Journal of Tropical Ecology (2004) 20:1–9. Copyright  2004 Cambridge University Press DOI:10.1017/S0266467404006121 Printed in the United Kingdom

Seed fate of two Sapotaceae species in a Guianan rain forest in the context of escape and satiation hypotheses Ste´phanie Chauvet*, Franc¸ois Feer and Pierre-Michel Forget Muse´um National d’Histoire Naturelle, De´partement Ecologie et Gestion de la Biodiversite´, UMR 8571 CNRS-MNHN, 4 Avenue du Petit Chaˆteau, F-91800 Brunoy, France (Accepted 2 December 2002)

Abstract: Seed removal by rodents was investigated for Manilkara huberi and Chrysophyllum lucentifolium in a French Guianan forest. According to the escape hypothesis, seed survival was expected to be greater in populations of low conspecific adult density, while on the contrary, under the satiation hypothesis, it was expected to be greater in populations of high density. The two plots under study showed opposite densities for the two studied tree species. Therefore, according to both hypotheses, seed survival at each plot was expected to be opposite between species. To assess seed fate, seeds were thread-marked in order to relocate them after removal and to determine whether they were consumed or scatterhoarded by rodents. Contrary to what was expected, our results showed that both M. huberi and C. lucentifolium had better survival in the same plot. This suggests that seed fate for both study species was not influenced by the density of conspecific adult trees, but was rather affected by other habitat characteristics, likely the global resource abundance. Variation in seed predation rates of both species seemed largely related to their respective fruiting period, while scatterhoarding rate seemed more affected by intrinsic seed characteristics. Key Words: Chrysophyllum lucentifolium, escape hypothesis, French Guiana, Manilkara huberi, rodents, satiation hypothesis, scatterhoarding, seed dispersal, seed predation

INTRODUCTION In tropical regions, plant species commonly suffer very high predation rates between seedfall and germination, that can have major impacts on plant recruitment (De Steven & Putz 1984, Howe et al. 1985, Janzen et al. 1976, Osunkoya 1994, Schupp 1990, Sork 1987). Many studies have focused on the mortality factors involved and have examined which environmental and/or physiological factors may reflect a variation in predator responses. Janzen (1970) proposed that seeds are consumed by many obligatory or facultative host-specific plant parasites and predators, whose negative effects on plant recruitment decline with increasing distance from the parent tree, and/ or with decreasing seed density. From this model referred to as the ‘Janzen–Connell model’ (Connell 1971, Janzen 1970) or as the ‘escape hypothesis’ (Howe & Smallwood 1982), predictions can be inferred at the individual- or the population-level. At the individual level, seed survival is expected to increase with distance to the parent tree (Howe 1993, Notman et al. 1996, Peres et al. 1997, Terborgh et al. 1993). At the population level, seed sur-

* Corresponding author. Present address: Umeå University, Department of Ecology and Environmental Science, SE-901 87 Umeå, Sweden. Email: chauvet [email protected]

vival is expected to be negatively correlated with the density of conspecific adult trees, as the proportion of habitat exposed to specific predators increases with tree density (Schupp 1992). However, Janzen (1971) suggested that a phenomenon of predator satiation by the plant species might be induced by the obligatory or facultative host specificity of seed parasites and predators. Indeed, it should be an evolutionary response of plant species to seed predators that allows survival of some of the seeds produced. Under this ‘predator satiation’ hypothesis, seed survival is expected to increase with seedfall density as predators are overwhelmed by the abundance of seeds (Schupp 1990). This predator satiation can operate at different spatial scales. At the crop level, plants that escape by predator satiation should invest much energy in seed production, by producing either numerous small seeds or fewer larger seeds. At the population level, predator satiation occurs for plant species whose fruiting period is highly synchronized between trees (Janzen 1971), and whose adult tree densities are large enough to satiate predators (Forget 1992, Schupp 1992). Lastly, predator satiation may also operate at the community level, e.g. when the plant community exhibits seasonal changes in seedfall (Schupp 1990), or during large synchronous fruiting events (Hammond & Brown 1998, Janzen 1974).

2

STE´PHANIE CHAUVET, FRANC¸OIS FEER AND PIERRE-MICHEL FORGET

Because frugivorous rodents are abundant and widespread, and since the proportion of seeds they remove is often very high (Brewer & Rejma´nek 1999, Holl & Lulow 1997), rodents have major effects on plant demography (Leigh et al. 1993). During the year, they feed on many seed species that are available, and can therefore be considered as generalist seed predators (Emmons 1982, Henry 1999). Nevertheless, they are able to discriminate seeds of different species and they show some distinct preferences (Jansen & Forget 2001, Price & Jenkins 1986, Vander Wall 1990, 1995). Thus, when a preferred seed species is abundant (i.e. during its fruiting period), rodents may switch to this seed species and become temporally specialized predators. In such cases, rodents can be considered as facultative host-specific predators (Janzen 1970), and their action on seed survival may thus be studied with regard to both escape and satiation hypotheses. Though studies investigating the action of rodents often concentrate on their negative effect on plant recruitment through seed predation (Holl & Lulow 1997, Howe et al. 1985, Notman et al. 1996, Schupp 1988, 1992; Terborgh et al. 1993), rodents also have a positive effect on plant fitness as seed dispersers when seed scatterhoarding occurs (Forget 1990, Hallwachs 1986, Jansen & Forget 2001, Smythe 1989, Sork 1983). This trade-off between limitation and enhancement of seed survival probability by rodents depends heavily on seed characteristics such as nutrient reward or toxicity (Price & Jenkins 1986). In this study, as scatterhoarding by rodents may occur, we examine seed fate rather than seed removal for two tree species belonging to the Sapotaceae family. Both seed species are highly desired by rodents and very abundant during their peak fruiting period owing to a high synchronization in fruiting among trees. Rodents may therefore act as facultative host-specific predators (sensu Janzen 1970) for these species. In a first experiment, we investigated variation in seed fate that occurred between two plots located in the same study area, and between both species whose respective peak fruiting periods are 1 mo apart. Results are discussed with regard to both escape and satiation hypotheses. In a second experiment, in order to discriminate which part of variation in seed fate between species is due to specific seed characteristics, we compare their respective seed fate when both species are tested together out of their respective peak fruiting periods.

MATERIALS AND METHODS Study site and plots The study was conducted at the Nouragues Biological Station (4°05′N; 52°40′W), French Guiana. This site is covered by a continuous tropical rain forest considered to be primary (Charles-Dominique 2001), and receives 2990

mm of rainfall per year, with a long dry season in September–November and a short dry season usually in March (Grimaldi & Rie´ra 2001). The fauna is intact owing to the effective protection of this site, and game species are present and abundant (Bongers et al. 2001). The study site lies at the base of a granitic dome (inselberg) and consists of two tablelands separated by a creek, called respectively Petit Plateau (PP) and Grand Plateau (GP). This study was carried out at two 9-ha plots located 500 m from each other, one on each of the tablelands (Figure 1a). These plots differ in forest composition (Forget et al. 2001a, Poncy et al. 2001), but show similar animal community composition. The abundance of caviomorph rodents was estimated during regular censuses along a 1500-m pathway around and across each plot in April–May 1995 (n = 5 censuses) and 1997 (n = 5). The number of Myoprocta exilis encountered was low (0.40 and 0.33 individual per km at PP and GP, respectively), but greater than the number of Dasyprocta leporina (0.20 and 0.07 individual per km at PP and GP, respectively) (P.-M. Forget, unpubl. data). Small rodents and especially spiny rats (Proechimys spp.) are also present on both plots, but no quantitative data are available (Feer & Forget, 2002). Study species Manilkara huberi (Ducke) Chevalier and Chrysophyllum lucentifolium Cronq. (Table 1) are both large canopy trees belonging to the Sapotaceae. Manilkara huberi fruits every 2 y and C. lucentifolium every year, both with yearto-year variations in seed production. For both species, fruiting is synchronous among trees, with a heavy crop of seeds produced per tree. As a consequence, frugivores and granivores intensively forage in and under the crown leading to their obligatory and facultative host-specific behaviour. Manilkara huberi fruits during the overall fruiting peak, i.e. in March–April, whereas C. lucentifolium fruits in April–May, when the number of fruiting species and overall resource abundance decrease (Sabatier 1985). Like other Sapotaceae (Julliot & Sabatier 1993, Simmen & Sabatier 1996), C. lucentifolium and M. huberi predominate in the diet of the two largest primates (Ateles paniscus and Alouatta seniculus), which swallow seeds without destroying them, and are therefore crucial primary seed dispersers of both species (Julliot 1996). The pattern of dispersal is either scattered along travel routes or clumped in faeces under trees used as defecation and sleeping sites. Following primary seed dispersal, seeds of both species may be consumed by granivorous mammals or be secondarily dispersed by scatterhoarding rodents before germination occurs (Spironello 1999). Video-recording experiments performed in April 1997 at Nouragues by Jansen and co-workers with marked M. huberi seeds revealed that seed removal was primarily by

Escape or satiation hypotheses?

(a)

3

(b)

Petit Plateau (PP)

Plot M Plot M

Cr ee k

N

Plot C Grand Plateau (GP)

200 m

Figure 1. Study plots (a) and experimental design of seed sets (b) at each of the 9-ha plots (300 × 300 m) at Nouragues, French Guiana. In the first experiment, five marked seeds were placed at each seed sets at both plots, on 21 March and 19 April, for M. huberi and C. lucentifolium respectively. In the second experiment, five marked seeds of each species were placed together on 9 May, at 10 out of the 20 seed sets at plot M.

acouchies (P. Jansen, pers. comm.). However, exclosure experiments conducted at the same period by Feer & Forget (2002) with C. lucentifolium seeds showed that spiny rats also removed seeds at our study plots. Despite peccaries having been seen under fruiting trees during the period of study (pers. obs.), the patterns of seed removal observed in our experiments strongly suggest that they were not foraging at our seed sets. Indeed, peccaries are known to crush and kill all seeds on the spot (Jansen & Forget 2001), which was rarely observed in our experiments, as at least part of the seeds were moved before consumption or caching. Overall, we may thus expect that seed removal in this study was mostly due to M. exilis and Proechymis spp. (Feer & Forget 2002; Jansen & Forget 2001). Densities of adult trees of both studied species contrast between the two tablelands at the Nouragues station, thus allowing examination of whether differences in seed Table 1. Characteristics of Manilkara huberi and Chrysophyllum lucentifolium (C. Julliot & B. Simmen, pers. comm.). Study species Phenology Fruiting Fruit size1 (cm) Wet fruit mass1 (g) No. seeds per fruit1 Seed size2 (cm) Wet seed mass2 (g) Germination delay

M. huberi

C. lucentifolium

Every 2 y March to April 3.7 × 3.3 20.5 ± 0.6 1.32 ± 0.06 2.4 × 1.3 × 0.7 1.13 ± 0.01 10–12 mo

Annual April to May 5.7 × 5.7 97.4 ± 5.4 4.01 ± 0.16 2.5 × 1.2 × 0.8 1.42 ± 0.03 1–2 wk

1 Mean values (± 1 SE) calculated for n = 60 fruits in M. huberi and n = 68 fruits in C. lucentifolium. 2 Mean values (± 1 SE) calculated for n = 300 seeds per species.

removal follow the prediction of the escape or satiation hypotheses. The PP area has a high density of M. huberi whereas the GP area has a dense population of C. lucentifolium (Forget et al. 2001a). The plot in the PP area is thus named plot M (for Manilkara), and the plot in the GP area is named plot C (for Chrysophyllum) (Figure 1a). The respective densities of both species at both plots are presented in Table 2, with the level of seed survival that is expected according to the escape and satiation hypotheses. Almost all M. huberi trees produced fruits in the study plots (i.e. 23 out of 26), while about half of C. lucentifolium trees did (i.e. 19 out of 33).

Spatial variation in seed fate The experiment was conducted with seeds collected under fruiting trees. We tested for seed viability by placing them in water, and excluding the floating unsound ones, which are likely to be parasitized or aborted. Following Brewer & Rejma´nek (1999) and Forget (1990), seeds were labelled with 1-m-long white nylon threads in order to relocate them after removal and to know whether they were consumed or scatterhoarded by rodents. For each species, 20 sets of five marked seeds were placed 25 m apart along two perpendicular 250-m transects (10 sets per transect) crossing the centre of each plot (Figure 1b, see Forget et al. 2001a for similar design). Seeds of M. huberi and C. lucentifolium were placed during the peak of their respective fruiting periods (i.e. when predator satiation at the species level may occur), 21 March and 19 April 1997, respectively. Removed threaded seeds were

STE´PHANIE CHAUVET, FRANC¸OIS FEER AND PIERRE-MICHEL FORGET

4

Table 2. Mean (± 1 SE) tree density per hectare (dbh > 30 cm) and level of seed survival expected at the population scale for Manilkara huberi and Chrysophyllum lucentifolium at two 9-ha forest plots at Nouragues, French Guiana. These predictions are based on the assumption that both seed species are consumed by specialist predators. Plot M Species Density Predicted level of seed survival Escape hypothesis Satiation hypothesis

Plot C

M. huberi 2.8 ± 0.5

C. lucentifolium 0

M. huberi 0.1 ± 0.1

C. lucentifolium 3.7 ± 0.5

low high

high low

high low

low high

relocated within a 5-m-radius circular area throughout 28 d after the experiment started. The removal distance between the point of relocation and the initial point of placement was measured. Seeds remaining untouched by terrestrial mammals were considered as intact seeds. Removed seeds were recorded as consumed when the thread was found alone, or scatterhoarded when the seed was retrieved cached under leaves or buried beneath ground level. Unretrieved seeds were considered lost, as we could not assess if they were hidden or consumed further away than 5 m or taken to burrows. Among seeds that were primarily cached by rodents, some could have been secondarily consumed before the end of experiment (Jansen et al. 2002, Vander Wall & Joyner 1998). Those seeds will be further included in the category ‘consumed seeds’. Seed fate as a function of seed species Seeds used in this second experiment were also collected under fruiting trees, tested for viability, and threadmarked. Five marked seeds of each species (10 seeds per set) were placed together on 9 May 1997 at 10 seed sets in plot M. This experiment allowed us to test for eventual differences in seed removal and fate that may be due to intrinsic characteristics of seed species. Indeed, as this experiment was performed out of the peak fruiting period of both species, results are not influenced by their respective fruiting period and the likely differential satiation between species. Fates of seeds were recorded 14 d later as previously described, and distance of removal was also measured. At the time of this experiment, the shortage of seeds of both species prevented us from conducting the experiment in both plots. We chose to conduct the experiment at plot M rather than at plot C because in experiment 1, seed removal was higher in this plot, allowing us to expect seed removal at each set of seeds. Statistical analyses The purpose of this work was to study the seed fate of both species, rather than providing and comparing their survival curves. Therefore, results were not analysed as right-censored data. Instead, three variables of seed fate

were considered: the proportions of seeds that were consumed, cached or lost per set. In both experiments, these proportions were arcsine square-root transformed before analysis in order to approach normality (Sokal & Rohlf 1981). Because these variables were correlated, they were analysed with Multivariate Analysis of Variance (MANOVA) with Systat 9.0 software (SPSS 1999). In experiment 1, both species were tested at two different periods. Thus, when comparing M. huberi and C. lucentifolium seed fate, we did not test for the effect of species, but for two confounded effects: the species effect and the period effect. We therefore call it the species/ period effect. All seed fate variables were analysed with a two-way MANOVA, with the main effects being plot and species/period, and with the interaction plot × species/ period. As plots show opposite densities for the two studied species (Table 2), we expected plot and plot × species/period effects to be significant if density of conspecific trees influenced the seed fate of both species in the same way (escape or satiation hypothesis). In experiment 2, we compared seed fates between M. huberi and C. lucentifolium when all seeds were tested at the same period. The species effect was analysed for all seed fate variables with a one-way MANOVA.

RESULTS Variations in seed fate between species and plots During the respective fruiting period of each species, most seeds were removed, namely 68 and 36% for M. huberi and 96 and 83% for C. lucentifolium, at plots M and C, respectively (Figure 2). MANOVA indicated that the overall seed fate was significantly different between M. huberi and C. lucentifolium as well as between plots, but that there was no effect of the interaction (Table 3), indicating that variation in seed fate between plots is comparable between species. Univariate tests associated with the MANOVA indicated that the significant effect of species on seed fate was due to the differences in proportions of consumed and cached seeds (Table 3). Indeed, seeds of M. huberi were less frequently consumed but more often cached than seeds of C. lucentifolium at either plot M or plot C (Figure 2). Univariate tests also indicated that the

Escape or satiation hypotheses?

5

Percentage of seeds

100 80 60 40 20 0 I Co Ca L

Plot M

I Co Ca L

I Co Ca L

Plot C

Plot M

M. huberi

I Co Ca L

Plot C

C. lucentifolium

Figure 2. Seed fate (mean ± 1 SE) of M. huberi in March and C. lucentifolium in April after 28 d, in a 5-m radius from the set of seeds. For each species, n = 100 seeds per plot. Black bars indicated seeds that were primarily cached then secondarily consumed. Legend: I = intact seeds; Co = seeds retrieved consumed with/without primarily caching; Ca = cached seeds; L = lost seeds (lost seeds may have been removed at a distance > 5 m, or taken to burrows).

significant effect of plots on seed fate was due to the difference in the proportion of consumed seeds between plots (Table 3), both M. huberi and C. lucentifolium being more often consumed at plot M than at plot C (Figure 2). For both species combined, 43.2% of seeds consumed were eaten at the site where seeds were initially placed, while 94.9% of seeds cached were hidden further away. When seeds were moved before consumption or caching, the mean removal distances observed for M. huberi were respectively for consumed and cached seeds: 1.5 ± SE 0.4 m (n = 14) vs. 1.5 ± 0.1 m (n = 35) at plot M, and 1.7 ± 0.4 m (n = 10) vs. 1.2 ± 0.3 m (n = 6) at plot C. The mean removal distances observed for C. lucentifolium were respectively for consumed and cached seeds: 2.0 ±

SE 0.2 (n = 47) vs. 1.9 ± 0.4 m (n = 10) at plot M, and 1.6 ± 0.3 (n = 27) vs. 1.3 ± 0.3 m (n = 6) at plot C. These removal distances did not differ significantly between consumed and cached seeds, either at plot M or at plot C, for both M. huberi (t-test for plot M: t[47] = 0.04, P = 0.97; plot C: t[14] = −0.95, P = 0.36) and C. lucentifolium (plot M: t[55] = −0.25, P = 0.80; plot C: t[31] = −0.41, P = 0.69). Seed fate as a function of seed species Nearly all M. huberi and C. lucentifolium seeds were rapidly removed within 14 d (90 and 92%, respectively), when seeds were placed at plot M in May, i.e. out of their respective fruiting period (Figure 3). Despite comparable

Table 3. Multivariate analysis of variance and associated univariate tests of the effects of species (M. huberi and C. lucentifolium) and plot (plots M and C) on variables of seed fate (consumed, cached and lost seeds) after 28 d. This experience was performed at Nouragues during the respective fruiting period of both species. The proportions of seeds consumed, cached and lost were arcsine square-root transformed before analysis. MANOVA Source Species Plot Species × Plot ANOVA: summary of F values Source Species Plot Species × Plot

df

Lambda

F

3, 74 3, 74 3, 74

0.619 0.854 0.969

15.2 4.22 0.783

df

Consumed

Cached

1, 76 1, 76 1, 76

*** P < 0.001, ** P < 0.01, ns = not significant.

30.5*** 8.15** 0.688 ns

7.99** 3.01 ns 1.17 ns

P < 0.001 0.008 0.507 Lost 3.04 ns 0.170 ns 0.007 ns

STE´PHANIE CHAUVET, FRANC¸OIS FEER AND PIERRE-MICHEL FORGET

6

Percentage of seeds

100

DISCUSSION Within-forest spatial variations in seed fate

80 60 40 20 0 I Co Ca L

I Co Ca L

M. huberi

C. lucentifolium

Figure 3. Seed fate (mean ± 1 SE) of Manilkara huberi and Chrysophyllum lucentifolium both tested in May at plot M, after 14 d, in a 5-m radius from the set of seeds. n = 50 seeds per species. Black bars indicated seeds that were primarily cached then secondarily consumed. I = intact seeds; Co = seeds retrieved consumed with/without primarily caching;Ca = cached seeds; L = lost seeds (lost seeds may have been removed at a distance > 5 m, or taken to burrows). The two species differ in percentage of cached seeds (P < 0.05) as well as in percentage of lost seeds (P < 0.05). According to measures of removal distance, the ratio cached/consumed among lost seeds is expected to be equivalent to the ratio observed among retrieved seeds.

proportions of seeds remaining intact, a significant effect of species on seed fate was detected by MANOVA (Table 4). Univariate tests indicated that this species effect was due to significant differences between species in the proportions of cached seeds and lost seeds (Table 4). Indeed, M.huberi seeds were more often cached and less frequently lost than C. lucentifolium seeds, while overall proportions of consumed seeds were equivalent between species (Figure 3). When seeds were moved before consumption or caching, the mean removal distance was respectively for consumed and cached seeds: 2.4 ± SE 0.4 (n = 14) vs. 1.4 ± 0.2 m (n = 12) for M. huberi (t[24] = −2.12, P = 0.04), and 1.3 ± 0.3 (n = 17) vs. 1.5 ± 0.2 m (n = 3) for C. lucentifolium (t[18] = 0.17, P = 0.87).

Depending on the assumption that rodents are facultative host-specific predators for M. huberi and C. lucentifolium seeds, and depending on the respective conspecific tree density of study species at each plot, seed survival was expected to be opposite between species and plots, according to both the escape and satiation hypotheses (Table 2). However, contrary to what was expected, our study did not show any opposite results between species, as both had better survival on the same plot (Figure 2). Seed survival was estimated in two different ways. A first estimate consists in the proportion of seeds remaining untouched by rodents until the end of experiment (i.e. intact seeds) after assuming that all seeds removed are or will be consumed before seed recruitment occurs. When M. huberi and C. lucentifolium seeds were placed and observed during their respective fruiting periods, proportions of intact seed were greater for both species at plot C than at plot M. A second estimate of seed survival consists in a measure of seeds remaining viable until the end of the experiment, namely intact plus cached seeds. If lost seeds are excluded from consideration, then this estimate is the complement of seed mortality (i.e. seed consumption). For both species and at both plots, the mean removal distance did not differ between consumed and cached seeds. This suggests that the fate of seeds, i.e. consumed or cached, may not affect the distance at which the rodents remove the seed. Therefore, if lost seeds are removed further away than 5 m, the ratio cached/consumed among lost seeds might be equivalent to the ratio cached/consumed observed among retrieved seeds. Furthermore, the proportion of lost seeds did not differ between species or between plots. Thus, lost seeds can be excluded from consideration without biasing the interpretations of the results. It therefore provides equivalent analyses to consider the proportions of viable seeds or seed mortality after 28 d. Both M. huberi and C. lucentifolium had a greater proportion of seeds consumed

Table 4. Multivariate analysis of variance and associated univariate tests of the effect of species (M. huberi and C. lucentifolium) on variables of seed fate (consumed, cached and lost seeds) after 14 d. This experience was performed at plot M in May, out of the fruiting period of both species. The proportions of seeds consumed, cached and lost were arcsine square-root transformed before analysis. MANOVA Source Species ANOVA: summary of F values Source Species * P < 0.05, ns = not significant.

df

Lambda

F

P

3, 16

0.622

3.24

0.050

df

Consumed

Cached

Lost

1, 18

0.068 ns

7.56*

4.68*

Escape or satiation hypotheses?

at plot M than at plot C, meaning a greater proportion of viable seeds at plot C than at plot M. Thus, for both the proportions of intact seeds and viable seeds, M. huberi and C. lucentifolium do not show opposite results as expected. Considering that only one or the other of the escape or satiation hypotheses should apply to both study species, the absence of any opposite results suggests that neither of these hypotheses seems appropriate to describe within-forest spatial variation in seed removal. This may result either from violation of assumption if rodents are actually not acting as facultative specialists, or from other differences or mechanisms driving variation in seed fate, such as differences in the abundances of rodents and/or resources. Effect of species on seed fate As neither the escape nor the satiation hypothesis can explain variation in seed removal between plots if only one of these hypotheses is supposed to apply to both study species, then we may advocate considering a ‘species’ effect, and thus for considering species separately. Indeed, rodents could remove M. huberi seeds as expected by the escape hypothesis, while they may remove C. lucentifolium seeds as expected by the satiation hypothesis. In this case, high densities of M. huberi seeds might stimulate feeding activities of rodents, while low densities of these seeds might not represent an attractive resource for them. Observations by Feer & Simmen (unpubl. data) showing that M. huberi was the most abundant fruit species encountered at plot M in 1997 are consistent with this hypothesis. Along a 900-m census transect, they encountered 31 and 38 patches of fruits (and/or seeds) respectively on the 3 and 25 April, among which respectively 12 and 19 were M. huberi fruit patches. Conversely, because of their chemical or nutritional properties (Price & Jenkins 1986), seeds of C. lucentifolium might be very attractive to rodents, even at low density. Indeed, C. lucentifolium seeds contain a high percentage of water-soluble carbohydrates (Kinzey & Norconk 1993), a component that positively affects preferences for seeds by heteromyid rodents (Lockard & Lockard 1971, Price 1983). But on the other hand, owing to their high nutritional value, high densities of C. lucentifolium seeds may satiate rodents. We know of no data on chemical or nutritional properties of M. huberi seeds that would allow a comparison of the two study species. When M. huberi and C. lucentifolium seeds were set together in May at plot M, they showed similar proportions of intact and consumed seeds, but differed in the proportion of cached seeds. This suggests that intrinsic seed characteristics influence seed scatterhoarding but not seed consumption for these two species. As Vander Wall (1990) has emphasized, rodents are influenced by traits of seeds, and particularly by perishable components such as

7

cotyledon nutrient contents (Jansen & Forget 2001). Thus, independent of the nutritional quality of seeds, it is likely to be advantageous for a rodent to hoard M. huberi seeds that present a 10–12-mo dormancy before germination rather than C. lucentifolium seeds which germinate in 1– 2 wk and thus quickly lose their nutritional value. One may argue that the greater proportion of lost seeds observed for C. lucentifolium may affect this interpretation if these lost seeds were indeed cached further away than 5 m (Figure 3). However, among the retrieved seeds of C. lucentifolium, the removal distance did not differ significantly between consumed and cached seeds, indicating that the fate of one seed did not determine the distance to which it was removed. We can therefore consider that the ratio of cached/consumed seeds is similar among lost seeds and retrieved seeds, which further supports the result that the proportion of cached seeds is indeed influenced by species. Satiation hypothesis at a larger scale: the community level Because seeds had greater survival at plot C than at plot M in both study species, the effect of forest type (Forget et al. 2001b) may be important, since it may overshadow the effect of M. huberi and C. lucentifolium tree density. Heterogeneity in plant composition and density may explain differences in the level of resources for rodents. Thus, along the census transect, the number of fruit patches (M. huberi and C. lucentifolium excluded) was almost twice as many at plot C (32 and 38) as at plot M (19 and 19), at the beginning and end of April 1997, respectively (F. Feer & B. Simmen, unpubl. data). If rodents behave as generalists rather than as facultative specific predators, then this difference in resource abundance may have consequences for seed fate of M. huberi and C. lucentifolium, because the foraging activity of rodents may be influenced by the total amount of resources available at one plot (Schupp 1990), rather than only by conspecific seed density. The greater survivorship observed for M. huberi and C. lucentifolium seeds at plot C might then be due to the greater global resource abundance at this plot. Thus, even if the satiation hypothesis does not seem to apply at the population level (see above), it may still be decisive at the community level, as rodents may be satiated by the whole seedfall in plot C. When seeds of M. huberi and C. lucentifolium were placed and observed during their respective peak fruiting periods (i.e. in March–April and April–May respectively), the proportion of seeds consumed was greater for C. lucentifolium than M. huberi, while there was no difference when seeds were placed at the same time at plot M in May. This suggests that the difference in seed survival observed during the first experiment was probably not due to species but rather to the time of the experiment. Indeed,

8

STE´PHANIE CHAUVET, FRANC¸OIS FEER AND PIERRE-MICHEL FORGET

seed survival was higher in March–April (period of experiment with M. huberi) than in April–May (period of experiment with C. lucentifolium) as observed by Forget (1996) in Carapa procera, a rodent-dispersed tree species. This is consistent with the hypothesis of satiation at the community level, since consumer satiation seems to occur during the whole fruit-peak (i.e. the peak of resource abundance) in March–April (Sabatier 1985), thus promoting better seed survival than later in the wet season (May), when resources are becoming scarcer (Forget et al. 2002). In conclusion, our results did not provide any support to the escape or satiation hypotheses at the population scale, while they suggest that predator satiation may act at the community level. This emphasizes the necessity of conducting ecological studies at different spatial and temporal scales, and at different levels of organization (individual, population and community) as already noted by Schupp (1992).

recruitment of a tropical canopy tree, Dipteryx panamensis, in Panama. Oikos 43:207–216. EMMONS, L. H. 1982. Ecology of Proechimys (Rodentia, Echimidae) in South-Eastern Peru. Tropical Ecology 23:280–290. FEER, F. & FORGET, P.-M. 2002. Spatio-temporal variations in postdispersal seed fate. Biotropica, 34: 555–566. FORGET, P.-M. 1990. Seed-dispersal of Vouacapoua americana (Caesalpiniaceae) by caviomorph rodents in French Guiana. Journal of Tropical Ecology 6:459–468. FORGET, P.-M. 1992. Seed removal and seed fate in Gustavia superba (Lecythidaceae). Biotropica 24:408–414. FORGET, P.-M. 1996. Removal of seeds of Carapa procera (Meliaceae) by rodents and their fate in rainforest in French Guiana. Journal of Tropical Ecology 12:751–761. FORGET, P.-M., FEER, F., CHAUVET, S., JULLIOT, C., SIMMEN, B., BAYART, F. & PAGE`S-FEUILLADE, E. 2001a. Post-dispersal seed removal in frugivore-dispersed tree species. Pp. 265–274 in Bongers, F., Charles-Dominique, P., Forget, P.-M. & The´ry, M. (eds). Nouragues: dynamics and plant–animal interactions in a neotropical rainforest. Kluwer Academic Publishers, Dordrecht.

ACKNOWLEDGEMENTS Many thanks to S. Brewer, J.-F. Cosson, A. Dalecky, R. Froissart, L. Gaume, B. Godelle, D. McKey, E. Notman, D. Pozner, E. Schupp and two anonymous reviewers for helpful comments on the manuscript; to P. Jansen and C. Julliot for interesting discussions and help in the field; to B. Simmen for providing unpublished data on fruit availability; and to C. Julliot and B. Simmen for providing unpubl. data on M. huberi and C. lucentifolium characteristics. Special thanks to C. Doutreland, R. Froissart, A. Leitao and D. Pozner for their helpful support. This study was granted by the Ministe`re de l’Environnement with a grant SOFT (Sol et Foreˆt Tropicaux) # 96042 to P.-M. Forget and B. Simmen and a research grant from the CNRS.

FORGET, P.-M., RANKIN-DE MERONA, J. M. & JULLIOT, C. 2001b. The effects of forest type, harvesting and stand refinement on early seedling recruitment in a tropical rain forest. Journal of Tropical Ecology 17:593–609. FORGET, P.-M., HAMMOND, D., MILLERON, T. & THOMAS, R. 2002. Seasonality of fruiting and food hoarding by rodents in Neotropical forests: consequences for seed dispersal and seedling recruitment. Pp. 241–253 in Levey, D. J., Silva, W. R. & Galetti, M. (eds). Seed dispersal and frugivory: ecology, evolution and conservation. CABI Publishing, Wallingford. GRIMALDI, M. & RIE´RA, B. 2001. Geography. Pp. 9–18 in Bongers, F., Charles-Dominique, P., Forget, P.-M. & The´ry, M. (eds). Nouragues: dynamics and plant–animal interactions in a neotropical rainforest. Kluwer Academic Publishers, Dordrecht. HALLWACHS, W. 1986. Agoutis (Dasyprocta punctata): the inheritors of Guapinol (Hymenaea courbaril: Leguminosae). Pp. 119–135 in Estrada, A. & Fleming, T. H. (eds). Frugivores and seed dispersal.

LITERATURE CITED BONGERS, F., CHARLES-DOMINIQUE, P., FORGET, P.-M. & THE´RY, M. 2001. Nouragues: dynamics and plant–animal interactions in a neotropical rainforest. Kluwer Academic Publishers, Dordrecht. 421 pp. ´ NEK, M. 1999. Small rodents as significant BREWER, S. W. & REJMA dispersers of tree seeds in a neotropical forest. Journal of Vegetation Science 10:165–174. CHARLES-DOMINIQUE, P. 2001. The field station. Pp. 1–7 in Bongers, F., Charles-Dominique, P., Forget, P.-M. & The´ry, M. (eds). Nouragues: dynamics and plant–animal interactions in a neotropical rainforest. Kluwer Academic Publishers, Dordrecht. CONNELL, J. H. 1971. On the role of natural enemies in preventing competitive exclusion in some marine animals and in rain forest trees. Pp. 298–312 in den Boer, P. J. & Gradwell, P. R. (eds). Dynamics of populations. Pudoc, Wageningen. DE STEVEN, D. & PUTZ, F. E. 1984. Impact of mammals on early

Dr W. Junk, Dordrecht. HAMMOND, D. S. & BROWN, V. K. 1998. Disturbance, phenology and life-history characteristics: factors influencing distance/densitydependent attack on tropical seeds and seedlings. Pp. 51–78 in Newbery, D. M., Prins, H. T. T. & Brown, N. D. (eds). Dynamics of tropical communities. Blackwell Science, Oxford. HENRY, O. 1999. Frugivory and the importance of seeds in the diet of the orange-rumped agouti (Dasyprocta leporina) in French Guiana. Journal of Tropical Ecology 13:291–300. HOLL, K. D. & LULOW, M. E. 1997. Effects of species, habitat, and distance from edge on post-dispersal seed predation in a tropical rainforest. Biotropica 29:459–468. HOWE, H. F. 1993. Aspects of variation in a neotropical seed dispersal system. Vegetatio 107/108:149–162. HOWE, H. F. & SMALLWOOD, J. 1982. Ecology of seed dispersal. Annual Review of Ecology and Systematics 13:201–228. HOWE, H. F., SCHUPP, E. W. & WESTLEY, L. C. 1985. Early con-

Escape or satiation hypotheses?

sequences of seed dispersal for a neotropical tree (Virola surinamensis). Ecology 66:781–791. JANSEN, P. & FORGET, P.-M. 2001. Scatterhoarding by rodents and tree regeneration in French Guiana. Pp. 275–288 in Bongers, F., Charles-Dominique, P., Forget, P.-M. & The´ry, M. (eds). Nouragues: dynamics and plant–animal interactions in a neotropical rainforest. Kluwer Academic Publishers, Dordrecht. JANSEN, P. A., BARTHOLOMEUS, M., BONGERS, F., ELZINGA, J. A., DEN OUDEN, J. & VAN WIEREN, S. E. 2002. The role of seed size in dispersal by a scatterhoarding rodent. Pp. 209–225 in Levey, D. J., Silva, W. & Galetti, M. (eds). Seed dispersal and frugivory: ecology, evolution and conservation. CABI Publishing, Wallingford. JANZEN, D. H. 1970. Herbivores and the number of tree species in tropical forests. American Naturalist 104:501–527. JANZEN, D. H. 1971. Seed predation by animals. Annual Review of Ecology and Systematics 2:465–492. JANZEN, D. H. 1974. Tropical blackwater rivers, animals, and mast fruiting by the Dipterocarpaceae. Biotropica 6:69–103. JANZEN, D. H., MILLER, G. A., HACKFORTH-JONES, J., POND, C. M., HOOPER, K. & JANOS, D. P. 1976. Two Costa Rican batgenerated seed shadows of Andira inermis (Leguminosae). Ecology 57:1068–1075. JULLIOT, C. 1996. Seed dispersal by red howling monkeys (Alouatta seniculus) in the tropical rainforest of French Guiana. International Journal of Primatology 17:239–258. JULLIOT, C. & SABATIER, D. 1993. Diet of the red howler monkey (Alouatta seniculus) in French Guiana. International Journal of Primatology 14:527–550. KINZEY, W. G. & NORCONK, M. A. 1993. Physical and chemical properties of fruit and seeds eaten by Pithecia and Chiropotes in Surinam and Venezuela. International Journal of Primatology 14:207– 228. LEIGH, E. G., WRIGHT, J. S., HERRE, E. A. & PUTZ, F. E. 1993. The decline of tree diversity on newly isolated tropical islands: a test of a null hypothesis and some implications. Evolutionary Ecology 7:76–102. LOCKARD, R. B. & LOCKARD, J. S. 1971. Seed preference and buried seed retrieval of Dipodomys deserti. Journal of Mammalogy 52:219– 221. NOTMAN, E., GORCHOV, D. L. & CORNEJO, F. 1996. Effect of distance, aggregation, and habitat on levels of seed predation for two mammal-dispersed neotropical rain forest tree species. Oecologia 106: 221–227. OSUNKOYA, O. O. 1994. Postdispersal survivorship of North Queensland rainforest seeds and fruits: effects of forest, habitat and species. Australian Journal of Ecology 19:52–64. PERES, C. A., SCHIESARI, L. C. & DIAS-LEME, C. 1997. Vertebrate predation of Brazil-nuts (Bertholletia excelsa, Lecythidaceae), an

9

agouti-dispersed Amazonian seed crop: a test of the escape hypothesis. Journal of Tropical Ecology 13:69–79. PONCY, O., SABATIER, D., PRE´VOST, M. F. & HARDY, I. 2001. The lowland high rainforest: tree structure and species diversity. Pp. 31–46 in Bongers, F., Charles-Dominique, P., Forget, P.-M. & The´ry, M. (eds). Nouragues: dynamics and plant–animal interactions in a neotropical rainforest. Kluwer Academic Publishers, Dordrecht. PRICE, M. V. 1983. Laboratory studies of seed size and seed species selection by heteromyid rodents. Oecologia 60:259–263. PRICE, M. V. & JENKINS, S. H. 1986. Rodents as seed consumers and dispersers. Pp. 191–235 in Murray, D. R. (ed). Seed dispersal. Academic Press, Sydney. SABATIER, D. 1985. Saisonnalite´ et de´terminisme du pic de fructification en foreˆt guyanaise. Revue d’Ecologie (Terre Vie) 40:89–320. SCHUPP, E. W. 1988. Factors affecting post-dispersal seed survival in a tropical forest. Oecologia 76:525–530. SCHUPP, E. W. 1990. Annual variation in seedfall, postdispersal predation, and recruitment of a neotropical tree. Ecology 71:504–515. SCHUPP, E. W. 1992. The Janzen–Connell model for tropical tree diversity: population implications and the importance of spatial scale. American Naturalist 140:526–530. SIMMEN, B. & SABATIER, D. 1996. Diets of some French Guianan primates: food composition and food choices. International Journal of Primatology 17:661–693. SMYTHE, N. 1989. Seed survival in the palm Astrocaryum standleyanum: evidence for its dependence upon its seed dispersers. Biotropica 21:50–56. SOKAL, R. R. & ROHLF, F. J. 1981. Biometry. (Second edition). W. H. Freeman and Company, New York. 859 pp. SORK, V. L. 1983. Distribution of pignut hickory (Carya glabra) along a forest to edge transect, and factors affecting seedling recruitment. Bulletin of the Torrey Botanical Club 110:494–506. SORK, V. L. 1987. Effects of predation and light on seedling establishment in Gustavia superba. Ecology 68:1341–1350. SPIRONELLO, W. R. 1999. The Sapotaceae community ecology in a central amazonian forest: effects of seed dispersal and seed predation. Ph.D. dissertation. University of Cambridge, Cambridge. SPSS Inc. 1999. SYSTAT 9.0 Statistics. Chicago, Illinois. TERBORGH, J., LOSOS, E., RILEY, M. P. & BOLANOS RILEY, M. 1993. Predation by vertebrates and invertebrates on the seeds of five canopy tree species of an Amazonian forest. Vegetatio 107/108:375– 386. VANDER WALL, S. B. 1990. Food hoarding in animals. University of Chicago Press, Chicago. 445 pp. VANDER WALL, S. B. 1995. The effects of seed value on the caching behavior of yellow pine chipmunks. Oikos 74:533–537. VANDER WALL, S. B. & JOYNER, J. W. 1998. Recaching of Jeffrey pine (Pinus jeffreyi) seeds by yellow pine chipmunks (Tamias amoenus): potential effects on plant reproductive success. Canadian Journal of Zoology 76:154–162.