Control of Gut Retention Time by Secondary Metabolites in Ripe Solanum Fruits

Ecology, 79(7), 1998, pp. 2309±2319 ᭧ 1998 by the Ecological Society of America

CONTROL OF GUT RETENTION TIME BY SECONDARY METABOLITES IN RIPE SOLANUM FRUITS

SOFIA A. WAHAJ,1 DOUGLAS J. LEVEY,1 ANNA K. SANDERS,2 AND MARTIN L. CIPOLLINI2 1Department of Zoology, University of Florida, Gainesville, Florida 32611-8525 USA 2Department of Biology, 430 Berry College, Mt. Berry, Georgia 30149-0430 USA

Abstract. We tested whether compounds in ripe Solanum americanum fruits affect gut retention time of S. americanum seeds in Cedar Waxwings (Bombycilla cedrorum). Gly- coalkaloids were of special interest because they commonly occur in ripe Solanum fruits and are associated with diarrhea in humans. Also, we determined the in¯uence of gut retention time and the presence/absence of two glycoalkaloids, ␣-solasonine and ␣-solamargine, on germination of S. americanum seeds. In one trial, we measured seed retention times of 10 waxwings fed three types of arti®cial fruits, Control (containing no secondary metabolites), Low Concentration (containing low concentrations of ␣-solasonine and ␣- solamargine, matching those in ripe S. americanum fruit), and Extract (containing an ethanol extract of ripe S. americanum fruits, with many unidenti®ed secondary metabolites). Seeds in Low Concentration fruits were not defecated more quickly than those in Control fruits, but seeds in Extract fruits were. Thus, ripe S. americanum fruits contain a chemical or chemicals with a laxative effect. Also, seeds from Control fruits were deposited in more defecations and at lower densities in each defecation than those from Extract fruits. In a second trial, we compared retention times of seeds in control fruits and in fruits with high concentrations of glycoalkaloids, typical in ripe fruits of other Solanum species. These high concentrations had a signi®cant constipative effect on seed passage. Percentage germination and mean germination time of defecated seeds were not in¯uenced by mean retention time in waxwing guts. However, proportionately fewer seeds germinated from Low Concentration fruits than from Control fruits. These results suggest that plants have more control over seed processing by frugivores than generally acknowledged. Secondary metabolites in ripe fruits can increase or decrease retention time and thereby in¯uence seed deposition patterns (e.g., number of defecations with seeds, number of seeds per defecation, and presumably, dispersal distance). Key words: Bombycilla; frugivory; fruit; glycoalkaloids; laxative; retention time; secondary metabolites; seed dispersal; seed germination; solamargine; Solanum; solasonine.

INTRODUCTION frugivore coevolution is perhaps most apparent in the Despite the importance of fruiting plants to frugi- one group of plants, mistletoes, that can control where vores (Foster 1982, Terborgh 1986, Levey 1988) and their seeds are defecated (Walsberg 1975, Restrepo of frugivores to fruiting plants (Howe et al. 1985, 1987, Reid 1991). Sticky viscin in their fruits can in- Schupp 1988, Chapman and Chapman 1995) reciprocal ¯uence quality of dispersal by causing birds to wipe selection pressures have not resulted in tight coevo- defecated seeds on branches of a diameter appropriate lutionary relationships between these groups. Morpho- for germination and establishment (Reid 1989, 1991, logical adaptations for frugivory in animals appear rare Sargent 1995). or inconsistent among taxa and/or studies (Herrera Recently, Murray et al. (1994) proposed another, 1984, Jordano 1987, Karasov and Levey 1990, Levey more general, mechanism by which plants may in¯u- and Duke 1992, Witmer 1994, Ricklefs 1996) and, like- ence quality of seed dispersal. They provided experi- wise, frugivores have not strongly in¯uenced many mental evidence that secondary metabolites in fruit morphological and nutritional traits of fruits (Herrera pulp in¯uence retention time of seeds in frugivore guts, 1989, 1992, Fischer and Chapman 1993, Jordano 1995, mediating a tradeoff between dispersal distance (reten- Tamboia et al. 1996). One reason for the apparent lack tion time is positively correlated with dispersal dis- of evolutionary responses of plants to seed dispersers tance) and seed viability (retention time is negatively may be their inability to in¯uence what happens to correlated with viability). This hypothesis is appealing seeds after they are ingested (Wheelwright and Orians because it provides an evolutionary explanation for 1982, Howe 1984). In support of this hypothesis, plant± secondary metabolites in ripe fruit (Janzen 1977, Her- rera 1982, Cipollini and Levey 1997a) and suggests Manuscript received 3 March 1997; revised 13 October 1997; accepted 15 October 1997; ®nal version received 10 that plants have more control over seed dispersal than November 1997. previously thought. 2309 2310 SOFIA A. WAHAJ ET AL. Ecology, Vol. 79, No. 7

Murray et al.'s (1994) results are dif®cult to interpret and maintained on a banana-based diet (Denslow et al. for two reasons. First, the method of adding secondary 1987) and water, both provided ad libitum. Birds were chemicals to arti®cial fruits by soaking them in aqueous held on a 14 h : 10 h light : dark cycle at 23ЊC. fruit extracts could have resulted in oxidation and A roll of plastic sheeting suspended behind each cage breakdown of metabolites, production of by-products provided a single layer of plastic that was pulled by microbes, and chemical conversions (cf. Cilliers and through slots in the back and front of the cage, then Singleton 1990). As such, the concentrations of sec- under the one-way mirror. This setup allowed us to ondary metabolites and fruit pulp nutrientsÐwhich observe birds and collect defecations with minimal dis- also affect consumption and gut passage rates (Witmer turbance to the birds. To control for diurnal rhythm of 1996a)Ðcould not be controlled or standardized. Sec- gastrointestinal motility (Duke 1982), all trials were ond, Murray et al. (1994) did not directly compare conducted ϳ3 h after lights came on in the morning. germination success of seeds that had passed through Solanum americanum, a native species in the south- the birds' guts with and without the fruit extract. Thus, eastern United States, is common in disturbed areas. they could not distinguish between direct effects of In our area it is an annual or short-lived perennial, pulp secondary metabolites on germination vs. indirect producing fruits from July to October. Ripe fruit are effects due to variation in retention time caused by the often abundant (Ͼ100/plant) on larger plants during same or other compounds. this time. Fruits are dark purple and ϳ0.5 cm in di- Here we test Murray et al.'s (1994) hypothesis on ameter when ripe. They contain 71 Ϯ 2 seeds (n ϭ 26; the role of secondary metabolites in ripe fruit, address- Tamboia et al. 1996), each ϳ1.5 mm in diameter. ing the confounding factors described above. Murray We have no records of waxwings consuming S. amer- et al. (1994) used fruits of the tropical solanaceaous icanum fruits. Nonetheless, we feel justi®ed using this shrub Witheringia solanacea (which is now thought to system because (1) captive waxwings readily consumed be W. meiantha; K. Murray, personal communication). S. americanum fruits and, in fact, preferred them to We focused our work on a related temperate species, their maintenance diet (D. Levey, unpublished data); Solanum americanum Mill., because this species con- (2) wild waxwings consume fruits that are essentially tains glycoalkaloids, which we suspected could pro- identical to S. americanum in size, color, and nutrient duce laxative effects. Glycoalkaloids are common in content (Martin et al. 1951, Witmer 1996b, Cipollini ripe Solanum fruits and when ingested in large amounts and Levey 1997a); (3) a close relative of waxwings, are known to cause diarrhea in mammals (Zitnak 1979). Phainopepla nitens, has been reported to consume So- Some species, like S. americanum, exhibit low con- lanum fruits (Martin et al. 1951); (4) lack of feeding centrations in ripe fruits, whereas others, like S. car- records does not mean waxwings avoid S. americanum; olinense L., have levels so high that their ripe fruits and (5) waxwings and S. americanum were the closest are considered toxic (Zitnak 1979, Van Gelder 1990, phylogenetic matches we had available to the species Cipollini and Levey 1997b). studied by Murray et al. (1994). We conducted three experiments. In the ®rst, we test- Glycoalkaloids are virtually restricted to the Sola- ed whether S. americanum seeds in arti®cial fruits con- naceae and are composed of C27 steroid aglycones to taining a low concentration of glycoalkaloids are def- which various branched carbohydrate moieties are at- ecated more quickly than seeds in fruits without gly- tached (Ripperger and Schrieber 1981). The best known coalkaloids. In the same set of trials, we also tested for compounds are the potato glycoalkaloids, ␣-solanine the presence of other secondary metabolites with lax- and ␣-chaconine (Gregory 1984, Van Gelder 1990). ative effects by measuring retention times of seeds in The structurally related compounds ␣-solasonine and fruits made with an extract from ripe S. americanum ␣-solamargine occur more broadly among Solanum fruits. At low concentrations, we found no effect of species (Ripperger and Schrieber 1981) and are the glycoalkaloid on retention time and so, in the second focus of our study. They occur in ripe S. americanum experiment, we conducted similar tests using an in- fruit in relatively low concentrations (Ͻ1 mg/g), but creased glycoalkaloid concentration in the arti®cial can be found in much higher concentrations in ripe fruits. The third experiment tested whether gut reten- fruits of other Solanum species (e.g., S. carolinense; tion time and presence of glycoalkaloids in¯uence ger- 10±30 mg/g; Cipollini and Levey 1997a). mination of S. americanum seeds. Extraction of glycoalkaloids and formulation of METHODS agar fruits Extraction and puri®cation of ␣-solasonine and ␣- Study organisms solamargine.ÐWe isolated ␣-solasonine and ␣-sola- Cedar Waxwings are seed dispersers that feed on a margine by column chromatography of crude glycoal- large variety of ¯eshy fruits (Witmer 1996b). We cap- kaloid extracts from oven-dried samples of S. aculea- tured waxwings in blueberry ®elds near Gainesville, tissimum (provided by M. Weissenberg, Volcani Center, Florida. They were housed separately in cages (di- Bet Dagan, Israel). S. aculeatissimum was used because mensions 0.5 ϫ 0.5 ϫ 0.5 m) behind one-way mirrors of the ease with which pure ␣-solamargine and ␣-so- October 1998 SECONDARY METABOLITES AND RETENTION TIME 2311 lasonine can be extracted and puri®ed from this species. require any added nutrients because the 7 mL of crude We extracted oven-dried fruit pulp with hot ethanol, extract included nutrients and secondary metabolites rotoevaporated the ®ltrate to dryness, and redissolved at levels equal to those of real S. americanum fruit. it in 1% acetic acid. We then shifted the pH to 11.5 Agar solutions were cooled to 70ЊC prior to the addition using 30% ammonia and held the extract at 70ЊC for of glycoalkaloid or extract. After secondary metabo- 20 min. After the extract had cooled overnight at room lites were added, the agar solutions were further cooled temperature, we extracted the crude glycoalkaloid pre- to 50ЊC for the addition of S. americanum seeds. Ap- cipitate using chloroform. After removal of the chlo- proximately 125 seeds, freshly extracted from fruits of roform by rotoevaporation, the crude glycoalkaloid was at least 10 S. americanum plants (to control for vari- taken up with water-saturated n-butanol and chroma- ation in seed quality among fruits and plants; see Ger- tographed on a neutral alumina column using water- mination Experiment) were placed into each well of a saturated n-butanol as the solvent. Pure fractions were plastic tissue culture plate (Falcon #3077; 0.5-mL authenticated by co-chromatographing them on thin wells; each well produced one arti®cial fruit) and cov- layer silica gel with authentic ␣-solasonine and ␣-so- ered with agar. Fruits were then allowed to gel at room lamargine (authenticated using thin-layer chromatog- temperature. We made new fruits each morning and raphy, high-performance liquid chromatography, and used them within 1±2 h. fast atom bombardment mass spectroscopy by M. Weis- senberg, Volcani Center, Bet Dagan, Israel). They were Low glycoalkaloid concentration and Solanum then collected, rotoevaporated to dryness, and dis- extract experiment solved in a minimal amount of 80% ethanol for addition We used this experiment to test whether mean re- to agar fruits. tention time of seeds was affected by (1) ␣-solasonine Preparation of crude S. americanum extracts.ÐWe and ␣-solamargine at the low concentrations occurring obtained a crude extract of S. americanum fruits by in ripe S. americanum fruits and (2) unknown second- heating 20 g of freeze-dried seedless pulp in 200 mL ary metabolites in ripe S. americanum fruits. of 70ЊC ethanol (95%) for 30 min. The extract was On the day of a trial, birds were allowed to feed on ®ltered through a glass ®ber ®lter, rotoevaporated to seedless fruits of one of the three agar types (Low near dryness, and kept frozen until time of use. This Concentration, Extract, or Control) for ϳ2 h. This as- broad-spectrum extraction procedure resulted in the ex- sured that their guts would be full of the appropriate traction of a wide range of primary (nutritional) and fruit type at the start of the trial, thereby eliminating secondary metabolites, including ␣-solasonine and ␣- possible effects of the maintenance diet on retention solamargine (Ieven et al. 1979, Dey and Harborne time. Also, we found that birds eating the test fruits at 1989). the start of a trial were more likely to continue eating Preparation of agar fruits.ÐArti®cial agar fruits them after the trial started. Fruit type on the ®rst two mimicked pulp chemistry of ripe S. americanum fruits. days of trials was determined randomly; all birds were They were made by dissolving the following in 92.7 given the same fruit type on each of those days. Trials mL of a hot 2% agar solution: sugars (5.37%; glucose : were often unsuccessful and had to be repeated. After fructose : sucrose; 2:2:1), lipids (0.7%; corn oil : peanut the second day of trials, fruit type was determined ran- oil; 1:1), complex carbohydrates (0.7%; starch : pectin; domly for each bird, unless the bird needed only a 3:1), and protein (0.6%; soy protein extract). Concen- single fruit type to complete the experiment. trations of these constituents were determined by prior At time zero (3 h after lights came on), birds were chemical analysis of ripe S. americanum fruits (Ci- force-fed two fruits with seeds. (Meal size was typi- pollini and Levey 1997a). We used citric acid to adjust cally two fruits when birds were feeding normally.) the pH of agar fruits to 5.7, which is the average pH They were then allowed to resume feeding on seedless of ripe S. americanum fruits (M. Cipollini, unpublished fruits, ad libitum. We collected defecations every 5 min data). for 90 min, which a pilot study (n ϭ 8 birds) con®rmed We prepared four types of agar fruits: ``Control'' was usually long enough for all seeds to pass. fruits had only the components listed above; they We force-fed because some birds refused to eat seed- lacked glycoalkaloids. ``Low Concentration'' fruits ed fruits until their guts were nearly empty, whereas also contained glycoalkaloid at 0.008% (␣-solamar- others ate fruits readily. This variation in feeding be- gine : ␣-solasonine; 5:3), which mimicked the naturally havior concerned us because waxwings normally main- low alkaloid concentration of ripe S. americanum tain a full gut (Levey and Grajal 1991, Witmer 1994), fruits. ``High Concentration'' fruits were identical to and their consumption rate highly in¯uences retention Low Concentration fruits but contained glycoalkaloids time (D. Levey, unpublished data). Thus, variation in at 0.600% (solamargine : solasonine; 5:3), a concentra- feeding behavior could override any effects of second- tion approaching that of ``toxic'' fruits such as S. car- ary metabolites on retention time. To validate that olinense (Cipollini and Levey 1997a). ``Extract'' fruits force-feeding did not itself affect retention time, we were prepared using 7 mL of S. americanum crude compared retention times of Control fruits between extract in 93 mL of 2% agar. Extract fruits did not eight force-fed birds and the same birds after they had 2312 SOFIA A. WAHAJ ET AL. Ecology, Vol. 79, No. 7 eaten an identical fruit voluntarily and quickly (within eliminate confounding effects of other food types in 20 min after the start of a trial). We found no difference the gut, birds were not given any other food during in mean retention time between force-fed and non- these trials, although they had water ad libitum. Control force-fed trials (32.9 Ϯ 9.6 min vs. 30.1 Ϯ 6.7 min, fruits were force-fed in the same manner. respectively; paired t ϭ 0.997, df ϭ 7, P ϭ 0.352). We also observed large temporal variation in con- Germination experiment sumption rate within birds, even within single trials. We used this experiment to test whether gut retention We thought it prudent to control for this variation, but time affected percentage germination and mean ger- could not measure consumption rate within trials with- mination time of seeds collected in the Low Concen- out disturbing the birds. Instead, we used analysis of tration and Control trials. Each defecation was rinsed defecation rate to decide when birds were consuming in a 0.5-mm sieve to separate seeds from agar. Care seedless fruits at a normal and consistent rate after was taken to minimize physical scari®cation. Once force-feeding. This approach is justi®ed by a strong counted, seeds were placed on top of a mixture of pot- positive correlation between consumption rate and def- ting soil, topsoil, and sand (4:2:1). Seeds from each ecation rate (r2 ϭ 0.81, P ϭ 0.006, n ϭ 10 trials; pilot retention period were evenly distributed in separate study). Based on pilot trials, we knew that continuously cells (approximately 3 ϫ 5 ϫ 7 cm) and were planted feeding waxwings averaged at least one defecation ev- the same day they were collected from the plants and ery 5 min (X Ϯ 1 SDϭ 1.7 Ϯ 0.6 defecations/5 min, n fed to the birds. Trays of seeds were placed in a green- ϭ 10 trials) and rarely had two or more consecutive 5- house and watered daily for 30 min at ϳ14:00 h. Trays min periods without defecations (X ϭ 3.7 Ϯ 0.7% of were rotated 180Њ daily to ensure an even distribution consecutive 5-min intervals without defecations, n ϭ of water to all cells. Germination was monitored daily, 16 trials). Also, continuously feeding birds defecated and all seedlings were removed. We de®ne germination most seeds between 15 and 35 min, so this time period as the appearance of cotyledons, although, technically, was most critical for normal defecation behavior. We germination occurs before this time. We monitored ger- used these data to de®ne a priori criteria for determin- mination for at least 62 d; the exact length of time ing when a bird was not eating and defecating at a varied because we sowed different trays on different normal rate, in which case the trial was rejected. Trials dates and stopped monitoring all trays on the same date, were repeated for each bird until a trial (1) averaged after germination had nearly ceased (Ͻ2% of all cells 1 defecation/5-min interval between 15 and 60 min, had newly germinated seeds during the ®nal 4 d of (2) did not have any consecutive 5-min periods without monitoring). Because the order of Low Concentration defecations between 15 and 35 min, and (3) did not and Control trials was randomized, seeds from both have more than two consecutive periods without def- types of fruits were allowed essentially equal time for ecations between 15 and 60 min. germination. Mean germination time is the integrated Seeds were carefully removed from defecations and average time between sowing and germination of seeds: counted. In calculations, the midpoint of each 5-min Mean germination time ϭ ft interval was used to represent the passage time for all ͸ ii seeds defecated within the interval. Mean retention where fi is the fraction of total seeds germinating at time is the integrated average time between ingestion time ti. and excretion of seeds (Warner 1981, Robbins 1993): Statistical analysis Mean retention time ϭ ft ͸ ii Retention times.ÐWe used a multivariate, repeated- where fi is the fraction of total ingested seeds excreted measures ANOVA, with fruit type (Low Concentration, at time ti. Extract, and Control) as a ``within'' variable and mean retention time as the dependent variable (Abacus Con- High glycoalkaloid concentration experiment cepts 1989). We used the same model to test for dif- We used this experiment to test whether high con- ferences among fruit types in two other descriptors of centrations of (␣-solasonine and ␣-solamargine, typical gut retention time, transit time (i.e., time of ®rst ap- in some ripe Solanum fruits, affected mean retention pearance of seeds) and the time at which the last seed time of seeds. Because birds would not voluntarily con- was defecated. All P values were adjusted by Green- sume any High Concentration fruits, we force-fed them house-Geisser epsilon. three times: 15 min before the trial, at time zero, and We performed two planned contrasts on fruit type. 15 min later. Two fruits were fed each time because The ®rst tested whether mean retention times differed this feeding rate approximately matched the waxwings' between Low Concentration and Control fruits. The natural consumption rate of 6 fruits/h (2.15 Ϯ 0.59 g/ second compared mean retention times of Extract and h, n ϭ 10 birds; one fruit ϭ 0.36 g). Fruits fed at time Control fruits. Again, P values were adjusted by Green- zero contained seeds; all others did not. We did not house-Geisser epsilon. force-feed fruits after 15 min for fear it would cause Mean retention times of seeds from High Concen- birds to defecate seeds in their colon prematurely. To tration and Control fruits were compared with a paired October 1998 SECONDARY METABOLITES AND RETENTION TIME 2313

TABLE 1. Mean retention time of Solanum americanum (F ϭ 2.95, P ϭ 0.08; repeated-measures ANOVA). seeds from three types of arti®cial fruits (low concentration 2,18 of glycoalkaloid, high concentration of glycoalkaloid, ex- After 20 min, seeds from Extract fruits started to appear tract of fruit pulp) in Cedar Waxwings. more quickly in defecations than seeds from Control or Low Concentration fruits. Between 20 and 40 min, Source df SS MS FP seeds from Low Concentration fruits were defecated Subject 9 1079.5 120.0 less frequently than seeds from Control fruits, although Fruit type 2 677.9 338.9 6.28 0.009 this difference did not persist and was not large enough Fruit type ϫ subject 18 54.2 3.0 to create a signi®cant difference in retention times of seeds from Control and Low Concentration fruits. However, this hint of longer retention times of Low t test. In all tests, alpha was 0.05, and we report means Concentration fruit foreshadowed a greater and signif- Ϯ 1 SD, unless otherwise noted. icant effect on mean retention time with higher levels Germination.ÐAlthough we originally intended to of glycoalkaloids. The difference in seed retention be- analyze germination data from each 5-min interval sep- tween Extract and Control fruits persisted until the last arately, it was necessary to combine intervals for two seed out (F2,18 ϭ 4.10, P ϭ 0.03; repeated-measures reasons. First, sample sizes were often low (Ͻ5 seeds), ANOVA). In summary, effects of fruit pulp constitu- especially in the early and late intervals of each trial. Second, we felt it important to use a multivariate, repeated-measures ANOVA because such a model con- trols for correlation among the repeated measures. Be- cause there were often several time intervals for each bird in which no seeds were defecated and because the multivariate approach cannot accommodate such missing values (Abacus Concepts 1989), we combined all 5-min intervals so that there were no missing values for any bird. The combined time intervals were 0±27.5 min, 32.5±42.5 min, 47.5±57.5 min, and 62.5±87.5 min. Percentage germination within each interval was arcsine-transformed to reduce positive mean-variance relationships. We used a multivariate, repeated-measures ANOVA with fruit type (Low Concentration, Control) and time interval as ``within'' factors and percentage germination as the dependent variable. The same ANOVA model was then used with mean germination time as the dependent variable.

RESULTS Low glycoalkaloid concentration and Solanum extract experiment Fruit type signi®cantly affected mean retention time of seeds (F2,18 ϭ 6.28, P ϭ 0.009; Table 1). Mean retention time of seeds from Low Concentration and Control fruits were essentially identical (P ϭ 0.77; planned comparison), but seed retention differed signi®cantly between Control and Extract fruits (P ϭ 0.009; planned comparison; Fig. 1A). Mean retention time of seeds from Extract fruits was ϳ75% that for control fruits. Also, the pattern of seed deposition differed between Control and Extract fruits; seeds in Ex- FIG. 1. Mean retention times (ϩ1 SE)ofS. americanum tract fruits were deposited in fewer defecations than seeds defecated by waxwings (n ϭ 10) fed arti®cial fruits. (A) Control fruits had no glycoalkaloids, Low Concentration seeds in Control fruits (X Ϯ 1 SD ϭ 12.6 Ϯ 3.5 and (Low Conc.) fruits had low concentrations of ␣-solasonine 16.1 Ϯ 4.8, respectively, paired t ϭ 2.9, df ϭ 9, P ϭ and ␣-solamargine, and Extract fruits had naturally occurring 0.02), which means seed density was higher in these concentrations of all compounds in ripe S. americanum fruit defecations than in those from control fruits. pulp and soluble in ethanol. Seeds in Extract fruits were def- Cumulative frequency curves for percentage of seeds ecated signi®cantly faster than seeds in Control or Low Con- centration fruits. (B) High Concentration (High Conc.) fruits defecated vs. time were essentially identical until 20 had high concentrations of alkaloids, and Control fruits had min (Fig. 2A). Transit time (i.e., time of ®rst appear- none. High concentrations of glycoalkaloids had a signi®cant ance of seeds) did not differ for the three fruit types constipative effect on waxwings. 2314 SOFIA A. WAHAJ ET AL. Ecology, Vol. 79, No. 7

ents were not re¯ected in transit time; they became evident only after ϳ15% of seeds were defecated. The difference from Control fruits persisted and was signi®cant for Extract fruits, but not for Low Concentra- tion fruits.

High glycoalkaloid concentration experiment High Concentration fruits signi®cantly increased mean retention time (paired t ϭ 2.3, df ϭ 9, P ϭ 0.025). Seeds from High Concentration fruits took 36% longer to pass than seeds from Control fruits (Fig. 1B). As with Low Concentration and Extract fruits, the time of ®rst seed defecation did not differ from Control fruits (t ϭ 1.94, df ϭ 9, P ϭ 0.08; Fig. 2B). Between 30 and 85 min, however, seeds from Control fruits ap- peared much more quickly than those from High Con- centration fruits. There was no difference between the two types of fruit in time of appearance of the last seed (t ϭ 1.52, df ϭ 9, P ϭ 0.16).

Germination experiment Seeds started to germinate 7±10 d after sowing. Sow- ing density (number of seeds/cell) was negatively cor-

related with percentage germination (F1, 276 ϭ 4.96, P ϭ 0.03). However, the variation in germination ex- plained by sowing density was extremely low (r2 ϭ 0.01) and we therefore decided that inclusion of sowing density in the ANOVA model was not worth the cost in lost degrees of freedom. Regardless, the effect of sowing density would be the same on seeds defecated FIG. 2. Cumulative frequency distribution of seeds from arti®cial fruits eaten and defecated by Cedar Waxwings. Fruit early and late because both sets of seeds were defecated types are as in Fig. 1A and 1B. Error bars are not shown and sown with few other seeds. because they obscure differences among the lines. Percentage germination was not affected by gut re-

tention time (F3,24 ϭ 2.37, P ϭ 0.096; Table 2). Despite the relatively low P value, there was no trend in percentage germination with increasing time. In particular, seeds that were defecated earlier did not have a higher percentage germination than seeds defecated later (Fig.

TABLE 2. Repeated-measures ANOVAs of (a) percentage germination and (b) mean germination time of Solanum americanum seeds from two types of arti®cial fruits (with and without glycoalkaloids) retained in waxwing guts for four nonoverlapping time periods.

Source df SS MS FP a) Percentage germination Subject 8 1.78 0.22 Fruit type 1 1.11 1.11 5.63 0.045 Fruit type ϫ subject 8 1.57 0.20 Time 3 0.71 0.24 2.37 0.01 Time ϫ subject 24 2.41 0.10 Fruit type ϫ time 3 0.10 0.03 0.63 0.60 Fruit type ϫ time ϫ subject 24 1.31 0.06 b) Mean germination time Subject 8 473.7 59.21 Fruit type 1 28.5 28.53 0.15 0.71 Fruit type ϫ subject 8 1542.1 192.8 Time 3 35.9 12.0 0.54 0.66 Time ϫ subject 24 530.1 22.1 Fruit type ϫ time 3 43.9 14.6 0.61 0.62 Fruit type ϫ time ϫ subject 24 579.1 24.1 October 1998 SECONDARY METABOLITES AND RETENTION TIME 2315

3B). Also, fruit type did not affect mean germination

time (F1,8 ϭ 0.15, P ϭ 0.71; Fig. 3B; Table 2).

DISCUSSION Secondary metabolites and gut retention time Glycoalkaloids at concentrations typical of ripe So- lanum americanum fruit did not decrease gut retention time of seeds in Cedar Waxwings. In fact, at high concentrations glycoalkaloids increased retention time by 36%. However, ripe S. americanum fruits do appear to contain a laxative chemical or chemicals (for ease of discussion, we will henceforth assume a single chemical), because seeds in arti®cial fruits with a pulp extract were defecated more quickly than seeds in control fruits. Compounds in fruit pulp are widely known to have laxative effects in humans (Willimott 1933, McMillan and Thompson 1979). For example, sorbitol in Rosa- ceous fruits (e.g., prunes) and glucosides of emodin in Buckthorn fruits (Rhamnus spp.) are often used by humans as natural laxatives (Meyers 1983). At issue is whether such compounds have similar effects on natural seed dispersers. Our results suggest they do and therefore support Murray et al.'s (1994) conclusion that a compound in Witheringia solanacea pulp decreases retention time of seeds in another frugivorous bird, Myadestes melanops. Murray et al.'s (1994) evidence for such a compound was questioned, since effect of dietary sugar concentration on retention time was not controlled (Witmer 1996a). Because our Control and FIG. 3. Percentage germination (A) and mean germination time (B) of S. americanum seeds from Control and Low Con- Extract fruits were essentially identical except for pres- centration (Low Conc.) fruits defecated by Cedar Waxwings ence of secondary metabolites in extract fruits, our during four time intervals. Percentage of germinated seeds study addressed this concern. was not in¯uenced by retention time but was signi®cantly We hypothesized that glycoalkaloids could produce reduced by presence of glycoalkaloids. Neither gut retention time nor presence of glycoalkaloids affected mean germi- laxative effects when consumed by frugivores. Our data nation time. do not support this hypothesis, as glycoalkaloids at low concentrations did not affect retention time and at high concentrations, increased it. Nonetheless, for two rea- 3A). Seeds from fruits containing glycoalkaloids had signi®cantly lower percentage germination than seeds from control fruits (F1,8 ϭ 5.63, P ϭ 0.045; Fig. 3A; Table 2). Percentage germination from Control fruits was approximately 50% higher than that of seeds from Low Concentration fruits. Rate of germination did not differ between seeds from Low Concentration and Con- trol fruits (Fig. 4). Although both curves reached an asymptote, which suggests seed germination was essentially complete, it is possible that ungerminated seeds had prolonged dormancy; we stress that our experiment tells us little about seed viability. Thus, while our results could suggest that glycoalkaloids inhibit germination permanently, they con®rm only that glycoalkaloids slow germination rate. Mean germination time was not affected by gut retention time (F ϭ 0.54, P ϭ 0.657; Table 2). Seeds 3,24 FIG. 4. Cumulative proportion of S. americanum seeds defecated within the ®rst time interval did not germi- from Control and Low Concentration (Low Conc.) fruits ger- nate sooner than seeds from the last time interval (Fig. minating over time. 2316 SOFIA A. WAHAJ ET AL. Ecology, Vol. 79, No. 7 sons we feel it premature to reject our hypothesis. First, Implications for seed dispersal we tested only a single species of seed-disperserÐa A central implication of our results is that plants bird. Mammalian seed dispersers could respond as pre- appear to have more control than previously thought dicted. Indeed, organisms differ in their response to over how quickly seeds are passed and, presumably, glycoalkaloids, depending upon the binding af®nity of how they are dispersed. Our results suggest that both the glycoalkaloid for sterols and upon the quantity and laxative and constipative compounds occur in S. amer- type of sterol within the cell membrane (Roddick et al. icanum, albeit not simultaneously present in high 1990, Keukens et al. 1992, Cipollini and Levey 1997c). enough concentrations to have opposing effects. In Second, we only tested the two most important gly- hindsight, presence of both types of compounds is not coalkaloids in ripe S. americanum and S. carolinense surprising; reports in the medical literature clearly doc- fruits. Other glycoalkaloids in these species might con- ument that fruit consumption can increase or decrease tribute to a laxative effect but their levels in fruit tissues defecation rate (Birnberg 1933, Willimott 1933, Joslin are so lowÐvirtually undetectable by HPLCÐthat we et al. 1938, Fries et al. 1950, McMillan and Thompson think it unlikely that they affect gut retention time. The 1979). Secondary metabolites in fruit may have many structurally similar potato glycoalkaloids (␣-solanine other functions, as well (reviewed in Cipollini and Lev- and ␣-chaconine), however, commonly occur in other ey 1997b). Nonetheless, by altering retention time of Solanum spp. and their saponin-like properties can seeds, some of them almost certainly affect patterns of cause diarrhea in mammals (Willimott 1933, McMillan seed deposition and may thus have consequences for and Thompson 1979, Dalvi and Bowie 1983, Hopkins seed dispersal and plant ®tness (Levey 1986, Murray 1995). 1988, Jordano 1992). We now discuss possible advantages and disadvantages to plants of variation in seed Retention time and germination retention time within frugivores and then discuss how Seeds are often altered by passage through vertebrate our results relate to the natural history of Solanum. guts. In particular, treatment in the gut often increases Retention time.ÐProbably the most widely recog- or decreases germination rate or success (Izhaki and nized consequence of longer retention times is in- Safriel 1990, Barnea et al. 1991, Ellison et al. 1993, creased dispersal distance (Murray 1988, Clench and Levey 1986, Lieberman and Lieberman 1986). Thus, Mathias 1992, Jordano 1992). Yet the link between it is reasonable to expect that the degree of treatment retention time and dispersal distance is not clear, es- (i.e., retention time) will in¯uence germination. Indeed, pecially if frugivores return frequently to a fruiting tree Murray et al. (1994) found a negative correlation be- (Lambert 1996). Furthermore, longer dispersal dis- tween gut retention time of seeds in Witheringia fruits tances are not necessarily bene®cial. For example, in and proportion of seeds germinating. Their interpre- plants that live in patchy environments, long retention tation of this result and their retention time experiment times may increase the risk of seeds being dispersed was that a compound in Witheringia fruit pulp reduced into adjacent areas of unsuitable habitat, because in a gut retention time, which in turn increased seed via- patchy environment the probability of encountering a bility. Note that Murray et al. (1994) did not consider similar habitat decreases with distance from the starting direct effects of secondary metabolites on seed viabil- point (Grace 1991, Lechowicz and Bell 1991, Bell ity; they inferred indirect effects via retention time. 1992). Also, once a seed is carried beyond the vicinity They could not test for direct effects because they of high mortality near its parent, dispersal distance like- lacked data on germination of seeds passed in the ab- ly becomes less important (Clark and Clark 1984, sence of secondary metabolites. Howe et al. 1985, Augspurger 1988, Merg 1994). In contrast to Murray et al.'s (1994) results, we found Consequences of variation in seed retention time no effect of retention time on seed germination, but may also depend upon the type of seed disperser. Many did ®nd a direct inhibitory effect of glycoalkaloids. In seed predators, for example, also disperse some seeds particular, ␣-solamargine and ␣-solasonine delayed (Janzen 1981, Levey 1986, Lambert 1989). In these seed germination, even at the low concentrations typ- species, shorter retention times are likely to increase ical of ripe S. americanum fruits. It is unclear from our the probability that seeds will survive gut passage (Jan- results whether the glycoalkaloids decreased viability zen 1984, Lambert 1989, Jordano 1992, Gardener et or simply caused a delay in germination. A reduction al. 1993). Movement rates of frugivores may likewise in viability would be puzzling, given that the glycoal- determine whether longer or shorter retention times are kaloids are naturally present in ripe fruits at the con- bene®cial to the fruiting plant. For instance, it has been centrations we used. A delay in germination would hypothesized (H. Horn, personal communication) that likewise be surprising because the Low Concentration because birds are generally more mobile than mammals seeds had such a short exposure time (Ͻ2 h) to gly- of equivalent size, shorter retention times are bene®cial coalkaloids in the fruit. This short exposure time was for bird dispersal (to avoid dispersal outside the proper the only difference between Control and Low Concen- habitat) and longer retention times are bene®cial for tration fruits. mammal dispersal (to ensure adequate dispersal dis- October 1998 SECONDARY METABOLITES AND RETENTION TIME 2317 tance within the habitat). Perhaps not coincidentally, these species may also tie into their dispersal ecology mammal-dispersed Solanum fruits have higher concen- and the constipative effects we observed on gut reten- trations of glycoalkaloids, which are constipative, than tion times of seeds. As discussed in the previous sec- bird-dispersed Solanum fruits, at least one of which tion, slowed seed passage could positively affect seed contains a compound that signi®cantly speeds seed pas- shadows produced by mammals, which are much less sage in a frugivorous bird (Cipollini and Levey 1997b, mobile than birds. Rapid seed passage, on the other this study). hand, is more likely to be bene®cial to bird-dispersed The picture that emerges is that variation in seed Solanum seeds. Indeed, Solanum spp. of the ®rst type retention time will likely affect dispersal distance, but have higher concentrations of glycoalkaloids than bird- that advantages and disadvantages of this variation will dispersed Solanum fruits, at least one of which contains differ widely, depending on plant habitat requirements, a compound that signi®cantly speeds seed passage in seed germination, behavior of dispersers, and the in- a frugivorous bird (Cipollini and Levey 1997b, this ternal treatment of seeds by frugivores. study). Variation in seed retention time is also likely to in- ¯uence the number of defecations with seeds and the CONCLUSION average number of seeds per defecation, from a given Our experiments support the work of Murray et al. meal of fruit. Indeed, we found that a 26% increase in (1994) by showing that Solanum americanum fruits gut passage time resulted in 28% more defecations con- contain a compound or compounds that enhance gut taining seeds. Because the number of seeds per fruit passage rates of seeds in a frugivorous bird. Contrary and number of fruits ingested were controlled in our to our expectations, glycoalkaloids are not responsible study, slower gut passage also resulted in fewer seeds for these effects, but may act instead to delay seed per defecation. A similar relationship between retention passage and perhaps to inhibit seed germination. Un- time and defecation pattern has been observed in other like Murray et al. (1994), we found no evidence that frugivorous birds (Levey 1986). In this context, in- gut passage rates have any in¯uence on germination. creased retention time is likely to be bene®cial for three It is possible, however, that under conditions less ideal reasons. First, it increases the number of dispersal than those encountered by our seeds, we might have events (i.e., sites to which seeds are dispersed; e.g., discovered a positive effect of rapid seed passage on Janzen 1981). Second, because defecations with fewer percentage germination of defecated seeds. seeds produce fewer seedlings, it reduces competition Our results provide an evolutionary explanation for among seedlings, which can be severe (Howe 1989, secondary metabolites in ripe fruit. They also suggest Loiselle 1990). Third, because seed predators are more that plants might exert some control over seed shadows likely attracted to high than low densities of seeds in produced by frugivores by affecting gut passage rates defecations (Janzen 1982, Kaspari 1993, Merg 1994, with speci®c laxative and/or constipative chemicals. Moegenburg 1994), it may decrease postdispersal seed We believe continued study of the ecological roles of predation. secondary metabolites of ripe ¯eshy fruits will uncover Glycoalkaloids and the natural history of Sola- mechanisms by which plant±frugivore interactions are num.ÐOur results relate to the natural history of two mediated. This underemphasized area deserves in- groups of Solanum spp. One group retains high levels creased attention by ecologists interested in under- of glycoalkaloids in ripe fruit, is dispersed by mam- standing and explaining the evolution of plant±frugi- mals, exhibits delayed germination, and has ripe fruits vore interactions. that persist on the plant for long periods. The other group has low levels of glycoalkaloids in ripe fruit, is ACKNOWLEDGMENTS dispersed primarily by birds, exhibits rapid germina- We thank R. Bodmer, C. Chapman, H. Horn, G. Murray, tion, and has fruits that are consumed quickly after M. Witmer, and J. Lambert for insightful discussions or com- ripening. We report elsewhere on these patterns and ments on the manuscript. Lincoln Brower provided cages. This work was funded by a Research Experiences for Un- describe in detail two species, one belonging to each dergraduates supplement to NSF grant DEB-9207920 and by groupÐS. carolinense and S. americanum, respectively the Berry College student work opportunity program. (Cipollini and Levey 1997a, 1997b). A proposed function of secondary metabolites in ripe LITERATURE CITED fruit is to prevent seed germination until after dispersal Abacus Concepts, I. 1989. SuperANOVA: Accessible gen- (Cipollini and Levey 1997b). Indeed, not only do gly- eral linear modeling, Berkeley, California, USA. Augspurger, C. K. 1988. Impact of pathogens on natural plant coalkaloids reduce percentage germination of Solanum populations. Pages 413±433 in A. J. Davy, M. J. Hutchings, seeds (this study), but their concentrations are highest and A. R. Watkinson, editors. Plant population biology. by far in Solanum spp. of the ®rst type. Presumably, Blackwell Scienti®c, Oxford, UK. species of the ®rst group (i.e., those with slow removal Barnea, A., Y. Yom-Tov, and J. Friedman. 1991. Does ingestion by birds affect seed germination? Functional Ecol- rates) have a greater need to inhibit germination of ogy 5:394±402. seeds within fruits than species whose fruits are re- Bell, G. 1992. Five properties of environments. Pages 33± moved rapidly. High glycoalkaloid concentration in 56 in P. R. Grant and H. S. Horn, editors. Mold, molecules, 2318 SOFIA A. WAHAJ ET AL. Ecology, Vol. 79, No. 7

and metazoa: growing points in evolutionary biology. acter syndromes in Mediterranean woody plants. American Princeton University Press, Princeton, New Jersey, USA. Naturalist 140:421±446. Birnberg, T. L. 1933. Raw apple diet in the treatment of Hopkins, J. 1995. The glycoalkaloids: naturally of interest diarrheal conditions in children. American Journal of Dis- (but a hot potato?). Food Chemical Toxicology 33:323± eases of Children 45:18±24. 339. Chapman, C. A., and L. J. Chapman. 1995. Survival without Howe, H. F. 1984. Constraints of the evolution of mutualism. dispersal? Seedling recruitment under parents. Conserva- American Naturalist 123:764±777. tion Biology 9:275±283. . 1989. Scatter- and clump-dispersal and seedling de- Cilliers, J. J. L., and V. L. Singleton. 1990. Auto-oxidative mography: hypothesis and implications. Oecologia 79:417± phenolic ring opening under alkaline condition as a model 426. for natural polyphenols in food. Journal of Agricultural and Howe, H. F., E. W. Schupp, and L. C. Westley. 1985. Early Food Chemistry 38:1797±1798. consequences of seed dispersal for a neotropical tree (Vi- Cipollini, M. L., and D. J. Levey. 1997a. Why are some rola surinamensis). Ecology 66:781±791. fruits toxic? Glycoalkaloids in Solanum and fruit choice by Ieven, M., D. A. Vanden Berghe, F. Mertens, A. Vlietinck, vertebrates. Ecology 78:782±798. and E. Lammens. 1979. Screening of higher plants for Cipollini, M. L., and D. J. Levey. 1997b. Secondary metab- biological activities. Planta Medica 36:311±321. olites of ¯eshy vertebrate-dispersed fruits: adaptive hy- Izhaki, I., and U. N. Safriel. 1990. The effect of some Med- potheses and implications for seed dispersal. American iterranean scrubland frugivores upon germination patterns. Naturalist 150:346±372. Journal of Ecology 78:56±65. Cipollini, M. L., and D. J. Levey. 1997c. Antifungal activity Janzen, D. H. 1977. Why fruits rot, seeds mold, and meat of Solanum fruit glycoalkaloids: implications for frugivory spoils. American Naturalist 111:691±713. and seed dispersal. Ecology 78:799±809. . 1981. Digestive seed predation by a Costa Rican Clark, D. A., and D. B. Clark. 1984. Spacing dynamics of Baird's tapir. Biotropica 13(suppl.):59±63. a tropical rain forest tree: evaluation of the Janzen-Connell . 1982. Removal of seeds from horse dung by tropical model. American Naturalist 124:769±788. rodents: in¯uence of habitat and amount of dung. Ecology Clench, M. H., and J. R. Mathias. 1992. Intestinal transit: 63:1887±1900. how can it be delayed long enough for birds to act as long- . 1984. Dispersal of small seeds by big herbivores: distance dispersal agents? Auk 109:933±936. foliage is the fruit. American Naturalist 123:338±353. Dalvi, R. R., and W. C. Bowie. 1983. Toxicology of solanine: Jordano, P. 1987. Frugivory, external morphology and di- an overview. Veterinary and Human Toxicology 25:13±15. gestive system in Mediterranean slyviid warblers Sylvia Denslow, J. S., D. J. Levey, T. C. Moermond, and B. C. spp. Ibis 129:175±189. Wentworth. 1987. A synthetic diet for fruit-eating birds. . 1992. Fruits and frugivory. Pages 105±51 in M. Wilson Bulletin 99:131±134. Fenner, editor. Seeds: the ecology of regeneration in natural Dey, P. M., and J. B. Harborne. 1989. Methods in plant plant communities. CAB International, London, UK. biochemistry, Vol. 1. Plant Phenolics. Academic Press, . 1995. Angiosperm ¯eshy fruits and seed dispersers: London, UK. a comparative analysis of adaptation and constraints in Duke, G. E. 1982. Gastrointestinal motility and its regula- plant-animal interactions. American Naturalist 145:163± tion. Poultry Science 61:1245±1256. 191. Ellison, A. M., J. S. Denslow, B. A. Loiselle, and D. BreneÂs Joslin, C. L., J. E. Bradley, and T. A. Christensen. 1938. M. 1993. Seed and seedling ecology of neotropical Me- Banana therapy in the diarrheal diseases of infants and lastomataceae. Ecology 74:1733±1749. children. Journal of Pediatrics 12:66±70. Fischer, K. E., and C. A. Chapman. 1993. Frugivores and Karasov, W. H., and D. J. Levey. 1990. Digestive system fruit syndromes: differences in patterns at the genus and trade-offs and adaptations of frugivorous passerine birds. species level. Oikos 66:472±482. Physiological Zoology 63:1248±1270. Foster, R. B. 1982. Famine on Barro Colorado Island. Pages Kaspari, M. 1993. Removal of seeds from neotropical fru- 201±212 in E. G. Leigh, A. S. Rand, and D. M. Windsor, givore droppings: ant responses to seed number. Oecologia editors. The ecology of a tropical forest: seasonal rhythms 95:81±88. and long-term changes. Smithsonian Institution Press, Keukens, E. A., T. de Vrije, C. H. Fabrie, R. A. Demel, W. Washington, D.C., USA. M. Jongen, and B. de Kruijff. 1992. Dual speci®city of Fries, J. H., N. J. Chiara, and R. J. Waldron. 1950. Dehy- sterol-mediated glycoalkaloid induced membrane disrup- drated banana in the dietetic management of diarrheas of tion. Biochemica et Biophysica Acta 1110:127±136. infancy. Journal of Pediatrics 37:366±372. Lambert, F.R. 1989. Pigeons as seed predators and dispersers Gardener, C. J., J. G. McIvor, and A. Jansen. 1993. Passage of ®gs in a Malaysian lowland forest. Ibis 131:521±527. of legume and grass seeds through the digestive tract of Lambert, J. E. 1996. Gut passage rate in Cercopthecus mon- cattle and their survival in faeces. Journal of Applied Ecol- keys: the in¯uence of body size and implications for seed ogy 30:63±74. dispersal. American Journal of Physical Anthropology Grace, J. 1991. Physical and ecological evaluation of het- 22(suppl.):144±145. erogeneity. Functional Ecology 5:192±201. Lechowicz, M. J., and G. Bell. 1991. The ecology and ge- Gregory, P. 1984. Glycoalkaloid composition of potatoes: netics of ®tness in forest plants. II. Microspatial hetero- diversity and biological implications. Phytopathology 57: geneity of the edaphic environment. Journal of Ecology 79: 497±501. 687±696. Herrera, C. M. 1982. Defense of ripe fruit from pests: its Levey, D. J. 1986. Methods of seed processing by birds and signi®cance in relation to plant-disperser interactions. seed deposition patterns. Pages 147±158 in A. Estrada and American Naturalist 120:218±241. T. H. Fleming, editors. Frugivores and seed dispersal. W. . 1984. Adaptation to frugivory of Mediterranean avi- Junk, Dordrecht, The Netherlands. an seed dispersers. Ecology 65:609±617. . 1988. Spatial and temporal variation in Costa Rican . 1989. Seed dispersal by animals: a role in angio- fruit and fruit-eating bird abundance. Ecological Mono- sperm diversi®cation? American Naturalist 133:309±322. graphs 58:251±269. . 1992. Historical effects and sorting processes as Levey, D. J., and G. E. Duke. 1992. How do frugivores explanations for contemporary ecological patterns: char- process fruit? Gastrointestinal transit and glucose absorp- October 1998 SECONDARY METABOLITES AND RETENTION TIME 2319

tion in Cedar Waxwings (Bombycilla cedrorum). Auk 109: Rodrigo, editors. The alkaloids. Academic Press, New 722±730. York, New York, USA. Levey, D. J., and A. Grajal. 1991. Evolutionary implications Robbins, C. T. 1993. Wildlife feeding and nutrition. Aca- of fruit-processing limitations in Cedar Waxwings. Amer- demic Press, New York, New York, USA. ican Naturalist 138:171±189. Roddick, J. G., A. L. Rijenberg, and M. Weissenberg. 1990. Lieberman, M., and D. Lieberman. 1986. An experimental Membrane-disrupting properties of the steroidal glycoal- study of seed ingestion and germination in a plant-animal kaloids solasonine and solamargine. Phytochemistry 29: assemblage in Ghana. Journal of Tropical Ecology 2:113± 1513±1518. 126. Sargent, S. 1995. Seed fate in a tropical mistletoe: the importance of host twig size. Functional Ecology 9:197±204. Loiselle, B. A. 1990. Seeds in droppings of tropical fruit- Schupp, E. W. 1988. Seed and early seedling predation in eating birds: importance of considering seed composition. the forest understory and in treefall gaps. Oikos 51:71±78. Oecologia 82:494±500. Tamboia, T., M. L. Cipollini, and D. J. Levey. 1996. An Martin, A. C., H. S. Zim, and A. L. Nelson. 1951. American evaluation of vertebrate seed dispersal syndromes in four wildlife and plants: a guide to wildlife food habits. Dover, species of black nightshade (Solanum sect. Solanum). Oec- New York, New York, USA. ologia 107:522±532. McMillan, M., and J. C. Thompson. 1979. An outbreak of Terborgh, J. 1986. Community aspects of frugivory in trop- suspected solanine poisoning in schoolboys: examinations ical forests. Pages 371±384 in A. Estrada and T.H. Fleming, of criteria of solanine poisoning. Quarterly Journal of Med- editors. Frugivores and seed dispersal. Dr. W. Junk, Dor- icine 48:227±243. drecht, The Netherlands. Merg, K. F. 1994. Utility of seed dispersal in escape from Van Gelder, W. M. 1990. Chemistry, toxicology, and occur- seed predation: an experiment in a lowland rainforest of rence of steroidal glycoalkaloids: potential contaminants of New Guinea. Thesis. University of Florida, Gainesville, the potato (Solanum tuberosum L.). Pages 117±156 in A. Florida, USA. F. M. Rizk, editor. Poisonous plant contamination of edible Meyers, N. 1983. A wealth of wild species. Westview Press, plants. CRC Press, Boca Raton, Florida, USA. Walsberg, G. E. 1975. Digestive adaptations of Phainopepla Boulder, Colorado, USA. nitens associated with the eating of mistletoe berries. Con- Moegenburg, S. M. 1994. Sabal palmetto seed-animal in- dor 77:169±174. teractions. Thesis. University of Florida, Gainesville, Flor- Warner, A. C. I. 1981. Rate of passage of digesta through ida, USA. the gut of mammals and birds. Nutrition Abstracts and Re- Murray, K. G. 1988. Avian seed dispersal of three neotrop- views 51B:789±820. ical gap-dependent plants. Ecological Monographs 58:271± Wheelwright, N. T. and G. H. Orians. 1982. Seed dispersal 298. by animals: contrasts with pollen dispersal, problems of Murray, K. G., S. Russell, C. M. Picone, K. Winnett-Murray, terminology, and constraints on coevolution. American W. Sherwood, and M. L. Kuhlmann. 1994. Fruit laxatives Naturalist 119:402±413. and seed passage rates in frugivores: consequences for plant Willimott, S. G. 1933. An investigation of solasonine poi- reproductive success. Ecology 75:989±994. soning. Analyst 58:431±439. Reid, N. 1989. Dispersal of mistletoes by honeyeaters and Witmer, M. C. 1994. Contrasting digestive strategies of fru- ¯owerpeckers: components of seed dispersal quality. Ecol- givorous birds. Cornell University Press, Ithaca, New York, ogy 70:137±145. USA. . 1991. Coevolution of mistletoes and frugivorous . 1996a. Do some bird-dispersed fruits contain natural laxatives? A comment. Ecology 77:1947±1948. birds? Australian Journal of Ecology 16:457±469. . 1996b. Annual diet of Cedar Waxwings based on Restrepo, C. 1987. Aspectos ecoloÂgicos de la diseminacioÂn U.S. Biological Survey records (1885±1950) compared to de cinco especies de mueÂrdagos por aves. Humboldtia 1: diet of American Robins: contrasts in dietary patterns and 65±116. natural history. Auk 113:414±430. Ricklefs, R. E. 1996. Morphometry of the digestive tracts Zitnak, A. 1979. Steroids and capsaicinoids of Solanaceous of some passerine birds. Condor 98:279±292. food plants. Pages 41±91 in N. F.Childers and G. M. Russo, Ripperger, H., and K. Schrieber 1981. Solanum steroid al- editors. Nightshades and health. Somerset Press, Somer- kaloids. Pages 81±192 in R. H. F. Manske and R. G. A. ville, New Jersey, USA.