smaller than elephants" with great speed and strength. Eastern humped cattle (Bos indicus) descended from aurochs that diverged from their western relatives hundreds of thousands of years ago (Diamond, 1999). The domestication of modern cattle is dated at about 8000 years from the present, 2000 years later than that of sheep and goats, occurring independently in India, western Asia, and North Africa, which suggests that their domestication presented a greater challenge than that of smaller livestock. Aurochs existed from Paleolithic times, but the much smaller and calmer species of cattle (Bos taurus) appeared suddenly in Neolithic times, with no identifiable intermediates, suggesting domestication occurred concomitantly with evolution. Montgomery (2004) suggests, based in part on the discovery of aurochs' skulls one-third smaller than normal, that miniature aurochs arose suddenly and developed into modern cattle. There is further support for this theory in a Canadian herd of miniature cattle collected in the 1950s, and the documentation of many well-formed miniature cattle fathered by a single Charolais bull at USDA-ARS research facility in Florida. Self-taming or auto-domestication also has been observed in Cape Buffalo (Syncerus caffer) and Bighorn Sheep (Ovis canadensis Canadensis), the latter example inspired by the proffering of salt by a human (Montgomery, 2004).

The Mediterranean climate of the Fertile Crescent, thought to be the location of the earliest domestication of cattle, extends across southern Europe. Because of the large area it encompasses and its great variation in altitude and climate, the Mediterranean region has enormous plant diversity, including a majority of the most productive grasses. Man spread throughout the Mediterranean, bringing ruminants that foraged on the available vegetation which included forest, maquis (broadleaved evergreen trees and shrubs), phrygana (undershrubs that are generally short-lived, mostly unpalatable, and aromatic), and steppe (dry, grass-dominated communities) (Grove and Rackham, 2001). A Mediterranean climate, characterized by mild, wet winters and long, hot, dry summers, favors annuals that germinate with the return of rain in the autumn. Steppe plants include annual and perennial grasses, annual legumes, annual and perennial compositae, and tuberous and bulbous perennials. Mediterranean regions are often savannas, described as areas where individual trees occur, but which are too dry for forests. Two distinct types of grasslands occur, between and under trees, particularly when ruminants redistribute nutrients while seeking shade. Trees alter light, soil moisture, temperature, and nutrient content, and increase the diversity of savannas. Typically, more than 100 plant species occur within a few square meters in a Mediterranean savannah (Grove and Rackham, 2001). Thus, ancestors of livestock selected their diets from a diverse array of plants, as do wild and domesticated livestock foraging on rangelands today (Provenza et al., 2003b). Indeed, given the choice, ruminants eat a variety of foods. Why do they prefer variety and what are the implications for managing landscapes for their benefits?

	SIMPLIFYING COMPLEX SYSTEMS

TOP ABSTRACT INTRODUCTION DESCENT OF LIVESTOCK AND... SIMPLIFYING COMPLEX SYSTEMS HEALTH BENEFITS OF PLANT... CONTRASTING VIEWS ON INTAKE,... DYNAMICS OF FORAGE SELECTION... CONCLUSIONS REFERENCES

 
Consider an animal foraging in an environment with 100 species of plants. These plants all differ in concentrations of energy, protein, vitamins, and minerals. They all contain toxins, but at the appropriate dose, many of these toxins may have medicinal benefits (Engel, 2002). Envision further that how much of any forage an animal can graze depends on the other forages selectively grazed. This is because at the biochemical level, nutrients and toxins interact: nutrients with nutrients, nutrients with toxins, and toxins with toxins (Provenza et al., 2003b). Now imagine three to seven foods will make up the bulk of the diet in any meal. Which plants should an animal choose? Clearly, given 100 species and their interactions, there are many possibilities.

In discussing this topic, we typically ascribe a critical role to genetics: environments continually select those animals with genes best able to cope with environmental challenges. Clearly, gene expression affects animal morphology and physiology, which surely influences its potential to live in a given environment. In the case of foraging, traits related to animal morphology and physiology create the bounds within which animals can use different foods and habitats by affecting the need for nutrients, the ability to cope with toxins, and the value of medicines. But there's more to it than that. Genes learn from the environment (Lipton, 2005), and in concert with the genome, animals learn which plants to mix and match (Provenza et al., 2003b), both of which influence the ability to survive in an environment. For example, mixing plants A, B, C, and D may lead to a much different result nutritionally, toxicologically, and medicinally from mixing plants A, C, F, and G. Thus, while genes set bounds within which animals must live, there are a great many ways to do that. Interestingly, animals may simplify extremely complex situations in ways that may ultimately overlook many possibilities.

Humans have dealt with the challenge of nature's cornucopia in various ways, depending on environment, but regardless of where they managed to make a living our ancestors acquired plant-based knowledge largely through non-random processes that involved targeting a few species that were abundant, "palatable," easily cultivated, and harvested for sampling and eventual use (Etkin, 1994). Of the roughly 200 000 wild plant species on earth, only a few thousand are consumed by humans, just a few hundred of these have been domesticated, and only a dozen species account for over 80% of the current annual production of all crops (Diamond, 1999). By focusing on some species and ignoring many others, people transformed the highly diverse and complex world of plants into a manageable domain that met needs for nutrients, limited over-ingestion of toxins, and made good use of nature's pharmacy to bolster health (Johns, 1994). In so doing, however, traditional and modern societies likely succeeded in discovering only a small fraction of the plant mixtures potentially useful in nutrition and medicine.

Undoubtedly, before the advent of food production, peoples' knowledge of wild species was much greater, as everyone on Earth still depended wholly on wild plants for food and medicine. Many foods preferred by our ancestors are considered by people today to be "unpalatable" because of the secondary compounds such as tannins, terpenes, saponins, oxalates, alkaloids, cyanogenic glycosides, and many others. At present, with ready access to processed foods high in sugar, carbohydrates, fat, and salt, young people no longer acquire preferences for important "unpalatable" foods: they lack the traditional cultural foundations to guide their selection of plants high in nutrients and medicines (Johns, 1994). Interestingly, babies reared on indigenous diets acquire a taste for so-called "unpalatable" plants due to their benefits (Johns, 1994).

In the case of forages developed for pastures and rangelands, breeders historically selected for plants low in secondary compounds, not realizing their potential health benefits in appropriate dosages (Engel, 2002). Virtually every species developed for use in pastures has been selected for lower concentrations of secondary compounds: saponins in alfalfa (Medicago sativa L.), alkaloids in tall fescue (Festuca arundinacea Schreb.) and reed canarygrass (Phalaris arundinacea L.), tannins in the trefoils (Lotus spp.), and cyanogenic glycosides in the clovers (Trifolium spp.) to name a few. The desire to reduce concentrations of secondary compounds was integrally linked with the emphasis on planting monocultures, and for good reason: secondary compounds limit intake of a single food. Under natural conditions where diversity of plants is the rule, not the exception, eating a variety of foods is how animals cope with, and likely benefit from, secondary compounds in their diets (Freeland and Janzen, 1974; Engel, 2002; Provenza and Villalba, 2006). Today, we are coming full circle with efforts to genetically engineer plants that produce compounds insects and large herbivores avoid eating in too high doses. We are doing so without fully appreciating the value of biodiversity for animals.

	HEALTH BENEFITS OF PLANT DIVERSITY FOR ANIMALS

TOP ABSTRACT INTRODUCTION DESCENT OF LIVESTOCK AND... SIMPLIFYING COMPLEX SYSTEMS HEALTH BENEFITS OF PLANT... CONTRASTING VIEWS ON INTAKE,... DYNAMICS OF FORAGE SELECTION... CONCLUSIONS REFERENCES

 
Variety, the Spice of Life
While animals may perform well grazing monocultures, there are three reasons eating a variety of foods is likely to increase nutrition, health, and well-being. First, no forage has the perfect balance of nutrients found in a variety of forages (Westoby, 1978). In addition, grazing a variety of forages is likely to provide health benefits not possible if animals graze monocultures (Engel, 2002). Finally, all plants, including garden vegetables, contain secondary compounds that at appropriate doses may have health benefits (Provenza and Villalba, 2006), but at too high doses can be toxic (Palo and Robbins, 1991; Osweiler et al., 1985; Cheeke, 1998). Given choices, herbivores seldom consume enough toxins to result in poisoning because they regulate their intake of foods that contain toxins (Provenza, 1995, 1996; Foley et al., 1999). For instance, oral gavage of toxins causes dose-dependent decreases in intake of foods that contain the toxins (Wang and Provenza, 1997). To obtain needed nutrients, herbivores must graze a variety of plant species that contain different kinds of toxins that may be complementary (Provenza et al., 2003b). In principle, herbivores should be able to consume more of a combination of foods with different kinds of toxins as different toxins have different effects in the body and they are detoxified by different mechanisms (Freeland and Janzen, 1974). When lambs choose between foods that contain either amygdalin or lithium chloride, consumptions are less when the food contains only one of these toxins; the same is true with two-way combinations of nitrate and oxalate (Burritt and Provenza, 2000), tannins and terpenes, terpenes and oxalates, tannins and oxalates, and with three-way mixtures of tannins, terpenes, and oxalates (Villalba et al., 2004). Mule deer also browse more when offered both big sagebrush (Artemesia tridentata Nutt.) and common juniper (Juniperus communis L.) [12.3 g kg^–1 body weight (BW)], which contain different kinds of tannins and terpenes, than when they are offered only sagebrush (4.2 g kg^–1 BW) or juniper (7.8 g kg^–1 BW) (Smith, 1959). Common brushtail possums (Trichosurus vulpecula), that select from two diets containing phenolics and terpenes, consume more total food than when they consume diets with only one of these toxins (Dearing and Cork, 1999); the same is true in principle for sheep (Mott, 2006) and gray squirrels (Sciurus carolinensis) (Schmidt et al., 1998). Conversely, lambs offered foods containing either sparteine or saponin consume no more of both foods than lambs offered foods containing only one of these compounds because these toxins are not complementary; the same is true with tannin and saponin for sheep (Burritt and Provenza, 2000), but not for mice (Freeland et al., 1985).

Finally, differences among individuals in food intake and preference depend in part on variations in how animals are built morphologically and how they function physiologically, and marked differences among individuals are common even among uniform groups of animals in the need for nutrients (Scott and Provenza, 1999) and abilities to cope with toxins (Provenza et al., 1992, 1999). Individuals can best meet their needs for nutrients and regulate their intake of toxins when offered a variety of foods that differ in nutrients and toxins than when constrained to a single food, even if that food is "nutritionally balanced" (Provenza et al., 2003b). Thus, feeding and grazing practices that allow producers to capitalize on the individuality of animals are likely to improve performance of the herd by enabling the uniqueness of individuals to be manifest.

The Satiety Hypothesis: Linking Variety and Foraging Behavior
Why do animals graze a variety of forages? Preference for forage is generally thought to be influenced by palatability, but what is palatability? It is a narrowly defined term with many meanings. Webster defines ‘palatable’ as pleasant or acceptable to the taste and hence fit to be eaten. Animal scientists usually deem palatability to be the hedonic liking or affective responses to a food's flavor and texture, or the relish an animal shows when consuming a ration. Conversely, plant scientists depict palatability as attributes of plants that alter preference, including chemical composition, growth stage, and allied plants. All definitions focus on either a food's flavor or its physical and chemical characteristics.

Palatability is a complex phenomenon that integrates odor, taste, and texture with the postingestive effects of nutrients and toxins in food (Provenza 1995; Provenza et al., 1998; Provenza and Villalba, 2006) (Fig. 1 ). The flavor of a forage results when sensory receptors in the mouth and nose respond to gustatory (sweet, salt, sour, bitter), olfactory (an array of odors), and tactile (astringency, pain) stimuli. These receptors interact with visceral receptors that respond to nutrients and toxins (chemo-receptors), osmolality (osmo-receptors), and distension (mechano-receptors). Collectively, these flavor-feedback interactions enable animals to discriminate among forages of distinct physiological utilities, and they encourage animals to eat a variety of forages (Pfister et al., 1997; Early and Provenza, 1998; Villalba and Provenza, 1999a,b), and to graze in a variety of locations (Scott and Provenza, 1998, 2000; Bailey and Provenza, 2006).

View larger version (20K): [in this window] [in a new window]   Fig. 1. Palatability is a complex phenomenon that integrates a food's odor, taste, and texture with the postingestive effects of nutrients and toxins in the food. The process of ingesting a food causes an animal to satiate on the foods it is eating, and the satiety hypothesis attributes changes in palatability to transient food aversions due to flavors, nutrients, and toxins interacting along temporal concentration gradients.

Functions of Plant Secondary Compounds
During the past 35 years, researchers have become progressively more aware of the importance of secondary plant metabolites (Rosenthal and Janzen, 1979). Scientists understood the roles of primary elements, such as N, P, and K, but during this era they discovered that a vast array of interactions in terrestrial and aquatic ecosystems are mediated by secondary compounds formerly thought of as waste products of plant metabolism (Rosenthal and Berenbaum, 1992). They also came to better understand how environmental factors such as nutrients, water, and light influence the evolution (Coley et al., 1985) and phenotypic expression (Bryant et al., 1983) of secondary compounds, and how secondary compounds influence herbivores. Plants that produce compounds that can quickly induce satiety in foragers stand a better chance of surviving, and secondary compounds limit how much of a particular species an herbivore can eat, which spreads the load of herbivory across many species in a plant assemblage (Freeland and Janzen, 1974; Foley et al., 1999). Thus, while we once thought only poisonous plants that are particularly troublesome for herbivores contain compounds that were potentially toxic (Provenza et al., 1992), there is growing awareness that all plants contain toxins.

At excessive doses, secondary compounds in plants can adversely affect herbivores through their negative actions on cellular and metabolic processes (Cheeke and Shull, 1985; Cheeke, 1998). At appropriate doses, however, secondary compounds can suppress production of the bacteria, parasites, and fungi that inhabit herbivores' bodies and cause impairment of herbivore health (Engel, 2002). The difference between a toxin and a medicine is merely a matter of dosage (Plotkin, 2000). Lamentably, we know very little about how herbivores might learn to use secondary compounds for health and medicinal benefits. While much remains to be learned, herbivores can learn to use medicines to attenuate the aversive effects of acidosis as well as tannin and terpene toxicosis. They also select diets that (i) provide necessary amounts of energy and protein, (ii) synchronize the supply of energy and protein, (iii) balance supplies of macronutrients and toxins, and iv) contain different kinds of complementary toxins (Provenza and Villalba, 2006). These examples show that herbivores learn associations among nutrients, toxins, and medicines.

Undoubtedly, foraging behavior evolves to meet needs for nutrients, but just as foraging behavior can be affected by predators and competitors, some responses may also be geared toward reducing disease (Lozano, 1998). Invasions of parasites and pathogens can adversely affect an individual, making the need to counteract such stressors crucial for survival (Huffman, 2003). If a mammal can learn, either evolutionarily or contemporarily, to avoid certain plants because they lower fitness, it may also learn to seek certain plants or other substance in the environment such as medicines because they raise its fitness (Janzen, 1978). People learn to take aspirin for headaches, antacids for stomach aches, and ibuprofen to relieve pain, and we obtain prescriptions from doctors for medications. Many of the drugs we use come from plants in nature, but what about other animals; can they, too, learn to write prescriptions from nature's pharmacy?

The study of self-medication in animals has led to the emergence of a new field, often referred to as zoopharmacognosy, to describe the process by which animals select and use plant secondary metabolites or other non-nutritional substances for the treatment and prevention of disease (Rodriguez and Wrangham, 1993). Green and Garcia (1971) showed for the first time a case of learned self-medication in animals wherein rats increased preference for a distinctive flavor ingested during recuperation from illness. Sheep increase intake of "medicines" such as polyethylene glycol (PEG), a substance that attenuates the aversive effects of tannins, as tannin concentrations in their diet increase (Provenza et al., 2000). They discriminate the "medicinal" effects of PEG from "non-medicinal" substances by selectively increasing intake of PEG after eating a meal high in tannins (Villalba and Provenza, 2001). They also forage in locations where PEG is present, rather than where it is absent, when offered nutritious foods high in tannins in different locations (Villalba and Provenza, 2002). Likewise, cattle foraging on endophyte-infected tall fescue readily use lick tanks containing MTB-100 (Alltech, Nicholasville, KY), while they ignore lick tanks without MTB-100 (Cathy Bandyk, personal communication, 2004). MTB-100 contains the cell walls of yeast, which adsorb the alkaloids in tall fescue, thus acting as a medicine that enhances consumption of tall fescue by cattle. Sheep fed acid-producing substrates such as grain subsequently ingest foods and solutions containing sodium bicarbonate, which attenuates acidosis (Phy and Provenza, 1998). Finally, in the most elaborate studies to date, sheep learned to selectively ingest three medicines, sodium bentonite, polyethylene glycol, and dicalcium phosphate, that lead to recovery from illness due to eating excessive amounts of grain, tannins, and oxalic acid, respectively (Villalba et al., 2006a). This first demonstration of multiple malaise-medicine associations supports the notion that herbivores learn to self-medicate.

Condensed tannins have anti-parasitic properties in herbivores directly through anthelmintic effects (Athanasiadou et al., 2000). They also indirectly increase resistance and resilience of animals to parasitic infections due to improved protein nutrition (Min and Hart, 2003), which may enhance immune response to the parasite infection. Beneficial effects of condensed tannins on parasites occur in the range of 45 to 55 g of condensed tannins kg^–1 DM (Min and Hart, 2003). Herbivores feeding on plants with tannins show lower nematode burdens, lower fecal egg counts, and higher body weight gains than those feeding on plants of similar nutritional qualities without tannins. Condensed tannins in sericea lespedeza [Lespedeza cuneata (Dum.-Cours.) G. Don] reduce fecal egg production from gastrointestinal nematodes by reducing hatch and development of larvae (Min et al., 2004), and condensed tannins in Sulla (Hedysarum coronarium L.) have anthelmintic effects and immunogenic benefits (Niezen et al., 1995; Niezen et al., 2002). Chicory (Cichorium intybus L.), which contains an array of condensed tannins, other phenolic compounds, sesquiterpene lactones, coumarin, and caffeic acid derivatives, reduces need for commercial anthelmintics in young farmed deer (Cervus elaphus) (Hoskin et al., 1999). Sesquiterpene lactones, a large and diverse group of biologically active compounds common in the Asteraceae family, provide anti-tumor and anti-ulcer activity (Robles et al., 1995). Thus, while much remains to be learned about the health benefits of plant secondary metabolites, we are realizing that biochemical diversity in plants may enable animals to combat parasite loads and other diseases, thereby reducing reliance on commercial agents that parasites and microorganisms come to resist.

	CONTRASTING VIEWS ON INTAKE, SATIETY, AND VARIETY

TOP ABSTRACT INTRODUCTION DESCENT OF LIVESTOCK AND... SIMPLIFYING COMPLEX SYSTEMS HEALTH BENEFITS OF PLANT... CONTRASTING VIEWS ON INTAKE,... DYNAMICS OF FORAGE SELECTION... CONCLUSIONS REFERENCES

 
There has been a long-standing debate in ruminant nutrition over whether chemical or physical signals are more important in limiting food intake of animals in confinement, on pastures, and on rangelands (Grovum, 1988). Some people argue distension of the gut from cell walls and fiber is most important, whereas others claim chemical signals emanating from nutrients and toxins are more important. With forage intake on pastures and rangelands, many people believe prehending and masticating forages are essential factors influencing food intake. Finally, the physical structures of different plants may influence intake and cause animals to eat a variety of plants while foraging.

Linking Satiety with the Physical and Chemical Bases of Food Intake
Some contend that the intake of roughages by ruminants is limited by distension of the gut due to restricted flow of digesta through the gastrointestinal tract (Balch and Campling, 1962; Campling, 1970). Distention of the reticulum and cranial sac of the rumen stimulates slowly adapting tension receptors that relay information to the central nervous system during a meal (Grovum, 1988). Afferent action potentials emanate from vagal nerve fibers that connect with tension receptors in the abomasum of sheep (Harding and Leek, 1973). Restricted flow increases distension of the abomasum, which is correlated with the fiber content and digestibility of forages. Some contend that intake is correlated with the neutral detergent fiber concentrations of forages, especially C3 plants, and they regard it as the best single chemical predictor of filling effects (Mertens, 1987) and intake in ruminants (Van Soest, 1965; Waldo, 1986). More recent evidence suggests fiber alone is not an accurate predictor of filling; other variables include particle size and fragility, as well as rate and extent of fiber digestion (Allen, 1995).

Plant structure only is not the exclusive predictor of food intake and preference. Animals also respond to a complex array of biochemical compounds in plants. Indeed, some argue for a threshold below which food intake is limited by physical fill and above which intake is limited as energy demands are met (Conrad et al., 1964). Adjustments in food intake by ruminants are closely related to energy requirements (Baile and Forbes, 1974), and intake is influenced by concentrations and flows of nutrients and energy, including volatile fatty acids (VFAs) produced by fermentation in the rumen (Illius and Jessop, 1996). Satiating levels of the VFA propionate, the major precursor of glucose in the ruminant liver, markedly depresses food intake, especially when present in the portal vein; blocking vagal and splanchnic afferent nerve signals abolishes this effect (Anil and Forbes, 1980). The liver also responds to changes in sources of energy, such as glycerol, malate, lactate, and pyruvate, and may also sense temperature and pain (Anil and Forbes, 1987). Acetate, the VFA produced and absorbed in greatest quantities, also influences food intake (Baile and Pfander, 1966; Baile and Forbes, 1974) through stimulation of neural receptors in the rumen wall (Martin and Baile, 1972; Forbes, 1995). More recently, the peptide leptin, secreted by adipocytes, has been proposed as another metabolic signal influencing energy balance and food intake (Houseknecht et al., 1998).

Satiety is also influenced by protein and energy balance. Ruminants respond to imbalances of energy and protein (Villalba and Provenza, 1997a,b 1999a), and protein intake may be under tighter control than energy intake (Webster, 1993). Excesses of N in food result in buildup of ammonia in the rumen, which induces aversive postingestive feedback (Kertz et al., 1982; Provenza, 1996). Sheep avoid an excess of urea in their diet (Kyriazakis and Oldham, 1993). High levels of urea cause sub-clinical ammonia toxicity (Chalupa et al., 1970), which depresses food intake (Wilson et al., 1975; Kertz et al., 1982). Sheep regulate intake of foods with high levels of non-protein N to maintain blood ammonia N levels below 2 mg L^-l (Nicholson et al., 1992). Satiety also can be induced by diets low in protein, though ruminants are usually less constrained than non-ruminants with regard to the quality and quantity of protein they must ingest. This is because their primary source of protein, rumen microbes, is synthesized in the rumen, and N (urea) is continually recycled (Owens, 1988). However, when protein synthesis in the rumen is limited or when protein requirements are high, certain microbial amino acids may fall below optimal levels for animal production. Lambs decrease intake of foods associated with amino acid deficits, and increase intake when the deficiency is rectified (Rogers and Egan, 1975; Egan and Rogers, 1978). Thus, ruminants can sense amino acid imbalances, as can non-ruminants (Gietzen, 1993). Recognizing and rejecting amino acid-deficient diets occurs through a general control system, triggered by accumulation of amino acid-depleted transfer RNA in cells of the anterior piriform cortex in the brain. This internal nutrient sensor involved in maintaining amino acid homeodynamics appears to be conserved across evolution from single-cell organisms to mammals (Hao et al., 2005).

Physical and Chemical Bases for Forage Selection by Animals
Visceral signals from distension are integrated with chemical signals in the central nervous system, which suggests that distension and chemistry act in concert such that no one signal is solely responsible for inducing satiety (Forbes, 1995; Provenza, 1995). As in any complex system, the whole is different from the sum of its parts and thus many signals interact to enhance or depress a response relative to the effects of any signal in isolation. For instance, the combined effects of acetate, propionate, and distension of the rumen by inflation of a balloon, depressed food intake more (62%) than the isolated effect of each signal (about 11%) in lactating dairy cows (Anil et al., 1993).

With regard to forage intake on pastures and rangelands, nutrients and toxins occur in a multi-dimensional matrix of plant physical attributes that include length, width, and depth, all of which change temporally and spatially. Research of the past 30 yr has determined that plant structure affects intake (I) directly through bite mass (BM) and bite rate (BR) and indirectly through time spent grazing (GT) so I = BM x BR x GT. These variables interact such that: i) intake per bite is influenced mainly by the sensitivity of bite depth to changes in sward height, ii) bite mass is less sensitive than bite depth to variation in sward conditions, iii) bite rate is inversely correlated with bite mass, reflecting the increasing importance of prehension and mastication as bite mass increases, iv) rate of intake typically increases with increasing bite mass, and v) as bite mass and bite rate decline, time spent grazing increases (Hodgson et al., 1999). Research in this area also suggests differences in macronutrient characteristics of foods, along with variations in their vertical and horizontal availabilities, interact with nutritional status to affect foraging behavior (Edwards et al., 1996).

Researchers have documented the independent effects of chemical and physical attributes of plants on forage intake, but little research has been done to elucidated how chemical and physical plant properties interact to affect intake and preference. The intake of nutrients also depends on bite mass, bite rate, and nutrient content such that nutrient intake = bite mass x bite rate x nutrients bite^–1, and combinations of plant physical and chemical characteristics that optimize nutrients bite^–1 are likely to be preferred. These factors are confounded under field conditions as foods with high nutrient concentrations are typically easy to harvest; green grass is highly nutritious and has low tensile strength. Conversely, foods with low nutrient concentrations are usually difficult to harvest; mature grass is low in nutrients and has high tensile strength. For the most part, it remains to be seen how plant chemical and physical attributes interact to affect preference.

Most grazing studies have examined the impact of the structure of a single plant species on intake rate, and have manipulated sward structure with little regard for the importance of plant biochemical composition in intake (Illius and Hodgson, 1996; Hodgson et al., 1999). Some propose that nutritional and toxicological factors being equal, intake rate is determined by bite mass (Ungar et al., 1991), which is influenced by herbage height and weight per unit canopy volume (Black and Kenney, 1984). Likewise, forage preferences have been regarded as a function of the rate at which forages can be grazed. Tender plants that can be eaten quickly are likely to be preferred to tougher species (Kenney and Black, 1984). Cattle (Distel et al., 1995) and sheep (Black and Kenney, 1984) select patches that allow higher intake rates from swards that differ in height and density. In heterogeneous pastures of similar quality, animals may assess the reward from a patch by sensing the rate at which food can be obtained and modifying preferences accordingly (Distel et al., 1995). While a structurally uniform pasture might be highly desirable for nutrient assimilation, grazed perennial pastures are rarely structurally or spatially uniform. The influence of structural diversity in intake must be better-understood to allow pastures to be managed for efficient utilization.

The influence of nutrients on rates of food consumption has been overlooked primarily because research on physical plant characteristics has controlled for plant quality and because experiments were conducted under conditions where nutritional and toxicological differences among plants were not pronounced. Nevertheless, rates of food intake also are determined by the nutritional quality of food. Groups of lambs exposed to the same forage—wheat straw—and thus the same structure and architecture for the same periods of time showed quite different intake rates, depending on the postingestive consequences experienced during straw ingestion. Lambs that received intraruminal infusions of starch just after feeding increased intake, intake rate, bite rate, and intake per bite compared with lambs that received only intraruminal infusions of the vehicle (water) (Villalba and Provenza, 2000). Thus, when physical structure is constant (same forage, same height and density) and nutritional effects are unequal (intraruminal infusions of starch vs. water) intake rate is largely influenced by the postingestive effects of food ingestion. Rates of food intake are elastic (Owen-Smith, 1993), and influenced by postingestive signals (Villalba and Provenza, 1999a,b). Sheep simultaneously fed straw, while straw was put through a fistula into the rumen, were reluctant to eat straw, but they ate markedly more straw when grass hay, a more nutritious food, was put into the rumen instead of straw (Greenhalgh and Reid, 1971). Thus, rates of food intake are due to the interaction between the structure and nutritional composition of food.

These notions are true for nutrient–toxin interactions, and they are relevant to functional-response theory, which has been applied widely to herbivore use of plants based on physical characteristics. When herbivory is analyzed based on functional-response theory (Holling, 1959), an increase in plant biomass is associated with an increase in the rate of food intake. Along a continuum of different relative abundances, a plant species faces a lower risk of herbivore-induced mortality as it becomes increasingly abundant; conversely, herbivores remove a disproportionately large percent of biomass from a plant species when it is rare. Because herbivores satiate, flavor–nutrient–toxin interactions set the asymptote of functional response curves that define the relationships between plants and herbivores (Provenza et al., 2003a). Thus, the abundance of a species influences these responses as nutrients and toxins in that species and its neighbors interact to satiate the detoxification capabilities of herbivores at critical thresholds of plant abundance. Above these thresholds, herbivory will favor domination by a species. Below it, local extinction is more likely as a species becomes less abundant.

According to the satiety hypothesis, toxic plants should be best defended when neighboring plants have (i) low levels of nutrients needed to mitigate toxicosis, or (ii) high levels of nutrients that interact adversely with toxins. If a plant community supplies appropriate nutrients, the threshold of toxin satiation will increase, which is consistent with the high levels of herbivory experienced by toxic species that constitute a minor proportion of the diet (Bryant et al., 1991a,b; Augustine and McNaughton, 1998). As a toxic plant becomes more abundant and the availability of nutrients declines, toxin-satiation thresholds and use both decline. Thus, utilization of a toxic plant may increase or decrease without any change in the toxin content of the plant.

Most studies of functional response have considered only physical limitations to grazing, such as travel time between feeding patches and time required to ingest plants (Lundberg, 1988; Lundberg and Astrom, 1990; Lundberg and Danell, 1990). However, within limits imposed by physical constraints (Spalinger et al., 1988; Lundberg and Astrom, 1990; Astrom et al., 1990; Illius et al., 1999), nutrient-toxin satiation influences the upper limit of biomass intake, and individual plants and plant species with potentially high intake rates may not be eaten due to their high concentrations of toxins (Bryant et al., 1991a,b; Villalba and Provenza, 1999b, 2000). For example, livestock use relatively unpalatable plant species, such as sagebrush, less as they become increasingly abundant; use increases when sagebrush is interspersed with other species (Heady, 1964; Smith, 1959). Mule deer (Cervus elaphus), moose (Alces Americana), and snowshoe hares (Lepus americanus) also exhibit disproportionately intense browsing of rare woody species. For example, Douglas fir [Pseudotsuga mensiesii (Mirb.) Franco] and eastern white cedar (Thuja occidentalis L.), moderately palatable evergreens, are staple foods of Columbian black-tailed deer (Odocoileus hemionus columbianus) (Cowen, 1945), eastern white-tailed deer (Odocoileus virginianus) (Bookhout, 1965), and snowshoe hares (Bookhout, 1965); browsing of these trees is inversely proportional to the density of each species (Cowen, 1945; Bookhout, 1965). Browsing of saplings of woody species by snowshoe hares in Alaska during winter is most severe when they are rare; similar severe browsing by snowshoe hares and other mammals occurs on woody species whose palatability is high (Alaska feltleaf willow; Salix alaxensis Anderss.), moderate (Alaska paper birch; Betula resinifera Britton), or very low (green alder [Alnus crispa (Ait.) Pursh]) (Bryant and Kuropat, 1980).

Several strategies can be used in concert to capitalize on these observations. Mixtures of forages on pastures should be sown in relative abundances that reflect the nutrient-toxin satiation capabilities of herbivores. Herbivores can learn to use a variety of foods, especially foods high in toxins, and four factors enhance success (Provenza et al., 2003a,b). First, maintain high stock densities for short periods to encourage livestock to learn to eat a variety of plants. Second, make use of all plants in moderation by grazing areas where plants occur in patches (spatially) and by moving animals to different forages (temporally), such that livestock are not forced to eat only one kind of plant high in toxins. Third, allow livestock access to a variety of plants that vary in kinds and amounts of nutrients and toxins, ideally ones that complement one another. Fourth, retain replacement heifers from the herd so such foraging behaviors become part of the culture. Undoubtedly, there are critical thresholds of human and herbivore knowledge about how to mix and match various plant species and plant biochemical diversity below which plant diversity and animal performance decline at ever increasing rates, and above which diversity and performance can beget greater diversity and enhanced performance. While these principles have been developed to utilize plant species that are relatively high in toxins, they also apply to the management of livestock on nutrient-dense, high-quality perennial pastures, particularly as we increase beneficial forage diversity.

Nutritional State and Intake Rate
A few studies have examined how physiological condition affects intake rate and preference for forages by herbivores. The most commonly explored relationship has been the effect of fasting on intake rate and diet selection (Edwards et al., 1994; Newman et al., 1994). Fasting increases intake by increasing motivation to eat (Greenwood and Demment, 1988; Dougherty et al., 1989). However, this response depends on the nutritional state of the animal. When fasted and non-fasted animals were offered a food rich in energy but low in N, fasted animals fed an N-rich basal diet had lower rates of intake than non-fasted animals fed the same basal diet (Villalba and Provenza, 1999b). This response was likely due to an increased need by non-fasted animals to balance the excess N, supplied by the N-rich basal diet, with carbohydrates in the energy-rich food.

When the same food is offered in different physical forms, lambs prefer whole to ground foods because the former have higher intake rates (Villalba and Provenza, 1999b). However, when nutritionally different foods are offered in different forms, lambs fed a basal ration high in energy and low in protein ate less grain (energy) and more alfalfa (protein) than lambs fed a basal ration low in energy and high in protein. This pattern occurred regardless of the physical form of the foods (Villalba and Provenza, 1999b). Differences in preferences were pronounced after pre-loads of the basal rations, but equally fasted lambs differed in degree of preference for foods of different physical forms depending on the quality of their basal ration. Lambs fed an energy-dense ration ate less grain more slowly than lambs fed a ration with a higher ratio of protein to energy.

In summary, intake rate is affected by food structure and fasting, as well as nutritional state with regard to energy and protein. Both structure and biochemical composition may affect food preferences. However, when the need for a particular nutrient increases or decreases and the foods available are of different quality, biochemical composition may be more important than structure in affecting food preferences (Villalba and Provenza, 1999b). Food quality and structure likely operate along a continuum to influence preference such that structure becomes increasingly important as food quality becomes similar. In contrast, when food quality is different and the need for a particular nutrient increases or decreases—state-dependent diet selection, balancing the supply of nutrients—the biochemical composition of a food becomes crucial for diet selection.

	DYNAMICS OF FORAGE SELECTION IN TIME AND SPACE

TOP ABSTRACT INTRODUCTION DESCENT OF LIVESTOCK AND... SIMPLIFYING COMPLEX SYSTEMS HEALTH BENEFITS OF PLANT... CONTRASTING VIEWS ON INTAKE,... DYNAMICS OF FORAGE SELECTION... CONCLUSIONS REFERENCES

 
Temporal Dynamics
The number of animals per unit area, stock density, can influence how animals learn to forage. Low to moderate densities encourage selective grazing, which may prevent learning complementarities among less palatable forages. This pattern of selection can create dissociation in time between the ingestion of potentially complementary high- and low-quality forages, which can be of different magnitudes depending on when forages are replenished. This temporal dissociation of forages can adversely affect preferences for less palatable forages. Lambs learn preferences for flavored straw after a delayed supply of nutrients, but preferences are stronger with short delays between forages ingestion and nutrient reward (Villalba et al., 1999). Ingesting low-quality forages without a daily "supplement" of high-quality food may decrease preference for the low-quality forages. Thus, dissociating high- and low-quality forages temporally may influence how herbivores use those foods when offered choices.

Sheep learned to eat low-quality forage with toxins (T) and high-quality forage (HQ) in two different temporal arrangements (Villalba et al., 2006b). In one case, sheep were fed HQ for 12 d followed by T for 12 d such that their synergistic effects were dissociated into distinct feeding periods. In the other case, sheep were fed HQ and T concurrently for 12 d so their effects were associated within a meal. Subsequently, all sheep could forage at locations containing HQ and T, only HQ, or only T. Lambs that initially ate both forages in a meal always ate more T than those that initially ate the foods in two distinct periods, even when HQ was available ad libitum. As HQ decreased in abundance, lambs that learned to mix both foods foraged more opportunistically and remained longer at locations with both HQ and T or with just T. Even when both groups spent about the same amount of time at locations with HQ and T, lambs that initially ate both forages in a meal ate more T and consumed more food. Thus, preference for particular forage depends not only on its chemical characteristics but also on the temporal context in which herbivores experience that particular forage and other forages.

The sequence of toxin ingestion also influences the preference of herbivores for the forages. Lambs offered a meal of tannins followed by a meal of terpenes consume twice as much food as lambs offered a meal of terpenes followed by a meal of tannins (Mott, 2006). Preference for terpenes was also higher in the former than in the latter group. In another study, lambs grazed sagebrush steppe vegetation at a high (H) or a low (L) stock density and both groups were moved to fresh pasture daily. Lambs in a third treatment (H3) had three times the area of treatment H, but they were moved every 3 d, making the total area grazed by H and H3 equal, but with a different temporal allocation of forage (Shaw et al., 2006). Lambs in H spent more time foraging on sagebrush (25%), a plant high in terpenes, than lambs in H3 (16%), and lambs in both H and H3 spent more time foraging on sagebrush than lambs in L (1%). For lambs in H3, foraging on sagebrush was cyclic and depended on the daily availability of herbs; they consumed sagebrush most on Day 3 of the cycle when herbs were depleted. When the three groups were tested at high stock densities, their use of sagebrush depended on treatment: H > H3 > L. Thus, the availability of alternative foods, manipulated through animal density and the temporal allocation of forage, affected how lambs learned to use sagebrush (Shaw et al., 2006).

The time between feeding bouts on toxin-containing plants may influence intake and preference for different plant species. In the study by Shaw et al. (2006), lambs in H ate a large amount of sagebrush each day, whereas lambs in H3 ate high amounts of sagebrush only on Day 3. Thus, lambs in H3 had more time to detoxify the doses of terpenes in sagebrush and consequently the detrimental impacts of terpenes apparently were diminished. Longer intervals between feeding bouts allow animals time to detoxify and eliminate ingested toxins and to resume feeding (Foley et al., 1999). The addition of terpenes to artificial diets causes sheep to eat smaller and less frequent meals, thereby providing adequate time to eliminate the ingested toxins (Dziba and Provenza, 2006; Dziba et al., 2006). Thus, the impact of toxin-containing plants on the nutritional well being of herbivores may not be linearly related to the amounts of food consumed; it may also depend on the temporal context in which the toxin-containing plant was ingested.

Spatial Dynamics
Geometric patterns are ubiquitous in natural and agricultural systems, and fractal geometry is leading to a synthetic understanding of the structure and function of biological systems across orders of magnitude of scale, as well as for a wide range of taxa, habitats, and ecosystems (West et al., 1997; Enquist and Niklas, 2001; Ernest et al., 2003). Just as the study of biological scaling, or the patterns of change correlated with an individual's body size, has yielded insights into the universal foundations that influence the physiology and ecology of a wide array of organisms, including plants, animals, and protists, fractal geometry is being used to understand and quantify spatial diversity in ecological systems (Mandelbrot, 1983; Milne, 1992; Ritchie, 1997, 1998). The basic principles encompass a diverse array of organisms, in terrestrial and aquatic systems, and are especially relevant when resources are distributed in a patchy manner.

The irregular shape of the natural world, especially the aggregation and dispersal of resources, impacts how herbivores behave (Holling, 1992). Some believe these patterns influence foraging behavior and may have great relevance for both natural and managed ecosystems. To quantify the patchy distribution of resources and to understand a forager's responses to these distributions, a number of investigators are using fractal geometry. Compared to Euclidean geometry, fractal geometry provides a simple formalism to describe complex environments. Euclidean geometry describes the world, objects, and their function in dimensions d, where d assumes an integer (whole number) value. For example, lines have one dimension (length), squares have two (length and width), and cubes have three (length, width, and height). The volume V of water displaced by a submerged cube is simply expressed as V = w^d where w is the length of one side and d is the number of dimensions (three) of the cube. So, a cube with side length 3 has a V of 27. Volume of alfalfa growing under a pivot can be approximated by multiplying the area under the pivot ( r²) by the height of the alfalfa (h): V = r²h. Good approximations of the Euclidean ideal include large, rectangular fields of soybeans [(Glycine max (L.) Merr.] or corn (Zea mays L.), or circles of alfalfa created by center-pivot irrigation. With a simple transformation we can convert the three-dimensional property of V to another property of a three-dimensional object such as mass.

Fractal and Euclidean geometry differ in that fractals do not require the value of dimensions (d) to be integers. Instead, fractal dimensions can be fractional values. Returning to the example of the volume of a cube, if the cube was made of Swiss cheese, it would not completely fill the three-dimensional space. Instead of d assuming a value of three, the fractal dimension of the Swiss cheese cube would fall between two and three. The Swiss cheese cube would fill more space than a two-dimensional object, but it falls short of a solid cube. Similarly, the black squares on a checkerboard do not completely fill two dimensions and a dashed line does not fill one dimension. These non-Euclidean objects have d values that are not integers (Fig. 2 ).

View larger version (15K): [in this window] [in a new window]   Fig. 2. Examples of objects with Euclidian and Fractal (non-integer) dimensions (adapted from Milne, 1997). Fractal and Euclidean geometry differ in that fractals do not require the value of dimensions (d) to be integers. Instead, they can be fractional values. For instance, if a cube was made of Swiss cheese, it would not completely fill the three-dimensional space. Instead of d assuming a value of three, the fractal dimension of the Swiss cheese cube would fall between two and three.

Mandelbrot and Milne both showed there is no absolute measurement of the perimeter of a fractal object because the estimate of perimeter is scale dependent. Mounting evidence suggests resource distributions in nature are fractal or fractal-like, exhibiting statistically similar patterns over two to three orders of magnitude of scale (Ritchie and Olff, 2000; Milne, 1992). The implication is that resource distributions are scale-dependent, and because animals of different sizes use the world at different scales, changes in resource densities affect animal distributions across scales. Even within familiar agricultural landscapes, we are intuitively aware that these large, apparently uniform areas are composed of complex geometric patterns that cannot be described by traditional geometry. Thus, where Euclidean geometry fails in biology and ecology, fractal geometry has been a very powerful tool, and it may be particularly useful in pasture and rangeland management because it provides a simple mathematical formalism that can describe the patchy distribution of resources in the environment (Ritchie, 1998).

A simple hypothetical example using a single resource and foragers of varying size demonstrates the change in perceived resource density in fractal-like systems. In Fig. 3 , we compare a forager in a uniform resource landscape (left column) to one in a fractal landscape (right column). Black cells within the grid represent food resources, whereas white cells do not contain resources. Change in forager size is represented by changes in the search area represented by the green squares over the top of the resource distributions. Forager size and search area increase down each column from a 3 x 3 square to a 9 x 9 square of arbitrary units. As search area increases, the number of cells occupied (N_u and N_f) increases in both distributions (columns), but it does so less quickly in the fractal distribution. Plotting N_u and N_f against search area in log-log space shows two power functions with different exponent values (Fig. 4 ). Realized resource density in the uniform distribution has an exponent of nearly 2 where y = 0.61x^1.91 (r² = 0.99), meaning resource density is not strongly scale-dependent. Conversely, realized food density for the fractal distribution has a much lower exponent y = 0.51x^1.26 (r² = 0.99), suggesting strong scale dependence. This illustrates how to quantify a fractal using a mass fractal approach, and the relevance of resource distributions and spatial complexity to foragers.

View larger version (73K): [in this window] [in a new window]   Fig. 3. Differential resource encounter by a forager when the environment is uniform (left column) or fractal (right column). The green box represents the area searched by a forager, while black boxes represent a resource such as food. N is a count of the number of resource cells encountered with the subscripts u and f representing uniform and fractal landscapes, respectively. A change in forager size and hence search area is represented by the green squares over the top of the resource distributions. Forager size, and therefore search area, increases down each column from a 3 x 3 square to a 9 x 9 square of arbitrary units. As search area increases, the number of cells occupied (Nu and Nf) increases in both distributions (columns), but it does so less quickly in the fractal distribution.

View larger version (14K): [in this window] [in a new window]   Fig. 4. Log-log plot showing the change in resource encounter by a forager with changing scale in a uniform and a fractal environment (see Fig. 3). The exponent in each regression approximates the resource dimension of the environment, which is close to 2 in the uniform environment and lower in the fractal environment. Realized resource density in the uniform distribution has an exponent of approximately 2 where y = 0.61x1.91 (r2 = 0.99), which indicates resource density is not strongly scale-dependent. On the other hand, realized food density within the fractal distribution has a much lower exponent y = 0.51x1.26 (r2 = 0.99), which suggests strong scale dependence.

Integrating Spatial and Temporal Facets of Foraging
On pastures and rangelands, different combinations of plants are likely to influence food intake and animal performance, and mixtures may be better than monocultures depending on the plant species involved in the mixture (Haskell et al., 2002; Chapman et al., 2007; Soder et al., 2006). For example, cattle in Missouri perform better on mixtures of endophyte-inflected tall fescue and white clover (Trifolium repens L.) than on legume-only pastures, in part because the mixtures contain complementary toxins (Provenza et al., 2003b). Forages such as white clover contain cyanogenic compounds (Cheeke, 1998), whereas endophyte-infected tall fescue produces alkaloids (Aldrich et al., 1993; Thompson and Stuedemann, 1993). Toxins in both plants limit forage intake, while a combination of plants enables animals to consume more. Cattle can consume more and have higher feed efficiencies when they graze mixtures of fescue, smooth bromegrass (Bromus inermis L.), alfalfa, and birdsfoot trefoil (Lotus corniculatus L.) than monocultures (Wiedmeier, unpublished data), and mixtures produce more forage than monocultures (MacAdam, 2002). The same may be true for combinations of plants with tannins, such as birdsfoot trefoil or serecia lespedeza (Lespedeza cuneata Dumont), and with plants high in alkaloids, such as endophyte-infected tall fescue and reed canarygrass, as tannins may have an affinity for the N contained in alkaloids. Either forage alone constrains intake, while both forages in combination are likely to yield much higher intakes and animal performance.

The spatial arrangement of forages also influences forage intake, and animal behavior and performance. For instance, sowing clover and grass in spatially separated strips enhances intake and performance compared to clover-grass mixtures. When grass and clover are planted in strips, as opposed to conventional homogeneous mixtures, intake of forage by sheep increased by 25% (265 g sheep^–1d^–1) and milk production by dairy cows increased by 11% (2.4 kg cow^–1d^–1) (Cosgrove et al., 2001). Choice allows each animal to balance the mix of grass and clover, and separation of species minimizes time selecting desired amounts of different forages. Planting forages in strips overcomes many difficulties inherent in establishing and maintaining mixed pastures, and also mimics what happens naturally as different plant species aggregate in response to environment.

The temporal availability of forages also influences foraging behavior and animal performance. Given the choice, sheep in the United Kingdom prefer to graze clover in the morning and grass in the afternoon, even though clover is more nutritious than grass (Newman et al., 1992; Parsons et al., 1994). The satiety hypothesis helps to explain this observation. Hungry sheep initially prefer clover because it is more digestible than grass. As they continue to graze clover, however, sheep acquire a mild aversion likely from the byproducts of nutrient fermentation, excess organic acids produced from soluble carbohydrates, and ammonia produced from highly digestible proteins (Cooper et al., 1995; Francis, 2003), and from the aversive effects of potentially toxic cyanogenic glycosides. The mild aversion causes them to eat grass, which is lower in nutrients and toxins than the clover, in the afternoon. During the afternoon, the aversion subsides as sheep recuperate from eating clover; by morning, they're ready for more clover.

Herders in France use an understanding of space, time, and biochemical diversity to stimulate forage intake and more fully use the range of plants available by herding sheep and goats in grazing circuits (Hubert, 1993; Meuret et al., 1994). The circuit includes a moderation phase, which provides sheep access to plants that are abundant but not highly preferred to calm a hungry flock; the next phase is a main course for the bulk of the meal with plants of moderate abundance and preference; then comes a booster phase of highly preferred plants for added diversity; and finally a dessert phase of palatable plants that complement previously grazed forages. Daily grazing circuits are designed to stimulate and satisfy an animal's appetite for different nutrients, and enable animals to maximize intake of nutrients and regulate intake of different toxins. These circuits and the ensuing patterns of forage ingestion are consistent with recent studies that show sequence influences intake of foods that vary in nutrients and toxins (Mott, 2006).

	CONCLUSIONS

TOP ABSTRACT INTRODUCTION DESCENT OF LIVESTOCK AND... SIMPLIFYING COMPLEX SYSTEMS HEALTH BENEFITS OF PLANT... CONTRASTING VIEWS ON INTAKE,... DYNAMICS OF FORAGE SELECTION... CONCLUSIONS REFERENCES

 
In "The Case Against Meat," Motavalli (2002) acknowledges "many animals graze on land that would be unsuitable for cultivation," but he then cites compelling reasons to stop livestock production altogether: (i) it takes 2.2 kg grain to produce 0.45 kg of beef; (ii) a 4 ha farm can support 60 people growing soybeans, but it can support only two people raising cattle; and (iii) reducing meat production by just 10% in the U.S. would free enough grain to feed 60 million people. These arguments, based on the poor rate of conversion of grain to meat, make the case that feeding grain to cattle is wasteful. We agree feeding grain is problematic. Ruminants are not well adapted to diets high in grain; rather, they thrive on mixtures of plants high in cellulose. Feeding large amounts of grain is an innovation of the latter half of the 20th century, and concentrates are fed to domestic ruminants in just a few developed countries where the cost is less per unit energy for concentrates than for fiber (Van Soest, 1982). Therein resides the case for using land unsuitable for cultivation of grain to produce meat, milk, wool, and hide.

Domesticated ruminants, grazing only grasses, forbs, and shrubs, have provided humans with high quality proteins for at least the last 8000 yr, and livestock can play an important role in sustainable food production on any sort of land. Well-managed grazing is a productive use of land that cannot be cultivated without being destroyed. How much land is unsuitable for cultivation? In the U.S., only 3% of farmland has no limitations to production, and even with the most careful management, row crops such as corn and soybeans should only be grown on 44% of all U.S. farmland. Another 12% can be used for "closely sown" crops such as wheat (Triticum aestivum L. emend. Thell.) or barley (Hordium vulgare L.) (Brady, 1974). Worldwide, the FAO (Bot et al., 2000) estimates only 27% of all land, rather than all farmland, is "potentially arable." So what should we do with land that is not arable and thus not suited to cultivation? No matter how much more efficient humans, or fish, or chickens are at using grain for food, if land that should not be cultivated is used for grain production, its productivity will continually diminish. However, this land can produce food sustainably through grazing, which is a natural component of all ecosystems. We have the opportunity to achieve year-round production of meat and milk on lands unsuitable for cultivation, which requires large inputs of fossil fuels, without using energy concentrates as supplements or rations.

We must also learn from mistakes of the past. We have a strong tendency to try to simplify complex, dynamic wildland and agricultural ecosystems to facilitate their management and the production of a valuable commodity such as a species of wildlife or specific products such as meat, fish, and lumber. Attempting to "simplify" ecological systems can be successful in the short term, but our efforts typically lead to disastrous long-term impacts, as shown repeatedly in marine, forest, and rangeland systems (Holling, 1995; Milne, 1997). By focusing on only one component of a system, we inevitably hasten the demise of that and other interrelated components of the system. Conversely, the structural and biological complexity inherent in natural systems increases productivity, species diversity, and system stability. A number of studies of biodiversity in the ecological and agricultural literature have documented reduced inter-annual variability in production and reduced risk of a large scale catastrophic event, such as wildfire, disease, and pest outbreaks (Holling 1995; Milne 1992, 1997). Thus, the key to re-creating systems resides in understanding complementarities in time and space among diverse plant species, including contrasting chemical compositions, growth forms and structures, phenologies, and tolerances of erratic temperatures and precipitation regimes.

The potential benefits are obvious from the standpoints of livestock nutrition and health, though there remains a great deal to be learned about how to reconstruct agro-ecosystems with plants that complement and enhance one another structurally, functionally, and biochemically not only for the benefits of herbivores, but for the well-being of soils, plants, herbivores, and people. We must learn to create mixtures of plant species that not only contain nutrients and secondary compounds, but that are agronomical