taught by animals: how understanding diet selection leads...

REVIEW: DIET SELECTION AND ZOO DIETS 43

Ullrey, D. E. (1997): Hay quality evaluation. Nutri-tion Advisory Group Handbook Fact sheet 001: 1–10.Ullrey, D. E., Crissey, S. D. & Hintz, H. F.(1997): Elephants: nutrition and dietary husbandry.

Nutrition Advisory Group Handbook Fact sheet 004:1–20.

Manuscript submitted 3 April 2002;accepted 23 June 2003

Int. Zoo Yb. (2005) 39: 43–61 © The Zoological Society of London

Taught by animals: how understanding diet selectionleads to better zoo dietsB. D. MOORE1, K. J. MARSH, I. R. WALLIS & W. J. FOLEYSchool of Botany and Zoology, Australian National University, Canberra, 0200, A.C.T.,Australia

1 Present address: School of Tropical Biology, James Cook University, Townsville, Queensland 4811, [email protected]

Wild animals invariably obtain their nutrient requi-rements, regulate their ingestion of toxins and evenself-medicate. This review suggests that, while sizeand morphology dictate gross diet, the ability toselect a diet is learnt. Animals learn to distinguishnutritious foods from less beneficial or toxic itemsthrough the positive and negative consequences ofingestion. In this process, early life experiencesappear to be critically important. Zoo animals canrarely be provided with their wild diets and care-takers substitute nutrients from other sources. Thus,a suitable range of ingredients should be provided togive the animals a stimulating and nutritious dietthat ensures excellent health.

Key-words: conditioned food aversion, diet selection,foraging, herbivore, herbivory, life experience, nutri-tion, plant secondary compounds, PSM, zoos

Although confronted with a vast array offood items, wild animals usually obtainthe nutrients they need. Some animals,particularly herbivores, do this while reg-ulating their ingestion of toxins. In con-trast, for both practical and economicreasons, humans decide what animals incaptivity should eat and this is donewithout the benefit of the sensorycapabilities of the animals and theirindividual and evolutionary experiencewith natural foods. If natural diet selec-

tion was fully understood, zoo environ-ments could be engineered to allow theanimals to choose their own diets from aselection of suitable components. Thiswould offer the added advantages ofallowing the animals to exhibit an impor-tant range of natural behaviours and pro-viding diets that are nutritionally similarto those with which the species haveevolved. In this article diet selection inanimals is reviewed and suggestions aremade as to how this knowledge can beapplied in order to provide better diets forzoo animals.

ANIMAL NUTRITIONThe subject of this review is not animalnutrition but the processes by whichanimals fulfil their nutritional require-ments. These subjects may be difficult todisentangle so we start with an overviewof nutrition that introduces diet selection.For an account of animal nutrition, spe-cific to zoo animals, the reader is referredto the review by Hume (1995) and severalchapters in the volume edited by Kleimanet al. (1996).

44 ZOO ANIMAL NUTRITION

The nutrient requirements of allanimals fall into two categories. The firstcontains the essential nutrients, like essen-tial amino acids and essential fatty acids,various vitamins and minerals; distinctmolecules that animals must obtain fromtheir diets. The second category is thenon-essential nutrients; compounds thatanimals must obtain in order to generatetheir energy requirements and to synthe-size specific molecules, like non-essentialamino acids. These nutrients are non-essential because many molecules canserve the same purpose. For example, ananimal may obtain its energy by oxidizingfat, carbohydrate or protein, while carbonchains from several molecules can beresynthesized into non-essential amino orfatty acids. Some dietary components areneeded for healthy digestive functioningeven though they appear to detract fromthe quality of the diet (e.g. fibre and grit)(Leus & Macdonald, 1995). For example,Giant anteaters Myrmecophaga tridactylain captivity frequently suffer from diar-rhoea if their diet is not supplementedwith considerable amounts of grit(Moeller, 1990). Likewise, a lack of fibrein the diets of herbivores in captivity candisrupt normal digestive function andallow teeth to overgrow (Hume & Bar-bosa, 1993). Similarly, many animals,particularly granivorous birds, deliber-ately eat stones that lodge in the digestivetract and aid physical digestion.

Evolution has led to an astoundingdiversity of animal life making it remark-able that all animal tissues, from single-cell organisms to vertebrates, have similarnutritional requirements (Moir, 1968).For example, most animals require thesame nine or ten essential amino acids.Irrespective of outward appearance, allanimals are composed of the same fun-damental components (i.e. cells), so thesimilarity of nutrient requirements is notcompletely remarkable. Evolution haslargely conserved the biochemical pro-cesses that allow cells to function. Thus,similar cells in all animals contain mito-

chondria for respiration, which requires,among other things, enzymes composed ofamino acids. In turn, the catalytic activityof enzymes requires co-factors, such as themetal atoms zinc, copper and iron, or co-enzymes, which are often vitamins.

Thus, the prime variable separatinganimals nutritionally is not the nutrientsthey need but the ability to meet theirrequirements from vastly differentsources. Evolution continues to createinfinite nutritional niches filled by animalswith profound adaptations in their diges-tive systems and in their methods of gath-ering food. Just considering vertebrates,this process has given rise to birds thatobtain their nutrition from mistletoe, vul-tures that consume decaying flesh, ant-eating mammals and reptiles, snakes thatswallow prey many times their size,whales and sharks that feed on krill, fish-eating bats, bats that feed on blood, andherbivores, like rabbits and ringtail pos-sums, that obtain substantial nutritionfrom eating their own excrement.

This diversity is an evolutionaryresponse to the variable availability ofnutrients in foods. Many dietary items donot contain all the nutrients required byan animal in the correct proportions andmany have properties that reduce theirnutritional quality. These ‘anti-qualityfactors’ (Launchbaugh et al., 2001)include toxic compounds, compoundsthat reduce the availability of nutrients bybinding to them (e.g. tannins) and com-pounds that are indigestible to mostanimals and thereby dilute the nutrients inthe food. Animals show two differentresponses on two different scales to thesechallenges. Individual animals can avoidfoods possessing anti-quality factors andchoose to eat something else (i.e. dietselection) but the scope of this selectivityis determined by the evolutionaryresponse of that species to anti-qualityfactors. One example of such a responseis provided by ruminants, which haveevolved digestive systems that countermany of the physical and chemical


defences evolved by plants on which theyfeed.

Plant cells may be divided on a func-tional basis. One part contains celluloseand other polysaccharides of the cell walls(fibre), and the other part, the cell con-tents, includes much of the protein,soluble carbohydrates and plant toxins.While some animals select plant parts forthe cell contents and avoid or discard thefibre (e.g. Giant panda Ailuropoda melan-oleuca), others utilize the abundantenergy source provided by plant cell walls.This metabolism requires a microbial syn-ergism because vertebrates do not haveenzymes for degrading plant fibre. Thismicrobial ecosystem has static require-ments, such as slow passage of digestathrough an anaerobic chamber kept atnear neutral pH. The ecological advan-tages this system confers upon the animalare enormous. Apart from making avail-able an otherwise unobtainable energysource and perhaps detoxifying someplant secondary metabolites (PSMs),microbial metabolism may exchangepoor-quality plant protein or even non-protein nitrogen for high-quality micro-bial protein. Inappropriate feeding canmake this ecosystem highly inefficient.For example, when high-quality protein isfed to foregut-fermenting herbivores, it isdegraded and resynthesized into microbialprotein of similar quality, causing largelosses of energy during the chemical trans-formations (Van Soest, 1982).

This example demonstrates the precar-ious nature of providing zoo animals witha single complete diet in lieu of an appro-priate choice of items from which they canselect their own diet. Likewise, it demon-strates the risks inherent in changing thediets of animals in captivity without theknowledge to support the practice.

Domestic ruminants, such as SheepOvis aries, cattle Bos sp, goats Capra spand Buffalo Bubalus bubalis, occur inmost human civilizations. In the lessdeveloped parts of the world theseanimals eat plant fibre that is of little use

to people. As long as the gut microbesobtain nitrogen and soluble carbohy-drates, the animals thrive and can even eatsubstantial amounts of cardboard, ascows do in many Indian cities. Furthersupplementation with a molasses and ureablock, which the animals voluntarily lick,can further stimulate the microbes andinduce feeding so that a working animalalso produces considerable amounts ofmilk (Preston & Leng, 1987). In moredeveloped countries, such animals usuallygraze but many are held in barns andfeedlots, where they eat concentratedcommercial diets. These diets supply thenutrient requirements of the animals in arather inefficient way. For example, rumi-nants are not physiologically nor morpho-logically adapted to cope withhigh-quality feeds, so they metabolizegrains inefficiently. The two feeding sys-tems, roughage or concentrates, are simul-taneously stable and precarious. Foregutfermenters that eat extremely poorroughage without a source of soluble car-bohydrates and nitrogen may reduce oreven cease feeding (Van Soest, 1982).Alternatively, if these animals are sud-denly offered a highly digestible diet likegrain, there will be serious disruption ofthe microbial population and the animalmay die. Furthermore, feeding diets withinsufficient fibre may cause metabolicdiseases because of the rapid microbialmetabolism (Van Soest, 1982). Thesemetabolic problems largely disappear ifthe animal is allowed to choose its dietfrom a selection of appropriate foods.

Some foods are inappropriate foranimals in captivity because digestiveadaptations and feeding strategies haveevolved in particular environments, andare not always optimal in other contexts.For example, the Giant panda naturallyfeeds on a highly specialized diet con-sisting almost exclusively of bamboo butoccasionally supplemented by smallanimals and carrion (Schaller et al., 1985),which provide a valuable supplement totheir poor-quality diet. However, if Giant


pandas in captivity are routinely offeredenergy-dense foods, such as rice andanimal products, obesity and digestive dis-orders can result (Bush & Montali, 1993;Kirkwood, 1993).

Related to metabolic problems are fouradverse nutritional states that are oftendifficult to distinguish. (1) Malnutritiondescribes a deficiency of a particularnutrient (Hume, 1995), including macro-nutrients, such as amino acids, or micro-nutrients, such as vitamins and minerals[e.g. calcium and vitamin D deficienciesthat lead to rickets and osteomalacia(Vickers, 1968; Morrisey et al., 1995)].(2) Undernutrition generally refers toinadequate intake of energy or of protein,caused by undereating or a poor-qualitydiet. (3) Overnutrition occurs when energyabsorption exceeds energetic demands[which are greatly reduced in captivitycompared with activity in free-living indi-viduals (Nagy, 2001)] and it can result inobesity, poor reproductive performanceand disrupted metabolic processes (e.g.Hume & Barbosa, 1993; Schwitzer &Kaumanns, 2001). (4) Toxicity. Herbi-vores, in particular, are faced with a stag-gering array of PSMs and different speciesand individuals have varied abilities todetect and tolerate them. Nutrients, suchas vitamin A (Schweigert, 1995),vitamin D [e.g. in horses (Salyi, 2002)],copper (Bremner, 1998) and selenium [e.g.in aquatic birds (Spallholz & Hoffman,2002)], may also become toxic if overing-ested, while rejection of foods containingtoxins may cause undernutrition.

Wild animals routinely select nutritiousdiets while regulating their intake of toxiccompounds within manageable limits and,given appropriate circumstances, zooanimals can also avoid these adversenutritional states.

THEORETICAL FRAMEWORKS FORORGANIZING AND SUMMARIZING THEUNDERSTANDING OF DIET SELECTIONHumans have probably tried to under-stand how other species choose their diets

since times preceding the domestication ofanimals. There are many theories on dietselection (Provenza & Balph, 1990; Moore& Foley, 2000) and five of these are out-lined here and compared as to their use-fulness in facilitating a betterunderstanding of how zoo animals can beoffered more appropriate diets.1. Euphagia refers to animals possessing‘innate hungers’ for nutrients and aninstinctive avoidance of anti-quality fac-tors. Together these abilities would allowanimals to detect nutrients and toxins,and assemble a balanced diet. A pre-requisite of this model is the ability ofanimals to sense individual compounds intheir food. Herbivores can discriminatefoods that differ in digestible energy ornutrients (Launchbaugh et al., 2001) butthis is a learnt ability. While evidence sug-gests that animals do have innate hungersfor sodium and calcium (Krieckhaus &Wolf, 1968; Coldwell & Tordoff, 1993,1996), euphagia is not useful as a generalmodel of diet selection and it is not feas-ible to assume that zoo animals are bornwith complete nutritional wisdom.2. Hedyphagia also relies upon instinctiveknowledge of what is ‘good’ and ‘bad’ toeat but postulates that animals choosetheir diet according to its hedonic value,or the pleasure it gives. The central tenetis that natural selection has caused nutri-tious foods to taste agreeable and lessnutritious or toxic foods unpleasant.Hedyphagia proposes, for example, thatsweet foods are pleasing to the sensesbecause they contain high concentrationsof simple carbohydrates, whereas bitterfoods are displeasing because they tend tobe toxic (e.g. many alkaloids). While thehedonic value of foods undoubtedly playsan important role in diet selection, hedy-phagia fails to explain diet selection ingeneral. Regardless of their initial reactionto a flavour, animals can develop aver-sions to sweet foods and preferences forbitter foods (Arnold & Hill, 1972; Glen-dinning, 1994; Pfister et al., 1996). Thereis little evidence that animals can innately


recognize the nutritional or toxic proper-ties of foods using their sensory abilitiesalone (Launchbaugh et al., 2001).3. Morphophysiology and size combine toimpose limits on an animal’s diet that ulti-mately define its dietary niche. Becauseeach species has its own array of anatom-ical and physiological adaptations that arerelated to diet selection, this provides asound theoretical basis for the nutrition ofanimals in captivity (e.g. Hume & Bar-bosa, 1993; Hume, 1995; Leus & Mac-donald, 1995).

A simple example, involving only con-spicuous morphological and physiologicaldifferences, can separate animals into twomajor trophic groups. Carnivores arespecialized at locating, capturing, sub-duing and eating other animals. Theirdentition allows them to tear and cut theirhighly nutritious diet, while a simpledigestive tract processes it. Herbivores eatless nutritious diets that frequently con-tain high proportions of indigestible plantcell-wall material and they possess exten-sive gastrointestinal adaptations to pro-cess these diets.

Within the herbivores, body size aloneexplains much of the variation in dietselection, with smaller herbivores usuallyselecting a more nutritious diet than largerherbivores. This trend occurs in severalmammalian groups, such as primates(Harvey & Clutton-Brock, 1981), rumi-nants (Hofmann, 1973) and macropodoidspecies (Sanson, 1978). The relationshipbetween diet and body size has a physio-logical basis. With increasing size, meta-bolic requirements become relativelysmaller, whereas gut capacity retains itsrelative size. Therefore, larger herbivoreshave a larger gut and frequently have ahigher herbivory rating; that is, they eatless nutritious food than smallerherbivores.

There are many exceptions to the body-size rule. Eland Taurotragus oryx(700 kg), Giraffe Giraffa camelopardalis(800 kg), Moose Alces alces (600 kg) andEuropean bison Bison bonasus (850 kg)

are all large browsers or intermediatespecies. In contrast, Oribi Ourebia ourebi(15 kg), Springbok Antidorcas marsupialis(30 kg), Speke’s gazelle Gazella spekei(15 kg), Common ringtail possum Pseu-docheirus peregrinus (1 kg), Greater gliderPetauroides volans (1·3 kg) and KoalaPhascolarctos cinereus (9 kg) all have highherbivory ratings. (Plates 1 and 2.) Howdo we explain these deviations from theclassical body size–herbivory ratingmodel? There is remarkable diversity inboth form and function within the gut ofherbivores (Hofmann, 1989; Clauss et al.,2002), as well as gut plasticity (Karasov &Hume, 1997), and similar variation inother anatomic, metabolic and behav-ioural features of herbivores. This is wellillustrated by: (1) the long neck of thegiraffe and the bipedal stance GerenukLitocranius walleri take when feeding,which allow both to select browse; (2) theability of species to store nutrients or tomigrate when conditions becomeunfavourable (Hofmann, 1989); (3) vari-ation in energy expenditure (Blaxter,1989); (4) feeding synergisms betweenpopulations of different species (De Boer& Prins, 1990). All of these factors pro-vide scope for animals to deviate from thebody-size rule.

A more subtle example exposing the fal-libility of relying solely on body size andmorphology to predict diet comes fromtwo similar-looking 4–6 kg wallabies, theTammar wallaby Macropus eugenii andthe Parma wallaby Macropus parma,which are both primarily grazers (Hume,1999). The Tammar wallaby occurs inmaritime semi-arid habitats with a Medi-terranean climate, whereas the Parma wal-laby occurs in wet sclerophyll forests withmoist or rainforest understories (Strahan,1995). Physiological adaptations separatethe species. The Tammar wallaby’snitrogen requirement is half that of theParma wallaby (Hume, 1986) and thespecies also has a capacity to concentrateurine, enabling it to drink sea water(Purohit, 1971). These adaptations allow


Plates 1 and 2. Body size explains much of the vari-ation in herbivore diet selection, as smaller herbivoresusually select more nutritious diets than larger ones,although there are many exceptions to this. For ex-ample, the Springbok Antidorcas marsupialis (top) isa small gazelle with a higher herbivory rating than theGiraffe Giraffa camelopardalis (bottom), which is amuch larger browser. Ben Moore.

it to survive in a far harsher environmentwhere high-quality foods and fresh waterare rare in summer.

An understanding of the morphologicalfactors influencing diet selection offerszoo animal caretakers a way to identifythe nutritional needs of an animal and thefoods that the animal might be able to useto assemble a healthy and balanced diet.What it cannot explain, however, is howanimals know what to eat and what toavoid. What mechanisms allow an animalto decide whether or not to eat a parti-cular food item, to know when to stopeating that food and to assemble a bal-anced diet from an array of possible fooditems?4. Optimal foraging theory models(reviewed by Stephens & Krebs, 1986) aremathematical models that are used in anattempt to identify the rules used byanimals to make foraging decisions. Theserules take various forms, depending onthe currency being optimized (e.g. energyor nutrient gain, time minimization), thedecisions made by foragers (e.g. whichitems to eat, how long to remain at a for-aging location) and the constraints of for-aging (e.g. thresholds for nutrients ortoxins). More advanced models incor-porate digestive physiology and inter-actions between the foraging animal and


the resource it eats (e.g. Whelan et al.,2000).

While it is useful to predict how ananimal will forage, two key aspects ofoptimal foraging theory models limit theirusefulness. First, decisions about the suit-ability of food items must be made by theresearcher before predictions can be madeand, second, these models classify foodsas either acceptable or unacceptablewithout grounds for accommodating thepartial preferences exhibited by mostanimals (Provenza et al., 2003). Finally,the theory ignores how an animal knowswhat it is foraging for and how it knowswhen it has found its optimal diet.5. The learning model of diet selection isthe most flexible explanation presentedhere because it can predict which fooditems animals show a preference for, theorigins of those preferences and how andwhy they change. This model can explainwhy animals do not always choose the‘best’ diet immediately and why prefer-ences fluctuate.

As the name suggests, the centralpremise of this explanation is learning.Animals learn from conspecifics and fromexperience. For example, by using thepositive and negative consequences ofingesting individual foods they learn todistinguish nutritious foods from lessbeneficial or toxic items. In keeping withthe concept of hedyphagia, each food canbe ascribed a value but in the learningmodel this value is not intrinsic butreflects the experience of the animal.Moreover, this value is dynamic and isconstantly being ‘tuned’. Animals developaversions to eating foods that elicit nega-tive effects, such as nausea or gastrointes-tinal malaise resulting from the action oftoxins or nutrient imbalance. In contrast,animals develop preferences for foods thatresult in a sense of well-being because theyare nutritious or ameliorate sickness byrectifying nutrient deficiencies or coun-tering the effects of toxins.

Aversions and preferences can vary instrength and lie at two ends of a con-

tinuum. Animals develop conditionedfood aversions and preferences as theyincorporate novel foods into their dietthroughout their lifetimes, and continu-ally adjust them so as to ensure a balanceddiet when faced with fluctuating avail-ability and quality of foods. To this end,animals display a range of sampling andforaging strategies (Provenza et al., 1998).

As the detailed examples presented willillustrate, a strong body of evidence sup-ports the learning model of diet selection.An understanding of the model is vital tothose designing zoo diets because, inmany instances, its predictions may runcounter to those of the previous modelsdiscussed. Animals choose what to eat,not on the basis of an innate nutritionalwisdom but on the basis of their ownindividual experiences with that fooditem. As a result, a certain degree ofpatience must be exercised while animalsundergo this learning process and anappropriate learning environment must beprovided. The flexibility of learning meansthat animals are certainly capable ofincorporating novel foods into their diet,even where those foods are ‘unnatural’ orlie outside the evolutionary experience ofthe species. However, morphological andphysiological adaptations mean that somepreferred foods would be more suitablethan others.

While most or maybe all animals canlearn from post-ingestive experiences andmodify diet selection accordingly (Dayet al., 1998), herbivores routinelyencounter foods with low concentrationsof nutrients and which also containtoxins.

PLANT SECONDARY METABOLITES ANDDIET SELECTION IN HERBIVORESPlants have evolved an array of defencesto avoid being eaten and these can bereferred to collectively as anti-quality fac-tors (Launchbaugh et al., 2001), includingphysical defences, such as thorns, andchemical defences or PSMs, which workin various ways. For example, anti-quality


factors may intoxicate the animal [e.g.cyanide (Cheeke, 1998)] or they may makeit difficult for herbivores to extract nutri-ents, such as when tannins bind to plantprotein.

Foley et al. (1999) pointed out thatrefractive and potentially toxic factors areubiquitous, so it is usually unrealistic forherbivores to avoid them. Animals selectnutritious items from plants that containthose negative factors that they cope withbest. In this way, herbivores regulate theamount they eat so that their intake ofPSMs does not exceed a threshold (Prov-enza, 1995a; Lawler et al., 1998a). Forexample, Common ringtail possums andCommon brushtail possums Trichosurusvulpecula stop eating when they haveingested a threshold amount of jensenone,a secondary metabolite of Eucalyptus(Lawler et al., 1998a). Likewise, Ruffedgrouse Bonasa umbellus regulate theirintake of coniferyl benzoate, a PSM ofQuaking aspen Populus tremuloides (Jak-ubas et al., 1993), cattle Bos taurus restricttheir consumption of the alkaloids in Talllarkspur Delphinium barbeyi (Pfister et al.,1997) and sheep fed diets containinglithium chloride (LiCl) regulate theirfeeding to avoid toxicity (Wang & Prov-enza, 1996a). The ability of herbivores toregulate their intake of potential toxins iscontingent upon them forming an associa-tion between the negative consequences ofthe toxin and a sensory stimulant, such asthe odour, taste or texture of the food(Provenza, 1995a). The system fails when:feedback is inappropriate for the animalto associate the toxin with the flavour,feedback occurs too long after ingestionof the food or no reliable sensory cuesexist. Although regulating the intake oftoxins is vital for all herbivores, the con-sequences of failure are variable, becausethe ability of an animal to cope withtoxins depends primarily on its capacityfor detoxifying them and that, in turn,depends on time. These are known askinetic constraints.

How are diets of animals limited by therate of detoxification of secondary com-pounds? Freeland & Janzen (1974) arguedthat the rate of detoxification is crucial indiet selection. This theory has two rami-fications as it suggests that, first, detoxi-fication capacity determines dietary nicheand, second, better detoxification allowsan animal to ingest more PSMs and hencemore food. A simple demonstration ofthis second point involves the PSM ben-zoic acid, the detoxification of whichrequires conjugation predominantly withglycine (Hutt & Caldwell, 1990). WhenCommon brushtail possums were offereddiets containing supplementary glycinethey ate more benzoic acid (G. J. Pass,W. J. Foley & S. McLean, unpublisheddata). Further ramifications of this theoryare that herbivores with few resources fordetoxification should eat mixed diets andspread a presumed array of PSMs, orcomplementary toxins, over several detox-ification systems. In other words, if PSMsimpose separate kinetic constraints thatan animal can cope with individually, itshould be able to eat more of the com-bined foods than it could of any singleitem. There is circumstantial evidence forthis concept. Common brushtail possumsallowed to feed simultaneously from dietscontaining ground Eucalyptus melliodoraleaves and ground Eucalyptus radiataleaves ate more than when they were pre-sented with either diet singly (Dearing &Cork, 1999). Similar results were observedin Mule deer Odocoileus hemionus whenoffered Sagebrush Artemisia spp andJuniper Juniperus occidentalis (Smith,1959). Of course, factors other than PSMsmay explain these results so, while thetheory is generally accepted, in theabsence of detailed toxicological studiesthere is little evidence for or against it(Foley et al., 1999).

Further evidence for kinetic differencesbetween species comes from those animalsthat eat Eucalyptus foliage, a food thatoften contains high concentrations ofPSMs. The four major Eucalyptus foli-


vores, Koalas, Greater gliders, Commonringtail possums and Common brushtailpossums, eat Eucalyptus foliage to dif-ferent extents. Greater gliders and Koalastend to eat only Eucalyptus, while the dietof the Common brushtail possum, a gen-eralist herbivore, includes variableamounts of Eucalyptus and other treefoliage (Kerle, 1984). According to Free-land & Janzen’s (1974) theory, thesespecies may be expected to have differentdetoxification capabilities and Boyle et al.(2001) found that the extent to which eachof these species is able to oxidize the Euca-lyptus terpenes, cineole and p-cymene,increases with the degree of specializationon Eucalyptus foliage. However, it is notclear whether these differences reflectvarying capacities for detoxification norwhether they impose limitations onfeeding.

There is ample evidence, from manytaxa, that the ability of an animal to copewith anti-quality factors depends on itsnutritional state and its diet (Illius &Jessop, 1995). For example, lambs atemore food containing LiCl when the die-tary energy concentration increased(Wang & Provenza, 1996a, 1997). Supple-menting the diet of sheep with extraenergy and protein enabled them to eatmore terpene-rich Sagebrush (Banneret al., 2000) but there was no effect onfeeding attributable to energy or proteinsupplementation alone (Burritt et al.,2000). Perhaps the most widespread PSMsare tannins, which by definition bind pro-tein. Blue jays Cyanocitta cristata fed low-protein diets with tannins lost body massbut feeding tannins to Blue jays in con-junction with high-protein diets elimi-nated this loss (Johnson et al., 1993).Finally, Witmer (2001) suggests that sup-plementary protein (Eastern cottonwoodPopulus deltoides catkins) allows Cedarwaxwings Bombycilla cedrorum to amelio-rate the effects of PSMs and meet the costof maintaining normal acid-base balancebetter when consuming the highly acidic

fruits of the Guelder rose Viburnumopulus.

Detoxification capacity is not neces-sarily static and, given time, animals mayincrease their ability to detoxify PSMs.One way this happens is through theinduction of specific detoxificationenzymes, such as the cytochrome P450group (Pass et al., 1999). Common brush-tail possums fed diets supplemented withEucalyptus terpenes contained higher con-centrations of cytochrome P450 in theirlivers than possums fed the diet withoutsupplementation (Pass et al., 1999).

Many ecologists believe that the sheerdiversity of PSMs and detoxification sys-tems prevents us from understanding howthese components interact to determinefood intake (Levey & del Rio, 2001).However, regardless of the PSM, there arecommon features of detoxification: ittakes time and there are negative conse-quences that may be shared by all animals(Foley et al., 1995). These attributes maybe the common currency explaining howanimals integrate a diversity of PSMs intotheir diets. The implications of kineticconstraints for the diets of animals areclear. First, when toxins are encounteredin mixed diets, animals can often eat moreand cope better with toxins. Second,animals have evolved strategies todetoxify toxins and need not avoid thementirely; in fact, exposure to these com-pounds is often necessary to optimizedetoxification systems.

THE IMPORTANCE OF AVERSIONS ANDLEARNING IN DIET SELECTIONThere are many examples of foods thathumans consume that exert unpleasanteffects even before they are eaten (e.g.blue cheese and certain fruits, such as thedurian) but which actually provide usefulnutrition; other foods seem innocuous tothe senses but may be extremely toxic (e.g.various fungi or tropical fish infected withciguatera). Foods consumed by animalsare no different and consequently animals


must integrate aversions and learningwhen choosing their diets.

Non-conditioned aversions: short-termavoidance as a result of strong tastes andflavours Pre-ingestive cues, such as taste,odour or irritant properties, neither gen-erate nausea nor result in long-termavoidance but may cause avoidance offood through non-conditioned aversions.A good example in mammals is theburning sensation caused by capsaicin in‘hot’ chilli peppers stimulating trigeminalnerve fibres in the mouth. This stimula-tion often causes the animal to reduce thesize or frequency of its meals (Pass &Foley, 2000) but animals will return tothat food later. Furthermore, the observa-tion that many bitter compounds stimu-late rat Rattus norvegicus trigeminalnerves (Liu & Simon, 1998) suggests abroad role for trigeminal stimulation innon-conditioned aversions. It has beensuggested this might include the detectionof plant terpenoids and the astringency oftannins (Foley et al., 1999). The impor-tance of trigeminal nerve stimulation isnot confined to mammals because birdsalso exhibit non-conditioned aversionsmediated this way (Jakubas & Mason,1991).

Salicin is a bitter phenolic glycosidethat induces a non-conditioned aversionin Common brushtail possums (Pass &Foley, 2000). Animals fed diets containingsalicin eat less but do not reduce theirintake further when additional salicin isplaced directly into their stomachs so theydo not detect its bitter taste. Salicin hadno effect on nitrogen balance or ureametabolism (Pass & Foley, 2000) andthere seems to be no other significant costsof detoxification (McLean et al., 2001),suggesting that the bitter taste andpossibly stimulation of the trigeminalnerve deters Common brushtail possumsrather than any post-ingestive effect.Common brushtail possums seemedunable to overcome this aversion entirely

suggesting that learning cannot eliminateit.

Glendinning (1994) put forward auseful hypothesis to explain the differingsignificance of non-conditioned aversionsto animals in different trophic levels. Heidentified a fivefold difference in bitternesstolerance across 32 mammal species andshowed that carnivores, which rarelyencounter toxins and have low toleranceto them, have low bitterness thresholds.However, herbivores face a continual bar-rage of toxins and have high bitternessthresholds. While carnivores can ‘afford’to reject bitter foods, the herbivorescannot reject all bitter foods, not leastbecause bitterness and toxicity are poorlycorrelated. Responses to pre-ingestivecues also differ within these larger trophicgroups. For example, despite its potencyas a mammalian trigeminal stimulant,birds are insensitive to capsaicin (Masonet al., 1991; Jordt & Julius, 2002). Salicinacts as a feeding deterrent to Commonbrushtail possums (Edwards, 1978; Pass &Foley, 2000) but not to hares Lepus tim-idus, which frequently browse on salicin-rich foods. Launchbaugh et al. (2001)question the existence of any innatesystem for recognizing the nutritional ortoxic properties of food based on flavouror other plant qualities. They point outthat unpleasant flavours are not univer-sally repellant (Glendinning, 1994) andthat ruminants can develop conditionedfood preferences and aversions for bothbitter and sweet flavours (Launchbaughet al., 1993). However, the evidence pre-sented suggests that, in some cases, pre-ingestive cues can influence diet selection.Such a role is not incompatible with thelearning model of diet selection. In fact, itadds an extra layer of complexity andshould not be ignored by those feedingzoo animals, as pre-ingestive cues mayimpose constraints on conditioninganimals to particular foods. It is alsopossible that pre-ingestive cues may playa greater role in diet selection by inexpe-


Plate 3. Folivores of Eucalyptus spp, like the KoalaPhascolarctos cinereus, develop conditioned flavouraversions to strongly flavoured essential oils basedupon the consequences of simultaneously ingestingother natural eucalypt toxins that occur in similar con-centrations. Ben Moore.

rienced captive animals than in free-livingindividuals.

Conditioned food aversions: long-termavoidance mediated by nausea Animalsform conditioned food aversions whenthey associate the taste or flavour of foodwith nausea or other gastrointestinal mal-aise (Garcia & Koelling, 1966; Garcia,1989; Du Toit et al., 1991; Garcia &Riley, 1998). For example, Buteo hawksavoided black mice when they had a dif-ferent flavour from white mice and werepaired with a dose of the nauseous com-pound LiCl (Brett et al., 1976). Similarly,animals ranging from slugs (Sahley et al.,1981) and rats (Garcia & Koelling, 1966)to Coyotes Canis latrans (Gustavsonet al., 1974) and sheep (Provenza et al.,1990) learn to associate flavours withemetic stimulation and then avoid theseflavours in subsequent encounters. Whydo animals form these associations? Prov-enza (1995a, 1996) proposed that byforming conditioned aversions animalscan regulate their ingestion of toxins andavoid foods that are nutritionally deficientor unbalanced.

Although the experimental use of dis-tinctive flavours, such as aniseed (e.g.Kyriazakis et al., 1997), confirms that her-bivores form conditioned flavour aver-sions, the implications are difficult togauge. The following example is perhapsmore compelling because the compoundsinvolved are secondary metabolites ofEucalyptus, which the animals may natu-rally encounter together. Wild-capturedCommon brushtail possums andCommon ringtail possums can form con-ditioned aversions to the taste of cineole,a strongly flavoured terpene, based uponthe consequences of simultaneouslyingesting the toxin jensenone (Lawleret al., 1999). (Plate 3.) Moreover, pro-viding possums with the antiemetic drug,ondansetron, allowed them to eat morejensenone (Lawler et al., 1998b); evidencethat possums use cues from the nausea

and emetic system to regulate their intakeof potentially harmful PSMs.

Animals not only form aversions to fla-vours paired with toxins but also willavoid flavours associated with nutrient-deficient foods. For example, rats learn toavoid diets that are deficient in essentialamino acids and this again appears to berelated, at least in part, to the emeticsystem (Hammer et al., 1990). After expo-sure to flavoured foods with differingenergy and protein contents, lambsdeprived of either energy or protein selectan artificial diet to rectify the deficiency(Provenza et al., 1996; Wang & Provenza,1996a,b). Thus, apart from regulating theconsumption of toxins, feedback from the


emetic system may also help animals selecta nutritious diet (Provenza, 1995a).

Once formed, conditioned aversionscan persist for a long time. Tall larkspuraversions in cattle persisted for more than2 years (Lane et al., 1990). The initial for-mation of aversions, however, appears todepend on an animal’s familiarity with thefood. If a food is novel then aversionsform readily, as shown by Burritt & Prov-enza (1991) who dosed lambs with LiCland fed them novel and familiar foodstogether. The animals developed aversionsto the novel foods but their preference forthe familiar foods remained unchanged.New foods should be introduced gradu-ally to animals in captivity because a badexperience may cause animals to refusethose foods in the future.

Dire consequences may result whenanimals fail to associate toxic compoundswith gastrointestinal malaise because ofthe delay between ingestion of a food andits post-ingestive feedback (Burritt &Provenza, 1991; Provenza et al., 1992) orwhen a toxin does not stimulate the emeticsystem. For example, pyrrolizidine alka-loids take months or even years to stimu-late the emetic system (Cheeke, 1998),while cyanide apparently fails to stimulateit (Ionescu & Buresova, 1977). Nonethe-less, Common brushtail possums in NewZealand appear to form conditioned aver-sions to cyanide in poison baits(O’Connor & Matthews, 1995, 1997) andthe Water vole Arvicola terrestris shows aclear preference for acyanogenic varietiesof White clover Trifolium repens over cya-nogenic ones (Viette et al., 2000), sug-gesting that other mechanisms may be atplay. Although these sorts of compoundsare exceptions to the rule, they can posea threat to captive and free-living animalsalike.

Importance of early life experience andlearning from other animals Early lifeexperiences, possibly the result of simul-taneous changes in neurological, physio-logical and morphological processes,

appear to have lasting effects on foodpreferences (Squibb et al., 1990). Forexample, 4–6 week-old lambs exposed toMountain mahogany Cercocarpus mon-tanus ate more of the shrub later in lifethan lambs exposed at any other time(Squibb et al., 1990) and goats reared onland dominated by Blackbrush Coleogyneramosissima ate more Blackbrush in a pentrial than goats reared on alfalfa pellets(Distel & Provenza, 1991). Similarly, Kuo(1967) concluded that cats, dogs andMynah birds raised on restricted dietsavoided new foods, while those raised onvaried diets ate novel foods more readily.A similar study conducted with ratsshowed that immature rats given earlyexperience with several flavours in waterwould accept a novel flavour more readilythan those reared with water with onlyone flavour, while mature rats given sev-eral flavours were no more willing toaccept the novel flavour than mature ratsgiven only one flavour (Capretta et al.,1975). Thus, the age of an animal when itis first exposed to a food can affect howpreferences form and how the animal for-ages in the future. The implication forfeeding animals in captivity is clear: ifpossible, introduce selections of fooditems when animals are young. Animalsthat have been exposed to dietary choicesfrom an early age will be more willing tosample new food items when they areintroduced to them and will be betterequipped to assemble a healthy, balanceddiet from a selection of ingredients.Animals that have spent their lifetimes incaptivity on a simple, unvaried diet, how-ever, may not demonstrate the sameability to learn and adapt to novel fooditems.

There are many examples, relevant tofeeding animals in captivity, of how socialsituations may impinge on feeding. Younganimals that receive maternal care or livein social groups, gain skills in diet selec-tion from other animals (Altbacker et al.,1995). This is one way of increasing ananimal’s acceptance of novel foods (Prov-


enza & Balph, 1988) or of reversing anindividual’s previous aversions. Forexample, sheep averted to a proprietarypellet ate more when feeding with non-averted animals than they did whenfeeding alone (Thorhallsdottir et al.,1990). There are similar observations forsocial interactions in rats (Galef, 1985b).In contrast, the same social conditioningcan prevent animals from acceptingfood—so-called socially-induced avoid-ance (Galef, 1985a). The mother, in parti-cular, can shape the food selection of heroffspring (Provenza, 1995b) in both posi-tive and negative ways. An extremeexample of the latter is protein-deficientkittens Felis catus that showed a prefer-ence for a protein-deficient diet if they hadoriginally eaten it with their mother (Wyr-wicka, 1981). Similarly, lambs offered achoice of two palatable shrubs showed apreference for the one they had eaten withtheir mother (Nolte et al., 1990). As withconditioned food aversions, these socially-induced preferences can last for manyyears. Lynch et al. (1983) discovered thatlambs given wheat when their motherswere present ate much more than lambsremoved from their mothers. Even 3 yearslater, a tenfold difference in wheat con-sumption remained between the groups(Green et al., 1984).

Perhaps one of the most interestingaspects of social conditioning is that someof the effects can be traced to pregnancyand lactation. For example, weanlingRabbits Oryctolagus cuniculus initiallyselected foods that their mothers ateduring pregnancy and lactation, evenwhen this food was more toxic than otherfoods offered (Altbacker et al., 1995).Experiments have suggested that experi-ence of tastes in utero can affect prefer-ences in rats (Stickrod et al., 1982;Hepper, 1988) and potentially in goats,sheep and cattle (Bradley & Mistretta,1973; Hill & Mistretta, 1990; Nolte et al.,1992) and that flavours in mother’s milkcan affect post-weaning preferences ofrats, mice Mus musculus, pigs Sus scrofa,

cattle, sheep, rabbits and humans (see ref-erences in Mennella & Beauchamp, 1997).

DIET SELECTION IN THE ZOO SETTINGThe fact that wild animals choose a nutri-tious diet while encountering toxins sug-gests that animals in captivity can do thesame if given the necessary raw ingredi-ents. As well as offering nutritional ben-efits, allowing animals to choose their owndiets also serves as an enrichment tech-nique (e.g. Zimmermann & Feistner, 1996;Masefield, 1999).

Some benefits of letting animals choosetheir diets Fedele et al. (2002) showed thatgoats, when allowed to select their owndiets, weighed more and were able to reg-ulate their intake of macronutrientsduring gestation and lactation better thangoats that had been fed a traditional no-choice diet. Likewise, lambs can selectdiets with consistent protein:energy ratioswhen presented with individual foods con-taining markedly different ratios (Wang &Provenza, 1997). The same is true of Seabass Dicentrarchus labrax, which canselect an appropriate diet, drawing allthree macronutrients from a range ofincomplete diets, each containing eitherone or two macronutrients (Aranda et al.,2000). Perhaps poultry provide the mostevidence on choice feeding. Broilerchickens Gallus gallus (meat birds) canmix foods containing different concentra-tions of protein to achieve an optimalaverage protein concentration (Shariat-madari & Forbes, 1993; Gous & Swatson,2000), while laying chickens allowed tochoose their own diet ate less, yet gainedmass faster, had better food-conversionratios and laid heavier eggs than hensreared on a commercial pullet growermash (Olver & Malan, 2000). Theseexamples are primarily concerned withmacronutrients and domesticated animalsbut, given the choice, it seems that allanimals can select diets that rectify defi-ciencies and counter toxins. In fact, if free-


living animals did not do this, they wouldnot survive.

We have shown that animals can selectdiets to regulate their intake of toxins butanimals can also select substances of nonutritional value that can ameliorate thetoxic and other negative effects of foods.Recently, there has been much interest inthe ability of livestock to ‘self-medicate’with polyethylene glycol (PEG) to allaythe protein-binding effects of tannins. Bybinding tannins, PEG prevents tanninsbinding protein, which then becomesavailable to the animal (Ben Salem et al.,1999a,b). A secondary effect is that theanimals eat more (Ben Salem et al., 2000;Landau et al., 2000). Not only willanimals self-medicate but also they willadjust the dose; for example, lambs thateat more PEG as the tannin content oftheir diet increases (Provenza et al., 2000;Villalba & Provenza, 2001). There areother examples of animals self-medicating.For example, lambs fed grain choose toingest solutions containing sodium bicar-bonate and lasalocid to help attenuate therumen acidosis caused by this diet (Phy &Provenza, 1998). Similarly, free-livingJapanese macaques Macaca fuscata willingest soils rich in clay minerals to buffergastric upset caused by eating supple-mented diets rich in soluble carbohydratesand protein (Wakibara et al., 2001).Indeed, many primates practice geophagyas a means of mineral supplementation,adsorption of toxins and treatment ofdiarrhoea (see review by Krishnamani &Mahaney, 2000). As with other phe-nomena discussed in this review, self-med-ication occurs across taxa. Diamond et al.(1999) concluded that 11 species of frugiv-orous New Guinea birds ingested soilbecause it bound poisonous and/or bitter-tasting PSMs in their diets. This is anexample of one way in which free-livinganimals can use diet selection to overcomelimitations imposed by the inability oftheir morphology and physiology to copewith certain foods, and thus expand theirdietary niche. Providing similar oppor-

tunities can allow some zoo animals to dothe same.

CONCLUSIONIf animals are given the appropriate fooditems to choose from, they can select dietsthat satisfy their nutrient requirementswithout poisoning them, rectify any nutri-tional deficiencies and increase theircapacity to detoxify plant toxins or ame-liorate their effects. This is not an abilitythat animals are born with but is one thatmust be learnt. A basic set of conditionsis required for a healthy psychologicaldevelopment in humans: exposure to abroad range of experiences facilitateslearning from mistakes and the develop-ment of innovative coping strategies. Theimpact of these experiences is greatestduring the early years when neural path-ways are still forming and behaviour pat-terns are established that will last alifetime. The developmental journey is notundertaken alone, as parents, siblings andsociety all have valuable parts to play asrole models and as protectors from poten-tially dangerous experiences. While noneof these conditions is essential to anindividual’s biological survival, they canall contribute to psychological well-being,the richness of life experience and theability to cope with changing environ-ments. An analogous set of conditionsallows zoo animals to develop rich, variedand flexible diets.

For animals to develop appropriatefood preferences and aversions and learnhow to obtain a nutritious diet from arange of food items, they must becomefamiliar with this range, ideally at an earlyage. In many cases, interactions with con-specifics will be invaluable but in the caseof zoo animals, it will often fall partiallyor entirely to the caretakers to promotethe consumption of some foods and towithhold others. If we understand how themorphology and physiology of a speciescombine to limit its dietary niche it shouldbe possible to identify its nutritional needsand provide an adequate diet in captivity.


To go one step further and ensure this dietis diverse, palatable and capable ofchanging with an animal’s needs, how-ever, we would do well to learn from thosewith a vested interest and a lifetime ofexperience: in other words, we should betaught by the animals.

ACKNOWLEDGEMENT

The authors wish to thank Jane DeGabriel forhelpful comments on the manuscript.

REFERENCESAltbacker, V., Hudson, R. & Bilko, A. (1995):Rabbit-mothers diet influences pups’ later foodchoice. Ethology 99: 107–116.Aranda, A., Sanchez-Vazquez, F. J., Zamora, S.& Madrid, J. A. (2000): Self-design of fish diets bymeans of self-feeders: validation of procedures.Journal of Physiology and Biochemistry 56: 155–166.Arnold, G. W. & Hill, J. L. (1972): Chemical fac-tors affecting selection of food plants by ruminants.In Phytochemical ecology: 71–101. Harborne, J. B.(Ed.). New York: Academic Press.Banner, R. E., Rogosic, J., Burritt, E. A. & Prov-enza, F. D. (2000): Supplemental barley and char-coal increase intake of sagebrush by lambs. Journalof Range Management 53: 415–420.Ben Salem, H., Nefzaoui, A., Ben Salem, L. & Tis-serand, J. L. (1999a): Different means of adminis-tering polyethylene glycol to sheep: effect on thenutritive value of Acacia cyanophylla Lindl. foliage.Animal Science 68: 809–818.Ben Salem, H., Nefzaoui, A., Ben Salem, L. & Tis-serand, J. L. (1999b): Intake, digestibility, urinaryexcretion of purine derivatives and growth by sheepgiven fresh, air-dried or polyethylene glycol-treatedfoliage of Acacia cyanophylla Lindl. Animal FeedScience and Technology 78: 297–311.Ben Salem, H., Nefzaoui, A., Ben Salem, L. & Tis-serand, J. L. (2000): Deactivation of condensed tan-nins in Acacia cyanophylla Lindl. foliage bypolyethylene glycol in feed blocks. Effect on feedintake, diet digestibility, nitrogen balance, microbialsynthesis and growth by sheep. Livestock ProductionScience 64: 51–60.Blaxter, K. L. (1989): Energy metabolism in animalsand man. Cambridge: Cambridge University Press.Boyle, R., McLean, S., Foley, W., Davies, N. W.,Peacock, E. J. & Moore, B. (2001): Metabolites ofdietary 1,8-cineole in the male koala (Phascolarctoscinereus). Comparative Biochemistry & Physiology C129: 385–395.Bradley, R. M. & Mistretta, C. M. (1973): Thegustatory sense in foetal sheep during the last thirdof gestation. Journal of Physiology 231: 271–282.

Bremner, I. (1998): Manifestations of copper excess.American Journal of Clinical Nutrition 67:1069S–1073S.Brett, L. P., Hankins, W. G. & Garcia, J. (1976):Prey-lithium aversions. III. Buteo hawks. BehavioralBiology 17: 87–98.Burritt, E. A. & Provenza, F. D. (1991): Abilityof lambs to learn with a delay between food inges-tion and consequences given meals containing noveland familiar foods. Applied Animal Behaviour Sci-ence 32: 179–189.Burritt, E. A., Banner, R. E. & Provenza, F. D.(2000): Sagebrush ingestion by lambs: effects ofexperience and macronutrients. Journal of RangeManagement 53: 91–96.Bush, M. & Montali, R. J. (1993): Medicalmanagement of giant pandas. In Zoo and wild animalmedicine. Current therapy: 408–411. Fowler, M. E.(Ed.). Philadelphia, PA: W. B. Saunders Company.Capretta, P. J., Petersik, J. T. & Stewart, D. J.(1975): Acceptance of novel flavours is increasedafter early experience of diverse tastes. Nature 254:689–691.Cheeke, P. R. (1998): Natural toxicants in feeds, for-ages, and poisonous plants (2nd edn). Danville, IL:Interstate Publishers Incorporated.Clauss, M., Lechner-Doll, M. & Streich, W. J.(2002): Faecal particle size distribution in captivewild ruminants: an approach to the browser/grazerdichotomy from the other end. Oecologia (Berlin)131: 343–349.Coldwell, S. E. & Tordoff, M. G. (1993): Latentlearning about calcium and sodium. AmericanJournal of Physiology: Regulatory, Integrative &Comparative Physiology 265: 1480–1484.Coldwell, S. E. & Tordoff, M. G. (1996): Imme-diate acceptance of minerals and HCl by calcium-deprived rats: brief exposure tests. American Journalof Physiology: Regulatory, Integrative & ComparativePhysiology 271: 11–17.Day, J. E. L., Kyriazakis, I. & Rogers, P. J. (1998):Food choice and intake: towards a unifying frame-work of learning and feeding motivation. NutritionResearch Reviews 11: 25–43.De Boer, W. F. & Prins, H. H. T. (1990): Largeherbivores that strive mightily but eat and drink asfriends. Oecologia (Berlin) 82: 264–274.Dearing, M. D. & Cork, S. (1999): Role of detox-ification of plant secondary compounds on dietbreadth in a mammalian herbivore, Trichosurus vul-pecula. Journal of Chemical Ecology 25: 1205–1219.Diamond, J., Bishop, K. D. & Gilardi, J. D. (1999):Geophagy in New Guinea birds. Ibis 141: 181–193.Distel, R. A. & Provenza, F. D. (1991): Experienceearly in life affects voluntary intake of blackbrushby goats. Journal of Chemical Ecology 17: 431–450.Du Toit, J. T., Provenza, F. D. & Nastis, A.(1991): Conditioned taste-aversions—how sick musta ruminant get before it learns about toxicity infoods. Applied Animal Behaviour Science 30: 35–46.


Edwards, W. R. N. (1978): Effect of salicin contenton palatability of Populus foliage to opossum (Tri-chosurus vulpecula). New Zealand Journal of Science21: 103–106.Fedele, V., Claps, S., Rubino, R., Calandrelli, M.& Pilla, A. M. (2002): Effect of free-choice and tra-ditional feeding systems on goat feeding behaviourand intake. Livestock Production Science 74: 19–31.Foley, W. J., McLean, S. & Cork, S. J. (1995):Consequences of biotransformation of plant sec-ondary metabolites on acid-base metabolism inmammals—a final common pathway? Journal ofChemical Ecology 21: 721–743.Foley, W. J., Iason, G. R. & McArthur, C. (1999):Role of plant secondary metabolites in the nutri-tional ecology of mammalian herbivores: how farhave we come in 25 years? In Nutritional ecology ofherbivores. Proceedings of the Vth international sym-posium on the nutrition of herbivores: 130–209. Jung,H.-J. G. & Fahey Jr, G. C. (Eds). Savoy, IL: Amer-ican Society of Animal Science.Freeland, W. J. & Janzen, D. H. (1974): Strategiesin herbivory by mammals: the role of plant sec-ondary compounds. American Naturalist 108:269–289.Galef Jr, B. G. (1985a): Direct and indirect behav-ioral pathways to the social transmission of foodavoidance. Annals of the New York Academy of Sci-ences 443: 203–215.Galef Jr, B. G. (1985b): Socially induced diet pref-erence can partially reverse a LiCl-induced diet aver-sion. Animal Learning & Behavior 13: 415–418.Garcia, J. (1989): Food for Tolman: cognition andcathexis in concert. In Aversion, avoidance and anx-iety: perspectives on aversively motivated behavior:45–85. Archer, T. & Nilsson, L. (Eds). Hillsdale, NJ:Lawrence Erlbaum Associates.Garcia, J. & Koelling, R. A. (1966): Relation ofcue to consequence in avoidance learning. Psychon-omic Science 4: 123–124.Garcia, J. & Riley, A. L. (1998): Conditioned tasteaversions and preferences. In Comparative psy-chology: a handbook: 549–561. Greenberg, G. &Haraway, M. M. (Eds). New York: GarlandPublishing.Glendinning, J. I. (1994): Is the bitter rejectionresponse always adaptive? Physiology and Behavior56: 1217–1227.Gous, R. M. & Swatson, H. K. (2000): Mixtureexperiments: a severe test of the ability of a broilerchicken to make the right choice. British Poultry Sci-ence 41: 136–140.Green, G. C., Elwin, R. L., Mottershead, B. E. &Lynch, J. J. (1984): Long-term effects of earlyexperience to supplementary feeding in sheep. Pro-ceedings of the Australian Society of Animal Produc-tion 15: 373–375.Gustavson, C. R., Garcia, J., Hankins, W. G. &Rusiniak, K. W. (1974): Coyote predation controlby aversive conditioning. Science 184: 581–583.

Hammer, V. A., Gietzen, D. W., Beverly, J. L. &Rogers, Q. R. (1990): Serotonin3 receptor antago-nists block anorectic responses to amino acid imbal-ance. American Journal of Physiology: Regulatory,Integrative & Comparative Physiology 259:R627–R636.Harvey, P. H. & Clutton-Brock, T. H. (1981): Pri-mate home-range size and metabolic needs. Behav-ioral Ecology and Sociobiology 8: 151–155.Hepper, P. G. (1988): Adaptive fetal learning: pre-natal exposure to garlic affects postnatal preferences.Animal Behaviour 36: 935–936.Hill, D. L. & Mistretta, C. M. (1990): Develop-mental neurobiology of salt taste sensation. Trendsin Neuroscience 13: 188–195.Hofmann, R. R. (1973): The ruminant stomach:stomach structure and feeding habits of East Africangame ruminants. Nairobi: East African LiteratureBureau.Hofmann, R. R. (1989): Evolutionary steps of eco-physiological adaptation and diversification of rumi-nants: a comparative view of their digestive system.Oecologia (Berlin) 78: 443–457.Hume, I. D. (1986): Nitrogen metabolism in theparma wallaby, Macropus parma. Australian Journalof Zoology 34: 147–155.Hume, I. D. (1995): General concepts of nutritionand nutritional ecology. In Research and captivepropagation: 90–98. Ganslosser, U., Hodges, J. K. &Kaumanns, W. (Eds). Furth: Filander Verlag.Hume, I. D. (1999): Marsupial nutrition. Cambridge:Cambridge University Press.Hume, I. D. & Barbosa, P. S. (1993): Designing arti-ficial diets for captive marsupials. In Zoo and wildanimal medicine. Current therapy: 281–293. Fowler,M. E. (Ed.). Philadelphia, PA: W. B. SaundersCompany.Hutt, A. J. & Caldwell, J. (1990): Amino acidconjugation. In Conjugation reactions in drug meta-bolism: an integral approach: 273–305. Mulder, G. J.(Ed.). London: Taylor and Francis Ltd.Illius, A. W. & Jessop, N. S. (1995): Modelingmetabolic costs of allelochemical ingestion by for-aging herbivores. Journal of Chemical Ecology 21:693–719.Ionescu, E. & Buresova, O. (1977): Failure to elicitconditioned taste aversion by severe poisoning.Pharmacology, Biochemistry and Behavior 6:251–254.Jakubas, W. J. & Mason, J. R. (1991): Role of aviantrigeminal sensory system in detecting coniferyl ben-zoate, a plant allelochemical. Journal of ChemicalEcology 17: 2213–2221.Jakubas, W. J., Karasov, W. H. & Guglielmo, C.G. (1993): Ruffed grouse tolerance and biotransfor-mation of the plant secondary metabolite coniferylbenzoate. Condor 95: 625–640.Johnson, W. C., Thomas, L. & Adkisson, C. S.(1993): Dietary circumvention of acorn tannins byblue jays. Oecologia (Berlin) 94: 159–164.


Jordt, S. & Julius, D. (2002): Molecular basis forspecies-specific sensitivity to ‘hot’ chili peppers. Cell108: 421–430.Karasov, W. H. & Hume, I. D. (1997): Vertebrategastrointestinal system. In Handbook of Physiology.Section 13. Comparative Physiology 1: 409–480.Dantzler, W. H. (Ed.). New York: Oxford Univer-sity Press.Kerle, J. A. (1984): Variation in the ecology of Tri-chosurus: its adaptive significance. In Possums andgliders: 115–128. Smith, A. P. & Hume, I. D. (Eds).Sydney: Surrey Beatty in association with AustralianMammal Society.Kirkwood, J. K. (1993): Medical management ofpandas in Europe. In Zoo and wild animal medicine.Current therapy: 404–407. Fowler, M. E. (Ed.). Phil-adelphia, PA: W. B. Saunders Company.Kleiman, D. G., Allen, M. E., Thompson, K. V. &Lumpkin, S. (Eds) (1996): Wild mammals in cap-tivity: principles and techniques. Chicago, IL: Uni-versity of Chicago Press.Krieckhaus, E. E. & Wolf, G. (1968): Acquisitionof sodium by rats: interaction of innate mechanismsand latent learning. Journal of Comparative Physi-ology and Psychology 65: 197–201.Krishnamani, R. & Mahaney, W. C. (2000):Geophagy among primates: adaptive significanceand ecological consequences. Animal Behaviour 59:899–915.Kuo, Z. Y. (1967): The dynamics of behaviourdevelopment: an epigenetic view. New York: RandomHouse.Kyriazakis, I., Papachristou, T. G., Duncan, A.J. & Gordon, I. J. (1997): Mild conditioned aver-sions developed by sheep towards flavours associatedwith plant secondary compounds. Journal ofChemical Ecology 23: 727–746.Landau, S., Silanikove, N., Nitsan, Z., Barkai,D., Baram, H., Provenza, F. D. & Perevolotsky,A. (2000): Short-term changes in eating patternsexplain the effects of condensed tannins on feedintake in heifers. Applied Animal Behaviour Science69: 199–213.Lane, M. A., Ralphs, M. H., Olsen, J. D., Prov-enza, F. D. & Pfister, J. A. (1990): Conditionedtaste aversion: potential for reducing cattle loss tolarkspur. Journal of Range Management 43: 127–131.Launchbaugh, K. L., Provenza, F. D. & Burritt,E. A. (1993): How herbivores track variable envi-ronments: response to variability of phytotoxins.Journal of Chemical Ecology 19: 1047–1056.Launchbaugh, K. L., Provenza, F. D. & Pfister,J. A. (2001): Herbivore response to anti-quality fac-tors in forages. Journal of Range Management 54:431–440.Lawler, I. R., Foley, W. J., Eschler, B. M., Pass,D. M. & Handasyde, K. (1998a): Intraspecific var-iation in Eucalyptus secondary metabolites deter-mines food intake by folivorous marsupials.Oecologia (Berlin) 116: 160–169.

Lawler, I. R., Foley, W. J., Pass, G. J. & Eschler,B. M. (1998b): Administration of a 5HT(3) receptorantagonist increases the intake of diets containingEucalyptus secondary metabolites by marsupials.Journal of Comparative Physiology B 168: 611–618.Lawler, I. R., Stapley, J., Foley, W. J. & Eschler,B. M. (1999): Ecological example of conditionedflavor aversion in plant–herbivore interactions: effectof terpenes of Eucalyptus leaves on feeding bycommon ringtail and brushtail possums. Journal ofChemical Ecology 25: 401–415.Leus, K. & Macdonald, A. A. (1995): Gastrointes-tinal anatomy, diet selection and digestion in mam-mals: a brief overview. In Research and captivepropagation: 99–114. Ganslosser, U., Hodges, J. K.& Kaumanns, W. (Eds). Furth: Filander Verlag.Levey, D. J. & Del Rio, C. M. (2001): It takes guts(and more) to eat fruit: lessons from avian nutri-tional ecology. Auk 118: 819–831.Liu, L. & Simon, S. A. (1998): Responses of culturedrat trigeminal ganglion neurons to bitter tastants.Chemical Senses 23: 125–130.Lynch, J. J., Keogh, R. G., Elwin, R. L., Green,G. C. & Mottershead, B. E. (1983): Effects of earlyexperience on the post-weaning acceptance of wholegrain wheat by fine-wool Merino lambs. Animal Pro-duction 36: 175–183.Masefield, W. (1999): Forage preferences andenrichment in a group of captive Livingstone’s fruitbats Pteropus livingstonii. Dodo. Journal of DurrellWildlife Conservation Trust 35: 48–56.Mason, J. R., Bean, N. J., Shah, P. S. & Clark, L.(1991): Taxon-specific differences in responsivenessto capsaicin and several analogs: correlates betweenchemical structure and behavioral aversiveness.Journal of Chemical Ecology 17: 2539–2551.McLean, S., Pass, G. J., Foley, W. J., Brandon, S.& Davies, N. W. (2001): Does excretion of sec-ondary metabolites always involve a measurablemetabolic cost? Fate of plant antifeedant salicin incommon brushtail possum, Trichosurus vulpecula.Journal of Chemical Ecology 27: 1077–1089.Mennella, J. A. & Beauchamp, G. K. (1997): Theontogeny of human flavour perception. In Tastingand smelling: handbook of perception and cognition:199–221. Beauchamp, G. K. & Bartoshuk, L. (Eds).San Diego, CA: Academic Press.Moeller, W. (1990): Modern Xenarthrans. InGrzimek’s encyclopedia of mammals: 583–626.Parker, S. P. (Ed.). New York: McGraw-Hill.Moir, R. J. (1968): Ruminant digestion and evolu-tion. In Handbook of physiology, section 6. Alimen-tary canal. V. (Bile; digestion; ruminal physiology):2673–2694. Code, C. F. & Heidel, W. (Eds). Wash-ington, DC: American Physiological Society.Moore, B. D. & Foley, W. J. (2000): A review offeeding and diet selection in koalas (Phascolarctoscinereus). Australian Journal of Zoology 48: 317–333.Morrisey, J. K., Reichard, T., Lloyd, M. & Ber-nard, J. (1995): Vitamin D-deficiency rickets in threecolobus monkeys (Colobus guereza kikuyuensis) at


the Toledo Zoo. Journal of Zoo and Wildlife Medi-cine 26: 564–568.Nagy, K. A. (2001): Food requirements of wildanimals: predictive equations for free-living mam-mals, reptiles and birds. Nutrition Abstracts andReviews, Series B 71: 21R–31R.Nolte, D. L., Provenza, F. D. & Balph, D. F.(1990): The establishment and persistence of foodpreferences in lambs exposed to selected foods.Journal of Animal Science 68: 998–1002.Nolte, D. L., Provenza, F. D., Callan, R. &Panter, K. E. (1992): Garlic in the ovine fetalenvironment. Physiology & Behavior 52: 1091–1093.O'Connor, C. E. & Matthews, L. R. (1995): Cya-nide induced aversions in the possum (Trichosurusvulpecula)—effect of route of administration, dose,and formulation. Physiology & Behavior 58: 265–271.O'Connor, C. E. & Matthews, L. R. (1997): Dura-tion of cyanide-induced conditioned food aversionsin possums. Physiology & Behavior 62: 931–933.Olver, M. D. & Malan, D. D. (2000): The effectof choice-feeding from 7 weeks of age on the pro-duction characteristics of laying hens. South AfricanJournal of Animal Science 30: 110–114.Pass, G. J. & Foley, W. J. (2000): Plant secondarymetabolites as mammalian feeding deterrents: sepa-rating the effects of the taste of salicin from its post-ingestive consequences in the common brushtailpossum (Trichosurus vulpecula). Journal of Com-parative Physiology B 170: 185–192.Pass, G. J., McLean, S. & Stupans, I. (1999): Induc-tion of xenobiotic metabolising enzymes in thecommon brushtail possum, Trichosurus vulpecula, byEucalyptus terpenes. Comparative Biochemistry andPhysiology Part C 124: 239–246.Pfister, J. A., Manners, G. D., Gardner, D. R.,Price, K. W. & Ralphs, M. H. (1996): Influence ofalkaloid concentration on acceptability of tall lark-spur (Delphinium spp.) to cattle and sheep. Journalof Chemical Ecology 22: 1147–1168.Pfister, J. A., Provenza, F. D., Manners, G. D.,Gardner, D. R. & Ralphs, M. H. (1997): Tall lark-spur ingestion: can cattle regulate intake below toxiclevels. Journal of Chemical Ecology 23: 759–777.Phy, T. S. & Provenza, F. D. (1998): Sheep fedgrain prefer foods and solutions that attenuate aci-dosis. Journal of Animal Science 76: 954–960.Preston, T. R. & Leng, R. A. (1987): Matchingruminant production systems with available resourcesin the tropics and sub-tropics. Armidale: PenambulBooks.Provenza, F. D. (1995a): Postingestive feedback asan elementary determinant of food preference andintake in ruminants. Journal of Range Management48: 2–17.Provenza, F. D. (1995b): Tracking variable envi-ronments: there is more than one kind of memory.Journal of Chemical Ecology 21: 911–923.Provenza, F. D. (1996): Acquired aversions as thebasis for varied diets of ruminants foraging on

rangelands. Journal of Animal Science 74:2010–2020.Provenza, F. D. & Balph, D. F. (1988): Develop-ment of dietary choice in livestock on rangelandsand its implications for management. Journal ofAnimal Science 66: 2356–2368.Provenza, F. D. & Balph, D. F. (1990): Applica-bility of five diet-selection models to various foragingchallenges ruminants encounter. In Behaviouralmechanisms of food selection: 423–459. Hughes, R.N. (Ed.). Berlin: Springer-Verlag.Provenza, F. D., Burritt, E. A., Clausen, T. P.,Bryant, J. P., Reichardt, P. B. & Distel, R. A.(1990): Conditioned flavour aversions: a mechanismfor goats to avoid condensed tannins in blackbrush.The American Naturalist 136: 810–828.Provenza, F. D., Pfister, J. A. & Cheney, C. D.(1992): Mechanisms of learning in diet selection withreference to phytotoxicosis in herbivores. Journal ofRange Management 45: 36–45.Provenza, F. D., Scott, C. B., Phy, T. S. & Lynch,J. J. (1996): Preference of sheep for foods varying inflavors and nutrients. Journal of Animal Science 74:2355–2361.Provenza, F. D., Villalba, J. J., Cheney, C. D. &Werner, S. J. (1998): Self-organization of foragingbehaviour: from simplicity to complexity withoutgoals. Nutrition Research Reviews 11: 199–222.Provenza, F. D., Burritt, E. A., Perevolotsky, A.& Silanikove, N. (2000): Self-regulation of intake ofpolyethylene glycol by sheep fed diets varying intannin concentrations. Journal of Animal Science 78:1206–1212.Provenza, F. D., Villalba, J. J. & Bryant, J. P.(2003): Foraging by herbivores: linking the bio-chemical diversity of plants to herbivore culture andlandscape diversity. In Landscape ecology andresource management: linking theory with practice:387–421. Bissonette, J. A. & Storch, I. (Eds). Wash-ington, DC: Island Press.Purohit, K. G. (1971): Absolute duration of sur-vival of tammar wallaby (Macropus eugenii, Mar-supialia) on sea water and dry food. ComparativeBiochemistry & Physiology 39A: 473–481.Sahley, C. L., Gelperin, A. & Rudy, J. (1981):One-trial associative learning modifies food odorpreferences of a terrestrial mollusc. Proceedings ofthe National Acadamy of Science USA 78: 640–642.Salyi, G. (2002): Toxicoses of horses. 3. Vitamins,drugs and other substances. Magyar AllatorvosokLapja 124: 5–10.Sanson, G. D. (1978): The evolution and significanceof mastication in the Macropodidae. AustralianMammalogy 2: 23–28.Schaller, G. B., Hu, J., Pan, W. & Zhu, J. (1985):The giant pandas of Wolong. Chicago, IL: Universityof Chicago Press.Schweigert, F. J. (1995): Comparative aspects ofvitamin A and carotenoid metabolism in exoticmammals. In Research and captive propagation:


130–146. Ganslosser, U., Hodges, J. K. & Kau-manns, W. (Eds). Furth: Filander Verlag.Schwitzer, C. & Kaumanns, W. (2001): Bodyweights of ruffed lemurs (Varecia variegata) inEuropean zoos with reference to the problem ofobesity. Zoo Biology 20: 261–269.Shariatmadari, F. & Forbes, J. M. (1993): Growthand food intake responses to diets of different pro-tein contents and a choice between diets containingtwo concentrations of protein in broiler and layerstrains of chicken. British Poultry Science 34:959–970.Smith, A. D. (1959): Adequacy of some importantbrowse species in overwintering mule deer. Journalof Range Management 12: 8–13.Spallholz, J. E. & Hoffman, D. J. (2002): Seleniumtoxicity: cause and effects in aquatic birds. AquaticToxicology 57: 27–37.Squibb, R. C., Provenza, F. D. & Balph, D. F.(1990): Effect of age of exposure on consumption ofa shrub by sheep. Journal of Animal Science 68:987–997.Stephens, D. W. & Krebs, J. R. (1986): Foragingtheory. Princeton, NJ: Princeton University Press.Stickrod, G., Kimble, D. P. & Smotherman, W. P.(1982): In utero taste/odor aversion condition in therat. Physiology and Behavior 28: 5–7.Strahan, R. (1995): The mammals of Australia: thenational photographic index of Australian wildlife.Chatswood, NSW: Reed Books and The AustralianMuseum.Thorhallsdottir, A. G., Provenza, F. D. &Balph, D. F. (1990): Social influences on condi-tioned food aversions in sheep. Applied AnimalBehaviour Science 25: 45–50.Van Soest, P. J. (1982): Nutritional ecology of theruminant. Corvallis, OR: O. & B. Books.Vickers, J. H. (1968): Osteomalacia and rickets inmonkeys. In Current veterinary therapy III: 392–393.Kirk, R. W. (Ed.). Philadelphia, PA: W. B.Saunders.Viette, M., Tettamanti, C. & Saucy, F. (2000):Preference for acyanogenic white clover (Trifoliumrepens) in the vole Arvicola terrestris. II. General-

ization and further investigations. Journal ofChemical Ecology 26: 101–122.Villalba, J. J. & Provenza, F. D. (2001): Prefer-ence for polyethylene glycol by sheep fed a que-bracho tannin diet. Journal of Animal Science 79:2066–2074.Wakibara, J. V., Huffman, M. A., Wink, M.,Reich, S., Aufreiter, S., Hancock, R. G. V.,Sodhi, R., Mahaney, W. C. & Russel, S. (2001):The adaptive significance of geophagy for Japanesemacaques (Macaca fuscata) at Arashiyama, Japan.International Journal of Primatology 22: 495–514.Wang, J. & Provenza, F. D. (1996a): Food depri-vation affects preference of sheep for foods varyingin nutrients and a toxin. Journal of Chemical Ecology22: 2011–2021.Wang, J. & Provenza, F. D. (1996b): Food pref-erence and acceptance of novel foods by lambsdepend on the composition of the basal diet. Journalof Animal Science 74: 2349–2354.Wang, J. & Provenza, F. D. (1997): Dynamics ofpreference by sheep offered foods varying in flavors,nutrients, and a toxin. Journal of Chemical Ecology23: 275–288.Whelan, C. J., Brown, J. S., Schmidt, K. A.,Steele, B. B. & Willson, M. F. (2000): Linking con-sumer-resource theory and digestive physiology:application to diet shifts. Evolutionary EcologyResearch 2: 911–934.Witmer, M. C. (2001): Nutritional interactions andfruit removal: cedar waxwing consumption ofViburnum opulus fruits in spring. Ecology 82:3120–3130.Wyrwicka, W. (1981): The development of food pref-erences: parental influences and the primacy effect.Springfield, IL: Charles C. Thomas.Zimmermann, A. & Feistner, A. T. C. (1996):Effects of feeding enrichment on ruffed lemursVarecia variegata variegata and Varecia v. rubra.Dodo. Journal of the Wildlife Preservation Trusts 32:67–75.

Manuscript submitted 21 May 2002;accepted 19 June 2003

taught by animals: how understanding diet selection leads...

Documents