Journal of Chemical Ecology, Vol. 20, No. 3, 1994

HOW GOATS LEARN TO DISTINGUISH BETWEEN NOVEL FOODS THAT DIFFER IN POSTINGESTIVE CONSEQUENCES

FREDERICK ELIZABETH

D. P R O V E N Z A , L* J U S T I N J. L Y N C H , 2 A. B U R R I T T , l and C O D Y

B. S C O T T 1

tRange Science Department, Utah State University Logan, Utah 84322-5230 2CSIRO Division of Animal Production, Private Mail Bag Armidale, NSW, Australia 2350 (Received August 23, 1993; accepted October 25, 1993)

Abstract--To better understand some of the mechanisms that control selection of novel foods differing in postingestive consequences, we offered goats current season's (CSG) and older (OG) growth twigs from the shrub btackbmsh (Coleogyne ramosissima). CSG is higher than OG in nitrogen (1.04% v. 0.74%) and it is more digestible in vitro in goat rumen fluid (48% v. 38%). Nevertheless, goats acquire a preference for OG because CSG contains much higher levels of a condensed tannin that causes a learned food aversion. When CSG and OG were offered to goat naive to blackbrush, the goats did not choose either OG or CSG exclusively, but when they finally (I) ate more CSG than OG within a meal (averages of 44 g and 16 g, respectively) and (2) ate enough CSG within the meal to acquire an aversion (average of 44 g), they ingested less CSG than OG from then onward. Accordingly, the change in food selection resulting from postingestive feedback was influenced by the amount of each food ingested within a meal. This was further shown when we varied the amounts of CSG and OG that goats ingested within a meal, and then gave them by garage the toxin lithium chloride (LiCl). They subsequently ate less of the food eaten in the greatest amount, regardless of whether it was CSG or OG. The salience of the flavor (i.e., taste and odor) of CSG and OG also played a role in the acquired aversion to CSG. Salience evidently was due to a flavor common to both OG and CSG that was more concentrated in CSG. We conclude that the relative amounts of different foods ingested within a meal, and the salience of the flavors of those foods, are both important variables that cause goats to distinguish between novel foods that differ in postingestive consequences.

*To whom correspondenceshould be addressed. 6O9 0098-033U9410300A)609507.0010© 1994PIcnurn Publishing Corporation

610

PROVENZAETAL. Key Words--Toxin, food selection, food aversion, secondary metabolite, nutrition, palatability, lithium chloride, ruminants, goat, Caprasp. INTRODUCTION

Secondary metabolites defend plants from herbivory, but the causal mechanisms underlying plant defense are poorly understood in marine (Hay and Fenical, 1988) and in terrestrial (Feeny, 1992) ecosystems. Some argue that toxicosis is the basis for plant defense (Freeland and Janzen, 1974; Berenbaum, 1986; Bryant et al., 1991), while others contend that deterrence (i.e., rejection based on odor and taste only) is often independent of toxicosis (Bernays and Chapman, 1987; Bemays and Cornelius, 1992). Studies designed to separate the effects of toxicosis from deterrence show that ruminants decrease intake of foods paired with toxicity (Olsen and Ralphs, 1986; Lane et al., 1990; Provenza et al., 1990, Pfister et al., 1990; duToit et al., 1991; Launchbaugh et al., 1993) and with nutrient deficits (Rodgers and Egan, 1975; Egan and Rodgers, 1978), and they increase intake of the foods as toxicity diminishes and nutritional value improves (Burritt and Provenza, 1992). The increases in intake were caused by positive postingestive feedback derived from nutritional value the food provided, and the decreases in intake resulted from aversive postingestive feedback caused by toxicosis and nutrient deficits, not taste and odor alone (Provenza et al. 1992, Provenza, 1994a, b). Our objective was to understand some of the variables that control selection of novel foods differing in postingestive consequences. We reasoned that insights into acquired preferences for and aversions to novel foods might be gained by analysis of food selection within meals. Combining this with an understanding of the way ontogeny (Provenza, 1994a, b) and variation among individuals (Provenza and Cincotta, 1993) affect food selection may help to clarify our views of an apparently complex process: How ruminants learn to distinguish between nutritious and toxic foods given a diverse array of plant species, individuals, growth stages, and parts that vary in nutritional value and chemical defenses (Provenza and Balph, 1990). The shrub blackbrush (Coleogyne ramosissima), studied by Provenza et al. [1990], provided an opportunity to examine the response of goats, on a mealby-meal basis, to foods that differed in postingestive consequences. Blackbrush, a small, evergreen shrub that averages less than 1 m in height, grows in dense stands on millions of hectares in the southwestern United States. In the absence of browsing, blackbrush twigs (i.e., older growth, OG) grow little, but branches produce current season's growth (CSG) when apical meristems are removed by browsing (Provenza et al., 1983a). CSG and OG differ in various characteristics. CSG is less "woody" than OG and its bark is red while that of OG is grey.

GOATS DISTINGUISH BETWEEN NOVEL FOODS

611

CSG is higher than OG in nitrogen (1.04% vs. 0.74%) and it is more digestible in vitro (48% vs. 38%) (Provenza et al., 1983b). Nevertheless, goats acquire a strong preference for OG because CSG contains much higher levels of a condensed tannin that causes a learned food aversion in goats (Provenza et al., 1990). In the research reported in this paper, we studied the responses of individually penned goats to CSG and OG because that provided a degree of experimental control not possible with free-ranging goats. We conducted four experiments. METHODS AND MATERIALS

Experiment 1: Novel CSG and OG. Goats introduced to blackbmsh-dominated rangelands learn to avoid CSG and to eat OG even though both foods are novel (Provenza and Malechek, 1984; Provenza et al., 1990). This experiment was designed to determine how long it takes goats to learn to distinguish between CSG and OG. During October 1989, 10 goats (average body weight 21 kg) naive to blackbrush were offered a choice between CSG and OG in two separate food containers from 0900 until 1600 hr daily for three days. Blackbrush twigs were harvested daily and chopped into 1- to 3-cm lengths in a wood chipper. CSG and OG were offered ad libitum throughout the day, but goats were without food from 1600 hr one day until 0900 hr the next. They had access to water ad libitum. The amount of each food ingested was measured daily. The repeated measures analysis of variance had 10 blocks (goats) crossed with two foods (Winer, 1971). Experiment 2: Hourly Monitoring of CSG and OG. Goats rapidly acquired an aversion to CSG in the first experiment, but how they distinguished between OG and CSG on day 1 is not clear from the first experiment. They may have initially ingested CSG, experienced aversive postingestive feedback, and then switched to OG, or they may have ingested more CSG than OG, experienced aversive feedback, and then switched to OG. To discover which of these feeding patterns was involved, we conducted a second experiment in which we observed the sequence of feeding events, and the amounts of CSG and OG ingested by goats on an hourly basis for three days. During September 1990, 12 goats (average body weight 17 kg) naive to blackbrush were offered 500 g of CSG and 500 g of OG in separate food containers. The foods were available from 0930 to 1730 hr on day 1, from 0815 to 1820 hr on day 2, and from 0800 to 1400 hr on day 3. The foods were weighed at hourly intervals; about 15-30 min per hour were required to weigh the food. No food was available for the remainder of the day. Thus, goats had food for a total of 5, 8 and 5 hr, respectively, on days 1, 2, and 3. Goats had access to water ad libitum.

612

PROVENZA ET AL.

We performed two analyses on these data. For one analysis, 12 goats (blocks) were crossed with two foods. A repeated measures analysis was used because the data were analyzed across hours (Winer, 1971). The data were analyzed separately for each day. The other analysis involved the intake of OG and CSG for the first hour when a goat ingested > 30 g of CSG and the intake of CSG and OG for the next three hourly meals > 30 g. The consumption (CSG + OG) of different animals ranged from 32 to 94 g. For this analysis, 12 goats (blocks) were crossed with two foods and repeated across four meals. Experiment 3: Equal CSG and OG. The data from experiment 2 suggested that salience (i.e., a greater tendency of the flavor of CSG than OG to be associated with malaise) and differences in the relative amount of food ingested in a meal (i.e., a greater amount of CSG than OG) might both play a role in the goats" acquisition of an aversion to CSG. To test hypotheses concerning salience and amount of food ingested, we allowed goats to ingest the same amount of CSG and OG. We reasoned that the acquisition of an aversion to CSG under these conditions would be consistent with the salience hypothesis. The acquisition of an aversion to both CSG and OG would be consistent with the amount-ingested hypothesis. The experiment was conducted in November 1990 and involved 12 goats (average body weight 25 kg) naive to blackbrush. All animals were given 50 g of CSG and 50 g of OG from 1530 to 1630 h r o n day 1; each animal received 1300 g of alfalfa pellets at 1730 hr. On day 2, goats were offered 100 g of CSG and 100 g of OG from 1630 until 1700 hr and were then fed alfalfa pellets. On day 3, all goats had access to 200 g of CSG and 200 g of OG from 1630 until 1730 hr. We measured the amount of CSG and OG ingested each day. The data were analyzed as a repeated measures with 12 blocks (goats) crossed with two foods and repeated across three days (Winer, 1971). In addition, we determined the relationship between mass (g) and volume (cc) for both CSG and OG. The analysis of variance had two treatments, and twig type was nested within treatments. Experiment 4: Different ratios of CSG and OG. We divided goats into five groups and offered them different proportions of CSG and OG on a volumetric basis: (1) 10:90, (2) 30:70, (3) 50 : 50, (4) 70: 30, and (5) 90: 10. After eating a small amount of CSG and OG, the goats were given lithium chloride (LiC1), a compound that causes learned food aversions in ruminants (Provenza et al., 1994c). We reasoned that if the volume of food ingested affected the acquisition of the aversion, goats should eat less OG than CSG in cases (1) and (2), should eat less CSG than OG in cases (4) and (5), and should avoid OG and CSG equally in case (3). If some salient characteristic of the odor and taste of CSG also influenced the aversion, goats in all treatments should eat less CSG than OG. During December 1992, 37 goats (average body weight 27 kg) naive to

GOATS DISTINGUISH BETWEEN NOVEL FOODS

6t3

blackbmsh were divided into five groups and were offered (for 30 min) the following volumes (cc) of CSG/OG mixed together: (1) 11 : 103 (10:90; n = 7), (2) 34 : 80 (30 : 70; n = 7), (3) 57 : 57 (50: 50; n = 9), (4) 80: 34 (30 : 70; n = 7), and (5) 103:11 (90: 10; n = 7). These mixtures were equivalent, on a mass (g) basis for CSG/OG, to the following ratios: (1) 4 : 3 5 , (2) 10:27, (3) 17: 19, (4) 12:23, and (5) 30:4. Goats were offered only a small amount of CSG and OG to minimize aversive effects of tannin, and to ensure that both foods were eaten quickly. After eating the OG/CSG mixture, all goats received LiC1 (150 mg/kg body wt in 100 ml water; duToit et al. 1991) by gavage. On the next day, goats were offered a choice between CSG (200 g) and OG (200 g) for 1 hr. The volume of CSG ingested by each goat, as a percentage of the total volume consumed (OG + CSG), was used as the dependent variable in the least-squares regression analysis. The regression equation was derived using the raw data from each of the five treatment groups (i.e., 10, 30, 50, 70, 90%). In an attempt to better estimate the slope and the intercept for the regression equation, the data from days 1 and 2 of experiment 3 were expressed on a volumetric basis for each goat and included in the analysis. This provided a greater sample size in the range of values from 40 to 60% CSG ingested in a meal. Neither the slope nor the intercept of the equation generated from the data in experiment 4 changed significantly (P > 0.10) as a result of adding the data from experiment 3 to the regression model. The 11 data points from experiment 3 were not included in the analysis of variance of the data from experiment 4 because they did not fit the qualitative categories (i.e., 10, 30, 50, 70, 90%) exactly.

RESULTS AND DISCUSSION

Experiment 1. Goats ingested more OG than CSG on the first day of the experiment (Figure 1). Intake of OG increased on day 2 and again on day 3, but intake of CSG was similar on all days. This caused an interaction between food and day (P < 0.001). Goats ate both OG and CSG on day 1, but they had consumed more OG by 1600 hr, which indicates most goats acquired an aversion to CSG on the first day. Goats ingested < 100 g of CSG on all three days, evidently because aversive feedback from the tannin caused them to limit intake of CSG. Nonetheless, it is important to note in this and the other experiments that goats did not eliminate CSG from their diets, and it is likely that if the concentration of the condensed tannin decreased, goats would have increased intake of CSG, as occurs in sheep when the toxin concentration of a nutritious food decreases (Launchbaugh et al. I993). The increase in intake of OG on day 2 and again on day 3 is similar to the

614

PROVENZA ET AL.

._E

== o E

1O0

O.

1

2

3

Day

FiG. 1. Average intake (g) of current season's (CSG) and older (OG) growth blackbrush (SEM = 19) by goats from 0900 to 1600 hr for three consecutive days during experiment 1. Means separated by more than 54 g differ (LSD0.05). response of sheep to novel foods (Provenza et al., 1994a). The reluctance of sheep to eat new foods apparently does not depend on the nutritional quality of the food, or on the level of food deprivation, but on fear of the novel food (neophobia). Experiment 2. During the first hour of day 1, goats ingested significantly more CSG than OG, but during the last hour of day 1 they ingested significantly more OG than CSG (Figure 2), which resulted in a significant food × hour interaction (P < 0.05). On day 2, goats ingested more OG than CSG throughout the day (P < 0.01; Figure 3). The same pattern was evident on day 3 (P < 0.001). Goats' choice of food changed dramatically after they ate more than 30 g of CSG within 1 hr. Thereafter, when goats ate a meal of > 30 g, they ate less CSG than OG (Figure 4). As a result of this change in food selection, the interaction between food and time was significant ( P < 0.001). There was no pattern concerning the time when a goat ate its first large meal of CSG. Six goats ate > 30 g of CSG during the morning of the first day, five ate > 30 g during the afternoon of the first day, and one ate > 30 g on the morning of the second day. The sequence of CSG and OG consumption also varied. Seven goats initially ate a greater proportion of CSG, and five initially consumed more OG. Goats ate both CSG and OG, and after they ate a substantial amount of CSG relative to OG (average of 44 g and 16 g), their intake of CSG declined (Figure 4). When goats eat two or more novel foods and experience malaise, it

GOATS DISTINGUISH BETWEEN

NOVEL FOODS

615

50-

40-

.,E 30-

20-

c o

E 10-

1

2

3 Hour

4

5

Fie. 2. Average intake (g) of current season's (CSG) and older (OG) growth blackbrush (SEM = 3.8) by goats from 0930 until 1730 hr on day 1 of experiment 2. Means separated by more than 8 g differ (LSD0,os). 50 ¸

40 ¸

.E .~ 30

E

1

2

3

4

5

6

7

8

Hour

FIG. 3. Average intake (g) of current season's (CSG) and older (OG) growth blackbrush (SEM = 4.3) by goats from 0815 until 1820 hr on day 2 of Experiment 2, Means separated by more than 8 g differ (LSD0.0~). is not clear which food(s) will be avoided in the following meal. As Garcia (1989) states, the critical feature involves two tastes (each with its potentiated odor) competing for association with one nauseous feedback in the gut. Thus, there are at least four explanations, which are not mutually exclusive, for the way in which flavor (taste and odor) might control the behavior of goats.

PROVENZA ET AL.

616

o E

1

2

3

4

Meal

FIG. 4. Average intake (g) of current season's (CSG) and older (OG) growth blackbrush (SEM = 4.4) by goats when they ate a meal of more than 30 g of CSG, and for the ensuing three large meals whenever they occurred during days I-3 of experiment 2. Means separated by more than 9 g differ (LSDo.05).

First, novelty might be important. For instance, Revusky and Bedarf (1967) successively paired two flavors, one novel, with a single toxin dose. The novel flavor became much more aversive, whether it was first or second. Moreover, Cannon et al. [1985] found that only 10 min of exposure to one novel solution (saccharin) was enough to cause rats to prefer saccharin to a completely novel solution (coffee) following toxicosis. Some goats initially ate more OG than CSG, which could make OG more familiar than CSG. Thus, the greater novelty of CSG to OG might account for the learned aversion to CSG by some of the goats, but it can not account for the aversion to CSG by the other goats that ate CSG prior to OG. Learned safety might also be a factor. For example, Kalat and Rozin (1973) showed that a rat that drank the same solution twice prior to poisoning learned less aversion than when it received the solution only at the second presentation. Some goats ingested more CSG than OG more than once before they acquired an aversion to CSG, which according to this hypothesis should have meant that their aversion to CSG was less than their aversion to OG. Nevertheless, all goats acquired an aversion to CSG, so this explanation can not account for the learned aversion to CSG. Salience, in this case the tendency of a novel food to be associated with malaise, often influences acquired aversions more than does temporal proximity of food ingestion to poisoning (Kalet and Rozin, 1970, t97t). For instance, Cannon et al. (1985) paired exposure to saccharin, followed by exposure to

617

GOATS DISTINGUISH BETWEEN NOVEL FOODS

either a low or a high concentration of a quinine solution, with a single toxin dose. They found that subsequent consumption of saccharin was a positive function of the concentration of the quinine solution drunk prior to toxicosis. Accordingly, if some characteristic of the flavor of CSG is more salient than that of OG, it could explain why goats acquired a stronger aversion to CSG. Finally, the amount of food ingested might influence aversions to CSG. For example, the strength of a rat's aversion to saccharin is a direct function of the amount of saccharin consumed prior to poisoning (Bond and DiGiusto, 1975). If this also occurs in goats, then an aversion to CSG might have resulted when a goat consumed enough CSG to experience malaise. OG consumption may not matter if it were a relatively small proportion of the meal. Experiment 3. Goats ingested similar amounts of CSG and OG on day 1, but they ate less CSG than OG on day 2 and day 3 (Figure 5). This caused an interaction between food and day (P < 0.05). The relationship between mass and volume differed for CSG and OG (P < 0.0001). CSG was less dense (0.25 g/cc; SEM = 0.006) than OG (0.35 g/cc; SEM = 0.006). Goats ate similar amounts of CSG (39 g) and OG (45 g) on day 1 of this experiment (P > 0.05), but on days 2 and 3 they ate less CSG than OG. Thus, these results are in agreement with the hypothesis that the aversion was caused by something salient about the flavor of CSG, and they are not consistent with the amount-ingested hypothesis. Nevertheless, the data do not rule out the amount-ingested hypothesis because when goats ate an average of 39 g of CSG

._~ "o go

E .<

1

2 Day

3

FIG. 5. Average intake (g) of current season's (CSG) and older (OG) growth blackbmsh (SEM = 6.0) by goats during experiment 3. Means separated by more than 12 g differ (LSDo,os).

618

PROVENZA ET AL.

and 45 g of OG on day 1 of the experiment, they consumed a larger volume of CSG (159 cc) than of OG (130 cc). Thus, we conducted a more definitive test of these two hypotheses in experiment 4 in which we accounted for the volumes of CSG and OG ingested by goats, and we included a wider range of volumes, similar to those ingested by goats in experiment 2 (Figures 2-4). Experiment 4. Virtually all of the goats (34 of 37) ingested all of the CSG/ OG mixture in 15 min when it was offered on day 1 of experiment 4, but during the 1-hr test on day 2 only 15 of 37 goats ate blackbrush (four goats in 10:90 CSG/OG, two goats in 30:70, two goats in 50:50, three goats in 70:30, and four goats in 90: 10). Nevertheless, there were differences ( P < 0.001) among treatments in the percentage of CSG ingested by goats on day 2 (Table 1). There was an inverse relationship (r = - 0 . 7 3 ; P < 0.0001) between the percentage of CSG ingested on day 2 and that ingested on day 1 for experiments 3 and 4 (Figure 6). The intercept of the regression equation was 74 (P < 0.0001 ; SE = 7.5) and the slope was - 0 . 8 9 X (P < 0.0001; SE = 0.17). The equation suggests that on day 2 the average goat in treatment (1) ate a meal of 65 % CSG and 35 % OG and that the average goat in treatment (5) ate only OG. When a quadratic term was added to the model, the term was not significant (P > 0.10). Any goat that ate a meal of more than 6 cc during testing was included in the regression analysis. The inverse relationship between the percentage of CSG ingested on day 1 and that ingested on day 2 and the fact that the slope of the line ( - 0.89) did not differ from that of the hypothetical line ( - 1.0) depicted in Figure 6 indicate that the volume of food ingested played a role in the acquisition of an aversion by goats. These data are consistent with the outcome of Experiment 2 in which

TABLE 1. RELATIONSHIPBETWEEN CURRENT SEASON'S GROWTH (CSG) INGESTEDON

DAY 1 ANDON DAY 2 OF EXPERIMENT4~ CSG Ingested % of meal, day 1

% of meal, day 2b

SE

10 30 50 70 90

56~ 20b t7b 10~ 5c

5 6 7 7 5

~Goats in differenttreatmentsconsumedmealsthat varied from 10% to 90%, on a volumetricbasis, of CSG and oldergrowth (OG) on day 1. Immediatelyafter ingestingthe meal, goats were garaged with the toxin lithiumchloride. On day 2, they were given a choice betweenCSG and OG. °abc, meansin the same columnwith differentletters differ significantly(LSDo.os).

619

GOATS D I S T I N G U I S H BETWEEN N O V E L FOODS

E

80]

+

70

!

60

"'"

5O

i

40

2O 0

ii

10

1'0

30 CSG Ingested

50 (% of meal

70 1)

9O

day

FIG. 6. The relationship between current season's growth (CSG) ingested on day 1 of experiments 3 and 4 and that ingested on day 2 (lower line). Goats in different treatments consumed meals that varied from 10% to 90%, on volumetric basis, of CSG and older growth (OG) on day I; immediately after ingesting the meal, goats were garaged with the toxin lithium chloride. On day 2, they were given a choice between CSG and OG. The "volume of food ingested" played a role in the acquisition of an aversion to CSG in goats because there was a significant inverse relationship (r = -0.73) between the percentage of CSG ingested on day 1 and that ingested on day 2. The slope of the regression line is -0.89X (P < 0.0001; SE = 0.17) and the intercept is 74 (P < 0.0001 ; SE = 7.5). The upper line is the expected relationship if only volume of food ingested controlled the ingestion of OG and CSG. "Salience" also played a role in the goats' acquisition of an aversion to CSG because the intercept for the lower line is significantly less (P < 0.01) than that for the upper line (see text for further discussion). goats ingested a meal o f 79% CSG and 21% OG (on a volumetric basis), and thereafter ate meals composed primarily of OG (Figure 4). The hypothetical (upper) line in Figure 6 represents the expected relationship if only volume of food ingested controlled the ingestion of OG and CSG by goats. Cleady, salience also influenced the goats' aversion to CSG because the intercept of the lower line (74) was significantly less than that of the hypothetical line (100). Thus, these data are consistent with the results of experiment 3 (Figure 5). We did not attempt to characterize the salient properties of the flavor, but CSG had a stronger (more salient) odor than did OG. If the odor, and perhaps the taste as well, were caused by a flavor common to both OG and CSG, but more concentrated in CSG, then goats that ate primarily OG (treatment 1) prior to receiving LiCl should have eaten less CSG during testing because CSG was

620

PROVENZAETAL.

the food with the highest concentration of the flavor. Likewise, goats that ate primarily CSG (treatment 5) should have subsequently ingested OG, the food with the lowest concentration of the flavor. The data are consistent with this hypothesis. Lambs generalized food aversions on the basis of a salient flavor when they were offered barley with a low and a high concentration of either a sweet flavor (sodium saccharin) or a bitter flavor (aluminum sulfate). The lambs initially consumed the same amounts of barley, regardless of flavor intensity, but after they ate the flavored barley and received a mild dose of LiCI, they subsequently preferred the barley with the lower concentration of either flavor (Launchbaugh et al., 1993). More than half of the goats (22 of 37) did not eat either OG or CSG during testing in experiment 4 because of the novelty of the food and the dosage of LiC1. Animals form much stronger aversions to novel foods than to familiar foods (Revusky and Bedarf, 1967; Cannon et al., 1985; Burritt and Provenza, 1989, 1991). Moreover, the higher the dose of LiC1, the stronger the aversion (duToit et al., 1991). Administering LiCI by gavage produced a much stronger aversion in goats than did ingesting the condensed tannin in CSG (e.g., Figure 4), because when the toxin is given by gavage (i.e., LiC1), goats could not limit intake of the food to minimize aversive feedback.

CONCLUSIONS When CSG and OG were offered to goats naive to blackbmsh, the goats did not exclusively select either OG or CSG, but when they finally ate enough CSG to experience malaise, they ingested less CSG than OG from that point onward. The change in food selection resulting from postingestive feedback occurred within hours and was influenced both by volume of food ingested and by salience. The greater the volume of food ingested and the more salient the flavor of the food, the stronger was the acquired aversion. This was true whether the aversion was caused by the tannin in CSG (Figure 4) or by LiC1 (Figure 6). On several occasions, we have observed food selection by goats when they were first introduced to blackbrush-dominated rangelands. The goats sampled all foods in the area during the first few hours, including potentially toxic plants like Juniperus osteosperma (bark and green leaves), Gutierrezia microcephala (a forb), and Marrubium vulgare (a forb), as well as more nutritious shrubs like Prunus fascicutata and Purshia tridentata. They also sampled both CSG and OG from blackbrush. Within a few days, goats limit intake of the potentially toxic foods and ingest meals composed primarily of blackbrush OG and Prunus fasciculata. We do not know how much of each food goats ingest during the

GOATS DISTINGUISH BETWEEN NOVEL FOODS

621

first few days, but we suspect that the amount is low initially and gradually increases (or decreases), depending on ensuing postingestive feedback. It would be an interesting extension of the present experiments to determine how the goats' aversions and preferences are acquired in that more complex situation. Postingestive feedback and the senses of taste, smell, and sight are interrelated through affective and cognitive processes (Garcia, 1989). Taste plays a prominent role in both processes. Affective processes integrate the taste of food and its postingestive consequences, and changes in the intake of food items depend on the degree to which postingestive consequences are aversive or positive. The net result is a change in incentive to eat a particular food. Cognitive processes involve use of the senses of smell, sight, and higher cortical centers to seek foods that provide positive consequences and to avoid foods that cause internal malaise. The net result is a change in behavior. Animals do not make conscious decisions concerning acquired food aversions based on the amount of food ingested or its salience, because postingestive feedback does not involve cognitive processes. This is illustrated by the fact that animals (Garcia et al., 1985; Garcia, 1989), including sheep (Provenza et al., 1994b), in deep anesthesia still acquire food aversions. Thus, when a goat eats CSG and OG, and subsequently acquires an aversion to CSG, the decrease in incentive to eat CSG occurs whether or not the goat is conscious. The resulting feedback is a function of the match between the animal's physiology at the time it ingests the food and the chemical characteristics of the food. Several variables have been shown to control the interrelationship between affective and cognitive processes in both monogastrics and ruminants. For instance, acquired aversions are dependent: (1) on temporal contiguity between food ingestion and toxicosis (Cannon et al., 1985; Provenza et al., 1993b), (2) on food novelty (Revusky and Bedarf, 1967; Burritt and Provenza, 1991), (3) on the intensity of a particular flavor (Cannon et al., 1985; Launchbaugh et al., 1993), (4) on the volume of food ingested (Bond and DiGiusto, 1975; this study), (5) on prior experience with illness (Cannon et al., 1975; Launchbaugh et al., 1993), and (6) on prior experience with a salient flavor (Kalat and Rozin, 1970, 1971; Launchbaugh and Provenza, 1993; this study). Thus, the results of these and other studies (reviewed by Garcia, 1989; Provenza, 1994a, b) suggest that deterrence (i.e., taste and odor of foods) and toxicosis are intimately related via feedback mechanisms. Acknowledgments--This paper is published with the approval of the Utah Agricultural Experiment Station as Paper No. 4410. Financial assistance was provided by the Utah Agricultural Experiment Station, the Cooperative States Research Service, the National Science Foundation (BSR-8614856), and a Quinney Fellowship to J.J.L.

622

PROVENZA ET AL. REFERENCES

BERENBAUM, M. 1986. Postingestive effects of phytochemicals on insects: On Paracelsus and plant products, pp. 121-153, in J.R. MILLER and T.A. MILLER (eds.). Insect-Plant Interactions. Springer-Verlag, New York. BERNAYS, E,A., and CHAPMAN, R.F. 1987. Evolution of plant deterrence to insects, pp. 159-173, in R.F, Chapman, E.A. Bemays, and J.G. Stoffolano (eds.). Perspectives in Chemoreception and Behavior. Springer-Verlag, New York. BERNAYS, E.A., and CORNELIUS, M. t992. Relationship between deterrence and toxicity of plant secondary compounds for the alfalfa weevil Hypera brunneipennis. Entomol. Exp. AppL 64 : 289-292. BOND, N., and DIGuIsTo, E. 1975. Amount of solution drunk is a factor in the establishment of taste aversion. Anita. Learn. Behav. 3 : 81-84. BRYANT, J.P,, PROVENZA, F.D., PASTOR, J. REtCHARDT, P.B., CLAUSEN, T . P , and DuTOIT, J.T. 199t. Interactions between woody plants and browsing mammals mediated by secondary metabolites. Annu. Rev. Ecol. Syst. 22:431--446. BuRRrrr, E.A. and PROVENZA,F.D. 1989. Food aversion learning: Ability of lambs to distinguish safe from harmful foods. J, Anita, Sci. 67: 1732-1739. BURRITT, E,A. and PROVENZA. F.D. 1991. Ability of lambs to learn with a delay between food ingestion and consequences given meals containing novel and familiar foods. Appl. Anita. Behav. Sci. 32: 179-189. BURRITT, E.A., and PROVENZA,F.D. 1992. Lambs form preferences for non-nutritive flavors paired with glucose, J. Anita. Sci. 70: 1133-1136. CANNON, D.S., BERMAN, R.F., BAKER,T.B., and ATKINSON,C.A. 1975. Effect of preconditioning unconditioned stimulus experience on learned taste aversions. J. Exp. Psycho. Anim. Behav. Processes 104 : 270-284. CANNON, D.S., BEST, M.R., BATSON,J.D., BROWN, E.R., RUBENSTEIN,J.A., and CARRELt., L.E. 1985. Interfering with taste aversion learning in rats: The role of associative interference. Appetite 6 : 1-19. DuTOtT, J.T,, PROVENZA, F.D, and NASTIS, A. 1991. Conditioned food aversions: How sick must a ruminant get before it learns about toxicity in foods? AppL Anirn. Behav. Sci, 30:35-46. EGAN, A.R., and ROGERS, Q.R. 1978. Amino acid imbalance in ruminant lambs. Aust. J. Agric. Res. 29 : 1263-1279. FEENY, P. 1992. The evolution of chemical ecology: contributions from the study of herbivorous insects, pp. I--44, in G.A. Rosenthal and M.R. Berenbaum (eds.). Herbivores: Their Interactions with Secondary Plant Metabolites, 2nd ed. Academic Press, New York. FREELAND, W.J., and JANZEN, D.H. 1974. Strategies in herbivory by mammals: The role of plant secondary compounds. Am. Nat. 108 : 269-289. GARCIA, J. 1989. Food for Tolman: Cognition and cathexis in concert, pp. 45-85, in T. Archer and L. Nilsson (eds.), Aversion, Avoidance and Anxiety. Erlbaum Hillsdale, New Jersey. GARCIA, J., LAStTER, P.A,, BERMUDEZ-RATTONI,F., and DEEMS, D.A. 1985. A general theory of aversion learning, pp. 8-21, in N.S. Braveman and P. Bronstein (eds.). Experimental Assessments and Clinical Applications of Conditioned Food Aversions. New York Academy of Science, New York. HAY, M.E., and W. FENICAL. 1988. Marine plant-herbivore interactions: The ecology of chemical defense. Annu. Rev. EcoL Syst. 19: 111-145. KALAT, J.W., and ROZIN, P. 1970. "Salience:" A factor which can override temporal contiguity in taste-aversion learning. J. Comp. Physiol. Psychol. 71 : 192-197. KALAT, J.W., and ROZIN, P. 1971. Role of interference in taste-aversion learning. J. Cutup. Physiol, Psychol. 77 : 53-58.

GOATS DISTINGUISH BETWEEN NOVEL FOODS

623

KALAT, J.W., and RozIN, P, 1973. "Learned safety" as a mechanism in long-delay taste-aversion learning in rats. J. Cutup. PhysioL Psychol. 83 : 198-207. LANE, M.A., RALPHS, M.A., OLSEN, J.D., PROWNZA, F.D., and PFISTER, J,A. 1990. Conditioned taste aversion: Potential for reducing cattle loss to larkspur. J, Range. Manage. 43: t27-131, LAUNCHBAUGH,K.L., and PROVENZA,F.D. 1993. Can plants practice mimicry to avoid grazing by mammalian herbivores? Oikos 66:501-504. LAUNCHBAUGH,K.L., PROVENZA, F.D., and BURRWTT,E.A. 1993. How herbivores track variable environments: Response to variability of phytotoxins. J. Chem. Ecol. 19: 1047-1056. OLSEN, J.D., and RAt.PHS, M.H. 1986. Feed aversion induced by intramminat infusion with larkspur extract in cattle. Am. J. Vet. Res, 47: 1829-1833. PFISTER, J.A., PROVENZA, F.D., and MANNERS, G.D. 1990. Ingestion of tall larkspur by cattle: Separating the effects of flavor from post-ingestive consequences. J. Chem. Ecol. 16: 16971705. PROVENZA, F.D. 1994a, A functional explanation for food selection and the nutritional wisdom of ruminants. J. Range Manage. Accepted. PROVENZA, F.D, 1994b. Ontogeny and social transmission of food selection in domesticated ruminants. In P. VALSECCHIand B.G. GALEF, Jr. (Yds.). Ontogeny and Social Transmission of Food Preferences in Mammals: Basic and Applied Research. In press. PROVENZA, F.D., and BALPH, D.F. 1990. Applicability of five diet-selection models to various foraging challenges ruminants encounter, pp. 423--459, in R.N. HUGHES (ed.). Behavioral Mechanisms of Food Selection. NATO ASI Series G: Ecological Sciences, Vol. 20. SpringerVerlag, New York. PROVENZA, F.D., and CINCOTTA, R.P. 1993. Foraging as a self-organizational learning process: Accepting adaptability at the expense of predictability, pp. 78-101 in R.N. Hughes (ed.), Diet Selection. Blackwell, London. PROVENZA,F.D., and MALECHEK,J.C. 1984. Diet selection by domestic goats in relation to blackbrash twig chemistry. J. Appl. Ecol. 21:831-841. PROVENZA, F.D., PFISTER, J.A., and CHENEY, C.D. 1992. Mechanisms of learning in diet selection with reference to phytotoxicosis in herbivores. J. Range Manage. 45:36-45. PROVENZA, F.D., MALECHEK,J.C., URNESS, P.J., and BOWNS, J.E. 1983a. Some factors affecting twig growth in blackbrush. J. Range Manage. 36:518-520. PROVENZA, F.D., BOWNS, J.E., URNESS, P.J., MALECHEK,J.C., and BUTCHER,J.E. 1983b, Biological manipulation of blackbrush by goat browsing. J. Range Manage. 36 : 513-518. PROVENZA, F.D., BURRITT, E.A,, CLAUSEN, T.P., BRYANT, J.P., REICHARDT,P,B., and DISTEL, R.A. 1990. Conditioned flavor aversion: A mechanism for goats to avoid condensed tannins in blackbrush. Am. Nat. 136 : 810-828. PROVENZA, F.D., LYNCH, J.J. and NOLAN, J.V,, 1993a. The relative importance of mother and toxicosis in the selection of foods by lambs. J. Chem. Ecol. t9:3t3-323. PROVENZA, F.D., LYNCH, J.J. and CHEHEY, C.D., 1994a. An experimental analysis of the effects of a salient flavor and food deprivation on the intake of novel foods by sheep. J. Anim. Sci. Submitted. PROVENZA, F.D., LYNCH, J.J., and NOLANJ.V., 1994b. Food aversion conditioned in anesthetized sheep (Ovis aries). Physiol. Behav. In press. PROVENZA,F.D., NOLAN,J.V., and LYNCH,J.J. 1993b. Temporal contiguity between food ingestion and toxicosis affects the acquisition of food aversions in sheep. Appl. Anim. Behav. Sci. 38:269-281. PROVENZA, F.D., ORTEGA-REYES,L., SCOTT, C.B., LYNCH, J.J., and BURRITT, E.A. 1994C. Antiemetic drugs attenuate food aversions in sheep. J. Anita. Sci. accepted.

624

PROVENZA ET AL.

REVUSKY,S.H., and BEDARF, E.W. 1967. Association of illness with prior ingestion of novel foods. Science 155:219-220. ROGERS, Q.R., and EGAN, A.R. 1975. Amino acid imbalance in the liquid-rod lamb. Aust. J. Biol. Sci. 28: 169-181. WINER, B.V. 1971. Statistical Principles in Experimental Design. McGraw-Hill, New York, 907 pp.