Citation
Interactions among attalea palms, bruchid beetles, and neotropical terrestrial fruit-eating mammals : implications for the evolution of frugivory

Material Information

Title:
Interactions among attalea palms, bruchid beetles, and neotropical terrestrial fruit-eating mammals : implications for the evolution of frugivory
Creator:
Silvius, Kirsten Mariana, 1965-
Publication Date:
Language:
English
Physical Description:
x, 290 leaves : ; 29 cm.

Subjects

Subjects / Keywords:
Beetles ( jstor )
Ecology ( jstor )
Eggs ( jstor )
Endocarp ( jstor )
Forests ( jstor )
Fruits ( jstor )
Larvae ( jstor )
Peccaries ( jstor )
Species ( jstor )
Trees ( jstor )
Animal-plant relationships ( lcsh )
Attalea -- Habitat ( lcsh )
Bruchidae -- Host plants ( lcsh )
Dissertations, Academic -- Zoology -- UF ( lcsh )
Frugivores -- Ecology ( lcsh )
Zoology thesis, Ph.D ( lcsh )
City of Gainesville ( local )
Genre:
bibliography ( marcgt )
non-fiction ( marcgt )

Notes

Thesis:
Thesis (Ph.D.)--University of Florida, 1999.
Bibliography:
Includes bibliographical references (leaves 271-289).
General Note:
Typescript.
General Note:
Vita.
Statement of Responsibility:
by Kirsten Mariana Silvius.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
The University of Florida George A. Smathers Libraries respect the intellectual property rights of others and do not claim any copyright interest in this item. This item may be protected by copyright but is made available here under a claim of fair use (17 U.S.C. §107) for non-profit research and educational purposes. Users of this work have responsibility for determining copyright status prior to reusing, publishing or reproducing this item for purposes other than what is allowed by fair use or other copyright exemptions. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder. The Smathers Libraries would like to learn more about this item and invite individuals or organizations to contact the RDS coordinator (ufdissertations@uflib.ufl.edu) with any additional information they can provide.
Resource Identifier:
030367458 ( ALEPH )
41936537 ( OCLC )

Downloads

This item has the following downloads:


Full Text











INTERACTIONS AMONG ATTALEA PALMS, BRUCHID BEETLES, AND
NEOTROPICAL TERRESTRIAL FRUIT-EATING MAMMALS: IMPLICATIONS
FOR THE EVOLUTION OF FRUGIVORY













By


KIRSTEN MARIANA SILVIUS


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA


1999























To Joe,

without whose imagination

creativity

determination

bullying cajoling

loving

laughing

and, of course, canoeing

none of this would have been possible

or even desirable.















ACKNOWLEDGMENTS


Many people contributed to all stages of this

dissertation, and without them both field research and writing-up would have been very difficult. Two people, however, provided emotional and logistic support without which the whole thing might well have been impossible. I am deeply grateful to Jan Sendzimir and Gisela Bosch for their friendship and unique definition of human community. Over the years their support took the form of computers, books, hot tubs, unending glasses of beer and wine, a dog to walk and hug, two wonderful daughters to remind me of the other important things in life, a beautiful wedding celebration and, most importantly, a temporary place to stay that became a home to return to even if some day it is transferred to some other place on this Earth. They are proof that permanence can come from this impermanent process called graduate school.

A similar role was played by two members of the World Doctors (Medecins du Monde) mission to the Yanomami Area in Brazil. Joseph Ceruetto and Mattieu Lena were incomparable companions and true friends under difficult circumstances. They saw the links between the conservation of nature and the prevention of ethnocide, and provided logistic support for my


iii









research that few others were in a position to provide. They introduced me to the Yanomami people, their land, and the indigenists that are struggling to save both, and thus provided me with the strongest possible motivation to work for tropical rain forest conservation. The entire World Doctors team, including the coordinators in France, is to be commended for its vision and hard work.

Four other people contributed over and beyond the call of duty to the completion of this dissertation:

My parents, Robert and Rosita Silvius, were critical to the endeavor not only by providing their usual rock-solid trust and encouragement (and strategically placed care packages), but also by taking care of bureaucratic, financial and other logistic requirements on the home front while I was in Brazil.

John Eisenberg enabled the completion of this

dissertation by believing that independence is a good thing despite the academic risks it carries. During those times when I felt that maybe I had it all wrong and the data were not that interesting or valuable, his boundless fascination with and insight into all things natural rekindled my enthusiasm for the work. His insight into human nature also helped me evaluate more clearly the personal and nonacademic situations that came up while I was writing.

Buzz Holling contributed more than he can imagine by providing a different, albeit lumpy, way of thinking about ecology. His was the most influential of all the classes I


iv









took during my graduate career. He also enabled my escape from my sixth-floor computerless cubicle by inviting me into the superbly equipped Arthur Marshall lab. Buzz and the other denizens of 111 Bartram--Lance, Garry, Jan, Paul and the irreplaceable Toni Carter--provided advice, inspiration, technical assistance and a surfeit of Canadian humour. Without them writing up would have been no fun at all, eh?

I would also like to thank all members of my committee for providing essential information in the form of discussions, references and editing. Colin Chapman always seemed to have the papers I needed. Walter Judd's Tropical Botany course has stayed with me over the years as a firm reminder that ecology without taxonomy is an exercise in futility. Richard Bodmer's focus on the interesting evolutionary questions my data brought up was refreshing. I will also always be grateful for his inspired introduction of a desiccated Mauritia palm fruit at a crucial junction in my qualifying exams.

The Wildlife Conservation Society provided most of the funding for my research, and I would like to thank John Robinson and Alejandro Grajal for supporting a plant-oriented project. The Lincoln Park Zoo provided additional funding through the Scott Neotropical Fund. CNPq provided additional support in the form a visiting researcher fellowship to my husband, Jos6 Fragoso, and by granting me permission to carry out research in Brazil.









The staff at Maraca Island Ecological Reserve ensured the smooth running of the station and comfortable living conditions in a remote place. Station Manager Guttemberg Moreno de Oliveira did his best to understand the nature of our research and relied on his love of Maracd to navigate the political storms that threaten research in places like Roraima. A consummate humanist, when push came to shove he always did the right thing for his friends.
















TABLE OF CONTENTS


ACKNOWLEDGMENTS ........................................... iii

ABSTRACT ................................................... ix

CHAPTERS

1 INTRODUCTION .............................................1

2 PALM FRUITS AND NEOTROPICAL RAIN FOREST MAMMALS: DIRECT
AND INDIRECT RELATIONSHIPS .............................3

Neotropical Palm Use by Mammals ........................3
Methods ................................................6
Results and Discussion: Palm-Vertebrate Interactions...8
Results and Discussion: Fruit-Vertebrate-Bruchid
Interactions.....................................17
Summary and Conclusions ...............................29

3 PULP HANDLING BY MAMMALIAN FRUIT-EATERS AFFECTS PALM
SEED PREDATION BY BRUCHID BEETLES. .....................47

Introduction....... ....................................47
Study Area and Species ................................51
Methods ...............................................55
Results ...............................................63
Discussion ............................................71

4 PALM ENDOCARP USE BY THREE NEOTROPICAL SEED-EATERS:
FRUGIVORY OR"GRUBIVORY?"...............................96

Introduction ..........................................96
Study Site ............................................99
Study Species ........................................100
Methods ..............................................102
Results ..............................................112
Discussion ...........................................123

5 AGOUTI (DASYPROCTA AGOUTI) HABITAT USE, DIET, AND HOME
RANGE ON MARACA ISLAND ECOLOGICAL RESERVE, NORTHERN
BRAZILIAN AMAZON .....................................141

Introduction .........................................141
Study Area ...........................................143
Methods ..............................................145


vii









Results ..............................................153
Discussion ...........................................165

6 SCALE-DEPENDENT DISTRIBUTION PATTERNS IN PLANTS USED
BY THREE TERRESTRIAL RAIN FOREST MAMMALS AT MARACA
ISLAND ECOLOGICAL RESERVE ............................198

Introduction .........................................198
Study Site ...........................................201
Methods ..............................................207
Results ..............................................211
Discussion ...........................................221


7 CONCLUSIONS AND IMPLICATIONS ...........................246

Phenology: Methods and Implications.................. 246
Palm-Mediated Bruchid-Vertebrate Interactions ........247 Complexity in Frugivory Systems. ......................248
Mosaics of Fruit Availability: Scale and Grain Size..252
Plant-Animal Interactions and the Aggregated
Distribution of Tropical Forest Plants.......... 255

APPENDICES

1 "EDIBLE" PLANT SPECIES ON PHENOLOGY TRANSECTS ......... 264

2 FLORISTIC PLOT DATA SUMMARIZED FROM MILLIKEN AND
RATTER (1989) AND NASCIMENTO (1994) .................266

REFERENCES ................................................271

BIOGRAPHICAL SKETCH .......................................290


viii















Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy

INTERACTIONS AMONG ATTALEA PALMS, BRUCHID BEETLES AND
NEOTROPICAL TERRESTRIAL FRUIT-EATING MAMMALS: IMPLICATIONS
FOR THE EVOLUTION OF FRUGIVORY BY

Kirsten Mariana Silvius

May 1999


Chairman: John F. Eisenberg
Major Department: Zoology


The fruits and seeds of arborescent palms are important food items in the diet of Neotropical terrestrial mammals. A literature review shows that seventy-seven species of mammals have been documented feeding on fifty-nine species of palms. Most of the records are of pulp rather than seed consumption. Among mammals, terrestrial consumers are the most important predators and dispersers of palm seeds. Bruchid beetles (Coleoptera: Bruchidae: Pachymerini) are specialized predators on palm seeds and are associated almost exclusively with the coconut-like palms (tribe Cocoeae).

Manipulation of Attalea palm fruits indicated that pulp removal by rodents, primates, and ungulates improves fruit quality for ovipositing bruchid beetles. Fruits that fall to the ground unhandled by any vertebrate have the best chance









of surviving beetle infestation: they are rapidly removed by tapirs (Tapirus terrestris) and are later defecated at sites where infestation by bruchids is lower than near parent trees.

Monitoring of palm endocarp use by white-lipped

peccaries (Tavassu pecari), agoutis (DasyvDrocta aqouti), and squirrels (Sciurus igniventris) indicates that all three species feed on bruchid beetle larvae. Squirrels are palm specialists that feed on both larvae and endosperm year round; agoutis feed on both larvae and seeds only at times and at sites where alternate foods are not available; and white-lipped peccaries feed preferentially on bruchid larvae rather than seeds. The three species should therefore respond independently and distinctly rather than as a community to changes in Attalea fruit availability.

Home range size of five radio-collared agoutis ranged

from 2.9 to 8.5 ha. All animals showed a shift in core area use as different individuals of the same tree species and individuals of different tree species came into fruit. Agoutis responded to their habitat at the scale of individual trees, and appeared to base foraging decisions on knowledge of food available on their entire home range. Preferred diet items were primarily in the Sapotaceae (Pouteria surumuensis and Pradosia surinamensis) and secondarily in the Palmae (Astrocarvum aculeatum pulp and seeds and Attalea maripa pulp).















CHAPTER 1
INTRODUCTION



Arborescent palms, a widespread and abundant group of plants with a fairly coherent fruiting biology, are an important source of food for Neotropical forest mammals. Palm fruits are staples in the diet of some species at some study sites (ungulates, Bodmer 1989a,b; squirrels, Glanz et al. 1996) and have been proposed as "key" or "keystone" food species for frugivorous mammals (Terborgh 1986a,b, Spironelo 1991, Peres 1994a) at other sites. Due to their abundnace and importance in frugivore diets, palms can serve as easily studied models of the evolution and ecology of frugivory systems. Their importance in the diet of terrestrial mammals in particular may help us understand the role of these poorly studied species in seed dispersal, seed predation and forest regeneration. However, the unique life history characteristics of palms must be understood before the results of palm ecology studies can be extrapolated to tropical frugivory systems in general.

Palm-mammal interactions have been studied in detail at only a few Neotropical sites. As an introduction to a detailed study of the interactions between one palm species and its mammalian consumers, in Chapter 2 I review the current state of knowledge and assumptions about the








importance of palms to fruit- and seed-eating Neotropical forest mammals. The remainder of the dissertation focuses on the ecology of one palm, Attalea mariPa, and examines 1) the implications of the spatial distribution of trees for fruit availability to animals operating at different spatial scales and 2) the role of insects in fruit-vertebrate interactions.

Chapters 3 and 4 focus on two aspects of the interaction among bruchid beetles, Attalea maripa palms, and vertebrate fruit and seed consumers. Chapter 5 addresses the effect of the spatial distribution of the palms, within several forested habitats, on the pattern of food availability for one species of terrestrial fruit and seed eating mammal: the agouti, Dasyprocta agouti. Throughout, I address three components of frugivory theory: the role of keystone food resources in structuring frugivore communities; the effect of patchiness in food distribution on animal movements and demography; and the contribution of plant-animal interactions to the patchy distribution of plant species.














CHAPTER 2
PALM FRUITS AND NEOTROPICAL RAIN FOREST MAMMALS: DIRECT AND INDIRECT RELATIONSHIPS



Neotrooical Palm Use by Mammals



Neotropical arborescent palms produce fruits that in

their size and color fit the classical mammal dispersed fruit syndrome, as defined by van der Pijl (1982). The fruits can be large (e.g., 5 cm diameter for Mauritia flexuosa, 7 cm length for Attalea maripa; van Roosmalen 1985), with a nutritious oily or sweet mesocarp (pulp). The seed is usually protected by a thick endocarp (nut wall) of up to 0.5 cm in thickness. The exocarp (husk) and mesocarp are usually yellowish, orange, reddish-brown, or brown in color (Uhl and Dransfield 1987, van Roosmalen 1985). Some medium sized (< 20 m tall) and understory palms also produce similar fruits. Other small palms produce fruits that seem better adapted for bird dispersal due to the small size of their seeds and the red color of the fruits (e.g., Bactris spp.), although their fruits are also ingested by primates and their seeds dispersed after passage intact through the gut (e.g., Desmoncus, van Roosmalen 1985).

Currently, the Neotropical palm fruit- and seed-eating

fauna disperses seeds either by ingesting them along with the








fruit and passing them intact through the gut (e.g., tapirs), by spitting out the seeds after eating the pulp (e.g., tapirs and some primates), or by scatter-hoarding seeds (e.g., Dasyprocta and other rodents; Smythe 1986, van Roosmalen 1985). Terrestrial rather than arboreal frugivores and granivores tend to be the most important seed predators and dispersers for palms, using the pulp or seeds or both after they have fallen to the ground. Few mammals except for some primates and occasionally squirrels seem to harvest the fruits or seeds from the tree itself. Macaws appear to be the only important avian consumers of large palm fruits and seeds (Munn 1988). The only true palm specialist among them, the Hyacinth macaw (Anodorhynchus hyacinthinus), also feeds on palm seeds from the ground in the open Pantanal habitat (Munn et al. 1990). The ecological interactions between large palm fruits and terrestrial consumers suggest that palm fruit use by terrestrial frugivore/granivores represents more than just the use of residual fruit production. Examination of the foraging ecology of terrestrial fruit eaters may bring new insight to the study of frugivory, since paradigms in this field are based almost completely on work done with birds and primates.

Seasonal and supra-annual variation in levels of fruit availability can determine the carrying capacity for frugivorous animals, as well as lead to the evolution of behavioral, physiological, and morphological adaptations to deal with food shortages (Eisenberg 1980, Terborgh 1986a,b,






5


Leighton and Leighton 1983, Malcolm 1990, Caldecott 1992, van Schaik et al. 1993, Leigh et al. 1996). Any mammals that have become palm specialists or that rely heavily on palm fruits will see their population dynamics affected by the spatiotemporal patterns of palm fruit availability in their habitat. Although there are few studies on the fruiting phenology of arborescent palms, three studies of palm fruiting at the community level (de Steven et al. 1987, Sist 1989, Peres 1994b) and one population level study (Pifero and Sarukhan 1982, Piiero et al. 1984) demonstrate that there is great seasonal and supra-annual variability in the likelihood of reproduction of individual palms and palm patches alike. The predictability of these production patterns is unknown, but palms do not appear to be more reliable a fruit source than other tropical fruiting trees. Recent data from Barro Colorado Island, Panama (BCI; J. Wright pers. comm.) indicate that large and small fruited palms participate along with dicots in forest-wide fruit peaks and failures following El Nifto Southern Oscilation events.

The same factors that make palm fruits attractive to other mammals--nutritious pulp and seeds--also make them highly attractive to humans. Native human populations traditionally have used palms for their fiber, stem tissue, and sap as well as for their fruits (Balick 1988). Thus their current ecology has been affected by at least several thousand years of interactions with humans, which can affect








distribution and fruit traits (e.g., domestication of Bactris aassipaes) (Balbe 1988, Anderson et al. 1991).

For this review I surveyed the Neotropical ecological literature to extract the following information: Which mammals are known to use palms? Numerically, how important are palms in the diets of these animals? Since seeds are a long lasting resource and pulp a more ephemeral one, which part of the palm fruit is most heavily used? What is the nature of the interaction among palms, bruchid beetles specialized on palms, and vertebrate palm consumers? What are the ecological and evolutionary implications of these interactions?



Methods



The literature summarized here is a biased sampling of the information available on palm-mammal interactions, because it does not include many articles published in Latin American journals. It does however include unpublished theses from Latin American universities and unpublished data from several Latin American researchers. From each source I obtained the following information: the palm species used by mammals, the mammal species using palms, the part of the palm fruit used, the location of the study, and notes on the behavior of the mammals, in particular whether they seemed to be using the palm as a keystone resource in the season of fruit scarcity (sensu Terborgh 1986a,b). For studies that








lasted about a year or more and gave information on the dietary composition of the study species I summarized the importance of palm fruits as the proportion contributed to the diet. Most of the studies reported here took place in humid forest, some in deciduous or dry forest, and a few in open habitats. I do not include information on use of any palm parts other than pulp and seeds, although meristem and flowers are reported as important resources for many mammals, as are the arthropod populations that shelter in palms and which are heavily used by insectivorous or omnivorous arboreal and scansorial mammals, particularly primates. The available literature on palm bruchids is limited to ecological studies at a few Neotropical sites (BCI, Maracd Island Ecological Reserve in northern Brazil, and a few brief studies in Costa Rica, Peru, and Bolivia.)



Results and Discussion, Palm-Mammal Interactions



General Trends



At least 77 species of mammals in 44 genera were

reported to feed on palm pulp or seeds from 59 species of palms (excluding potential synomyms and species identified to genus only; Table 2-1). Five of these mammals species are from non-forested habitats, leaving a total of 72 mammals reported using palm fruits in Neotropical forests, either humid or dry. It is likely that many of the mammals listed








in Table 2-1 will feed on palms whenever they co-occur with them, even though they have not been reported to do so in the literature (e.g., the larger marsupials are only described as eating palms at two sites, but they probably do so throughout their range).

The five most frequently cited genera in Table 2-1 are reported from a limited number of study sites: Cebus (4 sp., 10 sites), Ateles (3 sp., 7 sites), Tavassu (2 sp., 6 sites), Sciurus (7 sp., 7 sites), Dasyprocta (2-3 sp., 4 sites). The best represented orders are primates (19 species) and rodents (29 species), followed by ungulates (Perissodactyla, 2 species, Artiodactyla, 5 species; note that all Neotropical forest ungulate species are represented), marsupials and carnivores (6 species each, but only 2 forest carnivores and 4 forest marsupials). The Chiroptera are last with 2 species and Edentates are represented only by two species mentioned in two studies (one Edentate species in Peru was not identified (Allen 1997).

Conspicuously absent from Table 2-1 are the large fruit bats, Artibeus sp., which are reported only from Trinidad, and the smaller marsupials. In the case of bats, it is likely that their small body size and relatively weak jaws preclude them from using most palm species, as they would be enable to carry away large fruits or to tear away the resistant husks (J. Eisenberg pers. comm.). In the case of marsupials, they are probably absent from consumer lists because their nocturnal habits make them less susceptible to








detection by researchers. Although Bodmer's (1989a,b, 1990, 1991) studies show that all ungulates rely heavily on palms, because ungulates are seldom studied, their interactions with palms in the rest of their range are not reported. In particular, they are not reported from French Guiana, despite the abundance of studies on smaller mammals there.

Members of the Procyonidae probably feed on palms

throughout their range. Rainforest canids, if studied, might be found to feed on fallen palm fruits. Canids in two nonforest sites (Brazilian cerrado and Sonoran desert) used palm fruits. Although the most cited order, rodents are underrepresented relative to their richness, abundance, and frugivorous/granivorous diet. Only half of the neotropical squirrels are reported, one of three Heteromys, and a few of the Echimydae. The agouti (Dasvprocta sp.), despite its frequency in Central and South America, is reported only from French Guiana, BCI, Maracd, northeast Peru and Belize. In a captive study, Adler (1995) found that Proechimys semispinosus in Central America ate the pulp and seeds of all palm fruits offered, including seeds with very thick endocarps such as Attalea. However, Emmons (pers. comm.) believes that due to their weak jaw structure Proechymis rats are not likely to be significant predators on Attalea palms.

Primates, on the other hand, are well represented,

probably due to the large number of detailed studies that have been carried out on their behavior and social structure. At the same time, primates alone among the species listed









have the hand and mouth dexterity necessary to open palm nuts and exploit both the mature seeds and the liquid endosperm. This is illustrated by Izawa's (1979) report of Cebus apella using bamboo sticks to beat and poke open palm fruits, and Struhsaker and Leland's (1977) description of palm nut smashing by the same species.


Evidence for Mammalian Dependence on Palms--Staple Food vs.
Keystone Items


Terborgh (1986a) defines keystone resources in

Neotropical forests as species that are common in the diets of several members of a frugivore community during periods when little fruit is available, and proposes that such resources help structure Neotropical animal communities. This is a unique usage of the term, distinct from other definitions in the literature (Power et al. 1996, Gilbert 1980); throughout this paper I refer to "keystone species" sensu Terborgh. Based on the use of palm nuts by capuchin monkeys in Cocha Cashu, Peru during the season of food scarcity, Terborgh proposes palms as keystone food resources. He qualifies this proposal, however, by stating that keystone status would hold only "if other consumer species [besides capuchins] are relying on the same resources;" in that case, "the plants that produce these resources can be recognized as playing a critical role in the ecosystem (p. 337)." Here I review the frequency of palm nut versus pulp use during times of fruit scarcity by Ceu and other fruit-eaters.









Pulp was the most frequently cited fruit part eaten by mammals (96 citations, based on a simple count of the number of times each part is listed in Table 2-1). Mature or old seeds followed with 64 citations, then immature seeds and entire fruit with 16 and 20 citations respectively. Fifty two references to palm eating by mammals did not mention the part eaten. Contrary to the emphasis given in the literature to palm seed-rodent interactions and to long-lasting palm nuts as a keystone resource (Terborgh 1986a,b), pulp may be a more important resource than seeds for the frugivore community. However, it is difficult to estimate the true importance of each fruit part, since the amounts of each part eaten are seldom given in the literature.

Few studies of Neotropical rainforest mammals last long enough to cover seasonal changes in food use, let alone supra-annual variation. Even if the data are gathered, they are not always presented in a form from which it is easy to calculate the percent importance of a food item in the diet (e.g., Terborgh 1983). Table 2-2 summarizes the information from a few studies that examined dietary composition to some degree. For five species of ungulates (Mazama americana, M. gouazoubira, Tayassu pecari, T. tajacu, Tapirus terrestris).

Bodmer (1989a) derived the percent composition of the diet from stomach contents gathered over a 15 month study period by local hunters in northeast Peru. All species had at least one palm in the list of the top three diet items, and four of them (Tavassu taiacu was the only exception) had palms as the









top two diet items. It seems that in this area all these species are dependent on palms, but the fact that peccaries at least do well in areas where no palms occur (e.g., Olmos 1993) suggests that they have a fairly flexible diet and that in other areas they may have other major resources. Throughout Amazonia, however, reports from other researchers suggest that palms are important at least for peccaries (Anderson 1983, Fragoso 1993).

In her study of Ateles belzebuth belzebuth on Maraca Island, Roraima, northern Brazil, Nufies (1992, 1995) found that palms were not an important diet item for this species. One exception was Attalea maria, which during one of the months of fruit scarcity constituted 10.1% of the diet, but it represented only 1.23% of the yearly diet. Even though the fruit was consumed at a time of fruit low, it was the pulp rather than the mature or old endosperm that was consumed. Roosmalen (1980) found an even lower percent importance of palm fruits in the diet of another spider monkey, Ateles Daniscus, during a two year study in Suriname. On the other- hand, Castellanos and Chanin (1996) found that A. belzebuth in southern Venezuela relied heavily on palms and used them throughout the year in proportion to their abundance.

In general palm use by Ateles is low when compared to palm use by Cebus. Buckley's (1983) study of Cbu caoucinus, where one palm species constituted 58.2% of the diet in October and was the highest ranking diet item for the









entire year. It was eaten during seven months of the year, not just during the season of fruit scarcity. This last figure is comparable to the 64% of diet reported for three months for Cebus apella in Manu, and 56% for Cebus albifrons in the same period (Terborgh 1983). There may thus be a real difference in the degree of palm use by spider vs. capuchin monkeys, but a comparison needs to be made controlling for the abundance of palms in the habitat.

Izawa (1979) comments that palm fruits were usually

available for a longer period of time (more asynchrony) than other fruit species, and that in addition primates fed on them before, during, and after they ripened, which increased their period of availability. Primates' ability to use the liquid immature endosperm seems to be one of the traits that makes palms an important resource for them. Observations such as these highlight the importance of clearly documenting the availability of--and distinguishing between--different maturity stages of fruits in phenological studies.

Several studies report that Cebus in other parts of

their range turn to palm nuts during the season of scarcity. U. Bobadilla (pers. comm.) documented Chiropotes sp. and Cebus sp. feeding on the pulp of Attalea speciosa (as Orbianva phalerata) in the rain forests of western Pard, Brazil, during the period of forest wide fruit low. This case is unusual because the palm fruited during the dry rather than the wet season; similar phenological behavior is reported for Attalea chalerata in the Beni of Bolivia (D.








Quiroga pers. comm.), whereas Attalea palms at study sites north of the Equator tend to fruit in the wet season, generally the peak of forest wide fruit availability. That is the time when Ateles belzebuth fed heavily on Attalea maria pulp in southern Venezuela (Castellanos and Chanin 1996). Bobadilla suggests that Attalea soeciosa may be an asynchronous fruiter at his high rainfall site, but this may be an artifact of low palm densities. (Even when a plant population has a well-marked peak in fruit production, there often are some individuals that fruit outside of the peak. When a plant population occurs at low density, non-peak fruiters may be as well represented in the population or in the samples as peak time fruiters.)

Jessenia (Oenocarpus) bataua in the southwestern Amazon (Peres 1994a) produces fruits primarily in the dry season, when other fruits are not available at this site, but also produces some fruits year round. Among the primates, Cebus apella relies heavily on the immature seeds during this time and Peres (1994b) suggests that the palm acts as a key food item. However, he also indicates that Cbu as well as Pithecia and Lacothrix use the immature seeds throughout the year. In Spironelo's (1991) study, palms were also used most frequently in.the dry season. Jessenia pulp was used in the dry season, which was also the season of fruit scarcity at that site, while immature seeds were used during the wet season. In Roraima, in the northern Brazilian Amazon, Astrocarvum aculeatum also fruits in the mid-dry season, when








fleshy fruit production is low (Silvius, pers. obs.), but the species is an unpredictable fruiter (Fragoso unpublished data).

In all other cases where palms are used during the

period of scarcity, it is the palm seeds that are involved, because they remain available on the ground past the time of low food availability (Terborgh 1986a for Cebus, Smythe et al. 1996 for agoutis, Chapter 4 for agoutis and white-lipped peccaries).

It is clear from Izawa's (1979) and Defler's (1979)

studies in Colombia that many primate species seek out the immature and still liquid endosperm during the time that it is abundant and also feed on the pulp of the palms. Struhsaker and Leland (1977) report on another specialized behavior of Cebus, in this case beating the nuts (endocarps) of Astrocarvum chambira against tree trunks, branches, or other nuts, apparently in an attempt to obtain the endocarp contents. I have observed this same behavior by Cebus nigrivitatus on Astrocaryum aculeatum in northern Brazil, but in this case the primates were smashing large preripe intact fruits rather than nuts, and the purpose of the knocking appeared to be to make the hard endosperm easier to remove, because fresh endocarps would be nearly impossible to open by banging (infestation by bruchid beetles, which makes the seeds easier to crack, occurs only on the ground). Norconk et al. (1998) show that among Neotropical primate seed predators, only Cebus has been reported to feed on palm









endocarp contents. It thus appears that Cebus spp. across their range rely heavily on palms throughout the year, and so to some extent may be considered dependent on palms.

This review suggests that a) the use of palms either as staples or as "keystones" sensu Terborgh may be unique to the genus Cebus and b) that it is difficult to determine whether the fruits and seeds are being used as staples or "keystones." This does not diminish their importance to mammalian fruit eaters, but it does challenge the idea that "keystone resources" sensu Terborgh structure tropical forest communities. The review also suggests that the importance of palms as low season resources is heavily context- and habitat-dependent, and will change with the time of fruiting of the palms and the relative availability of other preferred fruit.

Squirrels appear to be the only true palm specialists in the Neotropics, using palms in large numbers throughout their range. Although they occasionally use the pulp (Table 2-1), most frequently they feed on seeds collected from the ground. On BCI, Astrocarvum standlevanum makes up 21% of the diet of Sciurus garanatensis (Glanz et al. 1996; Table 2-2). At Los Tuxtlas, Mexico, two species of Sciurus destroy 20-60% of the fruits of the understory palm Astrocarvum mexicanum in low palm density areas and 75-80% in high palm density areas (Pifiero and Sarukhdn 1982). In Campinas state, Brazil, Svaarus romanzoffiana alone made up 49% of the diet of Sciurus ingrami (study period unspecified; Galetti et al.









1992). Emmons (1984), based on an accumulation of observations, estimates that about three palm species (Attalea and Astrocarvum groups) make up about 95% of the diet of the large squirrels S. spadiceus and S. igniventris. Because squirrels do not occur at high densities in Amazonia and therefore are not frequently seen (Eisenberg pers. comm.), palm seed use by squirrels is not well documented in the literature. Furthermore, there are no detailed studies of squirrel ecology in the Neotropics (Emmons, pers. comm.).

The potential effect of palm fruit availability on reproductive rates and presumably population sizes of squirrels is suggested by a comparison between BCI and a nearby (30 km away) mainland site (Glanz et al. 1996). On BCI, there are three peaks of squirrel reproduction each year, occurring throughout the time of staggered fruit production by Diptervx, Astrocaryum and Scheelea. Mortality for juveniles is highest for the last cohort. On the mainland site, where Diptervx is absent and Astrocaryum scarce, a single reproductive bout coincides with the period of Scheelea .fruit availability. There is no evidence of a similar matching at the supra-annual level.



Results and Discussion: Fruit-Vertebrate-Bruchid Interactions



The previous sections reviewed interactions between vertebrates and palm fruits. For a subset of Neotropical palms, however, these plant-vertebrate interactions intersect









with the use of palm seeds by the palm-specific bruchid beetles (Family Bruchidae, Order Coleoptera). Adult female beetles lay their eggs on the endocarp or mesocarp of fallen fruit, usually at night and usually but not always once the fruits are on the forest floor. After approximately 7 days, the first instar larvae hatch and tunnel through the endocarp to reach the endosperm. The larvae then feed on the endosperm and grow inside the protection of the palm endocarp for approximately two months, finally pupating and emerging as adults through an exit hole carved by the last larval instar. Although several palm families are parasitized in this way, the most intensive parasitism seems to occur among cocosoid palms in the subfamily Arecoideae, which includes the genera Attalea and Astrocarvum.

Bruchid-palm interactions are relevant to the above review of palm use by mammals because there is mounting evidence that in many cases mammals feed on bruchid larvae rather than on endosperm when they crack open palm endocarps. The dynamics of larva availability are likely to be different from those of endosperm availability. Only a few authors have recorded vertebrate frugivores or omnivores feeding on bruchid beetle larvae (Robinson 1986, Fragoso 1994), although several have suspected it occurs (Janzen 1971, Strushaker and Leland 1977, Wright 1990). A captive white-lipped peccary offered larvae-infested Attalea maria endocarps readily cracked all those that were not too thick and ingested the larvae (Fragoso 1994). Terborgh (1983) notes that









individuals of Cebus albifrons in Manu descend to the ground, pick among fallen Astrocarvum seeds for bruchid infested ones, crack them open, and pick out the remaining endosperm. Robinson (1986), on the other hand, noted that Cebus nigrivitattus in the Venezuelan Llanos picked up larvae infested Copernicia tectorum seeds from the ground and ate the larvae. Strushaker and Leland (1977) suspected that the Astrocarvum nuts cracked by Cebus apella in Colombia contained larvae, because most nuts collected on the ground were infested. On MaracA, agoutis, squirrels, and whitelipped peccaries fed on bruchid larvae from Attalea maria seeds in the wild (Chapter 4). Terborgh (1983) believes that the Cebus chose infested seeds over uninfested ones because they were easier to open than solid seeds; this is probably also a factor on Maraca, especially for white-lipped peccaries.

Humans in the Neotropics also use bruchid larvae from palm seeds both as food and as fish bait. They seek these larvae out from seeds that are usually too hard to eat, although they prefer the larger larvae obtained from woodboring beetles that lay their eggs in rotten palm trunks (for the Yanomami of Brazil and riberefios in Peru, Silvius, pers. obs.). Balbe (1988) reports that the Arawat6 of the eastern Brazilian Amazon kept bruchid infested palm seeds in their homes to "raise" the larvae. They then either eat the larvae or extract oil which they use to soften bow strings.








Research on the biology of palm bruchids falls into

three areas: 1) taxonomy, with a recent revision by Nilsson and Johnson (1993); 2) biogeography, with examination of host ranges and overlaps by Delobel et al. (1995) and Johnson et al. (1995); and 3) ecology of plant animal interactions. In the latter area, research has focused on two palm groups: Astrocarvum palms with Carvoborus beetles, and Attalea palms with Speciomerus and Pachvmerus beetles. The two palm genera differ significantly in the thickness and texture of their pulp, and the strength and size of their endocarps. Janzen (1971) initiated the research sequence with Attalea palms, focusing on the effect of bruchids on palm demography, but also indicated that there would probably be interactions between beetles and both pulp eaters and seed eaters. Later work followed up on the bruchid-palm tree aspect of the interaction (Wilson and Janzen 1972, Bradford and Smith 1977, Wright 1983, 1990). Three-way interactions were always commented on as peripheral phenomena, but never explicitly studied (with the exception of one short term experiment by Delgado et al. (1997) with the Astrocarvum system.

Palms from the genera Attalea, Scheelea, Maximiliana, and Orbianva have recently all been grouped in the genus Attalea by Henderson (1995) and I use the revised taxonomy. The taxonomy of bruchid beetles has also recently been revised (Nilsson and Johnson 1993). The name changes of species frequently encountered in the literature are summarized in Table 2-3.









Most long-term ecological work on the palm-bruchid

system has been carried out on Attalea butvraceaa (Janzen 1971, Wilson and Janzen 1972, Bradford and Smith 1977, Wright 1983, 1990) and Attalea maria (Fragoso 1994, 1997), with two additional short term studies on Astrocarvum chambira (Delgado et al. 1997) and Attalea phalerata (Quiroga 1997). Attalea butvracea and A. maria are very similar in reproductive biology and fruit form, with one potentially key difference: A. butvracea has endocarp fibers and sunken germination pores protected by these fibers, whereas A. maripa has only a woody endocarp. The fruits of A. butyracea are also slightly larger (4.5 to 8.5 cm long vs. 3.5 to 6.5 cm long; Henderson 1995).



Taxonomy and Taxon-Specific Interactions



Out of 1,500 species of bruchid beetles, only 20 species in four genera (three species of Cayroborus, six species of Carvobruchus, seven species of Pachvmerus, and four species of Speciomerus) are palm bruchids (tribe Pachymerini, subfamily Pachymerinae). All are endemic to the New World, and all but one are tropical (Nilsson and Johnson 1993, Delobel et al. 1995, Johnson et al. 1995). On a palm subfamily basis, bruchids are known to associate extensively (many species, many and substantiated records of high infestation rates) with tribes in the Arecoideae and Coryphoideae, and less extensively with a few species in the









Ceroxyloideae, Phytelephantoideae, and Calamoideae (Johnson et al. 1995). Within the Arecoideae, the tribe Cocoeae represents the most intensive association for all Neotropical palms (59% of 22 genera infested). The subtribes Attaleinae (primarily the genus Attalea) and Cocoae (primarily the genus Astrocarvum) have the greatest number of reports of bruchid infestation (Johnson et al. 1995). These are the groups that combine large seeds with thick endocarps.

The most specific association between a palm group and a beetle group is between Attalea palms and two species of beetles, Pachvmerus cardo and SDeciomerus aiganteum. Fifteen out of 24 record for the species Pachvmerus cardo are from Attalea palms, as are all of nine records for Seciomerus (Johnson et al. 1995). Both species are widespread, ranging from Central America to the southern Amazon basin.



Ecological Studies



The palm-bruchid system is ideal for study of

invertebrate-fruit interactions because eggs, entry holes, larvae, and exit holes are easily counted and endocarps are easily protected from vertebrate predation in the field by wire exclosures. All studies have used some variation of these methods. I evaluate the results of these studies in some detail because misinterpretation of data in some papers has led to a perpetuation of erroneus conclusions in subsequent publications (c.f. Henderson 1995).









Janzen (1971, 1972) described the interactions between Scheelea buryracea, Soeciomerus aioanteus, and Pachvmerus sp. in lowland moist forest habitat of Costa Rica. In this system, beetles lay their eggs on the endocarp only after the husk and pulp have been almost completely removed by rodents or by rotting. Egg laying intensity is low, with 2-5 eggs found per seed after prolonged exposure to oviposition. Janzen reports a development time of 3 to 5 months; that is, beetles begin to emerge from the seeds after this amount of time. Larval infestation rates were greater than 77% for seeds remaining in the original fruit pile by the parent tree. Fruit removal by frugivores and endocarp removal by granivores appears to be low, based on the size of the fruit piles described as remaining by the parent tree at the end of the fruiting season. A key observation by Janzen was that by the time rodents opened the endocarps under the parent tree, most of them contained full-grown larvae. If the endocarp contains multiple seeds, they are usually all killed by a single larva. Janzen also found high larval, pupal, and adult mortality by an "unknown microbial disease," which resulted in adult emergence rates much lower than infestation rates.

Work by Wilson and Janzen (1972) with the same system in Costa Rica again highlights a long development time for the larvae ("most are more than half grown" after 4.5 months in the field). Infestation rate was low (35% of experimental seeds). Bruchids had difficulty finding endocarps placed









individually rather than in clumps on the forest floor. The authors suggest that adults emerge from the endocarps after a rainfall cue at the start of the next palm fruiting season. Although they timed development and emergence for two beetles at 66 and 43 days respectively, the authors suggest that the larvae remain within the endocarps through the dry season following their laying season, and emerge only at the start of the next rainy (= fruiting) season. However, they also suggests that beetles remain in the endocarp in the adult form, so it is unclear which effect is predominant, or if the two terms, larvae and beetle, were inadvertently confused in the paper. In this study, only Soeciomerus was present, whereas in the previous study, trees infested by two species of beetles were included in the sample of infested seeds.

Bradford and Smith (1977) compared endocarps from a

small sample of Attalea burvracea trees in Panama and Costa Rica, both presumably exposed only to predation by Speciomerus aicranteus. They concluded that rodent predation was high on BCI and low in Costa Rica, whereas the opposite was true for bruchid predation. This is an erroneous conclusion, because they classified any rodent-opened endocarp as preyed on only by rodents, disregarding that fact that most endocarps opened by rodents probably had a larva in them, which the rodent would have removed. They do not take seriously Janzen's suggestion that rodents are encountering larvae more frequently than endosperm, stating instead that rodents "possibly" feed on the larvae. They also concluded








that seeds from multiple-seed endocarps had a higher probability of escaping predation than those from single seed endocarps. However, individual trees tend to produce a typical number of seeds per endocarp, and individual trees may also differ strongly in both the degree of bruchid infestation and rodent predation (Chapters 4 and 5). Since Bradford and Smith (1977) only sampled five trees, they could have had a strong tree effect in their study.

Based on differential gnawing patterns at the two sites, the same study also suggested that there is an evolutionarily based correlation between the number of seeds per endocarp and the way in which a single rodent species opens the endocarps. This is based on the assumption that the same rodent species were opening the endocarps at both sites. However, the gnawing patterns they describe could also be interpreted as being caused by different rodent species: on BCI the five trees they sampled appear to have been fed on only by squirrels, whereas in Costa Rica there was also predation by agoutis.

Wright (1983) continued work with the A. butvraceae-E. aiganteus system on BCI. He again reports a low ovipostion rate (1-5 eggs, with a rare extreme of 11 eggs after 48 hours of exposure). He found that there was high mortality of either the eggs or the first instar larvae, because only 47 out of 127 endocarps with eggs developed larvae, and many of those without larvae had entry holes that indicated a first instar larvae had at least initiated tunneling through the









endocarp but then failed to develop. He also found that females could distinguish between endocarps with no eggs and those with eggs, and preferred to lay on endocarps with no previous eggs.

In a later study at the same site, Wright (1990)

measured an interval of 5 days from ovipositon to hatching, and of 6 months from ovipostion to adult emergence. He notes that some adults, rather than larvae, remain inside the endocarp until the next wet season. The intensity of ovipositon, low to start with (0.91 eggs per endocarp), declined during the fruiting season, and stopped altogether 2 months before the end of the fruiting season. He concludes that this decline is due to a lack of recruitment into the beetle population during one fruiting season: beetles have a long development time and they emerge at the start of each fruiting season. Like Janzen, he notes that rodents often consume seeds that were first infested by bruchids, but still classifies these as rodent-destroyed, even though the bruchid would undoubtedly have killed the seed in the absence of rodents.

Working with the Attalea marica-Pachvmeu$ cardo system in northern Brazil, Fragoso (1994, 1997) found that seeds ingested by tapirs (as fruit) and subsequently defecated at upland latrines had a much higher chance of surviving bruchid infestation than seeds remaining at the parent tree. This survival translated to a higher seedling density at latrine sites than at parent trees or at random locations in the








forest. He also found that bruchids are able to penetrate thicker endocarps than rodents. Like Bradford and Smith (1977), he found that endocarps with multiple seeds were more likely to have at least one seed survive beetle or rodent predation than single-seeded endocarps. However, because single seeded endocarps are thicker than multiple seeded ones, they are more likely to survive attack by white-lipped peccaries, which, unlike rodents and bruchids, cannot penetrate the thickest endocarps. Therefore, white-lipped peccaries could be maintaining selection for thick, single seeded endocarps, which appears to be opposite to the selection pressure for multiple seeded endocarps maintained by bruchids



Astrocarvum-Bruchid System



As for the Attalea-bruchid system, much of the work with Astrocarvum has focused on the role of vertebrate dispersers in protecting seeds from bruchids by burial (Smythe 1989) and on the role of bruchids in distance and density dependent seed predation (Terborgh et al. 1993). Terborgh et al. (1993) documented greater infestation rates near the parent tree than 25 m away for an unknown beetle on Astrocaryum macrocalyx in Peru, when rodents and other vertebrates were excluded.

Delgado et al. (1995) document bruchid predation rates on fruits not affected by vertebrate consumers and find more









rapid oviposition on fruits artificially cleaned of pulp and husk. They report multiple larvae per seed, and rejection of some fruits altogether by ovipositing females. Smythe (1989) found high mortality from bruchids for peeled, unburied Astrocarvum standlevanum seeds on BCI, a result to be expected in light of the protection offered by the husk and pulp (Janzen 1971, Wright 1983, Delgado et al. 1995, Chapter 3).



Bruchid-Invertebrate Interactions



Interactions between bruchids and invertebrate parasites are more rarely recorded than those between vertebrates and bruchids, although parasites can cause significant mortality for bruchid larvae and have seasonal patterns that can affect the evolution of both bruchid beetles and palm phenology (Fragoso and Silvius unpublished data). Quicke and Delobel (1995) described a parasitic wasp species that attacks the late instar larve and pupae of Cavoborus serripes on several species of Astrocarvum palms. Parasitism rates were as high as 20%. The same bruchid species has its eggs parasitized by a trichogrammatid wasp (Delobel et al. 1995), as does Pachymerus card on Attalea palms, where egg parasitism rates were as high as 92% (Delobel et al. 1995). Janzen (1971) reports an unknown parasite on Soeciomerus qiganteus eggs in Costa Rica, as well as an unknown agent of mortality, which he refers to as a pathogen, that kills larvae and pupae.








Given the widespread nature of the parasitism, the high mortality rates recorded, and the probable specificity of the relationship, invertebrate predators and/or parasites on bruchids are likely to be as important in bruchid beetle ecology and evolution as vertebrates. For example, bruchid beetles might chose seeds that offer escape from parasites rather than from rodents.



Summary and Conclusions



The issue of the relationship between fruit availability and frugivore species richness and abundance is undeniably a large one that needs to be tackled in bits and pieces, or by many researchers working together. However, by identifying one subset of frugivores and fruits that show a more or less tight relationship, it may be possible to develop methodologies and theoretical approaches that can then be applied to the larger community of fruits and frugivores.

The review presented here suggests that several

Neotropical forest species, namely white-lipped peccaries, tapirs, Cebus monkeys, agoutis, and most importantly squirrels, depend to varying degrees on palm fruit availability. Squirrels, and to a lesser degree capuchin monkeys, are specialized on palm use. The use of palms as a season of scarcity resource (Terborgh 1986a, b) may be restricted to members of the genus Cebus. By Terborgh's own








definition, this would then no longer constitute a case of keystone resource use.

In palm species that have ecological and evolutionary interactions with bruchid beetles, it is unclear and unstudied whether frugivores and granivores are feeding on fruit matter or insect matter, and what the effect of these interactions is on palm and frugivore ecology. The Attaleabruchid system is the best understood. Several studies indicate that beetles cause significant mortality for palms, and that both vertebrate predators and invertebrate parasites cause significant mortality for beetles. There is some evidence that seeds dispersed long distances from the parent tree experience reduced beetle predation.









Table 2-1: Results of literature review of palm use by Neotropical mammals. Latin binomials for palms and mammals are those used by the original authors; note that all palms in the genera Attalea, Maximiliana, Orbignya and Scheelea are currently placed in the genus Attalea. imm = immature, mat = mature, germ = germinating, fruit = author did not distinguish between part eaten, endocarp contents = either endosperm or bruchid larvae or both.


Palm Species Mammal Species Part Site Ref
Mauritia flexuosa Mazama americana pulp Maraca, 17 Brazil
Tayassu pecari preripe 19 seed,
germ,
seed
Tayassu tajacu germ. 19 seed
Tapirus pulp 18 terrestris
Ateles belzebuth mat Maraca, 33 pulp Brazil
Mazama americana N.E. 6 Peru
Mazama 6 qgouazoubira
Tayassu tajacu 6 Tayassu pecari 6 Tapirus 6 terrestris
Nectomys embryo Mato 40 squamipes from Grosso, seed, Brazil
pulp,
discards
endo
sperm
Oxymycteris pulp, 40 roberti caches seeds
Oligorysomys sp. pulp, 40 seed
coat
Didelphis pulp 40 albiventris I









Table 2-1, continued


Palm Species Mammal Species Part Site Ref
Tapirus pulp 40 terrestris
Agouti paca pulp 40 seed
(holds
with
fore
paws)
Chrysochyon pulp 40 brachyurus
Dasyprocta N.E. 1 fuliginosa. Peru Small rodents 1
Agouti paca 1 Eira barbara 1 Tapirus 1 terrestris
Mazama americana 1 marsupials 1 edentates 1 Marmosops N.E. 52 noctivagus Peru Marmosops sp. 52 Nectomys 52 squamipes
Oecomys 52 trinitatis
Oryzomys sp. 52 Oryzomys 52 yunganus
Proechimys sp. 52
Proechimys 52 steereei
Attalea maripa Sciurus imm. Rio 38 spadiceus seed Urucu, mat. Brazil
pulp
Ateles belzebuth mature 11 pulp Bolivar,
Vene
_zuela
Cebus apella mat pulp Amazonas 47 Brazil
Odocoileus pulp Maraca, 17 virginianus ..Brazil









Table 2-1, continued


Palm Species Mammal Species Part Site Ref Tavassu taiacu pulp 43 Tayassu pecari pulp 17, endocarp 42 contents
Ateles belzebuth mat pulp Maraca, 33 Brazil
Alouatta pulp 43 seniculus
Saimiri sciurus pulp 43 Sciurus seed 42 igniventris
Dasyprocta pulp 42, aqouti seed 43 Dasyprocta French 28 punctata Guiana Sciurus aestuans 28 Ateles belzebuth fruit Bolivar, 11 Vene
zuela
Attalea regia Cebus albifrons pulp Eastern 14 seed Colombia
Echimys armatus pulp French 12 Guiana
Philander pulp 12
opossum
Didelphis pulp 12 marusupialis
Sciurus aestuans seed 12 Dasyprocta seed 12 agouti
...Agouti paca pulp 12 Small rodents pulp 12 Attalea phalerata Tapirus pulp Beni, 27 terrestris Bolivia Scheelea sp. Saimiri sciurus seed Cocha 49 Cashu,
Peru
........Cebus apella seed 49 Cebus albifrons seed 49 Sciurus seed 49 spadiceus
Scheelea sp. Sciurus seed Western 15
igniventris Amazonia
Scheelea sp. Ateles paniscus Cocha 48 chamek Cashu, Peru









Table 2-1, continued


Palm Species Mammal Species Part Site Ref Scheelea sp. Tapirus S. E. 6 terrestris Peru Scheelea rostrata Sciurus pulp Costa 7 variegatoides seed Rica Agouti paca pulp 7 seed
Dasyprocta pulp 7 punctata seed Scheelea zonensis Nasua narica pulp BCI 55 Cebus capucinus pulp 34 Agouti paca pulp 7 Dasyprocta seed 45 punctata pulp Proechimys seed 7 semispinosus pulp Sciurus seed 21,
_granatensis pulp 7 Sciurus gerradi 55 Scheelea Cebus apella pulp Colombia 26 attaloides
Orbignya martiana Agouti paca pulp Maranhao 3 Brazil
Dasyprocta pulp 3 punctata
Coendu pulp 3 prehensilis
Coendu sp. pulp 3 Proechimys pulp 3 longicaudata
Mesomys hispidus pulp 3 Tayassu pecari pulp 3 Tayassu tajacu pulp 3 Aotus imm. 3 trikingatus seed Cebus apella imm. 3 seed
Orbignya Chiropotes sp. pulp Pard, 5 phalerata Brazil Orbignya cohune Cebus capucinus seed Honduras 9 Astrocaryum Sciurus imm seed Rio 36, murumuru spadiceus Urucu, 37 Brazil
Cebus apella pulp Cocha 13 Cashu,
Peru
Sciurus spp. seed 13









Table 2-1, continued

Palm Species Mammal Species Part Site Ref
Dasyprocta seed 13 variegata
Proechimys spp. seed 13 Orvzomis spp. seed 13 Tavassu pecari seed 13 Astrocaryum sp. Mazama americana pulp N.E. 6 Peru
Tayassu tajacu seed 6 Tayassu pecari seed 6 Tapirus 6 terrestris
Astrocaryum Ateles belzebuth imm pulp MaracA, 33 aculeatum Brazil Tayassu pecari seed 17 Tayassu tajacu pulp 43 Tapirus pulp 43 terrestris
Cebus imm pulp 43 nigrivitatus
Dayprocta agouti pulp 43 seed
Scirus fruit 43 igniventris
Astrocaryum sp. Cebus capucinus seed Cocha 49 (=murumuru?) Cashu, Peru
Cebus albifrons seed 49 Sciurus seed 49 spadiceus
Tayassu sp. seed 49 Astrocaryum Cebus apella imm seed La 26 chambira Macarena Colombia
Ateles belzebuth imm seed 26
Lagothrix imm seed 26 lagothrica
............Cebus apella "nut" Colombia 46
Astrocaryum Sciurus seed Mexico 39 mexicanum aureogaster Sciurus deppei seed 39 Heteromys seed Belize 8 desmarestianus
Astrocaryum Nasua narica pulp BCI 55 standleyanum imm
1- _ fruit









Table 2-1, continued


Palm Species Mammal Species Part Site Ref
Agouti paca pulp 44, 45
Cebus capucinus pulp 34 Ateles geoffroyi pulp 24 Dasyprocta pulp 45 punctata
Sciurus seed 21 granatensis
Didelphis pulp 44 marsupialis
Philander pulp 44
opossum
Caluromys pulp 44 derbianus
Tayassu pecari seed 44 Tayassu tajacu seed 44 Heteromys seed 44 desmarestianus
Proechimys seed 44, semispinosus 2, 25
Tapirus bairdii 55 Astrocaryum Philander pulp French 12 vulgare opossum Guiana
Caluromys pulp 12 philander
Dasyprocta seed 12 punctata
Coendu pulp 12 prehensilis
Agouti paca pulp 12 small rodents pulp 12 Sciurus aestuans seed 12
Astrocaryum Sciurus aestuans French 28 paramaca Guiana Proechimys 28 cuvieri
Astrocaryum Sciurus aestuans French 28 sciophilum Guiana Myoprocta exilis 28 Astrocaryum sp. Proechimys Cocha 15 semispinosus Cashu, Peru
Astrocaryum sp Sciurus seed Western 15
.........spadiceus Amazonia








Table 2-1, continued


Palm Species Mammal Species Part Site Ref
Sciurus seed 15 iqniventris
Chamaedorea Heteromys seed Mexico 35 tepejilote desmarestianus
Agouti paca 55 Socratea Ateles belzebuth Colombia 31 exorrhiza
Ateles belzebut fruit Bolivar, 11 Vene
zuela
small rodents seed Maraca, 43 Brazil
Tayassu pecari seed Maraca, 43 Brazil
Artibeus fruit Amazonia 55, jamaicensis French 12 Guiana,
Trinidad
Artibeus Amazonia 55, lituratus French 12 Guiana
Socratea Dasyprocta BCI 45 durissima punctata Sciurus 21
_granatensis
Cebus capucinus pulp 34 Ateles geoffroyi mat pulp 24
Alouatta pulp 32 palliata
Proechymis 55 semispinosus
Myoprocta exilis French 12 Guiana
Heteromys seed Costa 66 desmarestianus Rica Socratea sp. Tayassu tajacu Cocha 29 Cashu
Peru
Tayassu pecari 29 Socratea sp. Ateles paniscus pulp Cocha 48 chamek Cashu Peru
Socratea elegans Cebus apella imm. Colombia 26 seed
Ateles belzebuth imm. Colombia 26 seed









Table 2-1, continued


Palm Species Mammal Species Part Site Ref
Iriartea Ateles belzebuth Colombia 31 ventricosa
Tayassu pecari seed Cocha 30 ?pulp Cashu,
Peru
Cebus apella 49 Saguinus 49 imperator
Saguinus 49 fuscicolis
Callicebus 49 moloch
Ateles paniscus pulp 48 chamek seed Proechimys seed 15 semispinosus
Iriartea sp. Mazama americana seed N.E. 6 pulp Peru
Mazama seed 6 gouazoubira pulp Tayassu tajacu fruit 6 Tayssu pecari fruit 6 Wettinia Saguinus sp. 55 maynensis
Euterpe sp. Ateles belzebuth Colombia 31 Euterpe sp. Ateles paniscus pulp Cocha 48 Cashu,
Peru
Euterpe Cebus apella mat pulp Colombia 26 precatoria
Ateles belzebuth fruit Maracd, 43 Brazil
Oenocarus bacaba Ateles belzebuth mat.frui MaracA, 33 t Brazil
Oenocarpus mapora Cebus apella imm Colombia 26 seed,
mat pulp
Oenocarpus Sciurus BCI 23 panamus granatensis Cebus capucinus 34 Oenocarpus sp. Ateles belzebuth Colombia 31 Jessenia bataua Cebus apella imm seed Rio 37 Urucu,
SBrazil
... Cebus albifrons 37









Table 2-1, continued


Palm Species Mammal Species Part Site Ref
Pithecia imm 37 albicans seeds, mat pulp
Lagothrix 37 lagrothrica cana
Pithecia sp. mat pulp Para, 5 Brazil
Tayassu tajacu pulp N.E. 6 Peru
Tayassu pecari pulp 6 Mazama americana pulp 6 Tapirus pulp 6 terrestris
Pithecia 55 monachus
Cebus albifrons pulp Eastern 14 seed Colombia
Cebus apella imm seed Colombia 26 mat seed
mat pulp
Ateles belzebuth imm seed 26 mat seed
mat pulp
Ateles belzebuth fruit Bolivar, 11 Venezuela
Cebus Amazonia 28 nigrivitatus
Sciurus aestuans French 28 guiana
Roystonea Artibeus fruit Trinidad 55 oleracea lituratus Syiagrus Cebus albifrons pulp Eastern 14 orinocensis seed Colombia Syagrus Cebus apella mat pulp Campinas 20 romanzonffiana Brazil Alouatta fusca mat pulp 20
Sciurus ingrami imm seed 20 old seed
Acrocomia sp. Dasyprocta sp. Central 55 Brazil
Cerdocyon thous 55









Table 2-1, continued


Palm Species Mammal Species Part Site Ref
Acrocomia Didelphis dry 55 aculeata albiventris upland habitat,
Brazil
Nectomys 55 squam ipes
Cebus apella 55 Agouti paca 55 Euphractes 55 sexcintus
Dasyprocta sp. 55 Acrocomia Sigmodon pulp Costa 4 vinifera hispidus Rica Tayassu tajacu pulp 4 seed
Bactris barronis Sciurus BCI 21 granatens2s
Bactris sp. Mazama americana N.E. 6 Peru
Bactris sp. Artibeus fruit Trinidad 22 jamaicensis
triniatus
Altibeus fruit 22 lituratus
palmarum
Bactris sp. Ceus albifrons pulp Eastern 14 seed Colombia
Bactris gassipaes Heteromys pulp Costa 53 desmarestianus Rica
Hoplomys pulp 53 gymnurus
Dasyprocta pulp 53 punctataa
Bactris sp. Saguinus mat Rio 36 fuscicollis fruit Urucu, Brazil
Bactris sp. Cebus capucinus fruit Honduras 9 Paurotis wrightii Cebus capucinus fruit Honduras 9 Desmoncus Cebus capucinus pulp BCI 34
isthmius
Welfia georgii Potos flavus Costa 54 Rica
... Sciurus sp. 54 ......Ceus capucinus 54 Dasyprocta 54 punctata








Table 2-1, continued


Palm Species Mammal Species Part Site Ref
Heteromys 54 desmarestianus
Haplomys 54 gymnurus
Proechymis 54 semispinous
Aqouti paca? 54 Tayassu pecari ? 54 Phytlephas sp. Dasyorocta sp. 55 Ammandra sp. Dasvprocta sp. 55 Washingtonia Canis lastrans pulp Sonora 10 filifera
Urocyon pulp 10 cineroargentatus
Copernicia Cebus imm Llanos, 41 tectorum nigrivitatus fruit Venegerm zuela
seeds
endocarp
contents
Areca lutescens Phyllostomatidae 22 Coccothrinax sp. Phyllostomatidae 22

1-Allen 1997; 2-Adler 1995; 3-Anderson 1983; 4-Baker 1983; 5Bobadilla, pers. comm.; 6-Bodmer 1989; 7-Bradford and Smith 1977; 8-Brewer and Rejmanek 1998; 9-Buckley 1983; 10-Bullock 1980; 11-Castellanos and Chanin 1996; 12-Charles Dominique et al. 1981; 13-Cintra and Horna 1997; 14-Defler 1979; 15-Emmons 1982; 16-Fleming 1974; 17-Fragoso, pers. comm; 18-Fragoso 1997; 19-Fragoso 1994; 20-Galeti et al. 1992; 21-Glanz et al. 1996; 22-Gardner 1977; 23-Heany and Thorington 1978 24-Hladik and Hladik 1969; 25-Hoch and Adler 1997; 26-Izawa 1979; 27-Quiroga 1997; 28-Kahn and de Granville 1992; 29Kiltie 1981; 30-Kiltie and Terborgh 1983; 31-Klein 1972; 32Milton 1980; 33-Nudes 1992; 34-Oppenheimer 1996; 35-Oyama 1991; 36-Peres 1993b; 37-Peres 1994a; 38-Peres 1994b; 39Pifiero et al. 1984; 40-Prada 1994; 41-Robinson 1986; 42Silvius, this study; 43-Silvius pers. obs.; 44-Smythe 1989; 45-Smythe et al. 1996; 46-Spironelo 1991; 47-Struhsaker and Leland 1977; 48-Symington 1987; 49-Terborgh 1983; 50-Terborgh 1986a; 51-Terborgh 1986b; 52-Valqui unpublished data; 53Vandermeer 1983; 54-Vandermeer et al. 1979; 55-Zona and Henderson 1989.









Table 2-2. Palm-eating mammals for which the percent composition of the diet has been estimated.


Mammal Palm % of total Highest Reference Species Species or annual monthly % diet
Cebus apella Jessenia 17.6 53 Spironelo
(Oenocarpus) 1991
bataua and 1-year study
Attalea maripa
primarily,
and some
Oenocarpus
bacaba,
Astrocaryum
mumbaca, Mauri tia flexuosa,
and Syagrus inajai (data
combined)
Jessenia 24.5
bataua
Attalea 28.7
maripa
Astrocaryum 3.5
mumbaca
Ateles Attalea 1.23 10.1 Nufres 1992 belzebuth maripa 2-year study belzebuth
Astrocaryum 0.18
aculeatum
Oenocarpus 0.09
bacaba
Mauritia 0.09
flexuosa
Mazama Astrocaryum 3.4 (5) Bodmer americana sp. 1989*, 15 month study Bactris sp. 9.1
Euterpe sp. 43.6 (1) Iriartea sp. 34.4 (2)
Jessenia 1.1
(Oenocarpus)
bataua
Mauritia 9.1
flexuosa
Mazama Euterpe sp. 48.0 (1) gouazoubira









Table 2-2,


continued


Mammal Palm % of total Highest Reference Species Species or annual monthly % diet
Iriartea sp. 36.0 (2)
Mauritia 8.0
flexuosa
Tayassu Astrocaryum 45.3 (3) tajacu sp.
Euterpe sp. 7.5 Iriartea sp. 27.7
Jessenia 46.2 (2)
(Oenocarpus)
bataua
Mauritia 11.7
flexuosa
Tayassu Astrocaryum 47.7 pecari sp.
Euterpe sp. 15.9
Iriartea sp. 75.0 (1) Jessenia 59.0 (2)
(Oenocarpus)
bataua
Mauritia 45.0
flexuosa
Tapirus Astrocaryum 5.3 terrestris sp.
Jessenia 23.7 (2)
(Oenocarpus)
bataua
Mauritia 76.3 (1)
flexuosa
Attalea sp. 13.2
Dasyprocta Astrocaryum 22.7 Smythe et punctata standleyanum al. 1996, 1year study, 326 feeding observations
Attalea 11.04
butyracea
Oenocarpus 0.31
panamus
Socratea 0.31
durissima









Table 2-2, continued


Mammal Palm % of total Highest Reference Species Species or annual monthly % diet
Agouti paca Astrocaryum 6.7 Smythe et
standleyanum al. 1986, 1year study, 15 feeding observations
Sciurus Astrocaryum 20.92 Glanz et al. granatensis standleyanum 1986, 2-year study
Bactris 0.05
barronis
Bactris 0.05
ma or
Oenocarpus 0.68
panamus
Attalea 24.62
butyracea
Socratea 0.05
durissima
Cebus Attalea rank 1 14.3-58.2 Buckley capucinus cohune May-October 1983, 17 month study
Sciurus Syagrus 49.0 Galetti et ingrami romanzof- al. 1992
fiana
Cebus apella Syagrus 2.16
romanzoffiana
Alouatta Syagrus 0.25 fusca romanzoffiana
Ateles Euterpe sp. 1.0 61.0 Klein 1972, belzebuth 1-year study
Socratea 1+ 1+
exorrhiza
Iriartea 2.0 23.0
ventricosa
Oenocarpus 1+ 22.0
sp.
Saguinus Bactris sp. 0.03 Peres 1993a, fuscicolis 20-month study
Pithecia Mauritia 2.3 Peres 1993b, albicans flexuosa 20-month I study








Table 2-2, continued


Mammal Palm % of total Highest Reference Species Species or annual monthly % diet
Ateles Palmae 0.9 Roosmalen paniscus 1980; 2-year study
Euterpe 0.3
oleracea
Attalea 0.3
regia
(?=maripa)
Oenocarpus 0.2
bacaba
Euterpe 0.05
oleracea
Sciurus Palmae 98.0 % Emmons 1984 spadiceus,
Sciurus
igniventris
Sciurus Astrocaryum 9.0 22 Heany and granatensis standleyanum Thorington 1978; 4
months,
rainy and
_dry seasons Oenocarpus 1 7
panamus
Attalea 27 86
butyracea
Cebus apella Palmae 64 from May Terborgh
(5 sp) to July 1983, 12month study
Cebus Palmae 56 from May albifrons (5 sp) to July Saimiri Palmae 15 from May sciurus (3 sp) to July Saguinus Palmae < 1 from May imperator (1 sp) to July Saguinus Palmae < 1 from May fuscicolis (1 sp) to July

* In this study, % is the percent of stomachs in which the food item occurred, while in all the other studies, it is the percent of feeding observations, either instantaneous or over a defined sampling period, in which that food item was eaten. For clarification, in Bodmer's study I give in parentheses, next to the percentage, the rank in the diet based on the percent of total dry weight of all stomach contents combined, if the palm ranks between first, second or third in importance in the diet.










Table 2-3. Revisions to palm and bruchid taxa referred to in the text.

FOMER NAME CURRENT NAME REFERENCES

Palms
Maximiliana marica Attalea maripa Fragoso 1994, 1997 Scheelea zonensis Attalea butyracea Bradford and Smith 1977, Wright 1983, Wright 1990, Forget 1994
Scheelea rostrata Attalea butyracea Janzen 1971, Janzen and Wilson 1972, Bradford and Smith 1977
Orbignya phalerata Attalea speciosa Anderson 1983, Anderson et al. 1991
Bruchids
Caryobruchus Speciomerus Janzen 1971, Janzen bucksii qiqanteus and Wilson 1972 Unidentified Pachymerus cardo Fragoso 1994














CHAPTER 3
PULP HANDLING BY MAMMALIAN FRUIT-EATERS AFFECTS PALM SEED
PREDATION BY BRUCHID BEETLES



Introduction



Studies of frugivory systems (a fleshy-fruited plant population and its consumer set, Howe 1993) to date have focused primarily on the interactions between plants and vertebrate fruit-eaters (Estrada and Fleming 1986, Fleming and Estrada 1993). This bias has resulted in a dominance of vertebrate-based evolutionary scenarios for both fruit trait evolution and plant demography. Thus fruit traits are interpreted as evolving in response to the need to attract pulp eaters that disperse seeds to safe sites. A limited group or researchers, led by the early theoretical work of Herrera (1982) and Janzen (1977), has focused on the interactions.between fruits and invertebrates, including both seed eaters and pulp eaters. In this study I focus on interactions between pulp, insects, and vertebrates, leaving discussion of interactions involving seeds for the next chapter.

Several studies show that invertebrate pulp consumers or the fungal spores they introduce into pulp can degrade pulp and make it unattractive to vertebrate pulp eaters that might









otherwise disperse the seeds (Jordano 1987, Borowicz 1988, Krishik et al. 1989, Buchloz and Levey 1990, Sallabanks and Courtney 1992, Cipollini and Stiles 1993). In other cases, invertebrates may enhance pulp quality for the vertebrate disperser/pulp eaters by providing protein, lipids, and other nutrients (Redford et al. 1984, Piper 1986, Valburg 1992a,b). The opposite effect, alteration of fruit quality by vertebrate pulp consumers for invertebrate fruit consumers, has seldom been examined, although it has been noted that some insect species preferentially lay their eggs into damaged areas of the pulp (Sallabanks and Courtney 1992). In such a situation, presumed dispersers who scar the pulp could have a negative effect on seed survival by increasing the likelihood of infestation by invertebrate seed eaters.

Due to variation in body size and behavior, different species of fruit-eaters handle pulp in distinct ways, and remove varying amounts of pulp from the fruits they handle. Differences in the amount and pattern of pulp removal could affect seed predation rates by insects and ultimately plant demography. In the Neotropics, a diverse array of consumers eat the fruits of arborescent palms: primates, carnivores, rodents, ungulates, and marsupials among the mammals, psittacines, falconiformes, and icterids among the birds, and tortoises among the reptiles (Chapter 2). Several Neotropical palm species are also parasitized by about 20 species of bruchid beetles (Chapter 2). One of the most specific and widespread relationships is that between palms








in the genus Attalea (including Maximiliana, Orbvania, and Scheelea; Henderson 1995) and the beetles Pachymerus cardo and Soeciomerus aioanteus. Interactions among these taxa have been documented from Costa Rica to Bolivia (Fragoso 1997b, Quiroga 1997, Johnson et al. 1995, Janzen 1971, Wilson and Janzen 1972).

In the palm-bruchid system, interactions between

bruchids, vertebrate seed eaters, and nut walls (endocarps) have been examined in some detail (Fragoso 1994, Wright 1983, 1991, Bradford and Smith 1977). The fruit husk (exocarp) and the pulp (mesocarp), however, are the first barriers bruchids encounter when attacking a seed, long before they reach the endocarp. Both affect the bruchids' ability to reach a seed. Janzen and Wright (Janzen 1970, Wright 1983, 1990) noted that bruchids do not lay eggs on endocarps of Attalea butyracea in Central America until the husk and most of the pulp has been removed by vertebrates or fungus. The same was true for an Astrocaryum palm in Peru whose pulp was experimentally removed (Delgado et al. 1997). This indicates that pulp itself acts as a deterrent to oviposition.

Several researchers have suggested that pulp chemicals in plant families other than the Palmae may deter feeding by insects (reviewed in Cipollini and Levey 1997), but there is little consideration in the literature of pulp as a physical deterrent for seed-eating insects. The physical properties of pulp, however, including consistency and presence of mucilage or lipids, could impede pulp mining by insects. Extremely








thick pulp could also increase mortality of larvae mining their way into the seeds. Traits associated with the attraction of seed dispersing mammals could thus also function as repellents to insects. Because pulp has evolved many times in the Angiosperm lineage, in some of these cases pulp could be evolutionarily molded by selection by insect seed predators as well as by dispersers. Given the extremely high rates of mortality caused by Coleoptera (especially Bruchidae and Curculionidae, Janzen 1982a,b, 1983, Herrera 1989, Sallabanks and Courtney 1992) in the tropics, selection for deterrence of beetle larvae should be strong.

On Maracd Island Ecological Reserve, in the northern Brazilian Amazon, high rates of bruchid predation by Pachymerus cardo have been documented for the common arborescent palm Attalea maripa (= Maximiliana maria) (Fragoso 1994, 1997b). The present study experimentally examines the effect of differential pulp handling by vertebrate consumers on a) pulp-dependent egg-laying choices by bruchid beetles that parasitize the seeds, and b) subsequent survival rate of seeds that received differing numbers of eggs in response to different pulp characteristics.









Study Area and Species



Study Area



Maracd Island Ecological Reserve is a 1,100 km2 riverine island located on the forest side of the forest-savanna boundary in the northern Brazilian state of Roraima. The study site terrain is flat and low (elevation ranges from 110-160 meters above sea level.) The sandy soils are derived from quartz-biotite schists, quartz-feldspar gneisses, and tonalitic granites (Thompson et al. 1992).

Rainfall records on the eastern end of the island show variation of between 1750 and 2,300 mm annually; during El Nifio Southern Oscillation events, rainfall decreases markedly across the state. The wet season is between May and September, the dry season between October and April, with rains starting either in early or late April. The heaviest individual rains tend to fall in the dry season.



Study Species



Attalea maria is a canopy to subcanopy arborescent palm very common on the eastern end of Maracd island (Milliken and Ratter 1989). In the terra firme forest where this study took place, it occurs in dense aggregations (mean 20.2 trees per quarter hectare, range 12-32 trees; Fragoso 1997b). It reaches its highest densities and reproductive output at








ecotones between the forest and either $avannas, wetlands or rivers (Chapter 4), and occurs at low densities or singly in the forest interior. Individual infloresences can be bisexual or male, and between one and three fruit bunches develop simultaneously or over the course of a six month fruiting season (April - September). Each fruit bunch can have between a few hundred and more than 3,000 fruits (two average-sized pre-ripe fruit bunches contained 1,183 and 2,455 fruits, respectively). Although a large proportion of reproductive age trees produce flowers, a much lower proportion develops fruits each year (pers. obs., Fragoso unpublished data).

Individual fruits range from 3 to 6 cm in length, and are roughly egg shaped, with a narrow tip and broad base. The husk is about one mm thick, dry and resistant to tearing. The mesocarp is one centimeter thick, dense and slippery, with fibers near the endocarp but not on the outside. It is slightly sweet, soapy and astringent to the human taste, and clings to the mouth parts. Color ranges from pale yellow to bright orange among different trees, but is consistent within one tree. Endocarp thickness ranges from 1.5 to 6.5 mm. Each endocarp can contain one, two, or three seeds, but one seed is most common (Fragoso 1994).

Fruits drop intact to the ground when they are ripe, but many are knocked down (prior to abscission) by primates and bird, often with the pulp only partly removed. The pulp is eaten by brocket deer (Mazama spp.), white-tailed deer









(Odocoileus virainianus), collared and white-lipped peccaries (Tavassu tajacu and T. pecari), agoutis (Dasvorocta aqouti), pacas (Aaouti paca), tapir (Tapirus terrestris), capuchin (Cebus niqrivitatus, C. apella), squirrel (Saimiri sciureus), howler (Allouata seniculus), and spider monkeys (Ateles belzebuth), Amazon parrots (Amazona spp.), macaws (Ara spp.), fruit caracaras (Daptrius sp.), and oropendulas (Gymnostinoos sp.). Among the invertebrates, members of the Coleoptera, Diptera and Orthoptera feed on the new pulp, and Dipterans and Dermapterans lay eggs in the pulp. The mature endosperm is eaten by bruchid beetles, squirrels (Sciurus igniventris), agoutis (Dasyorocta aaouti), unidentified small rodent(s), and white-lipped peccaries (Tavassu Decari). The immature endosperm is occasionally eaten by an unidentified arboreal consumer that is able to penetrate the immature endocarp.

The bruchid beetle Pachvmerus cardo occurs from Costa

Rica to Bolivia. It most frequently parasitizes seeds of the genus Attalea, but has also been found on Acrocomia, Aichanes, Areca, Bactris, CoDernicia, Svagrus, and Elaeis (Johnson et al. 1995). Like most bruchid beetles, it is nocturnal. Its adult biology is unstudied, and it is presumed to feed on pollen and nectar, though not necessarily of the same palm parasitized by its larvae.

On Maraca, adult beetles lay eggs on the pulp or

endocarp of the fruits once they have fallen to the ground (though at nearby open pasture sites they sometimes parasitize intact fruits on the tree; Fragoso and Silvius








unpublished data). After seven to 10 days the first instar larvae emerge from the eggs and drill through the endocarp into the seed. Development rates are variable: larvae feed on the endosperm and remain within the endocarp for a period of between two to nine months, and possibly longer, before pupating. Adults emerge from the endocarp at all times of the year. Both seed size and moisture availability during development appear to influence the size of the larvae and of the emerging beetle.

Although many first instar larvae enter the endocarp, only one larvae develops per endosperm. In the case of multiple-seeded fruits, two or three larvae can develop in an endocarp, but usually a single larva eats all of the seeds.

Bruchid eggs in this study were parasitized by an

unidentified insect, and were also attacked by small ants. Late instar bruchid larvae are parasitized by a wasp (Hymenoptera: Braconidae) and by another, unidentified insect.

Beetles in this study were identified by Dr. J. Kingsolver of the Division of Plant Industry, Florida Agriculture and Consumer Services. Only P. card was found during this study, but a few individuals of Seciomerus anteus were found in Attalea seeds at a forest site 10 kilometers away.








Methods



One 2-part experiment with 1,400 fruits and one single experiment with 100 fruits were carried out from June to December 1996.


Part I: Oviposition Preferences on Fruits with Different
Amounts of Pulp



The first part of the experiment tested the hypothesis that the amount of pulp remaining on an endocarp affects the quality of oviposition sites on the endocarp. Female bruchids should lay eggs preferentially on the higher quality "fruit types" and this preference can be measured by counting the number of eggs laid.

Of several conditions in which fruits were found under the parent tree, five "types" were chosen as the most representative and distinct for this experiment. These are described below in order of increasing pulp removal, along with the method by which they were procured for the experiment.



1) Intact husk fruits. Fruits that fell to the ground naturally or were dropped by primates without being opened. For the experiment, these were easily obtained by collecting intact ripe fruits from the ground or tree.








2) Intact pulp fruits. Fruits whose husk was removed by primates, but either rejected or accidentally dropped before pulp removal. Capuchin monkeys often removed only a thin layer of pulp along with the husk, and fruits eaten in this way resembled intact pulp fruits. Howler monkeys also removed only thin layers of pulp, but tended to leave narrow tooth marks in the pulp, whereas capuchins left wider, shallower scrapes. Spider monkeys, which swallow the fruits after squeezing them from the husk, dropped many intact, husk-free fruits. For the experiment, Intact Pulp fruits were obtained by peeling the husk away from intact fruits, taking care not to gash the pulp.



3) Gashed fruits. Squirrel monkeys, macaws, and several ungulate species removed the pulp in deep gashes that exposed the endocarp but left strips of endocarp surrounded by a thick bed of pulp. For the experiments, I artificially gashed intact pulp fruits by biting at the pulp and tearing it away in four locations per fruit.



4) Rodent fruits. Agoutis, and presumably pacas, removed the pulp in a corncob pattern, squeezing most of the juicy pulp out from the fibers but leaving a layer of moist fibers over the endocarp. Because Attalea pulp is a favored fruit of agoutis, many fresh (few hours after being eaten) rodentchewed fruits could be obtained from the forest floor for the experiments. Fresh fruits were recognizable because the








remaining pulp was still moist, sticky, and shiny, retained its original color and odor, and the tooth marks were clearly visible in the stretched-up fibers. Older rodent eaten fruits began to ferment within a few hours of use, causing their color, odor, and fiber pattern to change. On several occasions I also set out intact fruits in areas were I knew agoutis were actively feeding on Attalea fruits, and collected the eaten fruits later in the day.



5) Bare endocarps: Ungulates occasionally removed all the pulp and fiber from endocarps (e.g., those endocarps spat out by tapirs). Fruits of any of the above types that were left to rot under parent trees, and seeds defecated by spider monkeys or tapirs, also had little or no fiber on them. Bare endocarps were obtained by scraping away most of the pulp from intact fruits with a knife, then soaking them in water for a few hours or overnight to remove the remaining fiber.

All fruits were collected and prepared the day before

each experimental set up was placed in the forest, then kept in a refrigerator until the experiment was set up in the afternoon of the following day. If fruits were collected from several different trees, they were mixed together and then placed haphazardly into the different fruit categories. Because fruit size varies markedly among trees, when different sizes had to be used in a single set up, an attempt was made to partition fruit sizes equally among fruit types. All endocarps were measured after the experiment was complete









and One-Factor ANOVA's were used to test for differences in size among fruit types within each season. There were no significant differences in endocarp size in any season (F =

0.63, p = 0.64 in June; F = 2.17, p = 0.0716 in July; and F =

1.77, p = 0.13 in Septemer).

Fruits were protected from vertebrate pulp and seed eaters by galvanized wire exclosures ("chicken wire" with mesh size 1.5 x 1.5 cm) 5 cm in height and approximately 23 cm on four sides, with a closed top. These cages were effective in deterring vertebrates from feeding or stepping on the fruits, and provided protection from falling debris. Five exclosures, one for each fruit type and each containing ten fruits, were placed in a star pattern (with one corner touching) at the edge of a fruiting tree's natural seed shadow (1-4 meters from the base of the three) to prevent disturbance from falling fruits (Fig. 3-1). All fruiting trees used had been dropping fruit for at least several days, and had at least 100 naturally fallen fruits in the fruit pile. Fruits were separated by 1-3 cm from each other within each exclosure and were set directly on the natural leaf litter (in preliminary trials with cleared ground, soil disturbance attracted ants, which covered the fruits with dirt).

The experiment was repeated three times at different

stages of the palm fruiting season: 10 set ups in June (early peak), 10 in July (late peak), and 8 in September (tail-end of season). Thus the sample size for each fruit type in each









season was 10 replicates (8 in September) of 10 fruits each (Fig. 3-1). Because few trees at the same fruiting stage were available at any one time, trees were used as they became available. Sometimes several trees could be set on one day, and sometimes only one. All trees were within 1.5 km of each other, and had several other Attalea trees within 50 meters.

All fruits were checked twice after each experiment was set up: once the on morning following the set up, so that beetles had only one night to lay eggs, and again 7 days later, when the beetles had seven nights for egg laying. For the first five set ups in June, an attempt was made to count eggs after 14 nights of exposure, but was abandoned because hatching larvae and eggs slid off easily from the rotting pulp when handled. Only the eggs on Intact Pulp fruits could be successfully counted after 14 days. At each count, endocarps were picked up, all eggs counted, and the endocarps returned to their original position. For endocarps with a lot of pulp, the entire fruit was palpated carefully to feel for eggs hidden in the pulp. On Intact Husk fruits, the husk was not lifted up after it started to rot, to avoid exposing more pulp to beetles, and therefore some eggs may have been missed. Eggs on the other fruits were easy to locate.

Intact Husks, but not the other fruit types, were

examined weekly for up to nine weeks following set up. The condition of the husk, the number of eggs, and the location of the eggs (on husk or pulp) were noted.








Data Analysis, Part I



Because of the large number of zeros for non-preferred fruit types, the data were summarized as the mean number of eggs per seed for each tree x seed type combination, yielding 10 seed type replicates in June and July, and 8 in September. Because the patterns in seed preference were the same for the first and second counts, even though the total number of eggs was lower after one night, only the second count was analyzed.

Because Intact Husk fruits received no eggs by the second egg count, data on this fruit type were analyzed separately from data on the other four fruit types. Timing of oviposition on Intact Husk fruits is described in detail, as this is the main factor bearing on seed survival for this fruit type.

The number of eggs per fruit within each of the

remaining four fruit types was not normally distributed, and variance was greater in June than in July and September. The data were normalized by log(l + x) transformation, and analyzed with a two factor ANOVA, with fruit type and season as factors. Post-hoc comparisons were made with Scheffe F tests at the alpha = 0.01 level to detect significant differences between pairs of seed types within each season. The statistical package JMP (SAS Institute 1989) was used for all these analyses.








Part II: Infestation Rate of Seeds by Bruchids: Number of
Entry Holes and Seed Mortality



This part of the experiment was established to determine whether differences in oviposition rates on different fruit types translated to differences in larval infestation rates. Because mortality can occur at several stages, both the entry and development stages were monitored:



Entry stage. If beetles prefer to lay eggs on certain

fruits because they have a better chance of surviving the egg stage and penetrating the endocarp, then the fruits that have more eggs should show a higher ratio of entry holes:eggs laid than non-preferred fruits.



Development stage. Both increased number of eggs and increased endocarp penetration should lead to an increased chance of seed infestation. The fruits preferred by bruchid beetles should also be the ones most likely to be destroyed by larvae.

The endocarps from Part I were left in their exclosures in the field for approximately three months. This was the time period that both lab experiments and field observations indicated was sufficient for most larvae to reach full size and consume all the endosperm, but not sufficient to allow pupation and emergence. The three month period was counted from the day each experiment was set up, so that different









set ups were collected on different days and in different months. The length of time endocarps remained in the field averaged 99 days in June (range 92-102 days), 103 days in July (99-106 days), and 86 days in September (78-93 days).

After collection, all seeds were measured along the long axis, and the number of visible entry holes counted (regardless of whether they penetrated to the endosperm or not). Each endocarp was then split open with a machete, and the following parameters assessed: number of seeds, amount (estimated proportion) and condition (germinating, with fungus, or intact) of seeds not eaten by larvae, number and developmental stage of bruchids, if any (larva, late larva, pupa, or adult), and identity of parasites on bruchids, if any.



Data Analysis, Part II



The number of entry holes and the proportion of infested seeds for each category were compared within each season with a Kruskal-Wallis test. Unlike the situation in Part I, endocarps derived from Intact Husk fruits were included in the analysis in Part II.

The stage (small larvae, large larvae, late larvae,

pupa, and adult) of all larvae obtained were compared among seasons with a Chi-Square test for all fruit types grouped (due to the large number of blank cells). For the September data set only, the larval stage was compared among fruit









types with a Chi-square test, because all larval stages were well represented. The statistical package Statview (Abacus Concepts 1986) was used for these analyses.



Control Experiment



To determine whether seven nights of exposure to

oviposition are sufficient to lead to high levels of seed infestation under the controlled conditions of the experiment, a small scale experiment was set up among the experimental trees. Fifty Gashed fruits were protected by a chicken wire exclosure and exposed to oviposition for seven nights, then covered with mosquito meshing to exclude bruchid beetles and left in the forest. Another 50 fruits were placed next to the above exclosure, protected only with chicken wire. After three months the endocarps were collected, the number of entry holes counted, and the endosperm and larvae described as in Part II above. A t-test was used to compare the number of entry holes on the two sets of seeds, and the proportion on infested seeds was compared visually.



Results



Intact Husk fruits received no eggs during the first and second checks, with the exception of two fruits that were bitten open by insects and mammals, respectively (Table 3-1).









Only by the sixth week in the field had at least one Intact Husk fruit in all set ups received eggs. This occurred when the husk had rotted sufficiently to split, or to lift away from the pulp, opening a gap at the seed's proximal end between husk and endocarp. Eggs were laid both on the husk and the pulp at this stage. The number of eggs remained low, with no fruit ever receiving more than 8 eggs, and the mean number of eggs per fruit ranging from 0.3 to 1.15 for different set ups (Table 3-2).

The number of eggs found on Gashed, Rodent, Intact Pulp, and Bare Seed fruit types after 7 nights (mean 6.86 eggs/fruit, range 0-47) was greater than the number of eggs after one night (mean 3.36 eggs/fruit, range 0-26), but the relative differences among seed types remained the same on both checks. Only the results of the second check are further described. The total number of eggs and the variation among fruits in number of eggs decreased markedly during the study, both among and within seasons, for the other fruit types (Fig. 3-2a-d, Fig. 3-3). The decrease was much more marked for Gashed and Rodent fruits than for Intact Pulp and Bare Seed fruits (Fig. 3-2a-d). I describe first the trends, and later the significant differences among fruit types.

In June, Gashed and Rodent fruits set at the end of the month had fewer eggs than those set earlier in the month (Fig. 3-2a-b), but the same was not true for Bare Seed and Intact Pulp fruits (Fig. 3-2c-d). Gashed and Rodent fruits received the highest number of eggs in all seasons, but in








June Rodent fruits had more eggs than Gashed fruits (18.9 vs. 15.9 eggs per fruit), a difference that was reversed in July and September. Intact Pulp fruits consistently received the lowest number of eggs in all seasons (4.6, 2.5, and 2.1 mean eggs per fruit respectively), followed by Bare Seeds (6.6,

3.0, and 2.7 eggs). For the five trees that were checked after 14 nights, Intact Pulp fruits averaged 8.0 eggs/fruit and very few of the original eggs had hatched, indicating that they were still receiving eggs and that the first eggs laid were not successful.

These differences were consistent among trees within each season (Fig. 3-4a-c). In June, a marked difference among trees in total number of eggs received, and a reduction in the difference between Gashed and Rodent fruits were due to the spread in time of experimental set ups, with the trees used later in the season having both fewer eggs and a similar number of eggs on Gashed and Rodent fruits (Fig. 3-4a).

A 2-factor ANOVA on log transformed data revealed

significant effects of both Fruit Type (df = 3, F = 39.94, p < 0.0001) and Season (df = 2, F = 43.74, p < 0.0001), and a significant interaction between Season and Fruit Type (df = 6, F = 2.26, p = 0.0437). Post hoc comparisons indicated that in all seasons, Gashed and Rodent fruits are statistically indistinguishable from each other, as are Bare Seed and Intact Pulp fruits. In June and July, Gashed and Rodent fruits receive significantly more eggs than both Bare Seed and Intact Pulp fruits. In September, the significant








difference between Rodent fruits and both Bare Seeds and Intact Pulp disappears, but Gashed retains significantly higher numbers of eggs than Bare Seed and Intact Pulp.

When the seasons are compared to each other, without regard to fruit type, in June significantly more eggs were laid than in both July and September, but the latter two are statistically indistinguishable from each other.





Control Experiment



After three months, there was no significant difference in the number of entry holes on endocarps exposed to beetles for 7 nights (15.3 +/- 6.31 holes) and those exposed for the full three months (16.1 +/- 5.72 entry holes; t = 0.42, p =

0.338). Both endocarp types had the same larva-induced mortality rate (Table 3-3). The only apparent difference between the two sets of endocarps was that nine of the larvae in the three month exposure set had been parasitized/killed by insects, whereas there was no indication of parasitism in the seven night exposure set.



Phase II: Entry Holes



The number of entry holes was significantly different

among fruit types in all seasons (Kruskal Wallis, H = 280.78, 135.64, and 75.33 respectively in June, July, and September,








p = 0.0001 in all cases). The trends in entry holes followed those expected from the number of eggs for all seed types in all seasons, with the exception of Intact Pulp fruits. Although this fruit type always had the lowest number of eggs, behind Bare Seeds, in June it had the second lowest number of entry holes, ahead of Bare Seeds, and in July and September it had the same number of entry holes as Rodent, and more than Bare Seeds (data are mean of all seeds per fruit type). This discrepancy suggests that once the pulp became degraded some time after the second check, beetles continued laying on Intact Pulp fruits, and this latter round of egg-laying produced larvae that successfully entered the seed. Therefore, as for Intact Husks, the number of eggs counted at the second check does not correspond to the actual egg laying intensity eventually experienced by the seeds.

For the three other fruit types, the second check egg count should be a good index of the larval cohort that will attempt the first entry to the seed. Excluding the Intact Husk and Intact Pulp fruits for the reasons explained above, the ratios of entry hole number:egg number in June, July and September for each seed type are as follows: 0.539, 0.367, and 0.390 for Bare Seed; 0.826, 0.754, and 0.610 for Gashed, and 0.788, 0.692, and 0.504 for Rodent. These larval success ratios reflect the laying preferences of beetles; fruit types with the highest success ratios in each season received the highest number of eggs. Additionally, the seasonal decrease








in egg:entry hole ratio indicates increasing egg mortality from June to September.



Part II: Proportion of Endocarps Infested by Larva



An endocarp was considered infested if it contained: any live or dead larval stage or adult beetle, the parasites of bruchid beetles, the track of a larva through the endosperm, or larva frass. Any endocarp that had at least one larva was classified as infested, even if, in the case of multiple seeded endocarps, one of the seeds was still intact. Although in a natural (unprotected by exclosures) setting one seed of multiple-seeded endocarps sometimes escapes predation (Fragoso 1994), in this experiment very few endocarps showed this situation (9 Intact Husk fruits, 3 Intact Pulp fruits, 2 Gashed fruits, 2 Bare Seed fruits, and 1 Rodent fruit). Large larvae used all seeds within an endocarp, and when one seed was still intact, the larvae was usually small and still had the chance of destroying all seeds. The difference between the experimental and the natural settings is that the wire cages protected seeds (and eggs) from trampling, shifting, and pulp removal, therefore giving the eggs and first instar larva a much better chance of hatching and entering, respectively.

All fruit types in all seasons had a nearly 100%

infestation rate with the exception of Intact Husk (mean infestation rate 60% +/- 32) in June. Thirty seven Intact









Husk and 9 Bare Seed endocarps survived in June, compared to only 1, 1, and 0 Intact Pulp, Gashed, and Rodent endocarps respectively (Table 3-4). The difference in survival rate among seed types was significant in June (Kruskal-Wallis test, H = 16.08, p = 0.0001; based on mean survival rate per seed type per tree), but not significant in July and September (p = 0.07 and 0.4, respectively). Note that the sample sizes for Intact Husk and Intact Pulp fruits were reduced in July and September compared to other fruit types due to fruit removal by vertebrates.

The seed survival rate for IH fruits was significantly higher in June than in July and September, and was not different between July and September (X2 = 67.3, p = 0.0001, DF = 8) (Table 3-5).



Bruchid Stace



In June and July, most of the larvae found in all fruit types were large, late instar larvae that had consumed all or most of the endosperm in their seeds (Table 3-6). Two exceptions were Intact Husk fruits in June and July, which had a high proportion of very small larvae, indicating that larvae were younger than those in the other fruit types, as would be expected from the late timing of egg-laying on these fruits.

In September, 65.5% of larvae found were large larvae, but now a high proportion of seeds contained late larvae,








pupa, and adult beetles (30% combined late stages) (Table 35). Within this season, Bare Seeds had a disproportionately high number of late stages (53.7% combined). The combined differences in larval stages were significantly different among seasons (X2 = 221.47, p = 0.0001, df = 8). Since all seeds were the same age when collected from the forest (in fact, the September seeds were younger by about two weeks than the June and July seeds), these differences indicate that a) larva from eggs laid later in the season developed at a faster rate than those earlier in the season for all fruit types; and b) larvae in Bare Seeds late in the season developed faster than larvae in other seed types in the same season. This early development in Bare Seed fruits was already noticeable in July, when a higher percentage of larvae in Bare Seeds than in any other fruit type were late instar larvae (Table 3-6)



Fruit and Seed Loss to Vertebrates and Invertebrates



Fruit disappearance from exclosures were greater late in the fruiting season than earlier (Table 3-7). In June, fruit disappeareance occurred without exclosures being breached. In July and September, some but not all exclosures with fruit loss were breached, and in some cases chewed fruits were found within the uplifted exclosure.

In one exclosure where four Intact Pulp fruits were lost in September, without exclosure disturbance, only three








endocarps were found on the surface in December. Persistent digging uncovered all seven missing endocarps, individually buried about 10 cm under the soil surface, within the area protected by the exclosure. Of these seven endocarps, four were intact, two had very small larvae with 80 % of the endosperm intact, and one had a large larvae that had consumed all the endosperm. The other 40 endocarps set at this tree were all infested or their seeds had rotted. At the time that the first four fruits were buried, at least two of them had received at least 1 or 2 eggs (counted after the first check). At the second check, the six remaining Intact Pulp fruits had 0, 1, 1, 2, 3 and 4 eggs respectively, so that when the final three were buried, an unknown time after the second check, at least two of them had at least one egg on them. In other words, a maximum of three fruits could have been buried with eggs already on them.



Discussion



Implications for Fruit Handlina by Animals



The most striking differences in both ovipostion and

seed survival rates occurred between Intact Husk fruits and all other fruit types. Female Pachymerus card at Maracd did not lay on fruit husks until these began to rot, and even then the number of eggs laid was low compared to the number laid on the other fruit types within the same season. Given








the consumer set that uses Attalea fruits at the Maracd site, the avoidance of intact fruits by bruchid beetles has significant implications for the palm's dispersal ecology and adult distribution pattern.

The seeds that had the best chance of surviving were

those that fell to the ground in fruits unhandled by arboreal or terrestrial pulp eaters. These are the fruits that are ingested by tapirs in large amounts and carried long distances from the parent tree (Fragoso 1997b). Because intact fruits do not receive eggs until at least four weeks after falling, they are uninfested at the time of ingestion by tapirs. After defecation, they are subject to infestation by bruchids in the feces (Silvius pers. obs.), but the infestation rate in the feces is much lower than in the original fruit piles (Fragoso 1997b).

The fruits with the highest probability of receiving eggs are those that have been handled by the presumed best dispersers (rodents, primates and spit-dispersing ungulates) and also by procyonids and psittacines. Although the increase in egg laying did not result in an increase in larval infestation in the experiment once Intact Husk fruits are excluded, it must be remembered that the fruits in the experiment were protected to an extent that would not occur naturally. Foraging tapirs, peccaries and deer trample seeds into the ground, and falling litter protects them from access by beetles (as does burial under the fruit pile). Therefore the longer a fruit remains without eggs, the more likely it









is to survive infestation. Rodent and Gashed fruits always received eggs after the first night. However, since eggs do not hatch for several days, it is unclear how well oviposition translates to first instar survival if seeds are subsequently trampled or buried. Eggs laid on Rodent and Gashed fruits are more likely to remain attached to the seed than those laid on intact pulp, however, and therefore more likely to resist partial burial.

These observations refer to endocarps remaining unburied near the parent tree. Squirrels probably bury many endocarps throughout the fruiting season, and agoutis bury some seeds late in the season (Fragoso 1997b). Fruits dispersed by primates even within several hundred meters of a tree are still exposed to high rates of oviposition, at least within a forest with high Attalea density (fruits placed up to 200 meters from the nearest fruiting palm received the same numbers of eggs as fruits placed at the base of fruiting and non-fruiting palms; Silvius, unpublished data). Therefore at this time the best chance of survival is to be an Intact Husk or Intact Pulp fruit that gets carried away by a tapir soon after falling. Thus the activities of vertebrate pulp eaters can mediate seed predation by invertebrates when the invertebrate is the primary predator on the seed, as it is in this case, where more than 95% of all seeds remaining on the surface near the parent plant are destroyed by bruchid beetles (Fragoso 1997b, Chapter 4).








Pulp-Oviposition and Pulo-Development Rate Interactions: The
Role of Moisture Levels



The amount of pulp remaining on endocarps, and the

pattern in which it was arranged, had a significant effect on the number of eggs laid on fruits by bruchid beetles. This effect translated into higher egg and first instar larva survival, as indicated by the dependence of number of entry holes on number of eggs. Fruiting season interacted with the effect of pulp on both number of eggs and number of entry holes: overall number of eggs laid, and differences among fruit types in number of eggs laid, were greater early in the fruiting season (June) than late (July and September). Differences in egg laying and egg survival also translated into differences in seed infestation rates (number of endocarps with at least one larva entering), but this effect was significant only in June, when larva failed to penetrate Intact Husk fruits.

The general trend in egg laying preference was that

endocarps with some pulp were preferred above endocarps with no pulp or with too much pulp. Fruits with intact husks were inaccessible to ovipositing females. Avoidance of seeds with large amounts of pulp may be due to at least two factors: a) first instar larvae are unable to penetrate the pulp for either physical or chemical reasons, and b) eggs do not stay attached very well to intact, slimy pulp. Both possibilities are supported by the observation that female bruchids often








lay eggs on pieces of litter stuck onto Intact Pulp fruits, on indentations in the pulp made by pulp eating invertebrates, or at the proximal end of the seed, where the pulp layer is thinnest. The second possibility is supported by the fact that eggs on intact pulp frequently fell off when handled or after heavy rainfall.

On Bare Seed fruits, females usually laid eggs on the underside of the endocarp. Wright (1991) observed the same behavior for a different palm bruchid species in Panama. Along with the above observations on pulp condition, this behavior suggests that moisture is a key factor in egg-laying choice: eggs on bare endocarps are likely to dry out, because there is no pulp to retain moisture. Eggs on intact pulp will not dry out, but neither do they stick well. Fruits gnawed by rodents and gashed by ungulates or monkeys, on the other had, provide a secure attachment site on the bare endocarp or on the juice-free fiber, while providing moisture-retaining pulp or fiber. On Gashed Fruits, eggs were nearly always laid on the endocarp but close to or under the edge of the pulp. It is unclear, though, why infestation of completely bare endocarps was lower early in season, when it rained frequently, than later in the season. One explanation may be related to the observed higher insect-related egg mortality on bare endocarps than on fruits with pulp.

The net effect of these seasonal differences was that

even though oviposition rates were higher early than late in the fruiting season, seed survival was also higher early than








later. In other words, the correlation between oviposition rates and survival is strong within a season but does not carry on across seasons. Early in the season, Intact Husk and Intact Pulp fruits resisted beetle infestation long enough that, had they not been protected by exclosures, they had better chances of being removed by a disperser before bruchid eggs could hatch or even be laid on them. Therefore trees fruiting late in the season, when bruchid activity is lower, do not necessarily have higher seed survival probabilities, as predicted by Forget et al. (1994) and Wright (1990) for a similar system in Panama. Those authors' conclusions were based on monitoring of egg numbers only, and not on the translation from egg to larval infestation rates. On the other hand, late in the season, when overall fruit pulp availability is lower in the forest (Chapter 6), fruit removal rates by frugivores may be higher, as indicated by the invasion of exclosures in the last experimental set up. In this case seed survival would depend on whether frugivores remove seeds to a site that protects them from bruchids. On MaracA, bruchid activity is so high (compared to the mean 1-2 eggs per endocarp documented during the peak fruiting season on BCI, Wright 1990) that fruits placed more than 100 meters from fruiting trees have the same likelihood of being found by bruchids as fruits near the tree (Silvius, unpublished data).

The role of moisture seems to be important in another

way. Several researchers have observed that rainfall may cue








the emergence of adult beetles(Janzen 1971, Wright 1983). The more rapid development of larvae late in the season on Maraci (endocarps from the September replicate were collected in December, two months into the dry season, whereas those from June and July were collected in September and October, respectively), and the more rapid development of larvae in Bare Seeds (the ones most likely to lose moisture) suggest that beetles may be responding to the moisture level in the environment by speeding development when moisture is low. Thus, most beetles emerge in the dry season, regardless of when the eggs that gave rise to them were laid. With a few exceptions, emergence from natural, non-manipulated endocarps in the field did not begin until November, and did not reach its peak until December-January (Chapter 4), supporting these observations. An alternative hypothesis is that there are at least two sub-populations of beetles, that differ genetically or behaviorally with respect to reproductive traits: one group that lays early in the season but emerges late, and another that lays late in the season but emerges early.



Pulp as Protection



The interaction between pulp eaters, bruchid beetles, and palm fruits also indicates that pulp, as well as husk, can be a protective mechanism against invertebrate seed predators. It may be under selective pressure for this reason as well as for its attractiveness to dispersers. The









role of husks (pericarp or exocarp) as physical deterrents against invertebrates has received some mention in the literature (e.g., Pritchard 1969, Sallabanks and Courtney 1992, Rodgerson 1998), and the role of chemical deterrents in pulp has been extensively investigated (see reviews in Herrera 1982, Sallabanks and Courtney 1992, and Cipollini and Levey 1997), but I have found no mention of pulp as a physical deterrent.

Work by Delgado et al. (1997) suggests a similar role for pulp in a different palm species. These authors found that oviposition by bruchid beetles on Astrocaryum chambira fruits, which have a much drier, thinner mesocarp than Attalea fruits, was more rapid on peeled than intact fruits, and that oviposition became high on unpeeled fruits only after the mesocarp had decayed. Because Astrocaryum seeds are prized by many granivorous rodents (Smythe 1986, Terborgh et al. 1993, Cintra and Horna 1997, Hoch and Adler 1997), if pulp protects a fruit on the ground until rodents scatterhoard them, then the fruit's chance of survival is increased. This type protection would also work with the putative megafauna dispersers that may have coevolved with large-fruited tropical trees (Janzen and Martin 1982), represented today only by the tapir (Fragoso 1997b).








Burial by Scarabid Beetles



The seven fruits "lost" from the September exclosure

were buried at a depth and in a pattern typical of burial for nesting purposes by large scarabid beetles of the burrower type (Halffter and Edmonds 1982). Several observations indicate that the seven fruits were buried by the dung beetle Oxysternon festivum (Scarabaeidae / Coprinae / Phanaeini) and that burial of Attalea fruits by this species is a common occurrence at this study site. Adult scarabid beetles have been recorded feeding on pulp (Hanski and Camberfort 1991), and one species is known to synthesize nesting balls out of pulp and other plant parts (Monteith and Storey 1981), but I know of no reports of scarabid beetles using entire fruits to provision their nests. During a study scarabid beetles on MaracA in June 1996, large numbers of the beetles were captured in traps baited with fox (Cerdocyon thous) dung (which at that time contained Guava pulp and seeds), and was considered a common species in the study area (F. Vaz de Mello, pers. comm).

On two occasions the beetles were observed to bury

Intact Pulp Attalea fruits shallowly in a network of tunnels under fruit piles, in a manner similar to that of shallow burial of dung balls for adult feeding by other dung beetle species. A beetle captured at a fruit pile and held in the lab for a few hours buried five Intact Pulp fruits to a depth of 5 cm, the maximun depth of soil provided in its container.








Throughout the five fruit type experiment described in

this paper, fruits were found within the exclosures shallowly buried in the beetle fashion. Only the fruits from the single tree in September, described above, were ever found deeply buried. Thus, Oxvsternon exhibited the two kinds of food burial typical of dung beetles: shallowly buried fruits on which adult beetles may feed, and deeply buried fruits potentially used as food for larvae.

The seven exclosure fruits deeply buried by the beetles had variable numbers of eggs on them when buried. It would appear that those that were buried early had their eggs scraped off or killed by the burial process. The others survived burial and bruchid larvae infested the seeds, but burial seems to have delayed the development of the larvae in at least two seeds. The higher survival rate of scarabidburied seeds than of other seeds set at the same tree suggests that burial by scarabids functions in the same way as removal of intact fruits by tapirs: they decrease the chance of oviposition and/or first instar survival. The germination biology of Attalea is unstudied, and it is therefore unclear whether its seeds are able to germinate from a 10 cm depth is unknown. Survival is likely, however, given that many palms have remote germination in which seedlings usually grow down into the soil before extending up to break the surface (Tomlinson 1990). Scarabid beetles could thus be an effective "planter," if not mover, of palm seeds, and many of the seedlings that germinate near the








parent plant on Maracd (Fragoso 1997b) may be the result of burial by beetles rather than by rodents.



Larae scale effects of Dulp-vertebrate-beetle interactions



This study documented a small-scale interaction at the

individual seed level. It further focused on a single forest type with a characteristic mammal community. The distribution of frugivores at a larger scale, however, is patchy and nonrandom, both on MaracA and at other Neotropical sites. On Maraca, spider monkeys do not use forests near the river (Nuies 1992), while squirrel monkeys prefer those forests as well as other edge habitats (Silvius pers. obs.). Tapirs prefer to feed in the highest density Attalea patches in the river forest, leading to high levels of intact fruit removal there. At a scale much larger than that of this study, patchiness in frugivore distribution could therefore result in patchiness in palm-beetle-vertebrate interactions and in palm seed survival and regeneration. Such patchiness in interactions could contribute to the clumped distribution of many palm species, a distribution that has most often been attributed to anthropogenic effects (Balbe 1988, Clark et al. 1995) or edaphic factors (Kahn and de Granville 1992).








Table 3-1: Timing of egg laying on Intact Husk fruits, presented as number of experimental setups (trees) out of the total set up that had at least one egg on at least one fruit at the time of checking. Data is cumulative. Check # 1 occurred one day after seeds are laid out, subsequent checks followed at 7-day intervals. A blank cell indicates fruits were not checked at that date. Egg laying does not begin until the husks begin to rot and the pulp or endocarp is exposed.

Check Number

1 2 3 4 5 6 7 8 9 Season
June 0 0 4 6 7 9 10 10 10 10 trees
July 0 1* 4** 6 6 9 10 trees
September 0 **
8 trees
* = husk chewed open by insects, exposing the pulp.
** = husk chewed open by mammals, exposing the pulp


Table 3-2: Mean number and range of eggs per fruit for trees that received eggs on the Intact Husk fruits during the June experiment, by check. n = 100 seeds per check (10 seeds at each of 10 trees).

IMean 0 0 10.3 10.28 10.87 10.83 11.15 0.9 0.92 Range 0 0 0-1 0-2 0-5 0-4 0-8 0-5 0-6


Table 3-3: Infestation rates for fruits exposed to bruchid oviposition for seven nights and three months, respectively.

7 night exposure 3 month exposure Large Larvae 45 46 Empty Endocarp 3 4 Rotten Endosperm 2 0 Total 50 50








Table 3-4: Seed survival per fruit type and season, presented as number of infested (at least one endosperm with larvae or other sign of bruchid beetle infestation), good (no seeds with larvae or sign of infestation) and other (empty endocarp or endosperm rotted by fungus, with no sign of bruchid infestation). Note reduction in sample size for IH fruits in all seasons and for IP fruits in July and September, due to fruit loss from the exclosures to pulp consumers. June (initial n = 100 fruits per fruit type.


Infested ( 1 seed infested)


7 -


Good
(no seeds infested)


Other
(empty, fungus)


Intact Husk 57 36 2 Intact Pulp 98 1 1 Bare Seed 86 9 5 Gashed 96 0 4 Rodent 95 1 4


July (initial n = 100 fruits per fruit type)
Infested Good Other
(1 1 seed (no seeds (empty, infested) infested) fungus) Intact Husk 91 3 0 Intact Pulp 90 3 4 Bare Seed 97 0 2 Gashed 99 0 1 Rodent 92 5 5


September (initial n = 80 fruits per fruit type)
Infested Good Other
( 1 seed (no seeds (empty, infested) infested) fungus) Intact Husk 58 0 4 Intact Pulp 68 4 1 Bare Seed 78 0 1 Gashed 76 0 2 Rodent 78 0 1


Table 3-5: Seed survival in Intact Husk fruits during the three seasons. Survival is significantly higher in June than in July and September (X2 = 67.3, p = 0.0001, df = 8)

Season Infested Good Other June 57 36 2 July 91 3 0 September 58 0 4








Table 3-6: Infestation stage of endocarps, separated by fruit type and season. "New larvae" are small larvae that have not had time to eat much of the endosperm; "larvae" are large larvae that have eaten all or most of the endosperm but have not metamorphosed into pupal stages, "late larvae" are larvae that have started to metamorphose, as indicated by a reduction in size, thickening of body and change in head shape; "pupa" are soft, white with legs, mouth parts and wings; adults are dark, fully formed beetles. Field age is the number of days the seeds remained in the forest between setting and cracking; for all endocarp types except Intact Husk, this corresponds to larval age.


June (mean fiel


New Larvae


d ace = 99 days)


Larvae


Late Larvae


Pupa


Adults


Total


Bare 4 75 0 1 0 80 Seed (5%) (94%) (1%) Intact 14 39 0 0 0 53* Husk (26%) (74%)
Intact 3 89 0 0 0 92 Pulp (3%) (97%)
Gashed 1 87 0 0 0 88
(1%) (99%)
Rodent 1 86 1 0 0 88
(1%) (98%) (1%)
Total 23 376 1 1 0 401 Larvae (5.7%) (93.7%) (0.25%) (0.25%)
* Low number of larvae in IH seeds is due to low infestation rate of seeds during the first part of the experiment.










Table 3-6, continued:


July (mean field age = 103


New
Larvae


Larvae


days)


Late Larvae


Pupa


Adults Total


Bare 5 80 10 2 0 97 Seed (5%) (83%) (10%) (2%) Intact 23 65 2 1 0 91 Husk (25%) (72%) (2%) (1%) Intact 4 86 0 0 0 90 Pulp (4%) (96%)
Gashed 2 91 0 0 0 93
(2%) (98%)
Rodent 1 86 0 0 0 87
(1%) (99%)
Total 35 408 12 3 0 458 Larvae (7%) (89%) (3%) (1%)






September (mean field age 86 days)
New Larvae Late Pupa Adults Total
Larvae Larvae
Bare 2 40 22 23 3 90 Seed (2%) (44.4%) (24.4%) (26%) (3.3%) Intact 8 29 16 7 0 60* Husk (13%) (48%) (27%) (12%) Intact 4 55 7 5 1 72 Pulp (6%) (76%) (10%) (7%) (1%) Gashed 1 66 8 9 0 84
(1%) (79%) (9.5%) (10.5%)
Rodent 2 55 6 4 0 67
(3%) (82%) (9%) (6%)
Total 17 245 59 48 4 373 Larvae (4.5%) (65.5%) (16%) (13%) (1%)
* . . . . . . . . . . - - _ - -- T - - - . . . - * -- _ _ L - ---


is aue to


loss or


Low number or larvae in intact HUSK sees seeds to frugivores during experiment.






86


Table 3-7: Fruit losses from exclosures after one week.

Season Fruit Type Number Lost
June Intact Husk 5 July Intact Husk 6 Intact Pulp 2 Bare Seed 1 September Intact Husk 18 Gashed 2 Rodent 1 Bare Seed 1 Intact Pulp 4





































S-lOFruits/Type
0


- 5 Fruit Types


One Tree= One "Experimental Set-Up" = One Replicate
X 10= One Season or Experiment






Figure 3-1. Experimental field design for Parts I and II. IH = Intact Husk fruits, IP = Intact Pulp fruits, BS = Bare Seed fruits, G = Gashed fruits, R = Rodent fruits. See text for fruit type descriptions.


IH
P R �S



















00

00
0 0O [] O]


0r1


-I


June


Time Set


a) Rodent


Figure 3-2. Mean number of eggs laid by bruchid beetles on sets of 10 fruits of 4 different fruit types at each of 10, 10, and 8 trees in June, July, and September, respectively. a) Rodent gnawed fruits; b) Primate or ungulate gashed fruits; c) Endocarps with all pulp removed; d) Fruits with pulp intact but all husk removed.


O


September
















O
O


0


0 13
13
13

[]3


-I


June


September


Time Set


b) Gashed


Figure 3-2, continued.












15




03
" 10 O


z
0
LT.
5- O
6O g0
z ]0

00

0- I I

June July Sept


Time Set

c) Bare Seeds


Figure 3-2, continued.




Full Text

PAGE 1

INTERACTIONS AMONG ATT ALE A PALMS, BRUCHID BEETLES, AND NEOTROPICAL TERRESTRIAL FRUIT-EATING MAMMALS: IMPLICATIONS FOR THE EVOLUTION OF FRUGIVORY r , By KIRSTEN MARIANA SILVIUS y A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY : UNIVERSITY OF FLORIDA . 1999 •

PAGE 2

To Joe, without whose imagination creativity -r determination bullying cajoling loving laughing and, of course, canoeing none of this would have been possible or even desirable.

PAGE 3

ACKNOWLEDGMENTS Many people contributed to all stages of this dissertation, and without them both field research and writing-up would have been very difficult. Two people, however, provided emotional and logistic support without which the whole thing might well have been impossible. I am deeply grateful to Jan Sendzimir and Gisela Bosch for their friendship and unique definition of human community. Over the years their support took the form of computers, books, hot tubs, unending glasses of beer and wine, a dog to walk and hug, two wonderful daughters to remind me of the other important things in life, a beautiful wedding celebration and, most importantly, a temporary place to stay that became a home to return to even if some day it is transferred to some other place on this Earth. They are proof that permanence can come from this impermanent process called graduate school. A similar role was played by two members of the World Doctors (Medecins du Monde) mission to the Yanomami Area in Brazil. Joseph Ceruetto and Mattieu Lena were incomparable companions and true friends under difficult circumstances. They saw the links between the conservation of nature and the prevention of ethnocide, and provided logistic support for my

PAGE 4

research that few others were in a position to provide. They introduced me to the Yanomami people, their land, and the indigenists that are struggling to save both, and thus provided me with the strongest possible motivation to work for tropical rain forest conservation. The entire World Doctors team, including the coordinators in France, is to be commended for its vision and hard work. Four other people contributed over and beyond the call of duty to the completion of this dissertation: My, parents, Robert and Rosita Silvius, were critical to the endeavor not only by providing their usual rock-solid trust and encouragement (and strategically placed care packages), but also by taking care of bureaucratic, financial and other logistic requirements on the home front while I was in Brazil. John Eisenberg enabled the completion of this dissertation by believing that independence is a good thing despite the academic risks it carries. During those times when I felt that maybe I had it all wrong and the data were not that interesting or valuable, his boundless fascination with and insight into all things natural rekindled my enthusiasm for the work. His insight into human nature also helped me evaluate more clearly the personal and nonacademic situations that came up while I was writing. Buzz Rolling contributed more than he can imagine by providing a different, albeit lumpy, way of thinking about ecology. His was the most influential of all the classes I . " iv

PAGE 5

took dur ing my graduate career. He also enabled my escape from my sixth-floor computerless cubicle by inviting m.e into the superbly equipped Arthur Marshall lab. Buzz and the other denizens of 111 Bartram--Lance , Garry, Jan, Paul and the irreplaceable Toni Carter-provided advice, inspiration, technical assistance and a surfeit of Canadian humour. Without them writing up would have been no fun at all, eh? I would also like to thank all members of my committee for providing essential information in the form of discussions, references and editing. Colin Chapman always seemed to have the papers I needed. Walter Judd's Tropical Botany course has stayed with me over the years as a firm reminder that ecology without taxonomy is an exercise in futility. Richard Bodmer ' s focus on the interesting evolutionary questions my data brought up was refreshing. I will also always be grateful for his inspired introduction of a desiccated Mauri tia palm fruit at a crucial junction in my qualifying exams. The Wildlife Conservation Society provided most of the funding for my research, and I would like to thank John Robinson and Alejandro Grajal for supporting a plant -oriented project. The Lincoln Park Zoo provided additional funding through the Scott Neotropical Fund. CNPq provided additional support in the form a visiting researcher fellowship to my husband, Jose Fragoso, and by granting me permission to carry out research in Brazil. "

PAGE 6

. . The staff at Maraca Island Ecological Reserve ensured the smooth running of the station and comfortable living conditions in a remote place. Station Manager Guttemberg Moreno de Oliveira did his best to understand the nature of our research and relied on his love of Maraca to navigate the political storms that threaten research in places like Roraima. A consummate humanist, when push came to shove he always did the right thing for his friends. vi

PAGE 7

TABLE OF CONTENTS ACKNOWLEDGMENTS iii ABSTRACT ix CHAPTERS ,_ , 1 INTRODUCTION 1 2 PALM FRUITS AND NEOTROPICAL RAIN FOREST MAMMALS: DIRECT AND INDIRECT RELATIONSHIPS 3 Neotropical Palm Use by Mammals 3 Methods 6 Results and Discussion: Palm-Vertebrate Interactions ... 8 Results and Discussion: Fruit-Vertebrate-Bruchid Interactions 17 Summary and Conclusions 2 9 3 PULP HANDLING BY MAMMALIAN FRUIT-EATERS AFFECTS PALM SEED PREDATION BY BRUCHID BEETLES 47 Introduction 47 Study Area and Species 51 Methods 5 5 Results 63 Discussion 71 4 PALM ENDOCARP USE BY THREE NEOTROPICAL SEED-EATERS: FRUGIVORY OR"GRUBIVORY?" 96 Introduction 96 Study Site 99 Study Species 100 Methods 102 Results 112 Discussion 123 5 AGOUTI ( DASYPROCTA AGOUTI ) HABITAT USE, DIET, AND HOME RANGE ON MARACA ISLAND ECOLOGICAL RESERVE, NORTHERN BRAZILIAN AMAZON 141 Introduction 141 Study Area 14 3 Methods I45 vii :>

PAGE 8

Results 153 Discussion 165 6 SCALE-DEPENDENT DISTRIBUTION PATTERNS IN PLANTS USED BY THREE TERRESTRIAL RAIN FOREST MAMMALS AT MARACA ISLAND ECOLOGICAL RESERVE 198 Introduction 198 Study Site 201 Methods 207 Results 211 Discussion 221 7 CONCLUSIONS AND IMPLICATIONS 246 Phenology: Methods and Implications 246 Palm-Mediated Bruchid-Vertebrate Interactions 247 Complexity in Frugivory Systems 248 Mosaics of Fruit Availability: Scale and Grain Size.. 252 Plant-Animal Interactions and the Aggregated Distribution of Tropical Forest Plants 255 APPENDICES 1 "EDIBLE" PLANT SPECIES ON PHENOLOGY TRANSECTS 264 2 FLORISTIC PLOT DATA SUMMARIZED FROM MILLIKEN AND RATTER (1989) AND NASCIMENTO (1994) 266 REFERENCES "\ . ,271 BIOGRAPHICAL SKETCH 290 viii

PAGE 9

Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy INTERACTIONS AMONG ATTALEA PALMS, BRUCHID BEETLES AND NEOTROPICAL TERRESTRIAL FRUIT-EATING MAMMALS: IMPLICATIONS FOR THE EVOLUTION OF FRUGIVORY BY Kirsten Mariana Silvius May 1999 •' Chairman: John F. Eisenberg Major Department: Zoology The fruits and seeds of arborescent palms are important food items in the diet of Neotropical terrestrial mammals. A literature review shows that seventy-seven species of mammals have been documented feeding on fifty-nine species of palms. Most of the records are of pulp rather than seed consumption. Among mammals, terrestrial consumers are the most important predators and dispersers of palm seeds. Bruchid beetles (Coleoptera: Bruchidae: Pachymerini) are specialized predators on palm seeds and are associated almost exclusively with the coconut-like palms (tribe Cocoeae) . Manipulation of Attalea palm fruits indicated that pulp removal by rodents, primates, and ungulates improves fruit quality for ovipositing bruchid beetles. Fruits that fall to the ground unhandled by any vertebrate have the best chance ix

PAGE 10

of surviving beetle infestation: they are rapidly removed by tapirs (Tapirus terrestris ) and are later defecated at sites where infestation by bruchids is lov/er than near parent trees. Monitoring of palm endocarp use by white-lipped peccaries (Tavassu pecari ) , agoutis ( Dasvprocta agouti ) , and squirrels ( Sciurus igniventris ) indicates that all three species feed on bruchid beetle larvae. Squirrels are palm specialists that feed on both larvae and endosperm year round; agoutis feed on both larvae and seeds only at times and at sites where alternate foods are not available; and white-lipped peccaries feed preferentially on bruchid larvae rather than seeds. The three species should therefore respond independently and distinctly rather than as a community to changes in Attalea fruit availability. Home range size of five radio-collared agoutis ranged from 2.9 to 8.5 ha. All animals showed a shift in core area use as different individuals of the same tree species and individuals of different tree species came into fruit. Agoutis responded to their habitat at the scale of individual trees, and appeared to base foraging decisions on knowledge of food available on their entire home range. Preferred diet items were primarily in the Sapotaceae ( Pouteria surumuensi s and Pradosia surinamensi s ) and secondarily in the Palmae (Astrocaryum aculeatum pulp and seeds and Attalea marioa pulp) ... X

PAGE 11

CHAPTER 1 INTRODUCTION Arborescent palms, a widespread and abundant group of plants with a fairly coherent fruiting biology, are an important source of food for Neotropical forest mammals. Palm fruits are staples in the diet of some species at some study sites (ungulates, Bodmer 1989a, b; squirrels, Glanz et al . 1996) and have been proposed as "key" or "keystone" food species for frugivorous mammals (Terborgh 1986a, b, Spironelo 1991, Peres 1994a) at other sites. Due to their abundnace and importance in frugivore diets, palms can serve as easily studied models of the evolution and ecology of frugivory systems. Their importance in the diet of terrestrial mammals in particular may help us understand the role of these poorly studied species in seed dispersal, seed predation and forest regeneration. However, the unique life history characteristics of palms must be understood before the results of palm ecology studies can be extrapolated to tropical frugivory systems in general. Palm-mammal interactions have been studied in detail at only a few Neotropical sites. As an introduction to a detailed study of the interactions between one palm species and its mammalian consumers, in Chapter 2 I review the current state of knowledge and assumptions about the 1

PAGE 12

importance of palms to fruitand seed-eating Neotropical forest mammals. The remainder of the dissertation focuses on the ecology of one palm, Attalea maripa , and examines 1) the implications of the spatial distribution of trees for fruit availability to animals operating at different spatial scales and 2) the role of insects in fruit-vertebrate interactions. Chapters 3 and 4 focus on two aspects of the interaction among bruchid beetles, Attalea maripa palms, and vertebrate fruit and seed consumers. Chapter 5 addresses the effect of the spatial distribution of the palms, within several forested habitats, on the pattern of food availability for one species of terrestrial fruit and seed eating mammal: the agouti, Dasvprocta agouti . Throughout, I address three components of frugivory theory: the role of keystone food resources in structuring frugivore communities; the effect of patchiness in food distribution on animal movements and demography; and the contribution of plant-animal interactions to the patchy distribution of plant species.

PAGE 13

CHAPTER 2 PALM FRUITS AND NEOTROPICAL RAIN FOREST MAMMALS: DIRECT AND INDIRECT RELATIONSHIPS Neotropical Palm Use by Mammals Neotropical arborescent palms produce fruits that in their size and color fit the classical mammal dispersed fruit syndrome, as defined by van der Pijl (1982). The fruits can be large (e.g., 5 cm diameter for Mauritia f lexuosa . 7 cm length for Attalea maripa : van Roosmalen 1985), with a nutritious oily or sv;eet mesocarp (pulp) . The seed is usually protected by a thick endocarp (nut wall) of up to 0.5 cm in thickness. The exocarp (husk) and mesocarp are usually yellowish, orange, reddish-brown, or brown in color (Uhl and Dransfield 1987, van Roosmalen 1985). Some medium sized (< 20 m tall) and understory palms also produce similar fruits. Other small palms produce fruits that seem better adapted for bird dispersal due to the small size of their seeds and the red color of the fruits (e.g., Bactris spp.), although their fruits are also ingested by primates and their seeds dispersed after passage intact through the gut (e.g., Desmoncus, van Roosmalen 1985) . Currently, the Neotropical palm fruitand seed-eating fauna disperses seeds either by ingesting them along with the ' ' ' 3

PAGE 14

4 . . fruit and passing them intact through the gut (e.g., tapirs), by spitting out the seeds after eating the pulp (e.g., tapirs and some primates), or by scatter-hoarding seeds (e.g., Dasvprocta and other rodents; Smythe 1986, van Roosmalen 1985) . Terrestrial rather than arboreal frugivores and granivores tend to be the most important seed predators and dispersers for palms, using the pulp or seeds or both after they have fallen to the ground. Few mammals except for some primates and occasionally squirrels seem to harvest the fruits or seeds from the tree itself. Macaws appear to be the only important avian consumers of large palm fruits and seeds (Munn 1988) . The only true palm specialist among them, the Hyacinth macaw ( Anodorhvnchus hvacinthinus ) . also feeds on palm seeds from the ground in the open Pantanal habitat (Munn et al . 1990). The ecological interactions between large palm fruits and terrestrial consumers suggest that palm fruit use by terrestrial f rugivore/granivores represents more than just the use of residual fruit production. Examination of the foraging ecology of terrestrial fruit eaters may bring new insight to the study of frugivory, since paradigms in this field are based almost completely on work done with birds and primates. Seasonal and supra-annual variation in levels of fruit availability can determine the carrying capacity for frugivorous animals, as well as lead to the evolution of behavioral, physiological, and morphological adaptations to deal with food shortages (Eisenberg 1980, Terborgh 1986a, b,

PAGE 15

Leighton and Leighton 1983, Malcolm 1990, Caldecott 1992, van Schaik et al . 1993, Leigh et al . 1996). Any mammals that have become palm specialists or that rely heavily on palm fruits will see their population dynamics affected by the spatiotemporal patterns of palm fruit availability in their habitat. Although there are few studies on the fruiting phenology of arborescent palms, three studies of palm fruiting at the community level (de Steven et al . 1987, Sist 1989, Peres 1994b) and one population level study (Pinero and Sarukhan 1982, Pinero et al . 1984) demonstrate that there is great seasonal and supra-annual variability in the likelihood of reproduction of individual palms and palm patches alike. The predictability of these production patterns is unknown, but palms do not appear to be more reliable a fruit source than other tropical fruiting trees. Recent data from Barro Colorado Island, Panama (BCI; J. Wright pers . comm.) indicate that large and small fruited palms participate along with dicots in forest-wide fruit peaks and failures following El Nino Southern Oscilation events. The same factors that make palm fruits attractive to other mammals--nutritious pulp and seeds--also make them highly attractive to humans. Native human populations traditionally have used palms for their fiber, stem tissue, and sap as well as for their fruits (Balick 1988) . Thus their current ecology has been affected by at least several thousand years of interactions with humans, which can affect

PAGE 16

distribution and fruit traits (e.g., domestication of Bactris gassipaes ) (Balee 1988, Anderson et al . 1991). For this review I surveyed the Neotropical ecological literature to extract the following information: Which mammals are known to use palms? Numerically, how important are palms in the diets of these animals? Since seeds are a long lasting resource and pulp a more ephemeral one, which part of the palm fruit is most heavily used? What is the nature of the interaction among palms, bruchid beetles specialized on palms, and vertebrate palm consumers? What are the ecological and evolutionary implications of these interactions? Methods The literature summarized here is a biased sampling of the information available on palm-mammal interactions, because it does not include many articles published in Latin American journals. It does however include unpublished theses from Latin American universities and unpublished data from several Latin American researchers. From each source I obtained the following information: the palm species used by mammals, the mammal species using palms, the part of the palm fruit used, the location of the study, and notes on the behavior of the mammals, in particular whether they seemed to be using the palm as a keystone resource in the season of fruit scarcity (sensu Terborgh 1986a, b). For studies that

PAGE 17

lasted about a year or more and gave information on the dietary composition of the study species I summarized the importance of palm fruits as the proportion contributed to the diet. Most of the studies reported here took place in humid forest, some in deciduous or dry forest, and a few in open habitats. I do not include information on use of any palm parts other than pulp and seeds, although meristem and flowers are reported as important resources for many mammal as are the arthropod populations that shelter in palms and which are heavily used by insectivorous or omnivorous arboreal and scansorial mammals, particularly primates. The available literature on palm bruchids is limited to ecological studies at a few Neotropical sites (BCI, Maraca Island Ecological Reserve in northern Brazil, and a few bri studies in Costa Rica, Peru, and Bolivia.) Results and Discussion, Palm-Mammal Interactions General Trends ' At least 77 species of mammals in 44 genera were reported to feed on palm pulp or seeds from 59 species of palms (excluding potential synomyms and species identified genus only; Table 2-1). Five of these mammals species are from non-forested habitats, leaving a total of 72 mammals reported using palm fruits in Neotropical forests, either humid or dry. It is likely that many of the mammals listed

PAGE 18

in Table 2-1 will feed on palms whenever they co-occur with them, even though they have not been reported to do so in the literature (e.g., the larger marsupials are only described as eating palms at two sites, but they probably do so throughout their range ) . The five most frequently cited genera in Table 2-1 are reported from a limited number of study sites: Cebus (4 sp., 10 sites), Ateles (3 sp., 7 sites), Tavassu (2 sp., 6 sites), Sciurus (7 sp . , 7 sites), Dasvprocta (2-3 sp . , 4 sites). The best represented orders are primates (19 species) and rodents (29 species), followed by ungulates ( Perissodactyla , 2 species, Artiodactyla , 5 species; note that all Neotropical forest ungulate species are represented) , marsupials and carnivores (6 species each, but only 2 forest carnivores and 4 forest marsupials) . The Chiroptera are last with 2 species and Edentates are represented only by two species mentioned in two studies (one Edentate species in Peru was not identified (Allen 1997). Conspicuously absent from Table 2-1 are the large fruit bats, Artibeus sp . , which are reported only from Trinidad, and the smaller marsupials. In the case of bats, it is likely that their small body size and relatively weak jaws preclude them from using most palm species, as they would be enable to carry away large fruits or to tear away the resistant husks (J. Eisenberg pers . comm.). In the case of marsupials, they are probably absent from consumer lists because their nocturnal habits make them less susceptible to

PAGE 19

9 detection by researchers. Although Bodmer ' s (1989a, b, 1990, 1991) studies show that all ungulates rely heavily on palms, because ungulates are seldom studied, their interactions with palms in the rest of their range are not reported. In particular, they are not reported from French Guiana, despite the abundance of studies on smaller mammals there. Members of the Procyonidae probably feed on palms throughout their range. Rainforest canids , if studied, might be found to feed on fallen palm fruits. Canids in two nonforest sites (Brazilian cerrado and Sonoran desert) used palm fruits. Although the most cited order, rodents are underrepresented relative to their richness, abundance, and f rugivorous/granivorous diet. Only half of the neotropical squirrels are reported, one of three Heteromvs . and a few of the Echimydae. The agouti ( Dasvprocta sp.), despite its frequency in Central and South America, is reported only from French Guiana, BCI, Maraca, northeast Peru and Belize. In a captive study, Adler (1995) found that Proechimvs semispinosus in Central America ate the pulp and seeds of all palm fruits offered, including seeds with very thick endocarps such as Attalea . However, Emmons (pers. comm.) believes that due to their weak jaw structure Proechvmis rats are not likely to be significant predators on Attalea palms. Primates, on the other hand, are well represented, probably due to the large number of detailed studies that have been carried out on their behavior and social structure. At the same time, primates alone among the species listed

PAGE 20

• 10 have the hand and mouth dexterity necessary to open palm nuts and exploit both the mature seeds and the liquid endosperm. This is illustrated by Izawa's (1979) report of Cebus aoella using bamboo sticks to beat and poke open palm fruits, and Struhsaker and Leland's (1977) description of palm nut smashing by the same species. Evidence for Mammalian Dependence on Palms--Staple Food vs . Keystone Items Terborgh (1986a) defines keystone resources in Neotropical forests as species that are common in the diets of several members of a frugivore community during periods when little fruit is available, and proposes that such resources help structure Neotropical animal communities. This is a unique usage of the term, distinct from other definitions in the literature (Power et al . 1996, Gilbert 1980); throughout this paper I refer to "keystone species" sensu Terborgh. Based on the use of palm nuts by capuchin monkeys in Cocha Cashu, Peru during the season of food scarcity, Terborgh proposes palms as keystone food resources. He qualifies this proposal, however, by stating that keystone status would hold only "if other consumer species [besides capuchins] are relying on the same resources;" in that case, "the plants that produce these resources can be recognized as playing a critical role in the ecosystem (p. 337)." Here I review the frequency of palm nut versus pulp use during times of fruit scarcity by Cebus and other fruit-eaters.

PAGE 21

Pulp was the most frequently cited fruit part eaten by maitimals (96 citations, based on a simple count of the number of times each part is listed in Table 2-1) . Mature or old seeds followed with 64 citations, then immature seeds and entire fruit with 16 and 20 citations respectively. Fifty two references to palm eating by mammals did not mention the part eaten. Contrary to the emphasis given in the literature to palm seed-rodent interactions and to long-lasting palm nuts as a keystone resource (Terborgh 1986a, b), pulp may be a more important resource than seeds for the frugivore community. However, it is difficult to estimate the true importance of each fruit part, since the amounts of each part eaten are seldom given in the literature. Few studies of Neotropical rainforest mammals last long enough to cover seasonal changes in food use, let alone supra-annual variation. Even if the data are gathered, they are not always presented in a form from which it is easy to calculate the percent importance of a food item in the diet (e.g., Terborgh 1983). Table 2-2 summarizes the information from a few studies that examined dietary composition to some degree. For five species of ungulates ( Mazama americana . M. qouazoubira, Tavassu pecari . T. taiacu . Taoirus terrestris ) . Bodmer (1989a) derived the percent composition of the diet from stomach contents gathered over a 15 month study period by local hunters in northeast Peru. All species had at least one palm in the list of the top three diet items, and four of them (Tayassu taiacu was the only exception) had palms as the

PAGE 22

12 top two diet items. It seems that in this area all these species are dependent on palms, but the fact that peccaries at least do well in areas where no palms occur (e.g., Olmos 1993) suggests that they have a fairly flexible diet and that in other areas they may have other major resources. Throughout Amazonia, however, reports from other researchers suggest that palms are important at least for peccaries (Anderson 1983, Fragoso 1993). In her study of Ateles belzebuth belzebuth on Maraca Island, Roraima, northern Brazil, Nunes (1992, 1995) found that palms were not an important diet item for this species. One exception was Attalea marioa . which during one of the months of fruit scarcity constituted 10.1% of the diet, but it represented only 1.23% of the yearly diet. Even though the fruit was consumed at a time of fruit low, it was the pulp rather than the mature or old endosperm that was consumed. Roosmalen (1980) found an even lower percent importance of palm fruits in the diet of another spider monkey, Ateles paniscus . during a two year study in Suriname. On the otherhand, Castellanos and Chanin (1996) found that A. belzebuth in southern Venezuela relied heavily on palms and used them throughout the year in proportion to their abundance . ' ' In general palm use by Ateles is low when compared to palm use by Cebus . Buckley's (1983) study of Cebus capucinus , where one palm species constituted 58.2% of the diet in October and was the highest ranking diet item for the

PAGE 23

13 entire year. It was eaten during seven months of the year, ^ not just during the season of fruit scarcity. This last figure is comparable to the 64% of diet reported for three months for Cebus aoella in Manu, and 56% for Cebus albif rons in the same period (Terborgh 1983). There may thus be a real difference in the degree of palm use by spider vs. capuchin monkeys, but a comparison needs to be made controlling for the abundance of palms in the habitat. Izawa (1979) comments that palm fruits were usually available for a longer period of time (more asynchrony) than other fruit species, and that in addition primates fed on them before, during, and after they ripened, which increased their period of availability. Primates' ability to use the liquid immature endosperm seems to be one of the traits that makes palms an important resource for them. Observations such as these highlight the importance of clearly documenting the availability of--and distinguishing between--dif f erent maturity stages of fruits in phenological studies. Several studies report that Cebus in other parts of their range turn to palm nuts during the season of scarcity. U. Bobadilla (pers. comm.) documented Chiropotes sp. and £ebus sp. feeding on the pulp of Attalea soeciosa (as Orbiqnya phalerata) in the rain forests of western Para, Brazil, during the period of forest wide fruit low. This case is unusual because the palm fruited during the dry rather than the wet season; similar phenological behavior is reported for Attalea phalerata in the Beni of Bolivia (D.

PAGE 24

14 Quiroga pers . comm.), whereas Attalea palms at study sites north of the Equator tend to fruit in the wet season, generally the peak of forest wide fruit availability. That is the time v;hen Ateles belzebuth fed heavily on Attalea maripa pulp in southern Venezuela (Castellanos and Chanin 1996) . Bobadilla suggests that Attalea speciosa may be an asynchronous fruiter at his high rainfall site, but this may be an artifact of low palm densities. (Even when a plant population has a well-marked peak in fruit production, there often are some individuals that fruit outside of the peak. When a plant population occurs at low density, non-peak fruiters may be as well represented in the population or in the samples as peak time fruiters.) Jessenia ( Oenocarous ) bataua in the southwestern Amazon (Peres 1994a) produces fruits primarily in the dry season, when other fruits are not available at this site, but also produces some fruits year round. Among the primates, Cebus apella relies heavily on the immature seeds during this time and Peres (1994b) suggests that the palm acts as a key food item. However, he also indicates that Cebus as well as Pithecia and Laaothrix use the immature seeds throughout the year. In Spironelo's (1991) study, palms were also used most frequently in. the dry season. Jessenia pulp was used in the dry season, which was also the season of fruit scarcity at that site, while immature seeds were used during the wet season. In Roraima, in the northern Brazilian Amazon, Astrocaryum aculeatum also fruits in the mid-dry season, when

PAGE 25

fleshy fruit production is low (Silvius, pers . obs . ) , but the species is an unpredictable fruiter (Fragoso unpublished data) . , . . In all other cases where palms are used during the period of scarcity, it is the palm seeds that are involved, because they remain available on the ground past the time of low food availability (Terborgh 1986a for Cebus, Smythe et al. 1996 for agoutis, Chapter 4 for agoutis and white-lipped peccaries ) . . . . It is clear from Izawa's (1979) and Defler's (1979) studies in Colombia that many primate species seek out the immature and still liquid endosperm during the time that it is abundant and also feed on the pulp of the palms. Struhsaker and Leland (1977) report on another specialized behavior of Cebus , in this case beating the nuts (endocarps) of Astrocarvum chambira against tree trunks, branches, or other nuts, apparently in an attempt to obtain the endocarp contents. I have observed this same behavior by Cebus nigrivitatus on .^astrocarvum aculeatum in northern Brazil, but in this case the primates were smashing large preripe intact fruits rather than nuts, and the purpose of the knocking appeared to be to make the hard endosperm easier to remove, because fresh endocarps would be nearly impossible to open by banging (infestation by bruchid beetles, which makes the seeds easier to crack, occurs only on the ground) . Norconk et al . (1998) show that among Neotropical primate seed predators, only Cebus has been reported to feed on palm

PAGE 26

16 endocarp contents. It thus appears that Cebus spp . across their range rely heavily on palms throughout the year, and so to some extent may be considered dependent on palms. This review suggests that a) the use of palms either as staples or as "keystones" sensu Terborgh may be unique to the genus Cebus and b) that it is difficult to determine whether the fruits and seeds are being used as staples or "keystones." This does not diminish their importance to mammalian fruit eaters, but it does challenge the idea that "keystone resources" sensu Terborgh structure tropical forest communities. The review also suggests that the importance of palms as low season resources is heavily contextand habitat-dependent, and will change with the time of fruiting of the palms and the relative availability of other preferred fruit . Squirrels appear to be the only true palm specialists in the Neotropics, using palms in large numbers throughout their range. Although they occasionally use the pulp (Table 2-1), most frequently they feed on seeds collected from the ground. On BCI, Astrocarvum standlevanum makes up 21% of the diet of Sciurus aranatensi s (Glanz et al . 1996; Table 2-2). At Los Tuxtlas, Mexico, two species of Sciurus destroy 20-60% of the fruits of the understory palm Astrocarvum mexicanum in low palm density areas and 75-80% in high palm density areas (Pinero and Sarukhan 1982). In Campinas state, Brazil, Syaqrus romanzof f iana alone made up 49% of the diet of Sciurus inqrami (study period unspecified; Galetti et al .

PAGE 27

17 1992) . Emmons (1984), based on an accumulation of observations, estimates that about three palm species ( Attalea and Astrocarvum groups) make up about 95% of the diet of the large squirrels S. spadiceus and S. ianiventris . Because squirrels do not occur at high densities in Amazonia and therefore are not frequently seen (Eisenberg pers. comm. ) , palm seed use by squirrels is not well documented in the literature. Furthermore, there are no detailed studies of squirrel ecology in the Neotropics (Emmons, pers. comm.). The potential effect of palm fruit availability on reproductive rates and presumably population sizes of squirrels is suggested by a comparison between BCI and a nearby (30 km away) mainland site (Glanz et al . 1996). On BCI, there are three peaks of squirrel reproduction each year, occurring throughout the time of staggered fruit production by Diptervx . Astrocarvum and Scheelea . Mortality for juveniles is highest for the last cohort. On the mainland site, where Diptervx is absent and Astrocarvum scarce, a single reproductive bout coincides with the period of Scheelea .fruit availability. There is no evidence of a similar matching at the supra-annual level. Results and Discussion : Fruit-Vertebrate-Bruchid Interactions The previous sections reviewed interactions between vertebrates and palm fruits. For a subset of Neotropical palms, however, these plant -vertebrate interactions intersect

PAGE 28

18 with the use of palm seeds by the palm-specific bruchid beetles (Family Bruchidae, Order Coleoptera) . Adult female beetles lay their eggs on the endocarp or mesocarp of fallen fruit, usually at night and usually but not always once the fruits are on the forest floor. After approximately 7 days, the first instar larvae hatch and tunnel through the endocarp to reach the endosperm. The larvae then feed on the endosperm and grow inside the protection of the palm endocarp for approximately two months, finally pupating and emerging as adults through an exit hole carved by the last larval instar. Although several palm families are parasitized in this way, the most intensive parasitism seems to occur among cocosoid palms in the subfamily Arecoideae, which includes the genera Attalea and Astrocarvum . Bruchid-palm interactions are relevant to the above review of palm use by mammals because there is mounting evidence that in many cases mammals feed on bruchid larvae rather than on endosperm when they crack open palm endocarps . The dynamics of larva availability are likely to be different from those of endosperm availability. Only a few authors have recorded vertebrate frugivores or omnivores feeding on bruchid beetle larvae (Robinson 1986, Fragoso 1994), although several have suspected it occurs (Janzen 1971, Strushaker and Leland 1977, Wright 1990). A captive white-lipped peccary offered larvae-infested Attalea maripa endocarps readily cracked all those that were not too thick and ingested the larvae (Fragoso 1994). Terborgh (1983) notes that

PAGE 29

individuals of Cebus albi f rons in Manu descend to the ground, pick among fallen Astrocarvum seeds for bruchid infested ones, crack them open, and pick out the remaining endosperm. Robinson (1986), on the other hand, noted that Cebus niqrivitattus in the Venezuelan Llanos picked up larvae infested Copernicia tectorum seeds from the ground and ate the larvae. Strushaker and Leland (1977) suspected that the Astrocarvum nuts cracked by Cebus apella in Colombia contained larvae, because most nuts collected on the ground were infested. On Maraca, agoutis, squirrels, and whitelipped peccaries fed on bruchid larvae from Attalea maripa seeds in the wild (Chapter 4). Terborgh (1983) believes that the Cebus chose infested seeds over uninfested ones because they were easier to open than solid seeds; this is probably also a factor on Maraca, especially for white-lipped peccaries . Humans in the Neotropics also use bruchid larvae from palm seeds both as food and as fish bait. They seek these larvae out from seeds that are usually too hard to eat, although they prefer the larger larvae obtained from woodboring beetles that lay their eggs in rotten palm trunks (for the Yanomami of Brazil and riberenos in Peru, Silvius, pers. obs.). Balee (1988) reports that the Arawate of the eastern Brazilian Amazon kept bruchid infested palm seeds in their homes to "raise" the larvae. They then either eat the larvae or extract oil which they use to soften bow strings.

PAGE 30

Research on the biology of palm bruchids falls into three areas: 1) taxonomy, with a recent revision by Nilsson and Johnson (1993); 2) biogeography , with examination of host ranges and overlaps by Delobel et al . (1995) and Johnson et al . (1995); and 3) ecology of plant animal interactions. In the latter area, research has focused on two palm groups: Astrocarvum palms with Carvoborus beetles, and Attalea palms with Speciomerus and Pachvmerus beetles. The two palm genera differ significantly in the thickness and texture of their pulp, and the strength and size of their endocarps . Janzen (1971) initiated the research sequence with Attalea palms, focusing on the effect of bruchids on palm demography, but also indicated that there would probably be interactions between beetles and both pulp eaters and seed eaters. Later work followed up on the bruchid-palm tree aspect of the interaction (Wilson and Janzen 1972, Bradford and Smith 1977, Wright 1983, 1990). Three-way interactions were always commented on as peripheral phenomena, but never explicitly studied (with the exception of one short term experiment by Delgado et al . (1997) with the Astrocarvum system. .„ ; Palms from the genera Attalea . Scheelea . Maximiliana. and Orbianva have recently all been grouped in the genus Attalea by Henderson (1995) and I use the revised taxonomy. The taxonomy of bruchid beetles has also recently been revised (Nilsson and Johnson 1993). The name changes of species frequently encountered in the literature are summarized in Table 2-3. •

PAGE 31

21 Most long-term ecological work on the palm-bruchid system has been carried out on Attalea butvraceaa (Janzen 1971, Wilson and Janzen 1972, Bradford and Smith 1977, Wright 1983, 1990) and Attalea marioa (Fragoso 1994, 1997), with two additional short term studies on Astrocarvum chambira (Delgado et al . 1997) and Attalea phalerata (Quiroga 1997). Attalea butvracea and A. marioa are very similar in reproductive biology and fruit form, with one potentially key difference: A. butvracea has endocarp fibers and sunken germination pores protected by these fibers, whereas A. maripa has only a woody endocarp. The fruits of A. butvracea are also slightly larger (4.5 to 8.5 cm long vs. 3.5 to 6.5 cm long; Henderson 1995). Taxonomv and Taxon-Soecif ic Interactions Out of 1,500 species of bruchid beetles, only 20 species in four genera (three species of Cavroborus . six species of Caryobruchus, seven species of Pachvmerus . and four species of Speciomerus ) are palm bruchids (tribe Pachymerini, subfamily Pachymerinae ) . All are endemic to the New World, and all but one are tropical (Nilsson and Johnson 1993, Delobel et al . 1995, Johnson et al . 1995). On a palm subfamily basis, bruchids are known to associate extensively (many species, many and substantiated records of high infestation rates) with tribes in the Arecoideae and Coryphoideae, and less extensively with a few species in the

PAGE 32

22 Ceroxyloideae, Phytelephantoideae , and Calamoideae (Johnson et al . 1995). Within the Arecoideae, the tribe Cocoeae represents the most intensive association for all Neotropical palms (59% of 22 genera infested). The subtribes Attaleinae (primarily the genus Attalea ) and Cocoae (primarily the genus .^strocarvum ) have the greatest number of reports of bruchid infestation (Johnson et al . 1995). These are the groups that combine large seeds with thick endocarps. The most specific association between a palm group and a beetle group is between Attalea palms and two species of beetles, Pachvmerus cardo and Speciomerus giaanteum . Fifteen out of 24 record for the species Pachvmerus cardo are from Attalea palms, as are all of nine records for Speciomerus (Johnson et al . 1995). Both species are widespread, ranging from Central America to the southern Amazon basin. Ecological Studies ' ; The palm-bruchid system is ideal for study of invertebrate-fruit interactions because eggs, entry holes, larvae, and exit holes are easily counted and endocarps are easily protected from vertebrate predation in the field by wire exclosures. All studies have used some variation of these methods. I evaluate the results of these studies in some detail because misinterpretation of data in some papers has led to a perpetuation of erroneus conclusions in subsequent publications (c.f. Henderson 1995).

PAGE 33

23 Janzen (1971, 1972) described the interactions between Scheelea burvracea . Speciomerus giaanteus , and Pachvmerus sp. in lowland moist forest habitat of Costa Rica. In this system, beetles lay their eggs on the endocarp only after the husk and pulp have been almost completely removed by rodents or by rotting. Egg laying intensity is low, with 2-5 eggs found per seed after prolonged exposure to oviposition. Janzen reports a development time of 3 to 5 months; that is, beetles begin to emerge from the seeds after this amount of time. Larval infestation rates were greater than 77% for seeds remaining in the original fruit pile by the parent tree. Fruit removal by frugivores and endocarp removal by granivores appears to be low, based on the size of the fruit piles described as remaining by the parent tree at the end of the fruiting season. A key observation by Janzen was that by the time rodents opened the endocarps under the parent tree, most of them contained full-grown larvae. If the endocarp contains multiple seeds, they are usually all killed by a single larva. Janzen also found high larval, pupal, and adult mortality by an "unknown microbial disease, " which resulted in adult emergence rates much lower than infestation rates . ' . ••; . Work by Wilson and Janzen (1972) with the same system in Costa Rica again highlights a long development time for the larvae ("most are more than half grown" after 4.5 months in the field) . Infestation rate was low (35% of experimental seeds). Bruchids had difficulty finding endocarps placed

PAGE 34

individually rather than in clumps on the forest floor. The authors suggest that adults emerge from the endocarps after a rainfall cue at the start of the next palm fruiting season. Although they timed development and emergence for two beetles at 66 and 43 days respectively, the authors suggest that the larvae remain within the endocarps through the dry season following their laying season, and emerge only at the start of the next rainy (= fruiting) season. However, they also suggests that beetles remain in the endocarp in the adult form, so it is unclear which effect is predominant, or if the two terms, larvae and beetle, were inadvertently confused in the paper. In this study, only Speciomerus was present, whereas in the previous study, trees infested by two species of beetles were included in the sample of infested seeds. Bradford and Smith (1977) compared endocarps from a small sample of Attalea burvracea trees in Panama and Costa Rica, both presumably exposed only to predation by Speciomerus aiaanteus. They concluded that rodent predation was high on BCI and low in Costa Rica, whereas the opposite was true for bruchid predation. This is an erroneous conclusion, because they classified any rodent-opened endocarp as preyed on only by rodents, disregarding that fact that most endocarps opened by rodents probably had a larva in them, which the rodent would have removed. They do not take seriously Janzen's suggestion that rodents are encountering larvae more frequently than endosperm, stating instead that rodents "possibly" feed on the larvae. They also concluded

PAGE 35

25 that seeds from multiple-seed endocarps had a higher probability of escaping predation than those from single seed endocarps. However, individual trees tend to produce a typical number of seeds per endocarp, and individual trees may also differ strongly in both the degree of bruchid infestation and rodent predation (Chapters 4 and 5) . Since Bradford and Smith (1977) only sampled five trees, they could have had a strong tree effect in their study. Based on differential gnawing patterns at the two sites, the same study also suggested that there is an evolutionarily based correlation between the number of seeds per endocarp and the way in which a single rodent species opens the endocarps. This is based on the assumption that the same rodent species were opening the endocarps at both sites. However, the gnawing patterns they describe could also be interpreted as being caused by different rodent species: on BCI the five trees they sampled appear to have been fed on only by squirrels, whereas in Costa Rica there was also predation by agoutis. Wright (1983) continued work with the A. butvraceae -S . qiqanteus system on BCI. He again reports a low ovipostion rate (1-5 eggs, with a rare extreme of 11 eggs after 48 hours of exposure) . He found that there was high mortality of either the eggs or the first instar larvae, because only 47 out of 127 endocarps with eggs developed larvae, and many of those without larvae had entry holes that indicated a first instar larvae had at least initiated tunneling through the

PAGE 36

endocarp but then failed to develop. He also found that females could distinguish between endocarps with no eggs and those with eggs, and preferred to lay on endocarps with no previous eggs. . .' In a later study at the same site, Wright (1990) measured an interval of 5 days from ovipositon to hatching, and of 6 months from ovipostion to adult emergence. He notes that some adults, rather than larvae, remain inside the endocarp until the next wet season. The intensity of ovipositon, low to start with (0.91 eggs per endocarp), declined during the fruiting season, and stopped altogether 2 months before the end of the fruiting season. He concludes that this decline is due to a lack of recruitment into the beetle population during one fruiting season: beetles have a long development time and they emerge at the start of each fruiting season. Like Janzen, he notes that rodents often consume seeds that were first infested by bruchids, but still classifies these as rodent-destroyed, even though the bruchid would undoubtedly have killed the seed in the absence of rodents. Working with the Attalea maripa Pachvmerus cardo system in northern Brazil, Fragoso (1994, 1997) found that seeds ingested by tapirs (as fruit) and subsequently defecated at upland latrines had a much higher chance of surviving bruchid infestation than seeds remaining at the parent tree. This survival translated to a higher seedling density at latrine sites than at parent trees or at random locations in the

PAGE 37

27 forest. He also found that bruchids are able to penetrate thicker endocarps than rodents. Like Bradford and Smith (1977), he found that endocarps with multiple seeds were more likely to have at least one seed survive beetle or rodent predation than single-seeded endocarps. However, because single seeded endocarps are thicker than multiple seeded ones, they are more likely to survive attack by white-lipped peccaries, which, unlike rodents and bruchids, cannot penetrate the thickest endocarps. Therefore, white-lipped peccaries could be maintaining selection for thick, single seeded endocarps, which appears to be opposite to the selection pressure for multiple seeded endocarps maintained by bruchids * . . Astrocarvum-Bruchid System As for the Attalea-bruchid system, much of the work with Astrocarvum has focused on the role of vertebrate dispersers in protecting seeds from bruchids by burial (Smythe 1989) and on the role of bruchids in distance and density dependent seed predation (Terborgh et al . 1993). Terborgh et al . (1993) documented greater infestation rates near the parent tree than 25 m away for an unknown beetle on Astrocarvum macrocalyx in Peru, when rodents and other vertebrates were excluded. Delgado et al. (1995) document bruchid predation rates on fruits not affected by vertebrate consumers and find more

PAGE 38

28 rapid oviposition on fruits artificially cleaned of pulp and husk. They report multiple larvae per seed, and rejection of some fruits altogether by ovipositing females. Smythe (1989) found high mortality from bruchids for peeled, unburied Astrocarvum standlevanum seeds on BCI, a result to be expected in light of the protection offered by the husk and pulp (Janzen 1971, Wright 1983, Delgado et al . 1995, Chapter 3> • BruchidInvertebrate Interact ions Interactions between bruchids and invertebrate parasites are more rarely recorded than those between vertebrates and bruchids, although parasites can cause significant mortality for bruchid larvae and have seasonal patterns that can affect the evolution of both bruchid beetles and palm phenology (Fragoso and Silvius unpublished data) . Quicke and Delobel (1995) described a parasitic wasp species that attacks the late instar larve and pupae of Cavoborus serripes on several species of Astrocarvum palms. Parasitism rates were as high as 20%. The same bruchid species has its eggs parasitized by a trichogrammatid wasp (Delobel et al . 1995), as does Pachypierus cardo on Attalea palms, where egg parasitism rates were as high as 92% (Delobel et al . 1995). Janzen (1971) reports an unknown parasite on Soeciomerus qiganteus eggs in Costa Rica, as well as an unknown agent of mortality, which he refers to as a pathogen, that kills larvae and pupae.

PAGE 39

Given the widespread nature of the parasitism, the high mortality rates recorded, and the probable specificity of the relationship, invertebrate predators and/or parasites on bruchids are likely to be as important in bruchid beetle ecology and evolution as vertebrates. For example, bruchid beetles might chose seeds that offer escape from parasites rather than from rodents. Summary and Conclusions The issue of the relationship between fruit availability and frugivore species richness and abundance is undeniably a large one that needs to "be tackled in bits and pieces, or by many researchers working together. However, by identifying one subset of frugivores and fruits that show a more or less tight relationship, it may be possible to develop methodologies and theoretical approaches that can then be applied to the larger community of fruits and frugivores. _ The review presented here suggests that several Neotropical forest species, namely white-lipped peccaries, tapirs, Cebus monkeys, agoutis, and most importantly squirrels, depend to varying degrees on palm fruit availability. Squirrels, and to a lesser degree capuchin monkeys, are specialized on palm use. The use of palms as a season of scarcity resource (Terborgh 1986a, b) may be restricted to members of the genus Cebus . By Terborgh ' s own

PAGE 40

30 definition, this would then no longer constitute a case of keystone resource use. In palm species that have ecological and evolutionary interactions with bruchid beetles, it is unclear and unstudied whether frugivores and granivores are feeding on fruit matter or insect matter, and what the effect of these interactions is on palm and frugivore ecology. The Attalea bruchid system is the best understood. Several studies indicate that beetles cause significant mortality for palms, and that both vertebrate predators and invertebrate parasites cause significant mortality for beetles. There is some evidence that seeds dispersed long distances from the parent tree experience reduced beetle predation.

PAGE 41

Table 2-1: Results of literature review of palm use by Neotropical mammals. Latin binomials for palms and mammals are those used by the original authors; note that all palms in the genera Attalea . Maximiliana . Orbignya and Scheelea are currently placed in the genus Attalea . imm = immature, mat = mature, germ = germinating, fruit author did not distinguish between part eaten, endocarp contents = either endosperm or bruchid larvae or both. Palm Species Mammal Species Part Site Ref Mauri tia flexuosa Mazama americana pulp Maraca , Brazil 17 Tayassu pecari .1 preripe seed, germ, seed 19 Tayassu tajacu germ. seed 19 Tapirus terrestris pulp 18 Ateles helzebuth mat pulp Maraca , Brazil 33 Mazama americana N.E. Peru 6 Mazama gouazouhira 6 Tayassu tajacu 6 Tayassu pecari 6 Tapirus terrestris 6 Nectomys squamipes embryo from seed, pulp, discards endo sperm Mato Grosso , Brazil 40 Oxymycteris roberti pulp, caches seeds 40 01 igorysomys sp. pulp, seed coat 40 Didelphis albiventris pulp 40

PAGE 42

32 Table 2-1, continued Palm Species Mammal Species Part Site Ref Tapirus cerrestris pulp 40 Agouti paca pulp seed (holds with fore paws ) 40 Chrysochyon brachyurus pulp 40 Dasyprocta ful igmosa . N. E . Peru 1 SrCiall rodents 1 Agouti paca 1 Eira bar bar a 1 Tapirus t errestri s 1 Mazama americana 1 marsupials 1 edentates 1 Marmosops noctivagus N.E. Peru 52 Marmosops sp. 52 Nectomys squamipes 52 Oecomys trini tatis 52 Oryzomvs sp. 52 Oryzomys yunganus 52 Proechimys sp. 52 Proechimys steereei 52 Attalea maripa Sciurus spadiceus imm. seed mat . pulp Rio Urucu , Brazil 38 Ateles belzebuth mature pulp Bolivar, Venezuela 11 Cebus apella mat pulp Amazonas Brazil 47 Odocoileus vi rginianus pulp Maraca , Brazil 17

PAGE 43

33 Table 2-1, continued Palm Species Mammal Species Part Site Ref Tavassu caiacu pulp 43 Tayassu pecari pulp endocarp contents 17, 42 Ateles helzehuth mat pulp Maraca , Brazil 33 Alouatta seniculus pulp 43 Saiwiri sciurus pulp 43 Sci urus igniventris seed 42 Dasyprocta agouti pulp seed 42 , 43 Dasyprocta punctata French Guiana 28 Sciurus aestuans 28 Ateles belzebuth fruit Bolivar, Venezuela 11 Attalea regia Cebus albifrons pulp seed Eastern Colombia 14 Echimys armatus pulp French Guiana 12 Philander opossum pulp 12 Didelphis marusupialis pulp 12 Sciurus aestuans seed 12 Dasyprocta agouti seed 12 Agouti paca pulp 12 Small rodents pulp 12 Attalea phalerata Tapirus terrestris pulp Beni , Bolivia 27 Scheelea sp . Saimiri sciurus seed Cocha Cashu, Peru 49 Cebus apella seed 49 Cebus albifrons seed 49 Sci urus spadiceus seed 49 Scheelea sp . Sciurus igniventris seed Western Amazonia 15 Scheelea sp . Ateles paniscus chamek Cocha Cashu, Peru 48

PAGE 44

34 Table 2-1, continued Palm Species Mammal Species Part Site Ref Scheelea sp. Tapirus Cerrestris S. E. Peru 6 Scheelea rostrata Sci urus variegatoides pulp seed Costa Rica 1 Agouti paca pulp seed 7 Dasyprocta punctata pulp seed 7 Scheelea zonensis Nasua narica pulp BCI 55 Cebus capucinus pulp 34 Agouti paca pulp 7 Dasyprocta punctata seed pulp 45 Proechimys semispinosus seed pulp 7 Sciurus granatensis seed pulp 21, 7 Sciurus gerradi 55 Scheelea attaloides Cebus apella pulp Colombia 26 Orbignya wartiana Agouti paca pulp Maranhao Brazil 3 Dasyprocta punctata pulp 3 Coendu prehensilis pulp 3 Coendu sd. pulp 3 Proechimys longicaudata pulp 3 Mesomvs hisoidus pulp 3 Tayassu pecari pulp 3 Tayassu taiacu pulp 3 Actus trikingatus imm . seed 3 Cebus apella imm. seed 3 Orbi gnya phalerata Chiropotes sp. pulp Para , Brazil 5 Orbignya cohune Cebus capucinus seed Honduras 9 Astrocaryum murumuru Sci urus spadiceus imm seed Rio Urucu , Brazil 36, 37 Cebus apella pulp Cocha Cashu , Peru 13 Sciurus spp. seed 13

PAGE 45

35 Table 2-1, continued Palm Species Mammal Species Part Site Ref Da syprocta variegata seed 13 Proechimys spo . seed 13 Oryzomis sop. seed 13 Tavassu pecari seed 13 Astrocaryum sp . Mazama americana pulp N.E. Peru 6 Tayassu tajacu seed 6 Tayassu pecari seed 6 Tapirus terrestris 6 Astrocaryum aculeatuw Ateles belzebuth imm pulp Maraca , Brazil 33 Tayassu pecari seed 17 Tayassu tajacu pulp 43 Tapirus terrestris pulp 43 Cebus nigrivi tatus imm pulp 43 Dayprocta agouti pulp seed 43 Scirus iqniventris fruit 43 Astrocaryum sp. ( =murumuru ?) Cebus capucinus seed Cocha Cashu, Peru 49 Cebus albifrons seed 49 Sciurus spadiceus seed 49 Tayassu sp. seed 49 Astrocaryum chambira Cebus apella imm seed La Macarena Colombia 26 Ateles belzebuth imm seed 26 Lagothrix lagothrica imm seed 26 Cebus apella "nut" Colombia 46 Astrocaryum Sciurus aureogas ter seed Mexico 39 Sciurus deppei seed 39 Heteromys desmarestianus seed Belize 8 Astrocaryum standleyanum Nasua narica pulp imm fruit BCI 55

PAGE 46

36 Table 2-1, continued Palm Species Mammal Species Part Site Ref Agouti paca pulp 44, 45 Cebus caoucinus pulp 34 Ateles geoffroyi pulp 24 Dasyprocta punctata pulp 45 Sciurus granatensis seed 21 Di delphis marsupialis pulp 44 Phi lander opossum pulp 44 Caluromys derbianus pulp 44 Tayassu pecari seed 44 Tayassu tajacu seed 44 Heteromys desmarestianus seed 44 Proechimys semi spinosus seed 44, 2, 25 Tapirus bairdii 55 Astrocaryum vulgare Philander opossum pulp French Guiana 12 Caluromys philander pulp 12 Dasyprocta punctata seed 12 Coendu prehensilis pulp 12 Agouti paca pulp 12 small rodents pulp 12 Sciurus aestuans seed 12 Astrocaryum paramaca Sciurus aestuans French Guiana 28 Proechimys cuvieri 28 Astrocaryum sciophil um Sciurus aestuans French Guiana 28 Myoprocta exilis 28 Astrocaryum sp. Proechimys semispinosus Cocha Cashu , Peru 15 Astrocaryum sp Sciurus spadiceus seed Western Amazonia 15

PAGE 47

37 Table 2-1, continued Palm Species Mammal Species Part Site Ref Sci urus igniventris seed 15 Chamaedorea tepejilote Heteromys desmarest ianus seed Mexico 35 Agouti paca 55 Socratea exorrhiza Ateles belzehuth Colombia 31 Ateles helzehut fruit Bol ivar , Venezuela 11 small rodents seed Maraca , Brazil 43 Tayassu pecari seed Maraca , Brazil 43 Artibeus jamaicensis fruit Amazonia French Guiana , Trinidad 55, 12 Artibeus lituratus Amazonia French Guiana 55, 12 Socratea durissima Dasyprocta punctata BCI 45 Sciurus granatensis 21 Cebus capucinus pulp 34 Ateles geoffroyi mat pulp 24 Alouatta palliata pulp 32 Proechymi s semispinosus 55 Myoprocta exilis French Guiana 12 Heteromys desmarest ianus seed Costa Rica 66 Socratea sp. Tayassu tajacu Cocha Cashu Peru 29 Tayassu pecari 29 joczaLsa sp , Ateles paniscus chamek pulp Cocha Cashu Peru 48 Socratea e lagans Cebus apella imm . seed Colombia 26 Ateles belzebuth imm. seed Colombia 26

PAGE 48

38 Table 2-1, continued Palm Species Mammal Species Part Site Ref Iriartea ventricosa Ateles belzebuth Colombia 31 Tayassu pecari seed ?pulp Cocha Cashu, Peru 30 Cebus apella 49 Saguinus imverator 49 Saguinus fuscicolis 49 Callicebus mo loch 49 Ateles paniscus chamek pulp seed 48 Proechimys semispinosus seed 15 Iriartea sp. Mazama americana seed pulp N.E. Peru 6 Mazama gouazoubira seed pulp 6 Tayassu tajacu fruit 6 Tayssu pecari fruit 6 Wettinia maynensis Saguinus sp . 55 Euterpe sp. Ateles belzebuth Colombia 31 Euterpe sp . Ateles paniscus pulp Cocha Cashu, Peru 48 Euterpe precatoria Cebus apella mat pulp Colombia 26 Ateles belzebuth fruit Maraca , Brazil 43 Oenocarus bacaba Ateles belzebuth mat . f rui t Maraca , Brazil 33 Oenocarpus mapora Cebus apella imm seed, mat pulp Colombia 26 Oenocarpus Sci urus yx anac en si s BCI . 23 Cebus capucinus 34 Oenocarpus sp. Ateles belzebuth Colombia 31 Jessenia bataua Cebus apella imm seed Rio Urucu , Brazil 37 Cebus albifrons 37

PAGE 49

39 Table 2-1, continued Palm Species Mammal Species Part Site Ref Pi thecia albicans imm seeds , mat pulp 37 Lagothrix lagrothrica cana 37 Pithecia sp . mat pulp Para , Brazil 5 Tayassu tajacu pulp N.E. Peru 6 Tayassu pecari pulp 6 Mazama americana pulp 6 Tapirus terrestris pulp 6 Pi thecia monachus 55 Cehus albifrons pulp seed Eastern Colombia 14 Cebus apella imm seed mat seed mat pulp Colombia 26 Ateles belzebuth imm seed mat seed mat pulp 26 Ateles belzebuth fruit Bolivar , Venezuela 11 Cebus nigrivi tatus Amazonia 28 Sciurus aestuans French guiana 28 Roystonea oleracea Artibeus lituratus fruit Trinidad 55 Syiagrus orinocensis Cebus albifrons pulp seed Eastern Colombia 14 Syagrus ynwr^ r) '7nn "F f i Cebus apella mat pulp Campinas Brazil 20 Alouatta fusca mat pulp 20 Sciurus ingrami imm seed old seed 20 Acroco/nia sp. Dasyprocta sp. Central Brazil 55 Cerdocyon thous 55

PAGE 50

40 Table 2-1, continued Palm Species Mammal Species Part Site Ref Acrocomia aculeata Didelphis albiventris dry upland habitat , Brazil 55 Nectomys squamipes 55 Cebus apella 55 Agouti paca 55 Euphractes sexcintus 55 Dasyprocta sp. 55 Acrocomia vinifera Si gmodon hispidus pulp Costa Rica 4 Tayassu tajacu pulp seed 4 Bactris barronis Sciurus granatensis BCI 21 Bactris sp . Mazama americana N. E. Peru 6 Bactris sp. Artibeus jamaicensis triniatus fruit Trinidad 22 Al tibeus lituratus palmarum fruit 22 Bactris sp. Ceus albifrons pulp seed Eastern Colombia 14 Bactris gassipaes He t eromys desmiarestianus pulp Costa Rica 53 Hoplomys gymnurus pulp 53 Dasyprocta punctata pulp 53 Bactris sp. Saguinus fuscicollis mat fruit Rio Urucu , Brazil 36 Bactris sp. Cebus capucinus fruit Honduras 9 Paurotis wriahtii Cebus capucinus fruit Honduras 9 Desmoncus i s thm i \i <^ Cebus capucinus pulp BCI 34 Welfia georgii Potos flavus Costa Rica 54 Sciurus sp. 54 Ceus capucinus 54 Dasyprocta 1 punctata | 54

PAGE 51

41 Table 2-1, continued Palm Species Mammal Species Part Site Re f Heteromys desmarestianus 54 Haplomys gymnurus 54 Proechymis semispinous 54 Agouti paca? 54 Tayassu pecari ? 54 Phytlephas so. Dasyprocta sp. 55 Ammandra so. Dasvprocta sp. 55 Washingtonia filifera Canis lastrans pulp Sonora 10 Urocyon ciner oar gent atus pulp 10 Copernicia tectorum Cebus nigrivitatus imm fruit germ seeds endocarp contents Llanos , Venezuela 41 Areca lutescens Phyllostomatidae 22 Coccotkrinax so. Phyl lost oma tidae 22 1-Allen 1997; 2-Adler 1995; 3-Anderson 1983; 4-Baker 1983; 5Bobadilla, pers . comm.; 6-Bodmer 1989; 7-Bradford and Smith 1977; 8-Brewer and Rejmanek 1998; 9-Buckley 1983; 10-Bullock 1980; 11-Castellanos and Chanin 1996; 12-Charles Dominique et al. 1981; 13-Cintra and Horna 1997; 14-Defler 1979; 15-Emmons 1982; 16-Fleming 1974; 17-Fragoso, pers. comm; 18-Fragoso 1997; 19-Fragoso 1994; 20-Galeti et al . 1992; 21-Glanz et al . 1996; 22-Gardner 1977; 23-Heany and Thorington 1978 24-Hladik and Hladik 1969; 25-Hoch and Adler 1997; 26-Izawa 1979; 27-Quiroga 1997; 28-Kahn and de Granville 1992; 29Kiltie 1981; 30-Kiltie and Terborgh 1983; 31-Klein 1972; 32Milton 1980; 33-Nunes 1992; 34 -Oppenheimer 1996; 35-Oyama 1991; 36-Peres 1993b; 37-Peres 1994a; 38-Peres 1994b; 39Pinero et al . 1984; 40-Prada 1994; 41-Robinson 1986; 42Silvius, this study; 43-Silvius pers. obs . ; 44-Smythe 1989; 45-Smythe et al . 1996; 46-Spironelo 1991; 47-Struhsaker and Leland 1977; 48-Symington 1987; 49-Terborgh 1983; 50-Terborgh 1986a; 51-Terborgh 1986b; 52-Valqui unpublished data; 53Vandermeer 1983; 54 -Vandermeer et al . 1979; 55-Zona and Henderson 1989.

PAGE 52

42 Table 2-2. Palm-eating mammals for which the percent composition of the diet has been estimated. Mamma 1 Species Palm Species % of total or annual diet Highest monthly % Reference Cebus apella Jessenia ( Oenocarpus ) bataua and Attalea maripa primarily, and some Oenocarpus bacaba, Astrocaryum mumbaca , Mauri tia fl exuosa, and Syagrus inajai (data combined) 17.6 5 3 Spironelo 1991 1-year study Jessenia bataua 24.5 Attalea maripa 28.7 Astrocaryum mumbaca 3 . 5 Ateles belzehuth belzebuth Attalea maripa 1.23 10.1 Nunes 19 92 2 -year study Astrocaryum aculeatum 0 . 18 Oenocarpus bacaba 0.09 Mauri tia flexuosa 0 .09 Mazama americana Astrocaryum sp. 3.4 (5) Bodmer 1989*, 15 month study Bactris sp. 9 . 1 Euterpe sp. 43.6 (1) Iriartea sp. 3 4.4 ( 2 ) Jessenia ( Oenocarpus ) bataua 1 . 1 Mauritia fl exuosa 9 . 1 Mazama gouazoubira Euterpe sp. 48.0 (1)

PAGE 53

43 Table 2-2, continued Mamma 1 Species Palm Species % of total or annual diet Highest monthly % Reference Iriartea sp. 36.0 (2 ) Mauri tia fl exuosa 8.0 Tayassu ta jacu Astrocaryum sp. 45.3 (3) Eutaroe sc. 7 . 5 Iriartea sp. 27.7 Jessenia ( Oenocarpus ) bataua 46.2 (2) Mauri tia fl exuosa 11.7 Tayassu pecari Astrocaryum so . 47 . 7 Euterpe sp. 15 . 9 Iriartea sp. 75.0 (1) Jessenia (Oenocarpus ) bataua 59.0 (2) Mauri tia fl e.xuosa 45.0 Tapirus terrestris Astrocaryum sp. 5.3 Jessenia (Oenocarpus ) bataua 23.7 (2) Mauri tia fl exuosa 76.3 (1) Attalea sp . 13.2 Dasyprocta punctata Astrocaryum standleyanum 22 .7 Smythe et al. 1996, 1year study, 326 feeding observations Attalea butyracea 11.04 Oenocarpus panamus 0.31 Socratea durissima 0.31

PAGE 54

Table 2-2, continued Mamma 1 Species Palm Spec ies % of total or annual diet Highest monthly % Reference Agouti paca Astrocaryum standleyanum 6.7 Smythe et al. 1986, 1year study, 15 feeding observations Sciurus granatensis Astrocaryum standleyanum 20.92 Glanz et al . 1986, 2-year study Bactris barronis 0.05 Bactris ma J or 0.05 . . Oenocarpus panamus 0 . 68 Attalea butyracea 24 . 62 Socratea durissima 0 . 05 Cehus capucinus Attalea cohune rank 1 14.3-58.2 May-October Buckley 1983, 17 month study Sciurus ingrami Syagrus romanzoffiana 49.0 Galetti et al. 1992 Cehus apella Syagrus romanzoffiana 2 . 16 Alouatta fusca Syagrus romanzoffiana 0.25 Ateles bel zebu th Euterpe sp. 1.0 61 . 0 Klein 1972, 1-year study Socratea exorrhiza 1 + 1 + Iriartea ventricosa 2.0 23.0 Oenocarpus sp. 1 + 22 . 0 Saguinus fuscicolis Bactris sp. 0.03 Peres 1993a, 2 0 -month study Pi thecia albicans Mauri tia fl exuosa 2.3 Peres 1993b, 20-month study

PAGE 55

Table 2-2, continued 45 Mamma 1 Species Palm Spec ies % of total or annual diet Highest monthly % Reference Ateles paniscus Palmae 0 . 9 Roosmalen 1980; 2-year studv • ' Euterpe oleracea 0 . 3 Attalea regia ( ?=maripa ) 0 . 3 Oenocarpus bacaha 0 . 2 Euterpe ol eracea 0 .05 Sci urus spadiceus , Sciurus iqniventris Palmae 98.0 % Emmons 19 84 Sciurus granatensis Astrocaryum standleyanum 9.0 22 Heany and Thorington 1978; 4 months , rainy and dry seasons Oenocarpus panamus 1 7 Attal ea butyracea 27 86 L-cjjus apeij. a Palmae (5 sp) ' 64 from May to July Terborgh 1983, 12month study Cebus albifrons Palmae .. (5 sp) 56 from May t" Ci lTi 1 1 \/ <^ i_i X y Saimiri sciurus Palmae (3 sp) 15 from May to July Saguinus imperator Palmae (1 sp) < 1 from May to July Saguinus fuscicolis Palmae (1 sp) < 1 from May to July * In this study, % is the percent food item occurred, while in all percent of feeding observations, a defined sampling period, in whi For clarification, in Bodmer's st next to the percentage, the rank percent of total dry weight of al if the palm ranks between first, importance in the diet. of stomachs in which the the other studies, it is the either instantaneous or over ch that food item was eaten, udy I give in parentheses, in the diet based on the 1 stomach contents combined, second or third in

PAGE 56

46 Table 2-3. Revisions to palm and bruchid taxa referred to in the text . .ii'FOMER NAME CURRENT NAME REFERENCES Palms Maximi liana marica Attalea maripa Fragoso 1994, 1997 Scheelea zonensis Actalea butyracea Bradford and Smith 1977, Wright 1983, Wright 1990, Forget 1994 Scheelea rostrata Attalea butyracea Janzen 1971, Janzen and Wilson 1972, Bradford and Smith 1977 Orbignya phalerata Attalea speciosa Anderson 1983, Anderson et al . 1991 Bruchids Caryobruchus bucksii Speciomerus Qiqanteus Janzen 1971, Janzen and Wilson 1972 Unidentified Pachymerus cardo Fragoso 1994

PAGE 57

CHAPTER 3 PULP HANDLING BY MAMMALIAN FRUIT-EATERS AFFECTS PALM SEED PREDATION BY BRUCHID BEETLES ' Introduction Studies of frugivory systems (a fleshy-fruited plant population and its consumer set, Howe 1993) to date have focused primarily on the interactions between plants and vertebrate fruit-eaters (Estrada and Fleming 1986, Fleming and Estrada 1993). This bias has resulted in a dominance of vertebrate-based evolutionary scenarios for both fruit trait evolution and plant demography. Thus fruit traits are interpreted as evolving in response to the need to attract pulp eaters that disperse seeds to safe sites. A limited group or researchers, led by the early theoretical work of Herrera (1982) and Janzen (1977), has focused on the interactions. between fruits and invertebrates, including both seed eaters and pulp eaters. In this study I focus on interactions between pulp, insects, and vertebrates, leaving discussion of interactions involving seeds for the next chapter. Several studies show that invertebrate pulp consumers or the fungal spores they introduce into pulp can degrade pulp and make it unattractive to vertebrate pulp eaters that might • . " 47 .., .

PAGE 58

48 otherwise disperse the seeds (Jordano 1987, Borowicz 1988, Krishik et al . 1989, Buchloz and Levey 1990, Sallabanks and Courtney 1992, Cipollini and Stiles 1993). In other cases, invertebrates may enhance pulp quality for the vertebrate disperser/pulp eaters by providing protein, lipids, and other nutrients {Redford et al . 1984, Piper 1986, Valburg 1992a, b). The opposite effect, alteration of fruit quality by vertebrate pulp consumers for invertebrate fruit consumers, has seldom been examined, although it has been noted that some insect species preferentially lay their eggs into damaged areas of the pulp (Sallabanks and Courtney 1992). In such a situation, presumed dispersers who scar the pulp could have a negative effect on seed survival by increasing the likelihood of infestation by invertebrate seed eaters. Due to variation in body size and behavior, different species of fruit-eaters handle pulp in distinct ways, and remove varying amounts of pulp from the fruits they handle. Differences in the amount and pattern of pulp removal could affect seed predation rates by insects and ultimately plant demography. In the Neotropics, a diverse array of consumers eat the fruits of arborescent palms: primates, carnivores, rodents, ungulates, and marsupials among the mammals, psittacines, falconi formes , and icterids among the birds, and tortoises among the reptiles (Chapter 2). Several Neotropical palm species are also parasitized by about 20 species of bruchid beetles (Chapter 2). One of the most specific and widespread relationships is that between palms

PAGE 59

49 in the genus Attalea (including Maximiliana , Orbvania . and Scheelea; Henderson 1995) and the beetles Pachvmerus cardo and Speciomerus giaanteus . Interactions among these taxa have been documented from Costa Rica to Bolivia (Fragoso 1997b, Quiroga 1997, Johnson et al. 1995, Janzen 1971, Wilson and Janzen 1972 ) . In the palm-bruchid system, interactions between bruchids, vertebrate seed eaters, and nut walls (endocarps) have been examined in some detail (Fragoso 1994, Wright 1983, 1991, Bradford and Smith 1977). The fruit husk (exocarp) and the pulp (mesocarp) , however, are the first barriers bruchids encounter when attacking a seed, long before they reach the endocarp. Both affect the bruchids' ability to reach a seed. Janzen and Wright (Janzen 1970, Wright 1983, 1990) noted that bruchids do not lay eggs on endocarps of Attalea butvracea in Central America until the husk and most of the pulp has been removed by vertebrates or fungus. The same was true for an Astrocarvum palm in Peru whose pulp was experimentally removed (Delgado et al. 1997). This indicates that pulp itself acts as a deterrent to oviposition. Several researchers have suggested that pulp chemicals in plant families other than the Palmae may deter feeding by insects (reviewed in Cipollini and Levey 1997), but there is little consideration in the literature of pulp as a physical deterrent for seed-eating insects. The physical properties of pulp, however, including consistency and presence of mucilage or lipids, could impede pulp mining by insects. Extremely

PAGE 60

50 thick pulp could also increase mortality of larvae mining their way into the seeds. Traits associated with the attraction of seed dispersing mammals could thus also function as repellents to insects. Because pulp has evolved many times in the Angiosperm lineage, in some of these cases pulp could be evolutionarily molded by selection by insect seed predators as well as by dispersers. Given the extremely high rates of mortality caused by Coleoptera (especially Bruchidae and Curculionidae , Janzen 1982a, b, 1983, Herrera 1989, Sallabanks and Courtney 1992) in the tropics, selection for deterrence of beetle larvae should be strong. On Maraca Island Ecological Reserve, in the northern Brazilian Amazon, high rates of bruchid predation by Pachvmerus cardo have been documented for the common arborescent palm Attalea maripa (= Maximi liana maripa ) (Fragoso 1994, 1997b). The present study experimentally examines the effect of differential pulp handling by vertebrate consumers on a) pulp-dependent egg-laying choices by bruchid beetles that parasitize the seeds, and b) subsequent survival rate of seeds that received differing numbers of eggs in response to different pulp characteristics.

PAGE 61

51 i' . . ' study Area and Species . Study Area -» ; < • . 1 . ' -i • Maraca Island Ecological Reserve is a 1,100 km^ riverine island located on the forest side of the forest-savanna boundary in the northern Brazilian state of Roraima. The study site terrain is flat and low (elevation ranges from 110-160 meters above sea level.) The sandy soils are derived from quartz-biotite schists, quartz-feldspar gneisses, and tonalitic granites (Thompson et al . 1992). Rainfall records on the eastern end of the island show variation of between 1750 and 2,300 mm annually; during El Nino Southern Oscillation events, rainfall decreases markedly across the state. The wet season is between May and September, the dry season between October and April, with rains starting either in early or late April. The heaviest individual rains tend to fall in the dry season. Study Species . ' '"' ,. Attalea marioa is a canopy to subcanopy arborescent palm very common on the eastern end of Maraca island (Milliken and Ratter 1989). In the terra firme forest where this study took place, it occurs in dense aggregations (mean 20.2 trees per quarter hectare, range 12-32 trees; Fragoso 1997b). It reaches its highest densities and reproductive output at

PAGE 62

52 ecotones between the forest and either savannas, wetlands or rivers (Chapter 4), and occurs at low densities or singly in the forest interior. Individual inf loresences can be bisexual or male, and between one and three fruit bunches develop simultaneously or over the course of a six month fruiting season (April September) . Each fruit bunch can have between a few hundred and more than 3,000 fruits (two average-sized pre-ripe fruit bunches contained 1,183 and 2,455 fruits, respectively). Although a large proportion of reproductive age trees produce flowers, a much lower proportion develops fruits each year (pers. obs . , Fragoso unpublished data) . ; ,^ Individual fruits range from 3 to 6 cm in length, and are roughly egg shaped, with a narrow tip and broad base. The husk is about one mm thick, dry and resistant to tearing. The mesocarp is one centimeter thick, dense and slippery, with fibers near the endocarp but not on the outside. It is slightly sweet, soapy and astringent to the human taste, and clings to the mouth parts. Color ranges from pale yellow to bright orange among different trees, but is consistent within one tree. Endocarp thickness ranges from 1.5 to 6.5 mm. Each endocarp can contain one, two, or three seeds, but one seed is most common (Fragoso 1994). Fruits drop intact to the ground when they are ripe, but many are knocked down (prior to abscission) by primates and bird, often with the pulp only partly removed. The pulp is eaten by brocket deer (Mazama spp . ) , white-tailed deer ;

PAGE 63

( Odocoileus virainianus ) , collared and white-lipped peccaries ( Tavassu taiacu and T. pecari ) , agoutis ( Dasvprocta agouti ) , pacas ( Agouti paca ) , tapir ( Tapirus terrestris ) . capuchin ( Cebus nigrivitatus . C. apella ) , squirrel ( Saimiri sciureus) . howler ( Allouata seniculus) . and spider monkeys ( Ateles belzebuth ) . Amazon parrots ( Amazona spp.), macaws (Ara spp.), fruit caracaras ( Daptrius sp.), and oropendulas ( Gvmnostinops sp.). Among the invertebrates, members of the Coleoptera, Diptera and Orthoptera feed on the new pulp, and Dipterans and Dermapterans lay eggs in the pulp. The mature endosperm is eaten by bruchid beetles, squirrels ( Sciurus igniventris ) . agoutis ( Dasvprocta agouti ), unidentified small rodent(s), and white-lipped peccaries ( Tavassu pecari ) . The immature endosperm is occasionally eaten by an unidentified arboreal consumer that is able to penetrate the immature endocarp. The bruchid beetle Pachvmerus cardo occurs from Costa Rica to Bolivia. It most frequently parasitizes seeds of the genus Attalea, but has also been found on Acrocomia . Aiphanes, Areca, Bactris . Copernicia . Svagrus . and Elaeis , (Johnson et al . 1995). Like most bruchid beetles, it is nocturnal. Its adult biology is unstudied, and it is presumed to feed on pollen and nectar, though not necessarily of the same palm parasitized by its larvae. On Maraca, adult beetles lay eggs on the pulp or endocarp of the fruits once they have fallen to the ground (though at nearby open pasture sites they sometimes parasitize intact fruits on the tree; Fragoso and Silvius

PAGE 64

54 unpublished data) . After seven to 10 days the first instar larvae emerge from the eggs and drill through the endocarp into the seed. Development rates are variable: larvae feed on the endosperm and remain within the endocarp for a period of between two to nine months, and possibly longer, before pupating. Adults emerge from the endocarp at all times of the year. Both seed size and moisture availability during development appear to influence the size of the larvae and of the emerging beetle. Although many first instar larvae enter the endocarp, only one larvae develops per endosperm. In the case of multiple-seeded fruits, two or three larvae can develop in an endocarp, but usually a single larva eats all of the seeds. Bruchid eggs in this study were parasitized by an unidentified insect, and were also attacked by small ants. Late instar bruchid larvae are parasitized by a wasp (Hymenoptera: Braconidae) and by another, unidentified insect . '' Beetles in this study were identified by Dr. J. Kingsolver of the Division of Plant Industry, Florida Agriculture and Consumer Services. Only P. cardo was found during this study, but a few individuals of Speciomerus giqanteus were found in Attalea seeds at a forest site 10 kilometers away.

PAGE 65

Methods One 2-part experiment with 1,400 fruits and one single experiment with 100 fruits were carried out from June to December 1996. ^ Part I: Oviposition Preferences on Fruits with Different Amounts of Pulp The first part of the experiment tested the hypothesis that the amount of pulp remaining on an endocarp affects the quality of oviposition sites on the endocarp. Female bruchids should lay eggs preferentially on the higher quality "fruit types" and this preference can be measured by counting the number of eggs laid. Of several conditions in which fruits were found under the parent tree, five "types" were chosen as the most representative and distinct for this experiment. These are described below in order of increasing pulp removal, along with the method by which they were procured for the experiment . ' . . 1) Intact husk frni tc; . Fruits that fell to the ground naturally or were dropped by primates without being opened. For the experiment, these were easily obtained by collecting intact ripe fruits from the ground or tree.

PAGE 66

2) Intact pulp fruits . Fruits whose husk was removed byprimates, but either rejected or accidentally dropped before pulp removal. Capuchin monkeys often removed only a thin layer of pulp along v;ith the husk, and fruits eaten in this way resembled intact pulp fruits. Howler monkeys also removed only thin layers of pulp, but tended to leave narrow tooth marks in the pulp, whereas capuchins left wider, shallower scrapes. Spider monkeys, which swallow the fruits after squeezing them from the husk, dropped many intact, husk-free fruits. For the experiment. Intact Pulp fruits were obtained by peeling the husk away from intact fruits, taking care not to gash the pulp. , • 3) Gashed fruits . Squirrel monkeys, macaws, and several ungulate species removed the pulp in deep gashes that exposed the endocarp but left strips of endocarp surrounded by a thick bed of pulp. For the experiments, I artificially gashed intact pulp fruits by biting at the pulp and tearing it away in four locations per fruit. 4) Rodent fruits. Agoutis, and presumably pacas, removed the pulp in a corncob pattern, squeezing most of the juicy pulp out from the fibers but leaving a layer of moist fibers over the endocarp. Because Attalea pulp is a favored fruit of agoutis, many fresh (few hours after being eaten) rodentchewed fruits could be obtained from the forest floor for the experiments. Fresh fruits were recognizable because the

PAGE 67

57 remaining pulp was still moist, sticky, and shiny, retained its original color and odor, and the tooth marks were clearly visible in the stretched-up fibers. Older rodent eaten fruits began to ferment within a few hours of use, causing their color, odor, and fiber pattern to change. On several occasions I also set out intact fruits in areas were I knew agoutis were actively feeding on Attalea fruits, and collected the eaten fruits later in the day, .5J Bare endocarps : Ungulates occasionally removed all the pulp and fiber from endocarps (e.g., those endocarps spat out by tapirs) . Fruits of any of the above types that were left to rot under parent trees, and seeds defecated by spider monkeys or tapirs, also had little or no fiber on them. Bare endocarps were obtained by scraping away most of the pulp from intact fruits with a knife, then soaking them in water for a few hours or overnight to remove the remaining fiber. All fruits were collected and prepared the day before each experimental set up was placed in the forest, then kept in a refrigerator until the experiment was set up in the afternoon of the following day. If fruits were collected from several different trees, they were mixed together and then placed haphazardly into the different fruit categories. Because fruit size varies markedly among trees, when different sizes had to be used in a single set up, an attempt was made to partition fruit sizes equally among fruit types. All endocarps were measured after the experiment was complete

PAGE 68

, . • • 58 and One-Factor AlJOVA's were used to test for differences in size among fruit types within each season. There were no significant differences in endocarp size in any season (F = 0.63, p = 0.64 in June; F = 2.17, p = 0.0716 in July; and F = 1.77, p = 0.13 in Septemer) . . Fruits were protected from vertebrate pulp and seed eaters by galvanized wire exclosures ("chicken wire" with mesh size 1.5 x 1.5 cm) 5 cm in height and approximately 23 cm on four sides, with a closed top. These cages were effective in deterring vertebrates from feeding or stepping on the fruits, and provided protection from falling debris. Five exclosures, one for each fruit type and each containing ten fruits, were placed in a star pattern (with one corner touching) at the edge of a fruiting tree's natural seed shadow (1-4 meters from the base of the three) to prevent disturbance from falling fruits (Fig. 3-1). All fruiting trees used had been dropping fruit for at least several days, and had at least 100 naturally fallen fruits in the fruit pile. Fruits were separated by 1-3 cm from each other within each exclosure and were set directly on the natural leaf litter (in preliminary trials with cleared ground, soil disturbance attracted ants, which covered the fruits with dirt) . • The experiment was repeated three times at different stages of the palm fruiting season: 10 set ups in June (early peak), 10 in July (late peak), and 8 in September (tail-end of season) . Thus the sample size for each fruit type in each

PAGE 69

season was 10 replicates (8 in September) of 10 fruits each (Fig. 3-1). Because few trees at the same fruiting stage were available at any one time, trees were used as they became available. Sometimes several trees could be set on one day, and sometimes only one. All trees were within 1.5 km of each other, and had several other Attalea trees within 50 meters. All fruits v;ere checked twice after each experiment was set up: once the on morning following the set up, so that beetles had only one night to lay eggs, and again 7 days later, when the beetles had seven nights for egg laying. For the first five set ups in June, an attempt was made to count eggs after 14 nights of exposure, but was abandoned because hatching larvae and eggs slid off easily from the rotting pulp when handled. Only the eggs on Intact Pulp fruits could be successfully counted after 14 days. At each count, endocarps were picked up, all eggs counted, and the endocarps returned to their original position. For endocarps with a lot of pulp, the entire fruit was palpated carefully to feel for eggs hidden in the pulp. On Intact Husk fruits, the husk was not lifted up after it started to rot, to avoid exposing more pulp to beetles, and therefore some eggs may have been missed. Eggs on the other fruits were easy to locate. Intact Husks, but not the other fruit types, were examined weekly for up to nine weeks following set up. The condition of the husk, the number of eggs, and the location of the eggs (on husk or pulp) were noted.

PAGE 70

. ' ' ; ^ 60 Data Analysis, Part I ' • . Because of the large number of zeros for non-preferred fruit types, the data were summarized as the mean number of eggs per seed for each tree x seed type combination, yielding 10 seed type replicates in June and July, and 8 in September. Because the patterns in seed preference were the same for the first and second counts, even though the total number of eggs was lower after one night, only the second count was analyzed. Because Intact Husk fruits received no eggs by the second egg count, data on this fruit type were analyzed separately from data on the other four fruit types. Timing of oviposition on Intact Husk fruits is described in detail, as this is the main factor bearing on seed survival for this fruit type. The number of eggs per fruit within each of the remaining four fruit types was not normally distributed, and variance was greater in June than in July and September. The data were normalized by log(l + x) transformation, and analyzed with a two factor ANOVA, with fruit type and season as factors. Post-hoc comparisons were made with Scheffe F tests at the alpha = 0.01 level to detect significant differences between pairs of seed types within each season. The statistical package JMP (SAS Institute 1989) was used for all these analyses.

PAGE 71

Part II: Infestation Rate of Seeds by Bruchids: Number of Entry Holes and Seed Mortality This part of the experiment was established to determine whether differences in oviposition rates on different fruit types translated to differences in larval infestation rates. Because mortality can occur at several stages, both the entry and development stages were monitored: Entry stage . If beetles prefer to lay eggs on certain fruits because they have a better chance of surviving the egg stage and penetrating the endocarp, then the fruits that have more eggs should show a higher ratio of entry holes: eggs laid than non-preferred fruits. Development stage . Both increased number of eggs and increased endocarp penetration should lead to an increased chance of seed infestation. The fruits preferred by bruchid beetles should also be the ones most likely to be destroyed by larvae. The endocarps from Part I were left in their exclosures in the field for approximately three months. This was the time period that both lab experiments and field observations indicated was sufficient for most larvae to reach full size and consume all the endosperm, but not sufficient to allow pupation and emergence. The three month period was counted from the day each experiment was set up, so that different

PAGE 72

62 set ups were collected on different days and in different months. The length of time endocarps remained in the field averaged 99 days in June (range 92-102 days), 103 days in July (99-106 days), and 86 days in September (78-93 days). After collection, all seeds were measured along the long axis, and the number of visible entry holes counted (regardless of whether they penetrated to the endosperm or not). Each endocarp was then split open with a machete, and the following parameters assessed: number of seeds, amount (estimated proportion) and condition (germinating, with fungus, or intact) of seeds not eaten by larvae, number and developmental stage of bruchids, if any (larva, late larva, pupa, or adult), and identity of parasites on bruchids, if any V -.^ ,..•:.» ^ i.; Data Analysis. Part II ^ The number of entry holes and the proportion of infested seeds for each category were compared within each season with a Kruskal-Wallis test. Unlike the situation in Part I, ' endocarps derived from Intact Husk fruits were included in the analysis in Part II. The stage (small larvae, large larvae, late larvae, pupa, and adult) of all larvae obtained were compared among seasons with a Chi-Square test for all fruit types grouped (due to the large number of blank cells). For the September data set only, the larval stage was compared among fruit

PAGE 73

63 types with a Chi-square test, because all larval stages v/ere well represented. The statistical package Statview (Abacus Concepts 1986) was used for these analyses. Control Experiment ' .\ ' To determine whether seven nights of exposure to oviposition are sufficient to lead to high levels of seed infestation under the controlled conditions of the experiment, a small scale experiment was set up among the experimental trees. Fifty Gashed fruits were protected by a chicken wire exclosure and exposed to oviposition for seven nights, then covered with mosquito meshing to exclude bruchid beetles and left in the forest. Another 50 fruits were placed next to the above exclosure, protected only with chicken wire. After three months the endocarps were collected, the number of entry holes counted, and the endosperm and larvae described as in Part II above. A t-test was used to compare the number of entry holes on the two sets of seeds, and the proportion on infested seeds was compared visually . Results Intact Husk fruits received no eggs during the first and second checks, with the exception of two fruits that were bitten open by insects and mammals, respectively (Table 3-1) .

PAGE 74

••64 " . Only by the sixth week in the field had at least one Intact Husk fruit in all set ups received eggs. This occurred when the husk had rotted sufficiently to split, or to lift away from the pulp, opening a gap at the seed's proximal end between husk and endocarp. Eggs were laid both on the husk and the pulp at this stage. The number of eggs remained low, with no fruit ever receiving more than 8 eggs, and the mean number of eggs per fruit ranging from 0.3 to 1.15 for different set ups (Table 3-2). r' The number of eggs found on Gashed, Rodent, Intact Pulp, and Bare Seed fruit types after 7 nights (mean 6.86 eggs/ fruit, range 0-47) was greater than the number of eggs after one night (mean 3.36 eggs/fruit, range 0-26), but the relative differences among seed types remained the same on both checks. Only the results of the second check are further described. The total number of eggs and the variation among fruits in number of eggs decreased markedly during the study, both among and within seasons, for the other fruit types (Fig. 3-2a-d, Fig. 3-3). The decrease was much more marked for Gashed and Rodent fruits than for Intact Pulp and Bare Seed fruits (Fig. 3-2a-d) . I describe first the trends, and later the significant differences among fruit types. In June, Gashed and Rodent fruits set at the end of the month had fewer eggs than those set earlier in the month (Fig. 3-2a-b), but the same was not true for Bare Seed and Intact Pulp fruits (Fig. 3-2c-d). Gashed and Rodent fruits received the highest number of eggs in all seasons, but in

PAGE 75

65 June Rodent fruits had more eggs than Gashed fruits (18.9 vs. 15.9 eggs per fruit), a difference that was reversed in July and September. Intact Pulp fruits consistently received the lowest number of eggs in all seasons (4.6, 2.5, and 2.1 mean eggs per fruit respectively), follov;ed by Bare Seeds (6.6, 3.0, and 2.7 eggs). For the five trees that were checked after 14 nights, Intact Pulp fruits averaged 8.0 eggs/fruit and very few of the original eggs had hatched, indicating that they were still receiving eggs and that the first eggs laid were not successful. These differences were consistent among trees within each season (Fig. 3-4a-c) . In June, a marked difference among trees in total number of eggs received, and a reduction in the difference between Gashed and Rodent fruits were due to the spread in time of experimental set ups, with the trees used later in the season having both fewer eggs and a similar number of eggs on Gashed and Rodent fruits (Fig. 3-4a). A 2 -factor ANOVA on log transformed data revealed significant effects of both Fruit Type (df = 3, F = 39.94, p < 0.0001) and Season (df = 2, F = 43.74, p < 0.0001), and a significant interaction between Season and Fruit Type (df 6, F = 2.26, p = 0.0437). Post hoc comparisons indicated that in all seasons. Gashed and Rodent fruits are statistically indistinguishable from each other, as are Bare Seed and Intact Pulp fruits. In June and July, Gashed and Rodent fruits receive significantly more eggs than both Bare Seed and Intact Pulp fruits. In September, the significant

PAGE 76

66 difference between Rodent fruits and both Bare Seeds and Intact Pulp disappears, but Gashed retains significantly higher numbers of eggs than Bare Seed and Intact Pulp. When the seasons are compared to each other, without regard to fruit type, in June significantly more eggs were laid than in both July and September, but the latter two are statistically indistinguishable from each other. Control Experiment After three months, there was no significant difference in the number of entry holes on endocarps exposed to beetles for 7 nights (15.3 +/6.31 holes) and those exposed for the full three months (16.1 +/5.72 entry holes; t = 0.42, p = 0.338). Both endocarp types had the same larva-induced mortality rate (Table 3-3). The only apparent difference between the two sets of endocarps was that nine of the larvae in the three month exposure set had been parasitized/killed by insects, whereas there was no indication of parasitism in the seven night exposure set. Phase II: Entry Holes The number of entry holes was significantly different among fruit types in all seasons (Kruskal Wallis, H = 280.78, 135.64, and 75.33 respectively in June, July, and September,

PAGE 77

p = 0.0001 in all cases). The trends in entry holes followed those expected from the number of eggs for all seed types in all seasons, with the exception of Intact Pulp fruits. Although this fruit type always had the lowest number of eggs, behind Bare Seeds, in June it had the second lowest number of entry holes, ahead of Bare Seeds, and in July and September it had the same number of entry holes as Rodent, and more than Bare Seeds (data are mean of all seeds per fruit type) . This discrepancy suggests that once the pulp became degraded some time after the second check, beetles continued laying on Intact Pulp fruits, and this latter round of egg-laying produced larvae that successfully entered the seed. Therefore, as for Intact Husks, the number of eggs counted at the second check does not correspond to the actual egg laying intensity eventually experienced by the seeds. For the three other fruit types, the second check egg count should be a good index of the larval cohort that will attempt the first entry to the seed. Excluding the Intact Husk and Intact Pulp fruits for the reasons explained above,' the ratios of entry hole number : egg number in June, July and September for each seed type are as follows: 0.539, 0.367, and 0.390 for Bare Seed; 0.826, 0.754, and 0.610 for Gashed, and 0.788, 0.692, and 0.504 for Rodent. These larval success ratios reflect the laying preferences of beetles; fruit types with the highest success ratios in each season received the highest number of eggs. Additionally, the seasonal decrease

PAGE 78

in egg: entry hole ratio indicates increasing egg mortality from June to September. „. , ; Part II: Proportion of Endocaros Infested by Larva An endocarp was considered infested if it contained: any live or dead larval stage or adult beetle, the parasites of bruchid beetles, the track of a larva through the endosperm, or larva frass. Any endocarp that had at least one larva was classified as infested, even if, in the case of multiple seeded endocarps , one of the seeds was still intact. Although in a natural (unprotected by exclosures) setting one seed of multiple-seeded endocarps sometimes escapes predation (Fragoso 19 94) , in this experiment very few endocarps showed this situation (9 Intact Husk fruits, 3 Intact Pulp fruits, 2 Gashed fruits, 2 Bare Seed fruits, and 1 Rodent fruit). Large larvae used all seeds within an endocarp, and when one seed was still intact, the larvae was usually small and still had the chance of destroying all seeds. The difference between the experimental and the natural settings is that the wire cages protected seeds (and eggs) from trampling, shifting, and pulp removal, therefore giving the eggs and first instar larva a much better chance of hatching and entering, respectively. ' All fruit types in all seasons had a nearly 100% infestation rate with the exception of Intact Husk (mean infestation rate 60% +/32) in June. Thirty seven Intact

PAGE 79

Husk and 9 Bare Seed endocarps survived in June, compared to only 1, 1, and 0 Intact Pulp, Gashed, and Rodent endocarps respectively (Table 3-4). The difference in survival rate among seed types was significant in June (Kruskal-Wallis test, H = 16.08, p = 0.0001; based on mean survival rate per seed type per tree) , but not significant in July and September (p = 0.07 and 0.4, respectively). Note that the sample sizes for Intact Husk and Intact Pulp fruits were reduced in July and September compared to other fruit types due to fruit removal by vertebrates. The seed survival rate for IH fruits was significantly higher in June than in July and September, and was not different between July and September (x2 = 67.3, p = 0.0001, DF = 8) (Table 3-5) . . ,, , ... • Bruchid Stage In June and July, most of the larvae found in all fruit types were large, late instar larvae that had consumed all or most of the endosperm in their seeds (Table 3-6) . Two exceptions were Intact Husk fruits in June and July, which had a high proportion of very small larvae, indicating that larvae were younger than those in the other fruit types, as would be expected from the late timing of egg-laying on these fruits . . . • In September, 65.5% of larvae found were large larvae, but now a high proportion of seeds contained late larvae.

PAGE 80

70 pupa, and adult beetles (30% combined late stages) (Table 35). Within this season, Bare Seeds had a disproportionately high number of late stages (53.7% combined). The combined differences in larval stages were significantly different among seasons (X^ = 221.47, p = 0.0001, df = 8). Since all seeds were the same age when collected from the forest (in fact, the September seeds were younger by about two weeks than the June and July seeds) , these differences indicate that a) larva from eggs laid later in the season developed at a faster rate than those earlier in the season for all fruit types; and b) larvae in Bare Seeds late in the season developed faster than larvae in other seed types in the same season. This early development in Bare Seed fruits was already noticeable in July, when a higher percentage of larvae in Bare Seeds than in any other fruit type were late instar larvae (Table 3-6) Fruit a nd Seed Loss to Vertebrates and Invertebrates Fruit disappearance from exclosures were greater late in the fruiting season than earlier (Table 3-7). In June, fruit disappeareance occurred without exclosures being breached. In July and September, some but not all exclosures with fruit loss were breached, and in some cases chewed fruits were found within the uplifted exclosure. In one exclosure where four Intact Pulp fruits were lost in September, without exclosure disturbance, only three

PAGE 81

71 endocarps were found on the surface in December. Persistent digging uncovered all seven missing endocarps, individually buried about 10 cm under the soil surface, within the area protected by the exclosure. Of these seven endocarps, four were intact, two had very small larvae with 80 % of the endosperm intact, and one had a large larvae that had consumed all the endosperm. The other 40 endocarps set at this tree were all infested or their seeds had rotted. At the time that the first four fruits were buried, at least two of them had received at least 1 or 2 eggs (counted after the first check) . At the second check, the six remaining Intact Pulp fruits had 0, 1, 1, 2, 3 and 4 eggs respectively, so that when the final three were buried, an unknown time after the second check, at least two of them had at least one egg on them. In other words, a maximum of three fruits could have been buried with eggs already on them. Discussion ' Implications for Fruit Handling by Animals The most striking differences in both ovipostion and seed survival rates occurred between Intact Husk fruits and all other fruit types. Female Pachvmerus cardo at Maraca did not lay on fruit husks until these began to rot, and even then the number of eggs laid was low compared to the number laid on the other fruit types within the same season. Given

PAGE 82

the consumer set that uses Attalea fruits at the Maraca site, the avoidance of intact fruits by bruchid beetles has significant implications for the palm's dispersal ecology and adult distribution pattern. The seeds that had the best chance of surviving were those that fell to the ground in fruits unhandled by arboreal or terrestrial pulp eaters. These are the fruits that are ingested by tapirs in large amounts and carried long distances from the parent tree (Fragoso 1997b) . Because intact fruits do not receive eggs until at least four weeks after falling, they are uninfested at the time of ingestion by tapirs. After defecation, they are subject to infestation by bruchids in the feces (Silvius pers . obs . ) , but the infestation rate in the feces is much lower than in the original fruit piles (Fragoso 1997b) . The fruits with the highest probability of receiving eggs are those that have been handled by the presumed best dispersers (rodents, primates and spit-dispersing ungulates) and also by procyonids and psittacines. Although the increase in egg laying did not result in an increase in larval infestation in the experiment once Intact Husk fruits are excluded, it must be remembered that the fruits in the experiment were protected to an extent that would not occur naturally. Foraging tapirs, peccaries and deer trample seeds into the ground, and falling litter protects them from access by beetles (as does burial under the fruit pile) . Therefore the longer a fruit remains without eggs, the more likely it

PAGE 83

is to survive infestation. Rodent and Gashed fruits always received eggs after the first night. However, since eggs do not hatch for several days, it is unclear how well oviposition translates to first instar survival if seeds are subsequently trampled or buried. Eggs laid on Rodent and Gashed fruits are more likely to remain attached to the seed than those laid on intact pulp, however, and therefore more likely to resist partial burial. These observations refer to endocarps remaining unburied near the parent tree. Squirrels probably bury many endocarps throughout the fruiting season, and agoutis bury some seeds late in the season (Fragoso 1997b) . Fruits dispersed by primates even within several hundred meters of a tree are still exposed to high rates of oviposition, at least within a forest with high Attalea density (fruits placed up to 200 meters from the nearest fruiting palm received the same numbers of eggs as fruits placed at the base of fruiting and non-fruiting palms; Silvius, unpublished data). Therefore at this time the best chance of survival is to be an Intact Husk or Intact Pulp fruit that gets carried away by a tapir soon after falling. Thus the activities of vertebrate pulp eaters can mediate seed predation by invertebrates when the invertebrate is the primary predator on the seed, as it is in this case, where more than 95% of all seeds remaining on the surface near the parent plant are destroyed by bruchid beetles (Fragoso 1997b, Chapter 4) .

PAGE 84

74 Pulp-Oviposition and ?ulp-Development Rate Interactions: The Role of Moisture Levels The amount of pulp remaining on endocarps, and the pattern in which it was arranged, had a significant effect on the number of eggs laid on fruits by bruchid beetles. This effect translated into higher egg and first instar larva survival, as indicated by the dependence of number of entry holes on number of eggs. Fruiting season interacted with the effect of pulp on both number of eggs and number of entry holes: overall number of eggs laid, and differences among fruit types in number of eggs laid, were greater early in the fruiting season (June) than late (July and September) . Differences in egg laying and egg survival also translated into differences in seed infestation rates (number of endocarps with at least one larva entering) , but this effect was significant only in June, when larva failed to penetrate Intact Husk fruits. The general trend in egg laying preference was that endocarps with some pulp were preferred above endocarps with no pulp or with too much pulp. Fruits with intact husks were inaccessible to ovipositing females. Avoidance of seeds with large amounts of pulp may be due to at least two factors: a) first instar larvae are unable to penetrate the pulp for either physical or chemical reasons, and b) eggs do not stay attached very well to intact, slimy pulp. Both possibilities are supported by the observation that female bruchids often

PAGE 85

lay eggs on pieces of litter stuck onto Intact Pulp fruits, on indentations in the pulp made by pulp eating invertebrates, or at the proximal end of the seed, where the pulp layer is thinnest. The second possibility is supported by the fact that eggs on intact pulp frequently fell off when handled or after heavy rainfall. On Bare Seed fruits, females usually laid eggs on the underside of the endocarp. Wright (1991) observed the same behavior for a different palm bruchid species in Panama. Along with the above observations on pulp condition, this behavior suggests that moisture is a key factor in egg-laying choice: eggs on bare endocarps are likely to dry out, because there is no pulp to retain moisture. Eggs on intact pulp will not dry out, but neither do they stick well. Fruits gnawed by rodents and gashed by ungulates or monkeys, on the other had, provide a secure attachment site on the bare endocarp or on the juice-free fiber, while providing moisture-retaining pulp or fiber. On Gashed Fruits, eggs were nearly always laid on the endocarp but close to or under the edge of the pulp. It is unclear, though, why infestation of completely bare endocarps was lower early in season, when it rained frequently, than later in the season. One explanation may be related to the observed higher insect-related egg mortality on bare endocarps than on fruits with pulp. : . ' ' The net effect of these seasonal differences was that even though oviposition rates were higher early than late in the fruiting season, seed survival was also higher early than

PAGE 86

later. In other words, the correlation between oviposition rates and survival is strong within a season but does not carry on across seasons. Early in the season, Intact Husk and Intact Pulp fruits resisted beetle infestation long enough that, had they not been protected by exclosures, they had better chances of being removed by a disperser before bruchid eggs could hatch or even be laid on them. Therefore trees fruiting late in the season, when bruchid activity is lower, do not necessarily have higher seed survival probabilities, as predicted by Forget et al . (1994) and Wright (1990) for a similar system in Panama. Those authors' conclusions were based on monitoring of egg numbers only, and not on the translation from egg to larval infestation rates. On the other hand, late in the season, when overall fruit pulp availability is lower in the forest (Chapter 6) , fruit removal rates by frugivores may be higher, as indicated by the invasion of exclosures in the last experimental set up. In this case seed survival would depend on whether frugivores remove seeds to a site that protects them from bruchids. On Maraca, bruchid activity is so high (compared to the mean 1-2 eggs per endocarp documented during the peak fruiting season on BCI, Wright 1990) that fruits placed more than 100 meters from fruiting trees have the same likelihood of being found by bruchids as fruits near the tree (Silvius, unpublished data) . ... The role of moisture seems to be important in another way. Several researchers have observed that rainfall may cue

PAGE 87

77 the emergence of adult beet les ( Janzen 1971, Wright 1983). The more rapid development of larvae late in the season on Maraca (endocarps from the September replicate were collected in December, two months into the dry season, whereas those from June and July were collected in September and October, respectively) , and the more rapid development of larvae in Bare Seeds (the ones most likely to lose moisture) suggest that beetles may be responding to the moisture level in the environment by speeding development when moisture is low. Thus, most beetles emerge in the dry season, regardless of when the eggs that gave rise to them were laid. With a few exceptions, emergence from natural, non-manipulated endocarps in the field did not begin until November, and did not reach its peak until DecemberJanuary (Chapter 4), supporting these observations. An alternative hypothesis is that there are at least two sub-populations of beetles, that differ genetically or behaviorally with respect to reproductive traits: one group that lays early in the season but emerges late, and another that lays late in the season but emerges early. Pulp as Protection The interaction between pulp eaters, bruchid beetles, and palm fruits also indicates that pulp, as well as husk, can be a protective mechanism against invertebrate seed predators. It may be under selective pressure for this reason as well as for its attractiveness to dispersers. The

PAGE 88

78 role of husks (pericarp or exocarp) as physical deterrents against invertebrates has received some mention in the literature (e.g., Pritchard 1969, Sallabanks and Courtney 1992, Rodgerson 1998), and the role of chemical deterrents in pulp has been extensively investigated (see reviews in Herrera 1982, Sallabanks and Courtney 1992, and Cipollini and Levey 1997), but I have found no mention of pulp as a physical deterrent. Work by Delgado et al . (1997) suggests a similar role for pulp in a different palm species. These authors found that oviposition by bruchid beetles on Astrocarvum chambira fruits, which have a much drier, thinner mesocarp than Attalea fruits, was more rapid on peeled than intact fruits, and that oviposition became high on unpeeled fruits only after the mesocarp had decayed. Because Astrocarvum seeds are prized by many granivorous rodents (Smythe 1986, Terborgh et al. 1993, Cintra and Horna 1997, Hoch and Adler 1997), if pulp protects a fruit on the ground until rodents scatterhoard them, then the fruit's chance of survival is increased. This type protection would also work with the putative megafauna dispersers that may have coevolved with large-fruited tropical trees (Janzen and Martin 1982), represented today only by the tapir (Fragoso 1997b) .

PAGE 89

79 Burial by Scarabid Beetles The seven fruits "lost" from the September exclosure were buried at a depth and in a pattern typical of burial for nesting purposes by large scarabid beetles of the burrower type (Halffter and Edmonds 1982). Several observations indicate that the seven fruits were buried by the dung beetle Oxvsternon f estivum ( Scarabaeidae / Coprinae / Phanaeini) and that burial of Attalea fruits by this species is a common occurrence at this study site. Adult scarabid beetles have been recorded feeding on pulp (Hanski and Camberfort 1991), and one species is known to synthesize nesting balls out of pulp and other plant parts (Monteith and Storey 1981), but I know of no reports of scarabid beetles using entire fruits to provision their nests. During a study scarabid beetles on Maraca in June 1996, large numbers of the beetles were captured in traps baited with fox ( Cerdocvon thous) dung (which at that time contained Guava pulp and seeds), and was considered a common species in the study area (F. Vaz de Mello, pers . comm) . On two occasions the beetles were observed to bury Intact Pulp Attalea fruits shallowly in a network of tunnels under fruit piles, in a manner similar to that of shallow burial of dung balls for adult feeding by other dung beetle species. A beetle captured at a fruit pile and held in the lab for a few hours buried five Intact Pulp fruits to a depth of 5 cm, the maximun depth of soil provided in its container.

PAGE 90

80 Throughout the five fruit type experiment described in this paper, fruits were found within the exclosures shallowly buried in the beetle fashion. Only the fruits from the single tree in September, described above, were ever found deeply buried. Thus, Qxvsternon exhibited the two kinds of food burial typical of dung beetles: shallowly buried fruits on which adult beetles may feed, and deeply buried fruits potentially used as food for larvae. The seven exclosure fruits deeply buried by the beetles had variable numbers of eggs on them when buried. It would appear that those that were buried early had their eggs scraped off or killed by the burial process. The others survived burial and bruchid larvae infested the seeds, but burial seems to have delayed the development of the larvae in at least two seeds. The higher survival rate of scarabidburied seeds than of other seeds set at the same tree suggests that burial by scarabids functions in the same way as removal of intact fruits by tapirs: they decrease the chance of oviposition and/or first instar survival. The germination biology of Attalea is unstudied, and it is therefore unclear whether its seeds are able to germinate from a 10 cm depth is unknown. Survival is likely, however, given that many palms have remote germination in which seedlings usually grow down into the soil before extending up to break the surface (Tomlinson 1990). Scarabid beetles could thus be an effective "planter," if not mover, of palm seeds, and many of the seedlings that germinate near the

PAGE 91

81 parent plant on Maraca (Fragoso 1997b) may be the result of burial by beetles rather than by rodents. Large scale effects of pulo-vertebrate-beetle interactions This study documented a small-scale interaction at the individual seed level. It further focused on a single forest type with a characteristic mammal community. The distribution of frugivores at a larger scale, however, is patchy and nonrandom, both on Maraca and at other Neotropical sites. On Maraca, spider monkeys do not use forests near the river (Nuhes 1992), while squirrel monkeys prefer those forests as well as other edge habitats (Silvius pers . obs . ) . Tapirs prefer to feed in the highest density Attalea patches in the river forest, leading to high levels of intact fruit removal there. At a scale much larger than that of this study, patchiness in frugivore distribution could therefore result in patchiness in palm-beetle-vertebrate interactions and in palm seed survival and regeneration. Such patchiness in interactions could contribute to the clumped distribution of many palm species, a distribution that has most often been attributed to anthropogenic effects (Balee 1988, Clark et al . 1995) or edaphic factors (Kahn and de Granville 1992). >'>. ' " . .

PAGE 92

82 Table 3-1: Timing of egg laying on Intact Husk fruits, presented as number of experimental setups (trees) out of the total set up that had at least one egg on at least one fruit at the time of checking. Data is cumulative. Check # 1 occurred one day after seeds are laid out, subsequent checks followed at 7-day intervals. A blank cell indicates fruits were not checked at that date. Egg laying does not begin until the husks begin to rot and the pulp or endocarp is exposed . Check Number Season 1 2 3 4 5 6 7 8 9 June 10 trees 0 0 4 6 7 9 10 10 10 July 10 trees 0 1* 4 * 6 6 9 September 8 trees 0 * * = husk chewed open by insects, exposing the pulp. ** = husk chewed open by mammals, exposing the pulp Table 3-2: Mean number and range of eggs per fruit for trees that received eggs on the Intact Husk fruits during the June experiment, by check. n = 100 seeds per check (10 seeds at each of 10 trees) . Mean 0 0 0 . 3 0 .28 0 . 87 0 . 83 1.15 0.9 0 . 92 Range 0 0 0-1 0-2 0-5 0-4 0-8 0-5 0-6 Table 3-3: Infestation rates for fruits exposed to bruchid oviposition for seven nights and three months, respectively. 7 night exposure 3 month exposure Large Larvae 45 46 Empty Endocarp 3 4 Rotten Endosperm 2 0 Total 50 50

PAGE 93

Table 3-4: Seed survival per fruit type and season, presented as number of infested (at least one endosperm with larvae or other sign of bruchid beetle infestation) , good (no seeds with larvae or sign of infestation) and other (empty endocarp or endosperm rotted by fungus, with no sign of bruchid infestation) . Note reduction in sample size for IH fruits in all seasons and for IP fruits in July and September, due to fruit loss from the exclosures to pulp consumers. June (initial n = 100 fruits per fruit type. . Infested Good Other (>. 1 seed (no seeds ( empty. infested ) infested) fungus ) Intact Husk 57 36 2 Intact Pulp 98 1 1 Bare Seed 86 9 5 Gashed 96 0 4 Rodent 95 1 4 July (initial n = 100 fruits 1 Der fruit type) Infested Good Other (>. 1 seed (no seeds ( empty, infested ) infested ) fungus ) Intact Husk 91 3 0 Intact Pulp 90 3 4 Bare Seed 97 0 2 Gashed 99 0 1 Rodent 92 5 5 September (initial n = 80 fruits per fruit type) Infested Good Other (>. 1 seed (no seeds ( empty. infested ) infested) fungus ) Intact Husk 58 0 4 Intact Pulp 68 4 1 Bare Seed 78 0 1 Gashed 76 0 2 Rodent 78 0 1 Table 3-5: Seed survival in Intact Husk fruits during the three seasons. Survival is significantly higher in June than in July and September (x2 = 67.3, p = 0.0001, df = 8) Season Infested Good Other June 57 36 2 Ju ly 91 3 0 September 58 0 4

PAGE 94

84 Table 3-6: Infestation stage of endocarps , separated by fruit type and season. "New larvae" are small larvae that have not had time to eat much of the endosperm; "larvae" are large larvae that have eaten all or most of the endosperm but have not metamorphosed into pupal stages, "late larvae" are larvae that have started to metamorphose, as indicated by a reduction in size, thickening of body and change in head shape; "pupa" are soft, white with legs, mouth parts and wings; adults are dark, fully formed beetles. Field age is the number of days the seeds remained in the forest between setting and cracking; for all endocarp types except Intact Husk, this corresponds to larval age. June (me an field age = 9 9 days New Larvae Late Pupa Adults Total Larvae Larvae Bare 4 75 0 1 0 80 Seed (5%) (94%) (1%) Intact 14 39 0 0 0 53* Husk (26%) (74%) Intact 3 89 0 0 0 92 Pulp (3%) (97%) Gashed 1 87 0 0 0 88 (1%) (99%) Rodent 1 86 1 0 0 88 (1%) (98%) (1%) Total 23 376 1 1 0 401 Larvae (5.7%) (93.7%) (0.25%) (0.25%) * Low number of larvae m IH seeds is due to low infes rate of seeds during the first part of the experiment.

PAGE 95

Table 3-6, continued: July (me an field age = 103 days) New Larvae Late Pupa Adults Total Larvae Larvae Bare 5 80 10 2 0 97 Seed (5%) (83%) (10%) (2%) Intact 23 65 2 1 0 91 Husk (25%) (72%) (2%) (1%) Intact 4 86 0 0 0 90 Pulp (4%) (96%) Gashed 2 91 0 0 0 93 (2%) (98%) Rodent 1 86 0 0, 0 87 (1%) (99%) Total 35 408 12 3 0 458 Larvae (7%) (89%) (3%) (1%) Septembe r (mean field age = 8 6 days) New Larvae Late Pupa Adults Total Larvae Larvae Bare 2 40 22 23 3 90 Seed (2%) (44 .4%) (24.4%) (26%) (3.3%) Intact 8 29 16 7 0 60* Husk (13%) (48%) (27%) (12%) Intact 4 55 7 5 1 72 Pulp (6%) (76%) (10%) (7%) (1%) Gashed 1 66 8 9 0 84 (1%) (79%) (9.5%) (10 . 5%) Rodent 2 55 6 4 0 67 (3%) (82%) (9%) (6%) Total 17 245 59 48 4 373 Larvae (4.5%) (65 . 5%) (16%) (13%) (1%) * Low number of larvae in Intact Husk seeds is due to loss of seeds to frugivores during experiment.

PAGE 96

86 Table 3-7: Fruit losses from exclosures after one week. S 6 3. s o n Fruit Type Number Lost June Intact Husk 5 J uiy Intact Husk r 0 XliLdCL iruxp Bare Seed 1 September Intact Husk 18 Gashed i-. Rodent 1 Bare Seed 1 Intact PuId 4

PAGE 97

87 Figure 3-1. Experimental field design for Parts I and II. IH = Intact Husk fruits, IP = Intact Pulp fruits, BS = Bare Seed fruits, G = Gashed fruits, R = Rodent fruits. See text for fruit type descriptions.

PAGE 98

88 •a o CO to Hi o Z c 40-1 302010an T June July "T 1. September Time Set a) Rodent Figure 3-2. Mean number of eggs laid by bruchid beetles on sets of 10 fruits of 4 different fruit types at each of 10, 10, and 8 trees in June, July, and September, respectively, a) Rodent gnawed fruits; b) Primate or ungulate gashed fruits; c) Endocarps with all pulp removed; d) Fruits with pulp intact but all husk removed.

PAGE 99

89 •J) 00 U d Z % 30-1 25 2015 105 5 June July T 1 September b) Gashed Figure 3-2, continued. Time Set

PAGE 100

90 (U u u C/5 1/3 00 op o Z e J3 U 15-1 105 s, s OP July Sept Time Set c) Bare Seeds Figure 3-2, continued.

PAGE 101

91 •a 6 1 _ -DJune July B "I 1 September Time Set d) Intact Pulp Figure 3-2, continued.

PAGE 102

92 Figure 3-3. Mean number of eggs on all fruits at all trees for each fruit type, with standard error, showing the high oviposition intensity in June, and the low oviposition intensity in July and September. IH = Intact Husk fruits, IP = Intact Pulp fruits, BS = Bare Seed fruits, G = Gashed fruits, R Rodent fruits.

PAGE 103

II Cn C/1 C — • PQ -H 0 cn • 4J CQ •H -l 4-1 >, D 4-1 4J a cn 1 — 1 4J cn 3 -H 0 cu p u u fO o 4J 4-1 (U fO ij 4-1 0 (d (0 d 4-1 i-H 1 — i d) •H 4-> a e M ?i •H QJ cn Oi 4J jj (U M OJ H QJ 3 QJ cn u) >-i 4-) *4-l -H cn 3 • t—i 0) u cn u a. 4-1 4J (0 >i •H J«i 3 cn ^ 4-1 4-> 3 4-1 u H X. 3 4J CO U 4-1 C 4-J U Q) •H 3 Di 4-) 0 C d P£i 14-1 O M e II 1—1 II i-H DS (0 cn 0) M > C 4J cn o 4-1 • -H cn •h 3 01 c 01 ^ Di 0 X! 4-4 0) •H e QJ T3 U-l •H 4-) QJ 0 cn ax; 0 QJ cn U aw (0 0) •H o Xi > e 0 U II 3 c c •H d c (13 (0 cn cn 0) 0) XJ u >1-H a M 3 <]) 3 U cn u 1-} 4-1 d (U 0 1 4-1 — TS •H ro 4-1 43 Q) 4J •H CU a 0) T3 C/D H U 1-1 3 4J Qj" QJ U di d ^ cn •H j:: 3 05 4-1 1^ CQ T)

PAGE 105

95

PAGE 106

CHAPTER 4 PALM ENDOCARP USE BY THREE NEOTROPICAL SEED-EATERS: GRANIVORY OR "GRUBIVORY?" Xr-'.} . % Introduction Palm fruits are an important component of Neotropical mammal diets. Primates feed on mature palm pulp and immature endosperm, and in some cases on old "nuts" (endocarp plus contents) remaining on the ground for up to a year after the fruiting season (Chapter 2). Terrestrial mammals, including ungulates, rodents, marsupials, and carnivores, use the pulp and crack or gnaw open fresh and old nuts (Chapter 2) . Because they are protected by thick endocarps, palm seeds may remain available on the ground for longer periods of time than softer fruits or seeds. By virtue of their thick endocarps, coccosoid palms in particular provide ideal sites for the growth and metamorphosis of palm bruchid beetles (Coleoptera: Bruchidae: Phachymerini) . Palm-bruchid interactions are especially well-developed between Attalea sp. and Astrocarvum sp . palms and Pachymerus sp., Carvoborus sp., and Soeciomerus sp. beetles (Delobel et a. 1995, Johnson et al . 1995, and Johnson 1993). The beetles' large, lipid-rich larvae feed on the endosperm and remain inside the endocarp for three to nine 96 '

PAGE 107

97 months (Chapter 2, 3). Beetle infestation rates can be very high for seeds remaining near the parent tree or within an aggregation of palms. For non-manipulated seed piles, Janzen 1971 and Fragoso 1997b documented 80% and 77% respectively of seeds killed by bruchids; for experimentally cleaned seeds, Smythe 1989 and Terborgh et al . 1993 documented 98% and 75% mortality, respectively. When vertebrates feed on endocarp contents, they are therefore likely to interact with bruchid beetles as well as with palm seeds. Not only do they kill bruchids and potentially affect the population dynamics of both the beetle and the palm (as suggested by Forget et al . 1994 for palm bruchids, and Herrera 1989, Traveset 1992 for other beetle species) , they also enter into an evolutionary relationship with the palm-beetle complex (c.f. Herrera 1984). There is evidence that in several frugivory systems, both pulp and seed eating vertebrates actively seek out invertebrate pests on fruits and seeds. Redford et al . (1984) experimentally showed that marmosets prefer fruit pulp infested by insect larvae. Through field and cage experiments, Vaulberg ( 1992a, b) showed that at least one species of understory bird preferred infested to non-infested pulp. Scott and Black (1981) demonstrated that cockatoos feeding on weevil-infested Banksia inf ructescences were searching for larvae. In other cases, consumers ingest fruits irrespective of the infestation state (Herrera 1989,

PAGE 108

98 Piper 1986); i.e., they do not reject them because they contain insect larvae. . In the case of thick-endocarp palms, several authors have suggested that vertebrates incidentally feed on larvae (Janzen 1971, Kiltie 1981). Others have suggested that they actively seek out bruchid larvae (Robinson 1986, Struhsaker and Leland 1977). Fragoso (1994) showed that white-lipped peccaries ( Tavassu pecari ) in captivity will crack infested endocarps and ingest the larvae. Terborgh (1983) describes Cebus apella preferentially opening endocarps that have been infested, because they are easier to open, but suggests that the primates are seeking endosperm remains rather than larvae. If vertebrate consumers feed on larvae from insectinfested fruits and seeds, then measures of food availability for fruit and seed eaters must take into account the availability and development patterns of these insects. In particular, estimates of seasonal food scarcity (c.f. Terborgh 1986a, b, Leigh et al . 1996) that rely on counts of mature fruits may not accurately reflect true availability in the environment. For one rain forest site in the northern Amazon, this study asks whether vertebrates seek out bruchid larvae in palm endocarps, and whether they do so at the time of low fruit availability. I examine the timing of availability of Attalea marioa endocarp contents--either endosperm or larvae--with respect to the availability of other fruits and seeds in the habitat. I monitor the timing

PAGE 109

of endocarp use by squirrels ( Sciurus igniventris ) , agoutis ( Dasvprocta agouti ) . and white-lipped peccaries, three species that differ both in their degree of dependence on seeds and in their ability to access a habitat mosaic that can provide a wide range of foods during the season of fruit scarcity. Study Site Maraca Island Ecological Reserve, a 1,100 km2 riverine island in the northern Brazilian state of Roraima, lies on the forest side of the forest -savanna boundary between the Rupununi savannas of Guyana and the Amazonian rain forest. Habitats grade from small savannas, transitional forests, and high diversity moist upland forest on the eastern end of the reserve, through an area dominated by the deciduous legume Peltogvne gracilioes in the center and center east, and moist upland mixed forests on the western half. The eastern terrain, where this work was carried out, is flat and low (110-160 meters above sea level) . The sandy soils are derived from quartz-biotite schists, quartz-feldspar gneisses, and tonalitic granites (Thompson et al . 1992). Rainfall records on the eastern end of the island show variation between 1,750 and 2,300 mm annually; during El Nino years, rainfall decreases markedly across the state. The wet season is between May and September, the dry season between October and April, with rains starting either in early or late

PAGE 110

100 , . April. The heaviest individual rains tend to fall in the dry season. There are no extensive flooded forested habitats or palm swamps in the Reserve, although the riverine forest floods in low pockets during high rainfall years (Fragoso 1994). Several small ponds and savannas are interspersed throughout the forest, and frequent strands of Mauritia palms run along stream courses and other periodically flooded habitats. The plant families with the greatest number of tree species on terra firme habitats are the Leguminosae, Rubiaceae, Moraceae, and Burseraceae (Milliken and Ratter 1989, 1998). Based on importance value (IV = relative density + relative dominance + relative frequency) rather than species richness, Sapotaceae is the dominant family in the eastern forests. Rubiaceae, Moraceae, Burseraceae, Annonaceae, Chrysobalanaceae, Palmae, and Bignonaceae also occur within the top 5 dominant families based on IV in transects on the eastern end. Study Species " : Individuals of the arborescent palm Attalea maripa (previously Maximi liana maripa : taxonomy revised by Henderson 1995) can produce from a few hundred to several thousand 3 to 6 cm long fruits per year (Chapter 3). Endocarp thickness ranges from 1.5 to 6.5 mm. The endocarps of palms in the genus Attalea are thicker and more resistant to rot than those of palms in the genus Astrocarvum ; to date, reports of

PAGE 111

101 „ potential bruchid larvae use by non-rodent vertebrates are only for the latter palm genus (Struhsaker and Leland 1977, Kiltie 1981, Terborgh 1983). White-lipped peccaries are the most omnivorous of the three study mammals, feeding on invertebrates, fish, fruits, seeds, and vegetative plant parts. However, fruits and seeds can make up as much as 60% of their diet (Bodmer 1989) . They live in large herds (30-200 animals documented on Maraca) , have large home ranges (20 to over 100 km^ , Fragoso 1998), and use a mosaic of habitat types, including terra firme forest, Mauritia palm swamps, savannas, and riverine forest (Fragoso 1994, in review). , . . . The only squirrel species on Maraca, the northern Amazon red squirrel, is a large (500-900 g) animal with terrestrial habits (Emmons and Feer 1997). On Maraca it appears to be common, but is also shy and its ecology has never been studied. Agoutis on Maraca have an ecology and behavior similar to that described by Smythe (1978) for Dasvorocta punctata in Panama, but have larger home ranges (2.5 to 8 hectares; Chapter 5) . Both white-lipped peccaries and agouties were radio-tracked concurrently with this study. There are at least three bruchid beetle species on Maraca, but only one, Pachvmerus cardo, parasitized Attalea palm seeds in this study. Bruchids lay eggs on fallen fruit whose exocarp has been removed by vertebrate consumers (Chapter 3). Within a week, first instar larvae hatch and

PAGE 112

102 . ' \ tunnel through the endocarp into the seed. Exclosure experiments and infested seeds kept in the lab indicate that bruchid beetles require a minimum of two months, and usually three months, to complete development from egg to adult. The last larval instar carves a round exit hole almost completely through the endocarp; after pupation and metamorphosis, the adult beetle emerges by pushing off the remaining thin "lid" of endocarp from the hole. Methods Overview Research involved monitoring, in three different habitats, of forest wide fruit availability, Attalea palm phenology, and Attalea endocarp use by mammals. Endocarp use was monitored at three different scales: 1) individual trees within one habitat, 2) community of trees within one habitat, and 3) across habitat types. These correspond to the probable scales of activity of the three seed-eating mammals studied: individual white-lipped peccary herds move among habitat types, while individual agoutis and squirrels move among individual trees in their small home ranges. The three habitat types studied--riverine forest, eastern terra firme forest, and central terra firme forest-differ in their tree species composition and diversity (Milliken and Ratter 1989). The two terra firme forests have

PAGE 113

similar species composition, but palms are more common in the eastern forest, and overall diversity is higher in the central forest. Both have high densities of trees in the Sapotaceae. In contrast, the riverine forest has high densities of Attalea palms, low tree diversity, and almost no trees in the Sapotaceae. Study sites were separated by 5 to 10 km. Monitoring was centered on one 2 km long transect in each habitat, but included additional trails in the eastern forest. Background Fruit Availability Fruit availability on the ground was monitored in the three habitats every two to three weeks (range 14-3 0 days, mean 2 0.76 days) from November 1995 to October 1996. In the eastern forest, monitoring continued until March 1997. All fruits and seeds seen on the ground along the 2 km by 6 m transects were counted and identified, with each individual item or cluster of items counted as a single patch (fruits separated by more than 10 meters were counted as separate patches if there were no other fruits of the same species in between; if the fruit piles from two adjacent trees overlapped, they were counted as a single patch) . Both ripe or newly fallen items and older seeds were counted, and their stage identified (immature, preripe, ripe, rotten, germinating; I define preripe fruits as full sized, immature color fruits) . The ground monitoring method was chosen

PAGE 114

104 because agoutis and white-lipped peccaries can only feed on fruit that falls to the ground; because there is a delay in the availability of food items to terrestrial f rugivore/granivores compared to arboreal consumers (Zhang and Wang 1995); and because agoutis and white-lipped peccaries feed on older seeds and germinating seeds whose availability would not be measured by canopy observations or fruit trap based sampling. The diet of squirrels on Maraca has not been studied, and the monitoring method may not be an accurate representation of food availability for them. However, dietary studies of a similar species in Central America suggest that this method would have captured many of the important large seeded species, including all palms (Glanz et al . 1996) . . Because fruits are rapidly removed by terrestrial animals (Zhang and Wang 1995), a patch-based enumeration was used rather than individual fruit counts, and no attempt was made to estimate the total number or mass of fruits and seeds falling to the ground. Fallen fruit were used as an index of the overall pattern of fruit availability. Although ground transects may fail to capture some fruits that fall to the ground in low numbers (Zhang and Wang 1995), because agoutis feed on large fruits for which remnants of feeding can be found and counted, and because white-lipped foraging leaves obvious signs, this error was minimized. White-lipped peccaries are able to move between the three habitat types in one or two days. During this study

PAGE 115

105 there were two herds resident in the study area; one herd included the main trail system in the eastern forest and the riverine trail in its home range, along with other habitats on the southeastern portion of the reserve. The second herd used the main trail system and the area of the central forest sampled by the transect, in addition to the riverine forest on the north and north east of the island (Fragoso unpublished data) . Fruit and seed availability at any one time for white-lipped peccaries is therefore represented by combining data from all three transects, whereas for agoutis and squirrels on the main trail system or in the riverine forest it is represented by just one transect. Attalea Fruit Availability In the eastern forest, I located all Attalea trees with fruits within 20 m of 15 km of trail, starting in April 1996. On the riverine forest transect, all fruiting trees within 20 m of the 2 km long transect were located and monitored. In the central terra firme forest, where Attalea is uncommon, only the few palms (n = 6) that fruited along the 6 m wide transect were monitored. Trails were checked every two to three days during the early and peak palm fruiting season (May-July) , to locate as many trees as possible with immature fruit bunches, and to pinpoint within a few days the start of fruit fall. After that, trees with near-ripe fruit were checked at a maximum of one week intervals to ascertain

PAGE 116

fruiting data. Based on monitored Attalea trees in all three habitats for which the date of first fruit fall was known or could be estimated, the number of trees starting to fruit during each fifteen day period from April to November 1996 was calculated. The period over which pulp was available was calculated from the median number of days that fruit fall lasted, based on trees for which both the start and end of fruit drop period were noted. ^ Endocarp Use by Squirrels. Agoutis, and White-lipped Peccaries: Treeand Within-Habitat Scales Endocarp use by squirrels and agoutis was studied in the eastern terra firme forest only, while endocarp use by whitelipped peccaries was monitored in all three forest types. In the eastern terra firme forest, endocarps opened by mammals were counted at fruiting trees on a biweekly schedule from April to August 1996, then monthly from August to December 1996, and one final time in February 1997. In December 1996 only a subset of the trees was sampled due to time constraints (20 out of 98 trees available; trees were chosen to be representative of the full spread of endocarp ages available). For the following check (February 1997), the number of endocarps counted at the sub sampled trees was added on to the new data, to match the length of time during which eaten endocarps accumulated for all other trees. Trees were dropped from the census list when treefalls covered the seed pile, when rooting by collared peccaries ( Tavassu

PAGE 117

taiacu ) made it impossible to count seeds, and once the seed pile had been completely depleted by feeding or bruchid beetle emergence. During each scheduled check I searched an area with a 5 m radius around the tree trunk and counted and removed all endocarps opened by squirrels, agoutis and white-lipped peccaries. Because squirrels ate endocarp contents from perches, if no suitable perches were located within 5 m of the tree, I extended the search radius out to include suitable perches (vines and branches three meters or less above the ground) . Both squirrels and agoutis fed on endocarp contents close to the tree, repeatedly using the same feeding site, and endocarp piles were often found in the same location over a period of several months. During a radio tracking study (Chapter 5), agoutis nearly always consumed seeds of all tree species near or under the parent tree, and the same pattern has been found by other researchers (Wenny 1998, Forget 1990, 1993, Smythe 1978). Both squirrels and agoutis fed on endocarps directly from the endocarp pile; squirrels additionally fed on previously buried endocarps, and there was evidence at some trees of digging by squirrels (messier, smaller holes than those dug by agoutis) . There was no evidence of scatterhoarding by agoutis near the monitored trees. The number of endocarps located near the parent tree is therefore a good index of endocarp use by squirrels and agoutis; white-lips do not carry endocarps away from the tree .

PAGE 118

108 Each consumer opened the endocarps in a characteristic fashion. On several occasions squirrels were observed feeding on Attalea endocarps, and the opened endocarps were picked up and examined. Squirrel tooth marks were in the form of long gashes, usually in a triangular pattern with a small round hole in the middle. They could be located at either the distal, narrow tip of the endocarp, or at the broad proximal base, by the germination pores. On multiple seeded endocarps, a hole was often made for each seed. Agoutis were never directly observed feeding on ' , . endocarps, but on several occasions I found fresh partially eaten endocarps at sites from which agoutis fled following my approach. The size of the tooth marks was larger than that on the squirrel seeds, and matched the tooth size on the skull of an adult agouti. Holes were in the middle of the endocarp, because agoutis hold onto the ends of the endocarp with their fore feet and gnaw them in a "corncob" fashion. On the side of the seed opposite the hole, the marks of the lower teeth could be seen, where they held the endocarp in place. No other rodents except pacas ( Agouti paca ) and acouchies ( Mvoorocta sp . ) have a gape this wide. Acouchies have not been recorded for the island. Pacas are generally assumed to feed on the pulp rather than the seeds of palms with hard endocarps (c.f. Smythe et al . 1996), but there is no direct information on this, and they are able to scrape away the hard endosperm of Mauritia palms ( Prada 1994) . Due to the consistency in tooth mark and hole appearance, all

PAGE 119

endocarps with centered holes were assumed to have been made by agoutis . ' White-lipped peccaries shattered endocarps with a rapid closing of the jaws, leaving splintered woody fragments. Only endocarp tips were counted, to avoid counting multiple fragments of a single endocarp. • Endocarps with tooth marks clearly belonging to small rodents were found inside and at the mouths of hollow logs (too narrow for agoutis), but the consumers were not identified and such endocarps were not searched out. Counts were not exhaustive, and some endocarps incidentally buried by trampling and other disturbance under shrubs and litter were missed. Endocarp fragments from white-lip feeding were especially difficult to recover; counts of white-lipped cracked endocarps thus always underestimate the actual number opened. Endocarps that could not be aged were not included in the counts. Freshly opened endocarps are lighter and redder in color than older endocarps, and often have endosperm remnants or bits of beetle larvae frass inside. Termites often built nests around and in these hollow endocarps, and this could happen soon after the seed was eaten, masking its age. Because trees were checked frequently, however, new piles were noticed and difficult to age endocarps were few in number. At the end of the study, all individuals of four tree species known to be important to the three mammals in the study site ( Pouteria surumuensis . Pradosia surinamensis .

PAGE 120

110 Astrocarvum aculeatum , and Attalea maripa ) were counted in a 25 m radius area around each monitored tree. The total number of endocarps eaten per tree, weighted by the number of days they were available (from the start of fruit fall to the last check) was calculated and correlated with the abundance of each tree species using the Spearman Rank Correlation test. Endocarp Use bv White-lipped peccaries: Across-Habitat Scale 1995 Attalea fruiting season . From November 199 5 to April 1996, endocarp piles remaining from the 1995 Attalea fruiting season within 6 m of the 2 km long transects in the three different habitats described above were checked every two to three weeks during the fruit availability transects (mean interval 20.76 days, range 14-30 days) for evidence of use by white-lipped peccaries. Endocarps were not counted or removed. By April 1996, the few old endocarps that remained were being covered by new fruits from the 1996 season. 1996 Attalea fruiting season . During biweekly and later monthly checks from April to November 1996, and again in February 1997, all seeds opened by white-lipped peccaries were counted in the riverine and central terra firme forests as described above for the eastern forest. As was the case for the eastern forest, in December 1996, only a subset of

PAGE 121

the trees available on the river trail was sampled due to time constraints (18 out of 73 trees available) . Endocarp Contents: Endosperm vs. Larvae To estimate seed infestation rates, larval development time under natural conditions, and the time period during which endosperm rather than larvae are available in the endocarps, endocarps were sampled several times during the study, as follows. On November 15 1996, 61 externally intact (no exit holes or signs of vertebrate use visible) endocarps were collected haphazardly from 12 trees that fruited in the early and peak Attalea fruiting season. They were cracked with a machete and the size of larvae determined. In December 1996 ten intact endocarps were haphazardly collected from all layers of the endocarp piles of the 6 most recently fruiting trees in the riverine habitat. The percent of endosperm remaining was estimated for each endocarp. For endocarps with multiple seeds, the percent remaining of each endosperm was estimated. In mid-February, 45 intact endocarps were drawn haphazardly from 9 early and peak fruiting trees on the eastern forest and riverine habitats. During the four final checks of endocarp piles (October 15 1996, November 15 199 6, December 5 1996 and February 17 1997), an index of bruchid emergence was obtained by counting the number of endocarps with exit holes on the main endocarp pile. For approximately 3 minutes the pile was searched by

PAGE 122

rapidly turning over endocarps . Endocarps with exit holes found while searching for mammal -opened seeds were also collected and counted. During the December count, only a subset of the trees was sampled. By the time beetles start to emerge, endocarp piles have often been depleted by squirrels and agoutis. Data were therefore used only as an index: no emerged endocarps, emerged endocarps present, and more than 10 emerged endocarps present (the latter category indicating a high rate of emergence) . Results Patterns of Fruit Availability Eastern terra firme forest . When all fruit and seed species except Attalea seeds are considered, there were two peaks of fruit and seed availability between November 1995 and October 1996: one short one in December January (mid to late dry season) , and one longer one from May through August (rainy season) (Fig. 4-1). If only "edible" fruits (known or suspected to be eaten by agoutis, white-lipped peccaries, and squirrels; Appendix 1) are included, the first peak disappears. Attalea endocarps were available on the ground for most of the year, except for March and April, immediately prior to the new Attalea fruiting season. In terms of numbers of patches, in this habitat they contributed little to edible

PAGE 123

113 fruit availability, but in terms of timing they were available when other edible fruits were at their annual low. Due to higher fruiting levels of the palm Astrocarvum aculeatum . Licania sp. (Chrysobalanaceae) , and Ecclinusa guianensis (Sapotaceae) in 1996-97 than in 1995-96, a midlate dry season peak in edible species did occur in 1996-97 in the eastern terra firme forest. The level and timing of fruiting of Attalea was similar in 1995 and 1996. Pouteria surumuensis . however, a species important at least to agoutis and white-lipped peccaries, failed to produce fruit in 1996 during its expected fruiting time (February-May) . This was the principal diet item for radio tracked agoutis in 1996 (Chapter 5) . / Riverine forest. Fruit and seed availability (number of patches) was lower overall in the riverine forest than in the eastern terra firme forest, but was relatively constant throughout the year, with the exception of a low period in January-February (Fig. 4-2). Edible foods other than Attalea endocarps were scarce, and peaked briefly in May and August. Both peaks were caused by Attalea pulp availability (ripe fruit falling) . Thus on the river transect the overall pattern of fruit and seed availability was driven by Attalea . Attalea endocarps were available at all times except MarchApril, and were most abundant during times when few other edible items were available. . ' „

PAGE 124

,114 Central terra firme forest . Fruit and seed availability (number of patches) was slightly lower in the central than in the eastern terra firme forest (Fig. 4-3). The same low, short peak in fruit availability as in the east occurred in DecemberJanuary , and a higher, more extended peak in late April to July, somewhat earlier than in the east. Unlike in the eastern forest, the pattern of edible fruit availability mirrored overall fruit availability. Attalea endocarps contributed even less than in the eastern forest to the edible fruit availability, but were maximally available at the time (August) when other fruits and seeds were scarce. Combined availability . For all three habitats combined, the period of lowest fruit availability was between February and March, and the period of highest availability was early April to late June (Fig. 4-4). Combining the three habitat types thus shortened the period of apparent fruit scarcity, and shifted it to the end of the dry season. Attalea Phenology , One hundred and nine fruiting Attalea trees were found along the 15 km of eastern terra firme forest, 77 along the 2 km of riverine forest, and 6 along the 2 km of central terra firme forest. The median length of the fruit drop period was 18 days on the river trail (n= 45 trees with known start and end fruit drop dates) and 15 days on the eastern forest trail

PAGE 125

(n = 26 trees) . Based on this information, and starting on April 1, the fruiting season was divided into 15 day intervals, and the number and proportion of trees starting to fruit during each interval was plotted (Fig. 4-5a) . The timing of Attalea fruit fall did not differ greatly between the riverine and eastern habitats. The fruiting period was slightly longer and the peak later by about 10 days on the river trail. Not shown on the graph are two trees that fruited on the main trail system in October 1995, and one tree that probably fruited in early April 1996; this would make the spread similar to that of the river trail. The shift to a later season on the river could be due to sampling bias, since the trees there were checked on average three days later than on the main trail, which would affect estimated ages. Sample size for the central forest was too low for comparison, but the six trees that fruited did so within the main fruiting season in the other habitats (Fig. 4-5a) . ' Endocarp Contents Beetle emergence. The pattern of emergence was similar in the riverine and eastern forests. Emergence was always gradual, spread over at least a few weeks and more commonly over several months within a single endocarp pile. In October 1996, endocarps at only 19% of all monitored trees showed emergence, and only 2% had more than 10 emerged seeds

PAGE 126

116 (Fig. 4-6) . The percentage rose above 40% in November and December 19 96, and by February 19 97 60% and 80% of endocarp piles respectively showed emergence. The percentage of trees with high emergence rates (> 10 endocarps) at any one time increased slowly, and never overtook the overall emergence rate. The proportion of endocarp piles with beetles of emergence age is probably higher than shown by sampling, since depletion of the endocarp pile by consumers reduces the chance of beetles emerging. Emergence of 10 or more endocarps occurred as early as seed age 71 days, and as late as seed age 265 days. Averaging the age at which endocarp piles first show > 10 endocarps with exit holes, for both riverine and eastern forests, the mean age at first emergence was 121 +/-46.2 days (n = 20 trees). Endospe rm availability . In December 1996, endocarps from three trees that started fruiting 62, 56 and 46 days earlier had an average of 45.2% (+/38.1%) of endosperm remaining per seed , and a combined infestation rate of 100%. Three trees sampled at the same time with seed ages 82, 82 and 86 days had only 3.9% (+/10.2%) endosperm remaining per seed and a 96.7% infestation rate. The high variation in amount of endosperm remaining exists because there can be a difference of as much as three weeks in the age of the endocarps of a single tree. In addition, oviposition only takes place after the exocarp has been removed, which occurs at a variable rate (Chapter 3). Therefore in a single

PAGE 127

117 endocarp pile both large and small larvae, or even small larvae and adult beetles, can co-occur, and the amount of endosperm remaining per seed can vary from 0 to nearly 100%. Of the 61 older intact endocarps collected in November 1996, 75.4% were infested with larvae or pupa; 67.2% had large larvae. Of the 45 seeds cracked in February 1997, 91% were infested, and 62% contained large larvae rather than pupae or adult beetles. . Endosperm content was thus greater than larvae content only for endocarps less than 1.5 months old. Since the peak in fruit fall occurred in mid-July (considering the peaks in both areas) , the peak in endosperm content should last until late August. Larvae would be maximally available 3 months after the peak, in mid-October. Because endocarps accumulate during the fruiting season, and are only gradually depleted by consumers and beetle emergence, endocarps with larvae remain available into the next rainy season (May) . Endocarp Use bv Squirrels. Agoutis, and Peccaries, Eastern Forest Out of 116 trees monitored in the eastern forest, squirrels fed on 99 trees during the course of the study (at least one endocarp opened), agoutis on 35 trees, and whitelips on 21 trees. A total of 6,801 squirrel-opened, 4,129 agouti-opened, and 258 white-lip-opened endocarps were collected. Squirrels began to open endocarps as soon as the first fruits fell (early May) . On the river and central

PAGE 128

118 forest transects, squirrels several times removed large numbers of pre-ripe fruits from the fruit bunch and opened them near the tree, removing the exocarp and still firm mesocarp in strips to reach the endocarp, but this behavior was never recorded in the eastern forest . Agoutis did not begin to open endocarps until late June, and white-lips first cracked endocarps in mid-October. Agoutis fed on the pulp of Attalea fruits from the time they began to fall and throughout the fruit fall season (Fragoso 1994, 1997b, Chapter 3, 5). They did not use fresh endocarps, instead opening endocarps whose mesocarp had already rotted or been removed several days, weeks or months earlier. Squirrel and Agouti Use of Endocarps: Temporal Patterns Squirrels . The proportion of available endocarp piles used by squirrels peaked early in the season, and then remained more or less stable (large fluctuations around a mean) for the rest of the study (Fig. 4-7) . The number of endocarps used by squirrels (weighted by number of days between checks and number of trees available) increased from May to August, peaking in late August/early September, and then declined to pre-peak levels for the last four months of the study (Fig. 4-8) . There were marked fluctuations among all checks in the number of endocarps opened, perhaps due to the use of other Attalea trees in the area not included in the study . '

PAGE 129

The peak in endocarp use by squirrels, for both % of trees used and number of endocarps opened, occurred prior to the period of low fruit availability. Squirrel use of Attalea endocarps also peaked prior to the period of maximum bruchid larvae availability. The forest-wide fruit low period coincides with an increase in the availability of Astrocarvum aculeatum fruit and seeds (Fig. 4-9), a favored food of squirrels throughout the Neotropics. In terms of numbers of patches, Astrocarvum did not contribute much to fruit availability; but because the fruits are large (8 cm diameter, second in size only to Pouteria surumuensis in this study) and easy to open, and because inf ructescences provide between 100 and 200 fruits each, they probably have a major impact on food availability for squirrels. Agoutis . The proportion of available Attalea endocarp piles used by agoutis increased steadily from late June to late August, decreasing slightly and reaching a plateau between August and November (Fig. 4-7). The number of Attalea endocarps opened increased steadily in July and August and leveled of from September through November (Fig. 4-8). The increase and peak in agouti use of Attalea endocarps coincided exactly with the decrease and low period in availability of "edible" fruits and seeds (August 6 to November 26; Fig. 4-1). The peak in use started prior to and

PAGE 130

120 continued during the period of maximum bruchid larvae availability. ''-.^i Squirrel and Agouti Use of Endocarps : Spatial Patterns I 'V Attalea trees used by squirrels were spread evenly throughout the eastern forest study site, whereas trees used by agoutis were clumped at the northwest and southeast ends of the site (Fig 4-10). These two areas correspond to high Attalea density, low tree species diversity areas. To the southeast is the "Angico slope" forest type (Milliken and Ratter 1989, 1998); to the northwest a less extensive species-poor habitat located around a Mauritia f lexuosa palm wetland embedded in the terra firme forest. The total number of endocarps used by squirrels at each tree, weighted by the number of days the endocarps were available (start of fruit drop to pile depletion or final check date) , was not significantly correlated with the number of either Attalea, Astrocarvum . Pradosia or Pouteria trees ' within a 25 m radius. For agoutis, the correlation was significant and negative with Pouteria (Spearman-Rank Correlation; r = -2.598, p = 0.0094), and significant and positive with Attalea (r = 4.544,p = 0.001). Thus agoutis used Attalea endocarps only when and where their preferred food, Pouteria surumuensis (Chapter 5) was not available. Squirrels used Attalea everywhere, indicating that it was a preferred food during the study.

PAGE 131

121 Endocarp Use bv White-Lips: All Habitats ; ' 1995 fruiting season . In the eastern forest, the first fruit transect was run on November 4 1995. On the second and third runs (November 2 0 and December 19) two trees in one patch had freshly cracked endocarps . No more cracked endocarps were found on the transect that year, but endocarps of a tree that dropped fruit in October 1995 and which was monitored every two weeks for a separate experiment were found cracked in January 1996. Thus white-lipped peccaries opened endocarps from mid-November 1995 to late January 1996. In the central forest, cracked endocarps were first found on December 21 1995, and freshly cracked endocarps were again found on January 29 1996. Thus white-lips used endocarps in this area in December and January, the same time period as in the eastern forest. On the river trail, a white-lipped peccary herd was first sighted in mid-October. The animals were feeding on Attalea endocarps and Phenakospermum (Musaceae) leaf bases. The first transect in this habitat was run on November 6 1995. Freshly cracked endocarps were found on all checks through January 2 8 1996. The period of endocarp use on the river trail thus started earlier than in the other habitats, but ended at about the same time. The overall period of endocarp use by white-lips in 1995 was therefore mid-October through late January.

PAGE 132

122 1996 fruiting season . The central forest transect was not checked between October 13 1996 and March 6 1997. During that time period, white-lips opened the endocarps at one of the six trees that fruited in 1996. In the eastern forest, white-lip-opened endocarps were first found at the monitored trees on October 15 1996, and continued to occur on all checks through February 12 1997 (Figs. 4-7 and 4-8). In the riverine forest, white1 ip-opened endocarps were first found on November 18 1996 (the previous check occurred on October 15), and continued to be found on all checks through February 17 1997. Thus the period of endocarp use in 1996-97 was nearly identical to that in 1995-96. White-lips used a relatively low proportion of available endocarp piles in the eastern and central forests (18% and 17% respectively) . The trees used on the eastern forest in 1996 (Fig. 4-10) were those located along a traditional travel route that connects two Mauritia palm swamps and one Pouteria patch (pers. obs . ) . White-lips used this route repeatedly during the study period, and cracked endocarps at the same trees over and over again as they passed through. No tree in the eastern forest ever had as many cracked endocarps as those on the river trail (47 vs. 239), even though the herd that used the river trail was the smaller of the two present in the study area (60 vs. 200 individuals, Fragoso unpublished data). On the river trail, peccaries used 89% of available trees, and visited the area repeatedly (at least every 2-3 weeks) for 3 months (Fig. 4-11) . Due to

PAGE 133

the dense aggregations of ticks that occur on endocarp piles after white-lip visits, for most trees in February 1997 the number of endocarps cracked was estimated rather than precisely counted. Based on actual counts and estimates, about 5,300 endocarps were opened by white-lips from early November to mid-February on the 2 km river trail. Endocarp use by white-lipped peccaries thus started later than use by squirrels or agoutis. If across-habitat food availability is considered (Fig. 4-4 , white-lip use of endocarps coincided with the low season only in February) . Unlike the situation with agoutis and squirrels, endocarp use by white-lipped peccaries corresponds exactly to the period of maximum larva availability. Discussion '' ' Consumption of Bruchid Larvae ' Squirrels, agoutis, and white-lipped peccaries all fed on bruchid larvae to some extent. After November, it was difficult to find Attalea endocarps without larvae, and endosperm content was low even in recently infested seeds. Endocarps incidentally or intentionally buried soon after ' fruit fall by vertebrates would probably escape infestation, but there was no evidence that consumers other than squirrels dug up previously buried endocarps. Early in the season,

PAGE 134

squirrels fed primarily on endosperm from freshly fallen endocarps. As late as November, some seeds opened by squirrels still contained endosperm (in recently fallen fruits, rodents are unable to remove all the endosperm, which is still moist and swollen, and fragments remain visibly attached to the inside of the endocarp) . Many endocarps opened by squirrels late in the season were filled with frass, indicating that the squirrel had removed a full-grown larvae rather than endosperm. Frass-filled endocarps opened by squirrels often had a single, small hole per seed, whereas endocarps with intact seeds had larger and sometimes multiple holes per seed. This suggests that it is easier to extract larvae than seeds from the endocarp. The relative energetic value of larvae vs. endosperm is unknown, but the fact that some human communities use them in preference of the seeds suggests that they may be a more digestible source of lipids and other nutrients (Chapter 2; Balee 1988). Agoutis, by virtue of the time they used endocarps, fed mostly on larva, and frass-filled agouti-endocarps were common. As for squirrels, the effort an agouti expends to open a larva-containing endocarp may be somewhat less than that for a seed-containing endocarp, especially since larvae thin away the endocarp by scraping from the inside. Both rodent species, however, essentially have to carve through a fixed endocarp thickness to reach the contents. White-lipped peccaries, on the other hand, open the endocarps by snapping the jaws shut. A hollow object is easier to

PAGE 135

12 5 . . crack in this way than a solid object, and a larvaecontaining endocarp is essentially a hollow object. When seeds were cracked with a machete, infested seeds broke on the first strike, while intact ones required several strikes. Other observations suggest that endocarp thickness may be a limiting factor for peccaries. Fragoso (1994) found that white-lips are unable to crack the thickest Attalea endocarps, whether or not they contain larvae. P. Crawshaw (pers. com.) reports finding two skulls of white-lipped peccaries where Attalea spp. endocarps embedded in the palate had apparently killed the animal. White-lipped peccaries may thus be able to use Attalea endocarps only at the time of maximum bruchid infestation. Infestation by bruchids may thus actually increase endocarp quality for consumers usually classified as ' granivores, and provide an abundant source of protein and lipids. It is not surprising that Neotropical squirrels should feed on bruchid larvae, given the degree of insectivory reported for temperate (Davis 1907, Steele et al'. 1996) and Old World tropical (Emmons 1980) squirrels, and the ability of at least temperate zone squirrels to distinguish between infested and non-infested seeds (Steele et al . 1996). Insectivory by agoutis is not as well documented, but several researchers report use of animal matter. In Mexico, 94% of 18 fecal samples contained insect parts (E. Naranjo and J. Croker, unpublished data) , and agoutis were observed feeding on vertebrate eggs, juveniles, and adults. In southern

PAGE 136

126 Brazil, captive agoutis readily killed and ate domestic chicks. (C. Castellano-Margarido pers . comm.), while Smythe (1978) predation on a Liomvs by a captive agouti. That agoutis at the Maraca study site do not hoard Attalea endocarps may be partially a result of a preference for infested seeds over uninfested ones, although the abundance of the endocarps probably also obviates the need for hoarding. Spatial and Temporal Pattern of Endocaro Use The three consumers studied used Attalea endocarps in very different ways. The northern Amazon red squirrel, as expected from its known specialization on palm seeds (Chapter 2), fed on endocarps year round. Agoutis used endocarps during the season of food scarcity, but only in specific microhabitats where preferred seeds were not available. White-lipped peccaries visited certain macrohabitats (forest types) specifically to feed on Attalea seeds: there are no other fruits besides Attalea available on the river trail in sufficient quantity to feed a large herd of white-lips. The fact that white-lips were feeding on difficult to obtain food (hard endocarps) and a probably low nutrient food (Phenakospermum leaf tissue) at the same time suggests that they were feeding on non-preferred, low value foods. On the other hand, the fact that they used Attalea palm endocarps during the time when bruchid larvae, a high-energy food, were

PAGE 137

maximally available (large numbers, and easy to open), suggests that Attalea endocarps were being used as a preferred food. A consideration of patterns of food availability weighs in favor of the second possibility. For agoutis, the time of food scarcity was probably adequately monitored by the 2 -kilometer transects. Palm wetland habitats were not monitored in this study, but they may be relatively unimportant as many agoutis in the terra firme habitats do not have access to this habitat (Chapter 5) . The simultaneous radio-tracking study (Chapter 5) indicated that for agoutis Attalea palms are not a preferred food; four agoutis tracked in terra firme forest along the phenology transect never fed on Attalea seeds, although they used Attalea pulp heavily. Agoutis relied instead on seeds in the Sapotaceae. For white-lipped peccaries, fruit availability was not adequately measured by the three transects, even when data from the three habitats are combined. White-lips rely heavily on Mauritia swamp habitats for palm pulp, seeds, and seedlings, and for invertebrates. At the end of the dry season, they may congregate at pools to feed on fish left stranded as the water recedes (Fragoso 1994) . They move over huge home ranges incorporating many macro (several kilometers in extent) and micro (several hundred meters or less in extent) habitat types (Fragoso 1994, in review). The two herds that visited the study in 1995-97 moved up to 20 kilometers from any one of the three transects (Fragoso

PAGE 138

12 8 unpublished data) . It would be difficult to monitor food availability simultaneously across their home range. Overall, they used Attalea in the scarcity season for the three forest types monitored on Maraca--the mid-dry season. However, this is also the time on Maraca when wetlands are beginning to dry and food is plentiful in the form of fish and germinating seeds. It is therefore unclear whether for an animal operating at a large scale habitat mosaic, the season of low fruit -production in one or a few habitats can be considered a season of food scarcity. For agoutis in the eastern forest, Attalea could be interpreted as a non-preferred, season of scarcity resource. Some Attalea trees that were not used by agoutis during this study period had endocarps opened by agoutis during the 1994 season. In years when preferred foods fail, as did Pouteria in 1997, agoutis may thus rely more heavily on Attalea endocarps. Smythe et al . (1996) found a similar pattern with use of Attalea endocarps by agoutis on Barro Colorado Island, Panama . " ' . .. ' . The differences in the ecological relationships between these three consumers and Attalea suggests that they will react very differently to environmental fluctuations; i.e., changes in food availability will lead to individual, species-specific rather than community level, changes in Attalea endocarp use.

PAGE 139

Figure 4-1. Contribution of Attalea endocarps to total and edible fruit and seed availability on the eastern terra firme forest transect . . .. y. .

PAGE 140

Figure 4-2. Contribution of Attalea endocarps to total and edible fruit and seed availability on the riverine forest transect.

PAGE 141

131 Figure 4-3. Contribution of Attalea endocarps to total and edible fruit and seed availability on the terra firme forest transect.

PAGE 142

132 Check Date Figure 4-4. Fruit availability summed across all three forest habitats (eastern terra firme, central terra firme, and riverine)

PAGE 143

133 40-1 Dates a) .. : Figure 4-5. Timing of fruit production by monitored Attalea trees along a 2 km long trail in each of three forest types, a) Number of trees that started fruiting during each two week period from April 1 1996 to March 30 1997. b) Percentage of trees monitored that started fruiting during the same period.

PAGE 144

134 40-1 b) Figure 4 -5, continued.

PAGE 145

135 Check Date Figure 4-6. Bruchid beetle emergence from Attalea endocarps the riverine and eastern terra firme forests. > 1 endocarp includes all endocarp pile with at least one exit hole; > 10 endocarp includes only those with at least 10 exit holes, indicating a high rate of emergence. m

PAGE 146

136 Check Date Figure 4-7. Proportion of available endocarp piles used by each of the three study animals in the eastern terra firme forest during each monthly check period from May 31, 1996 to February 17,

PAGE 147

137 Check Date Figure 4-8. Number of endocarps opened by each of the three study animals in the eastern terra firme forest during each monthly check period from May 31, 1996 to February 17, 1997, weighted by the number of days endocarps were available between consecutive checks. Dates on x-axis as in Figure 4-7.

PAGE 148

138 Check Date Figure 4-9. Availability of Astrocarvum aculeatum fruits and seeds on the eastern terra firme forest trail (Trail 2) from November 1995 to March 1997.

PAGE 149

139 Figure 4-10. Location of endocarp piles used by each of the three study animals for the eastern terra firme forest monitored Attalea population. ( • ) = Sciurus ianiventris : (o) = Dasvprocta agouti ; (A) = Tayassu oecari : (x) = not used by any consumer.

PAGE 150

Check Date Figure 4-11. Proportion of available endocarp piles used by white-lipped peccaries on the riverine forest trail during each monthly check period from May 31, 1996 to February 17, 1997.

PAGE 151

CHAPTER 5 AGOUTI ( DASYPROCTA AGOUTI ) HABITAT USE, DIET AND HOME RANGE ON MARACA ISLAND ECOLOGICAL RESERVE, NORTHERN BRAZILIAN AMAZON Introduction Agoutis are large terrestrial rodents (2-5 kg) , ubiquitous in Neotropical dry and moist forests. Eleven species range from southern Mexico to northern Argentina (Emmons and Peer 1997). Male-female pairs defend a territory and share it with their young until these reach reproductive age and disperse (Smythe 1978) . Although territories are well-delimited and agonistically defended, animals from several territories will come together to feed at a super-abundant food source such as a large fruiting tree or a bait station (Smythe 1978, Silvius pers . obs . ) . Such behavior suggests the social system could be flexible enough to shift with changing population densities and resource levels . , Agoutis are primarily granivorous, but also eat varying amounts of fruit pulp, leaves, fungi, and animal matter. Throughout their range they interact with large seeds in the Leguminosae, Palmae, Sapotaceae, Lecythidaceae , and Chrysobalanaceae (Smythe et al . 1986, Hallwachs 1986, Allen

PAGE 152

142 , 1997) but also with smaller-seeded species in the Myristicaceae , Moraceae and Burseraceae (Forget and Milleron 1991, Wenny 1998). They are reported to hoard some species of seeds when these are abundant on the forest floor (i.e. during ripe fruit fall), and to rely on them for food during seasons when they are less abundant (Smythe 1978). Hoarding, however, is habitat and season dependent (Forget and Milleron 1991, Smythe 1978) . Only one published study of agouties obtained home range sizes, from visual observations and trapping (1-2 ha, moist tropical forest in Panama, Smythe 1978) . There are no published studies of radiotracked agoutis, although papers on studies in Mexico (E. Naranjo, pers . comm.) and Colombia (0. Montenegro, pers. comm.) are currently in preparation. Approximate home range sizes are reported from two additional studies, but the methods by which they were obtained are not described (200 meter diameter area, rain forest in French Guiana, Dubost 1988; 4-6 ha in regenerating Costa Rican dry forest patches, Hallawachs 1986) . Reported population densities range from 1 to 63 animals per kin under free' ranging conditions; the reasons for these differences have been examined only in light of the role of predators in controlling agouti populations (Wright et al . 1994). This paper reports on the natural history of an agouti (Dasyprocta agouti ) population with respect to diet, habitat use, home range size, and seed handling. Five agoutis were radio tracked in the transitional forest between the Rupununi

PAGE 153

savannas and the Amazon rain forest as part of a study on palm use by Neotropical terrestrial f rugivore-granivores . The primary purpose of the tracking effort was a) to estimate the scale at which agoutis respond to fruit distribution (individual fruit, individual tree, tree patch, or habitat) and b) to describe the role of the palm Attalea marioa in their diet. Radiotracking secondarily provided information on home range size, seasonal changes in foraging patterns, dietary composition, and seed handling behavior. I use these observations to discuss the relative importance of plant distribution patterns, predation, and competition on agouti home range sizes and population densities. " Study Area Research was carried out on the main trail system on the eastern tip of Maraca Island Ecological Reserve, Roraima : ' ' state, Brazil. The Reserve and surrounding area have been described in detail in Milliken and Ratter (1998). The core trail system of about 40 kilometers in a 2 x 3 km area was established by Moskovits (1998) for a study of tortoises (Geochelone sp.). It was used again by Nunes (1992) for a study of spider monkeys ( Ateles belzebuth ) , and expanded by Fragoso (1994) in a study of collared ( Tavassu taiacu ) and white-lipped peccaries (T. pecari ) . Approximately 60 kms of trail were included in this study (Fig. 5-1) .

PAGE 154

Rainfall records on the eastern end of the island show variation between 1750 and 2,300 mm annually; during El Nino years, rainfall decreases markedly across the state. The wet season is between May and September, the dry season between October and April, with rains starting either in early or late April. The heaviest individual rains tend to fall in the dry season. During the study period covered in this paper (July 1996 to March 1997), rainfall conformed to the modal seasonality. . Floristics on Maraca are described in detail in Chapter 6. The most abundant large seeded tree species are primarily in the Sapotaceae and Palmae, and secondarily in the Leguminaosae and Lecythidaceae . Sixteen species of palm and 16 of Sapotaceae occur on the island, with Attalea maripa and Astrocarvum aculeatum the most common terra firme palms, and Pradosia surinamensis . Pouteria surumuensis . P. venosa, and Ecclinusa guianensis the most common species in the Sapotaceae. Two studies of plant reproductive phenology on the trail system (Moskovits 1985, Nunes 1992, 1995) indicate that peak forest-wide (all species) ripe fruit availability on the canopy occurs in the mid-to-late rainy season. Based on the proportion of trees sampled producing ripe fruit Nunes (1992) detected a peak from May through July, with high availability extending into September. Moskovits (1985), using counts of fruits on the canopy, detected peak fruit availability in May through June, with a marked decrease in August and September,

PAGE 155

145 lowest levels in October to December, and a gradual increase in January through April. • It Methods Radio -Tracking Agoutis were captured with agouti-hunting dogs trained by local Macuxi Amerindians. Hunter and dog walked through the forest until the dog gave chase to an agouti, which invariably ran into a hollow log. The dog stayed by the log until the hunters arrived. The agouti was extracted by cutting open the log or inducing the animal to run toward the opening, where it was caught in a bag. It was immediately mildly anesthetized with 3.33 mg/kg of Telazol (Tiletamine hydrochloride + Zolazepam) , weighed, radio collared with a Telonics® model radio collar, and replaced in the log to recover from sedation. A few days after the capture, the animal was located by radio and all the pre-existing trails potentially within its home range marked at 25-meter intervals. Subsequent locations were made by a) determining the general location of the animal from outside its home range (signal strength was an indicator of distance) and b) subsequent triangulation from the marked trail locations, usually at 50 meter intervals. Approaches to the home range were varied between tracking sessions, to avoid site-specific biases in radio

PAGE 156

signals. All bearings taken were immediately marked on a trail map. Because the tracker was very close to the animal (usually < 25 meters for final triangulation) , error triangles were small unless the animal was moving. If the animal was moving, it was tracked until a good triangulation was obtained. If it was moving very rapidly, and did not settle down within 20 minutes, tracking was stopped. Locations in these cases were included in the home range calculation, but no habitat description (below) was carried out . . ' . After a satisfactory location was obtained, the tracker walked a compass bearing to the location, searched for food items or signs of activity, and described the habitat according to the parameters listed below. Because the tracking method was intrusive, three days were allowed to elapse between tracking sessions in order to minimize disturbance to the animal. After the first few days of tracking, when they occasionally "stalked" or walked parallel to the tracker, the animals gave no indication of being disturbed by the tracker. They never startled or gave alarm calls, and usually remained at the site of the first location. Five animals were tracked between July and December 1996, and one of these continued to be tracked in February-March 1997. Because the main purpose of radio tracking was to examine habitat use during foraging, the animals were tracked at those times when they were most likely to be foraging

PAGE 157

rather than patrolling or resting: between 0830 and 1100 in the morning, and between 1430 and 1730 in the afternoon (sunrise and sunset were at 0600 and 1800 respectively) . The choice of times were based in information in Smythe (1978), modified by the results from initial tracking sessions. Twelve-hour daylight follows on Maraca confirmed that between 0600 and 0800 the agoutis moved rapidly between widely separate points in their home ranges, suggesting patrolling activity. By 1100 hours, they were often moving to or already at preferred resting areas in recent treefalls. Tracking sessions on individual animals were alternated between morning and afternoon, so that both times were equally represented for all individuals. Animals were not tracked during heavy rainfall, but all other weather conditions were represented. On 10 occasions, agoutis were tracked from about 0600 to 1800 ("twelve hour follows"), with radio locations every half hour or every hour, to confirm movement patterns and home range use. Locations were not inspected or habitats described during these all day follows, except on three occasions when they were used to augment the data set and the animal had not been tracked within three days of the session. Habitat Description at Location Food ea ten and available . The ground within of each accurate radio-location was searched for a 5 m radius partially

PAGE 158

148 • , . eaten food items or signs of digging and all fruit or seeds were noted. A distinction was made betv/een "eaten" foods, those that had bite marks on them or were associated with a hole or scrape mark or regular feeding area (a cleared area on the ground with food items around them) , and those that were "available," that is, in the immediate area were the animal had been but not eaten or associated with foraging signs. If an "eaten" item was clearly not fresh, but had v been used within the last three days, it was still recorded. Presence of focal tree species . The observer qualitatively determined whether the location was within a patch of one of four largefruited (5-20 cm diameter) tree species important to seed-eating animals in the study site: Attalea marioa . Astrocarvum aculeatum . Pradosia surinamensis and Pouteria surumuensis . A location was considered to be within a patch if trees of the focal species completely surrounded the location, with at least 3 trees within 10 meters . ' " 'if'Agouti Encounte r Transects In March, April, May, July, and August 1996, an experienced Macuxi tracker walked on 86 occasions through different parts of the trail system. He noted all mammals seen, including agoutis, and described their behavior. If the animal was feeding he identified the food item. All

PAGE 159

trails were covered every month, and sometimes every week. The walks lasted usually from 0730 to 1130, and from 1345 to 1715. • . ' . . Use of Attalea Endocarps ' Use of Attalea maripa endocarps by agoutis v/as monitored on a monthly schedule on the trail system (see Chapter 4 for detailed methods) . For Attalea palms that fruited within the radio-tracked agoutis' home ranges, all endocarps opened by agoutis within a 5 m radius of the trunk were counted and removed . Phenology . . A 2 km long by 3 m wide transect on a pre-existing trail that runs through or between all the marked agouti territories was walked at 2-3 week intervals from November 1995 through March 1997. All fruits and seeds seen on the ground were counted and identified, with each individual item or cluster of items counted as a single patch. Both ripe or newly fallen items and older seeds that remained on the ground, sometimes for months, were counted, and their stage identified (immature, preripe (full size but immture color), ripe, rotten, and germinating) . If the items showed signs of having been fed on (tooth marks, disturbance, or fruit fragments), the most likely consumer was noted (e.g., tortoises as well as agoutis fed on Pouteria seeds, but their

PAGE 160

150 ^ . : larger "beak" marks could be distinguished from rodent tooth marks, as could the irregular seed fragments left behind). , : The ground monitoring method was chosen for three reasons: a) agoutis can only feed on fruit that falls to the ground; b) because there is a delay in the availability of food items to terrestrial f rugivore/granivores compared to arboreal consumers (Zhang and VJang 1995); and c) because agoutis feed on older seeds and germinating seeds whose availability would not be measured by canopy observations or fruit trap sampling. Because fruits are rapidly removed by terrestrial animals (Zhang and Wang 1995), a patch based enumeration was used rather than individual fruit counts, and no attempt was made to estimate the total number or mass of fruits and seeds falling to the ground. Fallen fruit were used as an index of the overall pattern of fruit availability. Although ground transects may fail to capture some fruit species that fall to the ground in low numbers (Zhang and Wang 1995), because agoutis feed on large fruits for which remnants of feeding can be found and counted, this error was minimized. To avoid disturbing the agoutis whose home ranges overlapped the transect, the ground was not raked nor the counted fruits removed. The number of a) patches and b) species of fruits and seeds considered to be usable by agoutis was plotted for each transect sample (Appendix I lists the species considered) . For Pouteria surumuensis . Attalea marioa . Astrocarvum aguleatum, and Pradosia surinamensis . the availabilities of

PAGE 161

151 unripe, ripe, or rotting fruit, and old intact seeds were plotted individually. Data Analysis Because all nimals occupied similar habitat types, and all had roughly equal amounts of the four focal tree species available to them, data from all animals were combined for habitat analyses to increase sample size. The initial intention was to compare the habitat descriptions of agouti locations with randomly chosen locations within the home range of each animal. However, the home ranges were so small, and the locations so near to each other, that this proved impracticable. Instead, all locations from July to December were plotted on a single map for each animal. The three areas with highest density of locations ("core--"), and the three areas with lowest density of locations ("holes"), each with at least a 25 m radius, were selected visually from the maps. The core and hole areas were then located on the ground, and all reproductive size individuals of each of the four focal tree species located within 25 meters of the estimated center point were counted. The number of canopy gaps and hollow logs in the area where also counted, as these are two other variables that could affect the presence of agoutis. Center points were located bv measuring the distance on the map along a bearing from a 25 m mark to the center of the hole or core

PAGE 162

152 " ' / area, and then walking this distance along the bearing in the field. Data on the hole and core areas of all the animals were combined, and the number of trees of each of the focal species compared with a Mann-Whitney U test. Trees of the four focal species were counted along the 2 km X 3 m fruit availability transect. Their frequency of occurrence was compared with a Chi-Square test to the frequency of the same species in a forest inventory plot in the same area (Milliken and Ratter 1989). Since there was no difference between the two distributions, the relative frequency of trees of the four species on the transect was compared to the frequency with which radio tracked agoutis were located in a patch of each of the four tree species. Home Range Calculations ' . Home range size was calculated us^.^g the program Wildtrak® (ISIS Innovation Ltd. 1993). Exact locations on the field maps were translated to coordinates on a 7.9 m grid and the coordinates were entered in the program. 100% contour lines were used (i.e., all locations on the boundary of the range were included) because, due to the nontraditional tracking protocol described above, locations were considered minimum indicators of home range size. All locations, including repeat locations during single tracking sessions, and those obtained during all-day follows, were used for home range size calculations. Seasonal shifts

PAGE 163

153 using only the and locations from . ' ; Results Captures and Survivorship Six female and two male agoutis were captured between February and August 1996. Only one male and four female agoutis are included in this study. Individual agoutis were tracked for between 2 and 5.5 months, during the mid to late rainy season (July to November, 5 animals) , and the dry season (February and March, 1 animal) . Variation in length of period tracked was caused by death of animals and radio failure at different times. A total of 197 locations with habitat descriptions were obtained (Table 5-1). Each animal is henceforth identified by its radio frequency. Female 167 was found dead on November 23 1996. She had been killed the previous day, and had been tracked the day before that. The collar, still around the body, showed sharp bite marks, and the antenna was twisted by chewing, suggesting predation by an ocelot ( Felis pardalis ) . When the study was interrupted in the first week of December, the other four animals were still alive. In the first week of February, when the study was resumed, the signal for Male 148 withm a home range, however, are depicted first location from each tracking session, all day follows are not included. , ' •

PAGE 164

154 could not be located. That radio had been active for the longest period of time (ten months) and the batteries had probably failed. The radios for Females 158 and 170 were found with bite marks like those on 170, but no animal remains were associated with them. One radio was adjacent to a preferred denning site, the other had been carried into tangled undergrowth in a Mauritia palm stand outside the known home range of the animal. The lack of smell and remains, together with the condition of the collars, suggested the animals had been dead for at least 2 or 3 weeks. The fifth animal. Female 163, was still alive at the end of the study, 9 months after it had been radio collared. One female radio collared in February 1996, but not otherwise included in this study, was found dead and partly decomposed/scavenged in March, with no tooth marks on the collar or other signs of predation. The period from November to January, when animals 167, 158 and 170 died, corresponds to the mid-dry season and also the season of lowest fruit availability . Diet . -i ' Food items with agouti teeth marks or associated with agouti signs were found within a 5 m radius of the location at 140 out of 197 radio locations that were checked. During 86 walked transects from March 7 to August 30 1996, agoutis were encountered 152 times and 29 feeding

PAGE 165

observations were obtained. Eight items were recorded during casual observations, four of these from exotic plants planted near the research station buildings. Thirty eight food items from 19 wild and 4 cultivated plant species in 11 families were identified (Table 5-2). Plant parts included, in order of importance: seeds, pulp, cotyledons, leaves and flowers. In addition, agoutis also fed on bruchid beetle ( Pachvmerus cardo: Bruchidae ) larvae from palm nuts and on fungi. The most frequently eaten species during radio locations were Pouteria surumuensis (29.3%) and Pradosia surinamensis (20.7%), both in the Sapotaceae, and Attalea marioa (12.9%), in the Palmae. For the two Sapotaceous species only the seeds were eaten, while for the palm the pulp was the item most frequently eaten. During the mammal transects, Pouteria and Attalea were also the most frequently eaten items (74% and 22%, respectively), but Pradosia was not noted. The fruiting periods for Astrocaryum, Licania sp., Coueoia sp., and Licania kunthiana . the next most frequently noted species in radio locations, did not coincide with the transect period. Use of Attalea marioa Endocarns , . ' . " Female 163 had the highest density of Attalea palms in her home range. Five trees dropped fruit prior to or during the radio tracking study (6-20-96 to 9-1-96) . Agouti -opened

PAGE 166

endocarps were first noticed at two trees in early September 1996 (Table 5-3). Four trees had opened endocarps when checked in mid-October; and all five in mid-November 1996. In mid-February 1997, all trees again had agouti -opened endocarps. Peak endocarp use occurred in November through January (6.17 to 4.62 endocarps per day). On the tree most heavily used between November 1996 and February 1997, many of the opened endocarps were lying around a cleared area on the ground typical of agouti feeding sites (Smythe 1978) . The remaining agoutis had at least between one and five fruiting Attalea trees within their home ranges. No endocarps were opened by agoutis at any of these trees during the radio tracking study or by February 1997. Hole-Core Description : Core areas had a significantly higher number (MannWhitney U, p = 0.007) of Pouteria surumuensis trees (3.36 +/1.7) than did hole areas (1.57 +/1.34) (Table 5-4). All other parameters were virtually indistinguishable between hole and core areas, except that there was a non-significant tendency for hole areas to have more large gaps (Table 5-4) . Patch Choice • , For the July through December observations, there was a significant difference between the frequency of agouti

PAGE 167

157 locations within patches of each of the focal trees and the frequency of the occurrence of focal trees along the transects (x2 = 7.95, p = 0.05) (Table 5-4). Agoutis were more likely to be found near Pouteria patches and less likely to be found in Attalea patches than expected. Phenology _ With the exception of July, most of the agouti radio tracking study coincided with the usual environmental period of decreasing or low fruit production in the study site. The availability of species considered usable by agoutis in this study (Appendix I) followed the same general pattern found in the previous studies, but the pattern was more marked for the number of fruit and seed patches (Fig. 5-2a) than for the number of species of fruit (Fig. 5-2b) available. The much higher availability of species and patches in early 1997 than in early 1996 was due to heavy fruit production by two species in the Sapotaceae, Pouteria venosa and Pouteria hispida . and one palm, Astrocarvum aculeatum . The latter had a weak fruiting season the previous year, and the two former species aborted all their fruits at the immature stage in 1996 due to heavy insect infestation. The four core species used by agoutis had disjunct fruiting patterns. Attalea palms began to drop ripe fruit in late April, peak fruit production occurred in June-July, and some trees produced fruit until September (Fig. 5-3). The

PAGE 168

endocarps remained on the ground into the next year; in fact, Attalea maripa endocarps are plentiful on the ground all year except for the four months { February -May ) prior to and during the start of the fruiting season. " ' . " • Pouteria surumuensis (Fig. 5-4) did not drop ripe fruit until March, but preripe fruits with edible seeds (as indicated by agouti use) were available from January, and bare seeds with the pulp removed or rotted were plentiful on the ground until August. After that, most of the seeds found on the ground were partially eaten, germinating, or partially rotten. Tracking and transect observations were interrupted in December 19 95 January 1996. This was the time when Astrocarvum aculeatum ripe fruit was maximally available (Fig. 5-5), and it probably formed an important part of the agouti diet at that time, based on the number of seeds cleared of pulp and gnawed open that were found in February. Seeds remained abundant on the ground into April, when the study ended. ' Unlike the above species, Pradosia surinamensis germinated very quickly after falling and was available for a short time as germinating seeds (Fig. 5-6) . Heavy use by primates limited the fall of ripe fruit in this study area, though a few kilometers away more ripe fruit was available. In fact, most of the fruit that fell to the ground had been opened by primates, which fed on the pulp and dropped the seeds. Pradosia also had an irregular to biannual fruiting

PAGE 169

cycle, so that although seeds were available for short periods of time, they were produced frequently. Agoutis fed heavily on the seeds when they were available. When the fruiting patterns of the four primary food species are combined (Fig. 5-7), the period of maximal fruit availability occurred from May to August, with a second peak induced by Astrocarvum and Coueoia sp. in January-March. Agoutis were already feeding heavily on Pouteria in March 1996, based on the transect observations. Pouteria was abundantly available since January, and the first agoutignawed seeds and fruits appeared in February during the phenology transect. Agoutis continued to feed on Pouteria throughout the tracking period, to the end of November, at a fairly similar rate, indicating that they were feeding on buried seeds. The proportion of observations on Pouteria was lowered by data on the one agouti. Female 163, that had low availability of these fruits in its territory. Thus, agoutis used a single food item for nearly the entire year, and relied on it heavily for at least 7 months (February-August) . Seed handling , „ . ' • "i: Pradosia surinamensis . Agoutis always removed the endocarp and seed coat on all stages of Pradosia seeds. They foraged frequently among the germinating seedlings, uprooting them, but no evidence was found that cotyledons were being eaten. In one case, three clipped seedlings were found with

PAGE 170

160 only the root remaining, suggesting that the stem and leaves were eaten. Pouteria surumuensis . Initially, intact fruits or fruits with the entire seed removed were found on the phenology trail, with no evidence of seeds being eaten on the spot, suggesting they were being hoarded. By July, many seed chip piles or seed coat piles were found at these same sites, indicating that the seed was being manipulated prior to removal (seed coat removed) or eaten on the spot (seed .; chips ) . ' Pouteria seeds began to germinate about a month after ripe fruit drop. Some seeds left above ground developed a green layer under the seed coat, and these were ignored by the agoutis, which appeared to prefer white seeds. Several observation of germinated seed and seedling handling from July 1996 to March 1997 suggest that in addition to burying seeds for storage, agoutis were removing germinated tissue from Pouteria seedlings, without feeding on the seed at the time the seedling was cut. In three cases, seeds were dug up or partly eaten that had obviously had their roots and stem clipped at least once before and had then been reburied. In one case, the agouti clipped the stems of two seedlings in a clump of eight seedlings, without digging up the seeds. In two cases the seeds were dug up and removed/eaten from seedlings, with the stem remaining in the ground. In one case, the agouti was startled away from the seed just after

PAGE 171

161 clipping the seedling. In another, it had been digging at the base of the seedling and was startled before reaching the seed. In one case, the agouti dug up the seed, clipped it away from the seedling (root and stem remained in ground) and had a hole 45 cm away where it was about to bury the seed when startled. The seed oozed plentiful milky sap where it had been cut. Foraging near Pouteria trees was thus not limited only to finding, eating and burying seeds. Core Area Use and Observed Movements with Respect to Fruiting Trees " " r' ' . . ^ * / Movements and core areas of the radio tracked animals were strongly influenced by 1) the location of individual fruiting trees, and 2) the location of logs or tree-falls or areas of tangled branches/vines. Agoutis tended to forage near one or two Pouteria trees at a time during a period of from a few weeks to two months, creating within their home ranges a pattern of core areas centered on these trees. In several cases, and especially late in the tracking period when Pouteria seeds had been depleted, core areas focused on trees of other species (e.g., Licania kunthiana for Female 170, CQVgpi» sp. for Female 163). Distances between fruiting trees of all species were large enough to create distinct concentrations of locations on the maps. Because fruit species were available at different times, the shift of agoutis from one core tree to another created a seasonal shift in the area of the home range most heavily used.

PAGE 172

162 Figures 5-8 through 5-12 show these seasonal shifts when the tracking period for each animal is divided into two approximately equal time periods. . • , ^ In several instances preferred feeding trees were associated with shelter areas. Thus Female 158 (Fig. 5-8) fed consistently on a Pouteria tree in the eastern portion of its range, and sheltered in a large hollow log adjacent to it. When seeds were depleted, she started to use a Pouteria patch on the southern boundary of her range, where she was once chased by a neighboring agouti. Although seeds had been available there all along, she did not shift to that area until November. ' ' • Observations of the one agouti that was tracked across a change of season, Female 163 (Fig. 5-12), shows that this season shift extended into three different periods of fruit availability. Her home range use pattern also suggests that the fringe areas of the home range, as well as unused areas in the center of the range ("holes"), may be heavily used at other times of year as different trees come into fruit. Two areas that had previously been unused became centers of activity in November and February, due to the presence of fruiting Coueoia sp. and Tetraaastris panamensis trees. The use of preferred shelter areas affected the core area pattern. For Female 163, the heavy use of the area near the 50 meter mark on the clearing edge is due to the presence of a vine tangle where the agouti sheltered her young. The young animal was subsequently found dead near this area. On

PAGE 173

this and one other occasion when an agouti had young, its movements became very restricted for about two weeks, . focusing on a shelter area. Prior to a range shift in October-November Female 167 (Fig. 5-9) used a den in August in the southern part of her range, even though the rest of her activities were concentrated to the north. During one all day watch, this female did not move more than 50 meters from the den site. , With the above exception, locations obtained during all day watches usually covered one half to two thirds of the animals' previously known home ranges, and at most extended the known range by 10 or 2 0 meters, without adding completely new areas (Table 5-6, Figs. 5-13 through 5-15). During a single day the animals spent most of their time in one or two areas, with occasional movements to the edges of their range once or twice a day. Only one agouti, Male 148, had a Mauritia f lexuosa palm wetland in its home range, and it began to use that area when Mauritia was dropping flowers and immature fruit. This animal and one other had large Licania kunthiana trees in their home ranges, and both changed their movement patterns at one time to use them. Male 148 had Attalea seeds available in its preferred Pouteria core area, but never used them--there were no gnawed seeds and no sign of digging. When Hymenea courbaril and H. parvif olia trees in three of the home ranges flowered, all three agoutis began spending time near the tree, where they had not been before. Both

PAGE 174

164 Hvmenea species usually drop the previous season's fruits at the same time that it flowers (Miller 1991), but during the study period only one of the trees dropped a few fruits. Although the agoutis fed on other items besides Pouteria while Pouteria was available, their movements depended on the location of the Pouteria trees. They fed heavily on Attalea pulp, but, with the exception of Female 163, did not have core areas in Attalea patches, perhaps because they do not hoard Attalea seeds. . , Home Range Size and Densities Home range size, calculated from both the foraging locations and the all day watch locations for each animal, ranged from 3 ha to 8.5 ha (Table 5-6). Due to the tracking methodology, which did not include all times nor the times of maximum patrolling activity, this is a minimum home range size estimate. Observations of unmarked animals in areas in between monitored home ranges suggest that, at least in the immediate study area of the main trail system, agouti home ranges are closely packed together. Based on an average home range size of 5 ha, and assuming there are at least two adult agoutis per home range, the density of agoutis in the study site was estimated at 40 animals/km^. This is at best a slight underestimate, because juvenile agoutis share territories with their parents.

PAGE 175

165 Discussion Food Availability and Use of Space ' ' ' The use of space by agoutis in this study was similar to that described by Smythe (1978), whose study was based on trapping records and observations of animals near a baiting station. Agoutis in both cases had core areas in their home ranges (called "territories" by Smythe), and in both cases these were determined by the location of productive trees. When Pouteria seeds became scarce, agoutis on Maraca moved away from their previous core areas. The results of both studies indicate that agoutis respond to their habitat at the individual tree scale, and that, despite their ability to hoard seeds, they require several trees of different preferred species within their home range to buffer both seasonal and supra-annual variation in food supply. .' At least two of the fruit species important in agouti diets (Pouteria and Astrocarvum ) alternated years of heavy fruit production with seasons of low fruit production during this study. Moskovits' (1998) 18 month fruiting phenology study also shows that the height of peaks in fruit availability can differ greatly between years, as different species fail to fruit. Licania sp . , which was heavily used by agouti 163 in 1997, was also not available during the same season the previous year. Rare trees, like Swartzia aoetala

PAGE 176

166 . and Hvmenea spp., which were not recorded as being heavily used during the study, are probably very important during certain years. Spatial as well as temporal differences in fruit availability affected the foraging behavior and diet of individual agoutis in the tracking study. Female 163, whose home range was located in the low diversity, high Attalea density "Angico slope" forest type (Chapter 4), relied heavily on Attalea endocarps from October 1996 through at least January 1997. The fact that none of the other agoutis, in higher diversity habitats and with high Pouteria availability, used Attalea endocarps indicates that this palm is not a preferred food species. In a separate study of Attalea use by three mammalian species (Chapter 4), agoutis used endocarps only in the area adjacent to Female 163 's home range, and in one other species-poor area bordering a Mauritia palm swamp. Both habitat types or plant , ... . associations were larger in extent than an agouti's home range, forcing animals established in that area to rely on non-preferred food items, at least during the study period. An experimental study with white-winged trumpeters (Pgophia leucoptera : Aves) in the Peruvian Amazon (Sherman and Eason 1998), showed that, for that fruit-eating species, the permanently defended territory size was determined at least in part by the need to provide sufficient food resources during the season of lowest fruit-availability, even if during the season of high availability resources

PAGE 177

became superabundant. This should be true of any fruit and seed eating animal with a well-delimited and defended home range. In the case of agoutis, however, given their ability of to hoard seeds and to rely heavily on one tree species for several months, territory size must ensure an adequate food supply for supra-annual periods of food scarcity rather than seasonal periods. These would be times, for example, when palms and species in the Sapotaceae simultaneously fail to fruit. In both cases, the spatial distribution of trees and tree patches of different species in a habitat will have a • strong effect on the home range size of fruit-eating animals. Because home range size influences other aspects of ecology and behavior, such as population density and the intensity of territory defense possible, the spatial distribution of trees can have an important impact on the structure of a community of fruit and seed eaters by determining the density and possibly competitive interactions among the members of the community. Several species of fruit and seed eaters respond to fruit availability at the same individual tree scale as agoutis. Enunons (1980) describes use of "one to three localized areas" centered around fruiting trees for days at a time by African squirrels. Adler (1994) showed that spiny ^^ts (Proechvmis semisoinosus ) in Panama responded in a very localized way to fruit and seed availability. In his study, rat populations on small islands (1.7 to 3.5 ha, smaller than an agouti home range) showed among-island differences in '

PAGE 178

168 breeding season that corresponded to differences in the time when the dominant tree species on each island produced fruit. For small-bodied fruit and seed eaters, whose home ranges are smaller than the average extent of a plant association in the area they inhabit, differences among home ranges in the tree species available should alter demographic parameters among individual animals. One animal may breed earlier or have higher offspring survival than a neighboring animal, simply because of the identity of the trees that occur in its home range. Such a demographic response might tend to dampen population fluctuations in tropical forests, because mortality due to seasonal and minor supra-annual fluctuations in food availability would be spread out among individuals and among years. The extent to which this happens will be determined by the floristics of the area--species richness, number of rare vs. dominant species, and clumping tendencies of different species. Home range size and population density Home range size can affect population density by determining area used per individual or social unit. Other factors, such as predation, resource patchiness at the larger than home range scale (which would affect spacing between home ranges), food productivity levels, and competition with other species for food resources can affect population density. Comparisons of sites where population densities

PAGE 179

169 have been estimated or calculated illustrate the difficulty of separating among these factors. . ^ , Home range sizes on Maraca were similar to or slightly larger than those reported by Hallwachs (1986; 4 ha) and Dubost (1988; 200 meter diameter area), but larger than those reported for Barro Colorado Island (BCD by Smythe (1978; 1 2 ha). Agoutis were not radio-tracked in those studies, and methodology may have affected the range estimates. Additionally, on BCI home ranges were obtained for animals that used a baiting station; home ranges more distant from the bait site may have been much larger. It is also unclear whether Smythe includes all the area used by an animal in his territory estimates, or just the well-defended core areas. Some of Smythe 's animals were juveniles, which may have smaller or less well-established ranges than adults; all animals in this study were adults. (This age difference may be a function of the capture methods, because dogs are likely to remain with an animal that goes into a shelter hole, and adults with established ranges are more likely to have reliable shelters than juveniles.) It is also possible, however, that the number of young animals is reduced by predation on Maraca relative to BCI, given the depleted predator population at BCI (Wright et al . 1994). To the ' extent that home range size is affected by copulation density, this would tend to both increase home range size and decrease apparent population densities on Maraca relative to BCI.

PAGE 180

17 0 , A comparison of densities reported from dry and humid , forest in Central and South America shows surprising \\ . consistency and a few marked contrasts (Table 5-7). The lowest documented agouti densities are for the four sites in Peru. In the absence of information on home range sizes at these sites, it is difficult to determine the causes for density differences. As suggested by Janson and Emmons (1990), the low densities at Cocha Cashu, Peru, may be explained by periodic flooding of the study site; on Maraca, agouti sightings decreased markedly in riverine forest during the flood season, even though only a small portion of the transect itself was affected by flooding. This implies that either extreme shifting of habitat size or actual dispersal away from flooded areas occurs in these habitats, behaviors that have not yet been documented in tracking or capturerecapture studies. The remaining densities are within the same order of magnitude, from 27 animals per km^ in Mexico (Coates-Estrada and Estrada 1986) to 64 in French Guiana (Dubost 1988). The most anomalous density is that of 63 agoutis per -km^ reported by Eisenberg et al . (1979) for a seasonal forest in northern Venezuela. • ' Several methodological biases may affect the reported densities, masking real ecological differences. Low ' • densities at Bodmer ' s (unpublished' data) sites in Peru may occur because animals are completely unaccustomed to humans other than hunters, and would tend to freeze at the approach of a human. At BCI, densities may be inflated relative to

PAGE 181

other sites due to extreme habituation to humans (Wright et al. 1994). The time of sampling, in relation to seasonality of fruit production, can also affect density estimates. On Maraca, there was a high frequency of feeding observations on Pouteria and Attalea for transects occurring when these preferred foods were in peak fruit fall, whereas there were no feeding observations on rarer food items. This suggests that at such times agoutis are very active around fruiting trees, making them more likely to be spotted. When food is more dispersed or less abundant, they are less conspicuous in their foraging behavior. This variation in behavior could affect density estimates obtained through transect sampling. Peres et al . (1997) do not cite the duration or season of their transect study, so it is impossible to judge the effect of either food availability or length of study on their data. In French Guiana, agouti home ranges may not be as closely packed as on Maraca and BCI, due to the presence of acouchies; the density estimate based on home ranges may therefore be an overestimate. At Cocha Cashu, adding agouti and acouchy densities doubles the density of large, seedeating diurnal rodents and brings it closer to other estimates. It may also be important to consider the effects of competitors such as white-lipped peccaries on agouti densities. The small home ranges on BCI and higher densities in hunted than unhunted areas in Peru (Chapter 4; Bodmer et al. 1997) suggest that competitive release may allow an

PAGE 182

. . 172 increase in agouti densities. In Peru, decreased primate populations may result in an increase in the availability of some species in the Sapotaceae to agoutis. On BCI, the absence of white-lipped peccaries would decrease some " pressure on seed resources. Agouti densities on Maraca may currently be high due to a low white-lipped peccary population level; at times of high white-lipped peccary densities, anecdotal information points to a depletion of seeds and seedlings (J. Thompson, pers . comm.). During the current study, one agouti was noted to shift its core area after a white-lipped peccary herd passed through its home range and fed on Pouteria seeds in its core area. Both collared and white-lipped peccaries, as well as Echimyid and Murid rodents, feed on palm and Sapotaceae seeds, and their population densities should also be considered when examining competitive effects on agouti densities. This interaction with members of the terrestrial seed-eating guild may be as important as the presence or absence of predators in determining the population patterns of agoutis (c.f. Wright et al. 1994) . . . * . • Di^t; •. Agoutis on Maraca relied on a few plant species during the study period. This is similar to the situation on BCI (Smythe et al . 1996), where agoutis relied on the seeds of three plant species throughout the year, even though they

PAGE 183

included a total of 36 species in their diet. On BCI, agoutis also fed heavily on non-preferred fruits such as Tetraaastris panamensis in years when those were abundant, indicating that important diet items may vary among years. Smythe (1978) also recorded agoutis feeding on fungi, and E. Naranjo (unpublished data) recorded frequent use of invertebrate animal matter. On Maraca, reliance on animal matter resulted from the use of beetle larvae in infested palm seeds (Chapter 4). Sites that do not offer an array of a few abundant large-seeded items whose availability (phenology) is spaced throughout the year may not provide adequate habitat for agoutis, despite their ability to store seeds. This could be a partial explanation for low agouti densities at the Peruvian sites listed in Table 5-7. The high agouti densities in Peres et al . ' s (1997) southeastern Brazil site may be due to the availability there of Brazil nuts (Bertholletia excelsa ) in conjunction with Attalea fruit and nuts. This extreme patchiness and availability of largefruited species may be a characteristic of transitional forests (between moist forest and savanna) , which occur at both the southeastern Brazil and the Maraca study sites, making such habitat particularly suitable for agoutis.

PAGE 184

174 Seed Handling .'v. The seedling clipping behavior exhibited by agoutis in this study is typical of many rodent species in a variety of habitats and biomes, and is probably a way of controlling seed degradation due to germination (Vander Wall 1990) . Such behavior by agoutis supports the hypothesis that large size in Netropical seeds is an adaptation for surviving herbivory (Balling et al . 1997, Harms and Balling 1997, Mack 1998). But it may also be related to toxicity or ease of digestion of dead or older seeds. Collared peccaries were observed to scar Pouteria seeds and then drop them, and to uproot Pradosia seedlings and then return to feed on them a week later. Agoutis also uprooted Pradosia seedlings that they did not eat, although they frequently ate freshly fallen or older, non-germinating seeds. Agoutis may thus manipulate seeds for reasons not directly related to food storage.

PAGE 185

Table 5-1: Number of locations per animal. Animal ID Sex Tracking Period # of Log . Allday follows Fate 148 m 7-9-96 to 126-96 37 1 radio failed 158 f 7-9-96 to 11J u y 0 38 2 killed 163 f 7-6-96 to 124-96; 2-10-97 to 331-97 41 20 2 3 survived 167 f 7-3-96 to 1123-96 37 1 killed 170 f 9-28-96 to 11-28-96 17 1 killed

PAGE 186

176 Table 5-2: Eaten food items, with number of observations per item, from radio locations, transects, and casual observations . Species Fami ly Part Radio lor* Tran Q o t~ a o C ^ L> O CaS U. a X Pouteria surumuensis Sapotaceae pulp 1 0 0 seed* 37 20 1 ing * * -> j> Pradosia surinamen sis seed 22 1 seed1 ina 7 Pouteria venosa s eed— ling 1 Attalea maripa Palmae c; ~j nut * * * 7 Astrocaryum aculeatum duId 5 1 X imm . endosperm 2 nut * * * 3 Mauri tia fl exuosa f ruit+ 1 Chrysobalanaceae sp. Chrysobalanaceae f ruit+ 3 pulp 4 seed 6 Licania kunthiana seed 1 seed1 ing 9 Hymenea sp. Lequminosae fruits1 seed 1 seedling 1 flower 1 Inga sp. imm. fruit 1 1 Tetragastris panamensis Burseraceae f ruit+ 1

PAGE 187

177 Table 5-2, continued. S D e c i e a ^ (3 m T 1 \7 C GL 111 X X It cl U R^d i o loc . Transects Casual T~M 1 1 r~^ Puip Brosimu/n sp@ Moraceae seed 2 Eschweieira W 'w' ^ ^ L.^ t-* L. w f ruit+ Passi flora SD Passi f lor^ £11 ;3 in CI C. CI C f ruit+ 3 J. X xxxdv^cdc — — t ru 1 1 + Anopvn r* sp . seed— 1 1 n rr 1 Cashew# Anacardi 6.ceae (aril ) A seed 7 Mangi f era indica# seed z Apple* Rosaceae seed 1 Guava# f ruit+ 1 Tin ]
PAGE 188

178 Table 5-3: Number of Attalea maripa endocarps opened by agoutis on Female 163 's home range during the radio tracking study. Number of Endocarps Opened Tree Start Drop 9-7-96 10-15-96 11-15-96 2-12-97 D2 6-20-96 0 1 33 14 Dl 7-10-96 0 73 87 146 D3 7-20-96 9 52 46 16 D4 8-15-96 2 8 13 22 AN2 9-1-96 0 0 6 204 Total Opened 11 134 185 402 Period Length 3 9 days 3 0 days 87 days Endocarps /day 3 . 44 6.17 4.62 Table 5-4: Number of four focal tree species, gaps, and logs in hole vs. core areas. Species or Parame t er P -value Core mean Core s . d . Hole mean Hole s . d. Attalea 0 .44 2.21 1.76 2.21 3 . 17 Astrocarvum 0.40 1.86 1.79 1.36 1.69 Pouteria 0 . 007 3 .36 1 .74 1.57 1.34 Pradosia 0 . 84 2 . 57 2.24 2.29 1 . 86 Small gap 0.4 1. 14 0 . 86 0 . 93 1 . 07 Large gap 0.11 0.29 0.61 0.79 0 . 97 Log 0 . 98 0 . 5 0.23 0 .43 0 . 17

PAGE 189

Table 5-5: Trees of four key species found along the 2 km fruit availability transect, compared with the frequency with which agoutis were located in patches of the same four species. Attalea Astroca ryum Pouteria Pradosia Transect Trees 66 26 44 44 Radio Locations 44 29 63 43 Table 5-6. Home range size, day ranges, seasonal ranges, and number of locations used to calculate range sizes for the five agoutis tracked in this study. Ag out i Time Period Range Size ( na ) No. of locat ions ±^ O [ill ) n 0 Q £^ 1 O 1 Q £Z /-y-yo IZ-Z-yb 0 . 50 47 /-y-yo y-ib-yb 7.69 22 9-16-96 12-2-96 4 .79 25 158 (f) 7-9-96 11-30-96 5 95 D D 7-9-96 9-15-96 3 .36 19 9-16-96 11-30-96 3 . 08 25 12 -hour range 2 .27 24 163 (f) 7-6-96 3-31-97 2 . 94 123 7-6-96 9-15-96 1.53 26 9-16-96 12-4-96 2.21 25 2-13-97 3-31-97 1 . 13 25 12-hour range, 1121-96 2 .24 28 12 -hour range, 2-397 0 .41 14 12 -hour range, 221-97 1 . 60 15 12 -hour range, 312-97 0.71 13 167 (f) 7-11-96 11-23-96 5.73 44 7-11-96 9-15-96 3 . 97 22 9-16-96 11-23-96 2 . 91 22 170 (f) 9-28-96 11-27-96 5 . 18 57 9-28-96 10-31-96 1 . 92 10 11-1-96 11-27-96 1.67 12 12 hour range, 1126-96 2 .31 36

PAGE 190

180 Table 5-7: Agouti density at forest sites. Site Dens i ty ( ind/km2 ) Method Reference Peru: Samiria, unhunted, flooded 1.8 strip census Bodmer, pers . comm. Peru: Blanco, terra firme, hunted 2 . 3 strip census Bodmer, pers. comm. Peru: Terra f i rme , unhunted 1 . 8 strip census Bodmer, pers. comm. Maraca 40 radio tracked home ranges Silvius, this study Southeastern Amazonia 34 strip census Peres et al 1997 BCI, old forest 58.5 strip census Wright et al . 1994 Santa Rosa 38 . 8 observed home ranges Hallawachs 1986 (in Wright et al . 1994) Los Tuxtlas 27 strip census Coates Estrada and Estrada 1986 (in Wright et al . 1994) Sierra de Chame 30 strip census Hendrichs 1977 (in Wright et al. 1994) Tikal 7 . 8 strip census Cant 1977 Guatopo 63 trapping , strip census, road kill, correlations , other Eisenberg et al. 1979 (in Wright et al. 1994) Cocha Cashu, Peru 5.2 trapping, strip census, other Janson and Emmons 1990 French Guiana 63 .7* observation of known animals Calculated from Dubost 1988 * A "200 meter diameter" area = 3.14 ha per home range; this yields 31.2 home ranges per Km2 , and assuming two adult animals per home range, this yields adult density of 63.7 animals per Km^ .

PAGE 191

181 0 ;>ouon > I I a) . •• • . _ Figure 5-1. Trail systems on the eastern end of Maraca Island Ecological Reserve. a) Trails covered in mammal transects; the center box includes a network of trails, shown in part b; b) Close up of main trail system, with radiotracked agouti home ranges.

PAGE 192

182 0 too 300 aoofli 1 ' ' I I N i b) Figure 5-1, continued.

PAGE 193

183 1 j 60-1 50u E 3 z 40302010I r T vO fN T — r \0 \0 m On so — tN TJ7 o vo vo \o \o \o o ^ ^ ^ o> ^ OS 00 o 00 a\ — o 7
PAGE 194

184 15-1 105 i"i I — r 3o >o ON a\ ^ s ^ (N — m ON 00 (N m TT 1 — I — I — I — r NO sO NO NO nO ^ ^ On On On On O O vO NO On — ro — ~i o ^ ON ON T — I — T 00 o — 00 ON — i5 NO
PAGE 195

185 10-1 EJ pulp "O old endocarps O newendocaqis 7.5o a a. B 3 z Month Figure 5-3. Temporal pattern of availability of Attalea "^^^iP^ pulp, new endocarps (from fruit dropped zero to six months previously, and old endocarps (from fruit dropped more than six months earlier) .

PAGE 196

186 15-1 t/! u 2 10S 3 5 V, On > O Z 5 c 53 ON immature fruit preripe fruit ripe fruit seeds germinating seeds Month Figure 5-4. Temporal pattern of availability for Pouteria surumuensis fruits and seeds. , ;» > r-.' 1 I \ 1 .1 i 4

PAGE 197

187 15-1 immature fruit no mature fruit/new seeds "O-" old seeds e/! x: o a 10XI E z 5 T 1 t Month Figure 5-5. Temporal pattern of availability for Astrocarviim aculeatum fruits and seeds.

PAGE 198

188 immature fruit preripe fruit ripe fruit germinating seed Month Figure 5-6. Temporal pattern of availability for Pradosia surinamensi s fruits and seeds.

PAGE 199

189 U CIS a. B 20-1 15100 {2j Pouteria Seeds m Pradosia Seeds and Seedlings ...Q — Attaleapulp 6 Astrocaryum Pulp and Seeds vO ON ON > c o 2 ' — > ON I 2 3 BO 3 < ON t C/2 OS I > o 2 Month Figure 5-7. Congruent availability of the four principal food items for agoutis on Maraca ( Pouteria surumuensis seeds, Pradosia surinamensi s seeds and seedlings, Attalea marioa pulp, and Astrocarviim aculeatum pulp and seeds) .

PAGE 200

>; 0 80 m Figure 5-8. The seasonal range pattern for agouti 148 (does not include locations for 12 hr tracking period): °= the July 9 to September 15, 1996 tracking period, and °= the September 16 to December 2, 1996 tracking period. The different time periods correspond to periods of high and low Pouteria surumuensis seed availability, respectively.

PAGE 201

Figure 5-9. The seasonal range pattern for agouti 158 (does not include locations for 12 hr tracking period): the July 9 to September 15, 1996 tracking period, and " = the September 16 to November 30, 1996, tracking period. The different time periods correspond to periods of high and low Pouteria surumuensi s seed availability, respectively.

PAGE 202

192

PAGE 203

193 V..' -J ...^ t '>. * . ^ ? . .. / *; 0 '/ \ o o (a o \ . a \ _ '' ' / i u o \ D \. ' -^^^ a ' ' — • t 0 a ^ " 1 ' "'^-•B 0 80 m Figure 5-11. The seasonal range pattern for agouti 167 (does not include locations for 12 hr tracking period): " = the July 11 to September 15, 1996 tracking period, and " = the September 16 to November 23, 1996 tracking period. The different time periods correspond to periods of high and low Pouteria surumuensi s seed availability, respectively

PAGE 204

194 \ .' \ M ° / I \ / / \ ° / /\ A i \ / \ " \ .. a a A 1 .•' o . . . • • c 0 80 m Figure 5-12. The seasonal range pattern for agouti 170 (not including locations for 12 hr tracking periods): »= September 28 to October 31, 1996, and °= November 1-27, 1996

PAGE 205

195 m Figure 5-13 Four, 12 hr contiguous tracking range patterns tor agouti 163, presented as subsets of the entire home range (1 in diagram): 2= 12 hr range on 21 November 1996, 3=3 February 1997, 4 = February 21 1997, and 5 = 12 March 1997

PAGE 206

196 Figure 5-14. The 12 hr range pattern for agouti 158 (° ) presented as a subset of its total range includes locations for 12 hr tracking on 11-21-96) from July to November 1996.

PAGE 207

197 Figure 5-15. The 12 hr range presented as a subset of its locations for 12 hr tracking November 19 96. pattern for agouti 170 ( total range {°; includes on 11-21-96) from July to

PAGE 208

CHAPTER 6 SCALE-DEPENDENT DISTRIBUTION PATTERNS IN PLANTS USED BY THREE TERRESTRIAL RAIN FOREST MAMMALS AT MARACA ISLAND ECOLOGICAL RESERVE Introduction . Fruitand seed-eating animals in tropical forests encounter patchiness in food distribution at many scales. Not only are fruits discretely distributed in space on individual trees or shrubs (Chapman 1988, Fleming 1992), individual tree species are also often distributed in a . ,; clumped fashion within a few hectares or even a few meters (Hubbell 1979, He 1993, Tuomisto et al . 1995, He et al . 1996, Thorington et al. 1996). At the landscape scale (tens of kilometers), species diversity, plant associations, or habitat types can turn over rapidly within a few kilometers (Milliken and Ratter 1989, 1998, Tuomisto et al . 1995, Foster and Brokaw 1996, Chapman et al . 1997). Because distribution patterns are a function of the scale at which they are measured (Pielou 1977, Wiens 1995, Bellehumeur and Legendre 1998) , different animal species with different home range sizes and movement abilities will perceive different patterns of distribution of the same resources at one site. As the absolute scale or extent of the area used increases, the 198

PAGE 209

199 • • grain size with which the animal measures it should also shift (c.f. Ims 1995, Kozakiewicz 1995). Thus, an individual tapir ( Tapirus sp.) with a home range of several square kilometers, feeding on the ripe fruits of a species that is randomly distributed within one habitat type but patchily distributed among habitats, is able to move between fruit rich habitats. It perceives a high level of food availability at the time when the tree species in question is fruiting. A spiny rat ( Proechimvs sp. ) or squirrel (Sciurus sp . ) restricted to a home range of one • hectare, on the other hand, may not encounter a single fruiting tree of this species and experience a time of food shortage at the same time that the tapir--or even a neighboring spiny rat (c.f. Foster and Brokaw 1996) -is finding adequate food. The small scale to which individual rats are restricted may lead to source sink dynamics at the rat population level, whereas similar dynamics may not occur for the more mobile tapir. Scale dependency of food availability holds true for all animals in all biomes. The situation is particularly interesting for studies of tropical fruit-eaters, however, because in tropical forests patchiness involves a much higher diversity of tree species and functional groups, and is prevalent at all scales (e.g. Tuomisto et al . 1995, Thorington et al . 1996). In studies where resource distribution is an important variable, such as in correlations between foraging behavior, social structure, or

PAGE 210

. . , : 200 • „ . , interspecific competition and resource distribution, it is therefore critical that resource distribution be measured at the same scale at which the animal or population under study perceives it. This is seldom done, because a) the scale and grain size at which the animals sample the environment is not known prior to the study; b) in community level studies, it is too cumbersome to measure resource distribution at a different scale for each species; and c) it is labor intensive to quantify tree distribution in high diversity tropical forests. In studies of sympatric primate species (Terborgh 1983), for example, or even sympatric frugivorous species in different orders (Chapman et al . 1989), the same plots or fruit trap layouts have been used to measure resource availability and indices of tree dispersion for several species with different home range sizes, territoriality, and movement abilities. As long-term studies at single sites become more common, and information accumulates on several locally sympatric species or on individuals with overlapping territories, an ' examination of resource use at different scales and its implication for ecological interactions becomes feasible. In this paper I take advantage of the existence of forest inventories and mammal tracking studies over the last two decades at Maraca Island Ecological Reserve, in the northern Brazilian Amazon, to describe animals that use the same resources but have very different home range sizes and may perceive food availability differently in their environment.

PAGE 211

201 r:"; I use the known home ranges and foraging distances of three species of Neotropical terrestrial fruit eaters studied at one site as "meter-sticks" with which to measure patchiness in their environment. The home ranges of whitelipped peccaries ( Tavassu pecari -20-2Q0 km^ ) collared peccaries (Tavassu taiacu -10 km2 ) , and agoutis ( Dasvorocta aqouti-0 . 05 km^) differ by several orders of magnitude. I summarize the overall species diversity differences among terra firme habitat types, and then focus on terra firme canopy and sub-canopy tree species that are important in the diet of all three animals at this site. I predict that each animal species will encounter a different distribution pattern of the same tree species, and that the animal ' , species' ranging behavior will in part reflect this perception of resource distribution. Although I focus on the individual animal's or herd's perception of spatial variation the environment, I speculate on potential differences at the population level. Study Site Maraca Island Ecological Reserve is a 1,100 km2 riverine island located on the forest side of the forest-savanna boundary that marks the transition from the Rupununi savannas of Guyana to the Amazonian rain forest in the northern Brazilian state of Roraima. Floristic, geomorphological , and faunal inventories were described on the reserve during

PAGE 212

202 an intensive multidisciplinary survey by the Royal Geographic Society and the Brazilian National Institute for Amazonian Research (INPA) in 1987-88. The entire reserve, which lies parallel to the transition gradient, grades from small savannas, transitional forests and high diversity moist upland forest in the east, through an area dominated by the deciduous legume Peltoavne aracilioes in the center and center east, and moist upland mixed forests in the west. There is ongoing discussion as to the degree to which forests on the eastern end of the reserve, where most studies have taken place, have undergone human disturbance (Proctor and Miller 1998). Disturbance probably involved clearing of small patches of forest; no forest clearing has taken place during the last 110 years. Maraca is located on the Guiana Complex portion of the Guiana Shield, and includes the transition between a PlioPleistocene substrate at the eastern side and a late Tertiary formation on the western portion (Thompson et al . 1992). The eastern terrain is flat and low, while the western terrain is higher and more dissected by hills and ridges. Elevation ranges from 110-160 meters above sea level on the eastern half of the island, and from 330-350 meters above sea level on the western portion of the island. The soils are sandy and derived from quartz-biotite schists, quartz-feldspar gneisses, and tonalitic granites (Thompson et al . 1992). Rainfall records on the eastern end of the island show ' variation between 1750 and 2,300 mm annually (Thompson et al .

PAGE 213

203 1992); during El Nino Southern Oscillation events, rainfall decreases markedly across the state (unpublished data, Brazilian airforce records) . The wet season is between May and September, the dry season between October and April, with rains starting in either early or late April. The heaviest individual rains tend to fall in the dry season (Thompson et al. 1992, Milliken and Ratter 1998). Floristics . ' ' ' The plant families with the greatest number of tree species on terra firme habitats are the Leguminoseae , • . ^• Rubiaceae, Moraceae, and Burseraceae (Milliken and Ratter 1989, 1998; all vegetation descriptions are from these sources) . Based on importance value (IV = relative density + relative dominance + relative frequency) rather than species richness, Sapotaceae is the dominant family in the eastern forests (always within the top 5 families in each transect) . Rubiaceae, Moraceae, Burseraceae, Annonaceae, Chrysobalanaceae, Palmae and Bignonaceae also occur within the top 5 dominant families based on IV in transects on the eastern end. Sixteen species of palm and 16 of Sapotaceae occur on the island, with Attalea marioa and Astrocaryum aculeatum the most common terra firme palms, and Pradosia surinamensis , Pouteria surumuensi s . p. venosa, and Ecclinusa quianensis the most common species in the Sapotaceae.

PAGE 214

204 There is a gradual increase in diversity and turnover of species from east to west on the island. Only 26% and 40% of the tree species encountered on the eastern and central terra firme forests respectively were also encountered 60 km away on the western-most areas sampled. Family composition also changes from east to west: the Sapotaceae, so important in the east, contributes little to IV on the west. In contrast, the Euphorbiaceae becomes common on the west, though it was very poorly represented in the east. None of the top five families and only four out of the top ten families are common to the easternand westernmost transects (Milliken and Ratter 1989, 1998) . There are no extensive flooded forest habitats or palm swamps on the Reserve, although the riverine forest floods in low pockets during high rainfall years (Milliken and Ratter 1989, Fragoso 1994). Several small ponds and savannas are interspersed throughout the forest, and frequent strands of Mauritia f lexuosa palms run along stream courses and other periodically flooded habitats. ' . Study Animals • . White-lipped peccaries and collared peccaries were • tracked by Fragoso (1994, 1998, in review, unpublished data) during two study seasons. A herd of 40 white-lipped peccaries tracked for 12 months in 1991-9 had a home range of 21 km2, while a herd of 130 animals tracked for six months

PAGE 215

2 05 • ' , / . j( had a home range of 109 km^ . Two herds of collared peccaries tracked for six months (21 individuals) and eight months (36 individuals) had home ranges of 11 and 10 km^ respectively. In 1995-97, a herd of 70 white-lipped peccaries tracked for 17 months, and a herd of 200 tracked for 19 months had home ranges of approximately 50 and 150 km^ . During the same time period, three herds of collared peccaries with between 20 and 30 individuals had home ranges of approximately 10 15 km2 (Fragoso unpublished data) . Herds were usually located aerially twice a week. Casual encounters with the animals on the trail system increased the sample size for locations there. Agoutis were radio tracked as part of this study to determine the scale at which they used their habitat (Chapter 5) . Home range size for six radio collared animals tracked for between 2 and 8 months was 0.03 to 0.08 km2 (Chapter 5), Both collared peccaries and agoutis are territorial, engaging in agonistic behavior to defend home ranges against conspecifics (Fragoso 1998) . White-lipped peccaries, on the other hand, do not defend their large home ranges, and in this study the ranges of the two known groups overlapped extensively (in space but less frequently in time). They also overlapped with the home ranges of all three collared peccary herds. Unlike both peccary species, agoutis patrol most of their home range at least once a day. However, each day they concentrate foraging (including feeding and burying seeds) at only a few individual trees (Smythe 1978, Chapter 5) . Collared peccaries tend to feed at one or a few patches

PAGE 216

2 06 . of trees, or at one very large tree, such as an Enterolobium . during one day. White-lipped peccaries will forage over several kilometers of forest in one day, and frequently move between habitat types in a single day. For this study, the scale at which the animals operate is set at their home range size (known a priori), and the grain size at which they measure the environment is assumed to be equivalent to the food patch size (one tree for agoutis, one group of trees or patch for collared peccaries, one habitat type for whitelipped peccaries; Table 6-1) . Both species of peccaries and the agoutis on the trail system relied heavily on the palm Attalea marioa . and the sapotaceous species Pradosia surinamensis and Pouteria surumuensis (Chapter 4, Fragoso 1994). Other large fruited tree species known or suspected to be used by the three species are included in this study: all species in the Lecythidaceae and Sapotaceae, the palm Astrocarvum aculeatum . Couepia sp. and Licania kunthiana (Chrysobalanaceae) , Enterolobium spp . and Hvmenea spp. (Leguminosae) , and one unidentified species in the Icaniaceae (Appendix 1). The only small-fruited plant species used by all three animals was Brosimum lactescens (Moraceae) .

PAGE 217

^ 207 : I ^ . '' V Methods Data Collation from Previous Studies , The home ranges of white-lipped peccary herds radio tracked by Fragoso in 1991-93 and 1995-97, and collared peccary herds tracked in 1995-97 were plotted on a map of the study site. The location of all forests plots inventoried on Maraca by Milliken and Ratter (1989) and Nascimento (1994) in terra firme forest (upland) habitats were plotted on the same maps. Only terra firme habitats were considered, to avoid the distinct patchiness and moisture-adapted species associations created by poor drainage conditions in lowland and flooded areas. For each forest inventory plot, I noted the dominant species (more abundant by at least 10 individuals than the next most common species) and noted the number of trees of each species known to be important in the diets of agoutis and peccaries. Information on the habitat type, area sampled, overall tree density, and diversity are also given for each plot when available. In most cases trees were sampled at both the 3 0 and 10 cm dbh categories; I chose to use the 10 cm dbh category in order not to under-sample the abundance of Pouteria surumuensis . (individuals with less than 30 cm dbh bear fruit and are common) and of several arborescent palms that never attain 30 cm dbh. Additionally, equivalent plot size for 10-cm dbh in Milliken and Ratter (1989), who used a point center quadrat method, was most

PAGE 218

2 08 similar to the 0.25 hectare plot size used by Nascimento (1994). A visual comparison with data for 30 cm dbh plots showed no obvious differences in abundance for any of the species being considered. Fruit and Seed Availability on the Ground in Three Habitats Because I was working with terrestrial consumers, fruit and seed availability was monitored on the ground rather than on the tree. Three two-kilometer long transects were marked along two existing and one newly cut trails in three terra firme forest types (riverine ("river trail"), eastern ("trail 2"), and central ("km 7")) known to be used by a single herd of white-lipped peccaries and to be f loristically distinct based on Milliken and Ratter (1989) (Chapter 4). The two terra firme transects were separated from each other by 6 kilometers, and from the riverine forest by 3 and 9 kilometers respectively. Each transect was marked at 50 m intervals, yielding forty 3 X 50 m quadrats or segments per transect. Every two to three weeks (mean sampling interval 20.76 days, range 30 14 days) from November 4 1995 to October 13 1996, the ground along a 3 m width was surveyed for fruits and seeds. Sampling started two weeks later in the central terra firme forest, and two scheduled samples were missed. The location (segment number) of each fruit patch was noted. Species were identified using on-site reference collections and van

PAGE 219

• 209 . ^ " Roosmalen (1985). Fruit items were not collected, and were recounted on subsequent transects if they were still in edible condition. While ground surveys do not give an estimate of the quantity of food available on the ground, because of rapid fruit removal by consumers, they do capture the same spatio-temporal patterns of fruit production as arboreal surveys (Zhang and Wang 1995, Stevenson et al . 1998). In addition, ground transects document post-dispersal availability (older seeds and seedlings; Chapter 5). In this study a species was considered available through all of its edible stages, not just when it was ripe. For example, in the case of Pradosia surinamensis . unripe seeds are eaten by collared and white-lipped peccaries, unripe and ripe seeds are eaten by agoutis, and germinating seeds are eaten by all, while Attalea endocarps remain edible for up to a year after fruit fall (Chapter 4, 5). All fruits, seeds, and germinating seeds encountered were recorded. Seed patches produced by animal defecations were also counted; in the case of spider monkey sleeping trees and tapir latrines (Fragoso 1997), accumulated defecations resulted in large patches of seeds in the same location over several transect runs. For each transect, adult trees of Attalea marioa were counted in each segment, within 2.5 m of the center of the trail, yielding forty 5 x 50 m plots per transect. Additionally, Pouteria surumuensi .g trees were counted along the transect in eastern terra firme forest. -,.

PAGE 220

210 . The three transects are compared in terms of fruit species diversity, number of fruit patches available per segment and per transect over one year, differences in timing of peak fruit availability, and contribution of the "edible" species to overall patterns of availability. At the smallest scale, the index of dispersion of edible food patches was calculated for aggregated quadrat sizes from 50 m long to 1000 m long, and the clumping pattern compared among transects. Distribution indices based on an aggregation of quadrats cannot be tested for statistical significance, due to lack of independence of adjacent and summed quadrats (Pielou 1977). However, patterns shown by variance peaks at aggregate quadrate sizes less than half the length of the total quadrat are generally accepted as indicative of real pattern free from sample size influence (Ludwig and Reynolds 1988, Campbell 1998). No index of dispersion can be calculated for the floristic plots from Milliken and Ratter (1989) and Nascimento (1994), because data were gathered using different methodologies and on plots of different sizes .

PAGE 221

211 Results Floristic Plots All species The 3 0 terra firme forest quadrats sampled in the floristic plots ranged in area from 0.06 to 0.5 ha (Appendix 2) . They were concentrated on the central trail of the main trail system, in a Peltoovne rich area to the southeast, and on two trails in the center west and west (Figs. 6-2 and 63) . No plots have been established in riverine forest on Maraca. Number of trees sampled ranged from 40 to 2 00, and number of species per plot from 15 to 54 (Appendix 2). For purposes of comparison between general areas on the Reserve I have grouped the plots into eastern trail system (plots 111), center east / Peltogyne (plots 12-24), center west (plots 25-29) and west end (plots 29 and 30) (Figs. 6-2, 6-3). Twenty one out of the 30 plots showed clear dominance by at least one and usually several equally abundant species (species more abundant by at least 10 trees than the next most abundant species), and five more showed a trend for one species to dominate. Information on dominance in terms of numbers of individuals is used here to show turnover across the landscape of these species; abundant species are more reliably tracked than rare ones. The three smallest plots showed no species dominance, suggesting an area effect. Six

PAGE 222

species dominated in more than one plot: Peltoavne (6 plots), Pradosia (5 plots), Tetraaastri s (3 plots), Licania (3 plots), Ecclinusa (6 plots) and Attalea (2 plots); while three species were dominant in one plot each: Himatanthus . Oenocarpus, and Alseis . Frequency of dominance does not reflect frequency in the Reserve, because plots were not placed at random. Edible species Sapotaceae trends. Pradosia was common everywhere on the eastern areas (Table 6-2). Despite its ubiquitous presence, pattern at the scale of hundreds of meters is revealed by the plots: plots 3,4, and 5 have almost no Pradosia . whereas this species is dominant in plots 1 and 2, located about 300 meters away. This concurs with my observation that Pradosia occurs in frequent long swaths through the forest, interrupted by Pradosia-f ree areas. Pradosia and Ecclinusa are the only sapotaceous species that remain common in the central/eastern Peltoqyne forests, before disappearing in the west. Ecclinusa is most common in the center east, and also exhibits small scale patchiness: plots 15, 16, and 17 are adjacent to each other, but only 17 has a high abundance of Ecclinusa . Pouteria surumuensi .g appears to have the widest range of any of the Sapotaceae, occurring in the center west before dropping out in the west. It is associated with everywhere but in the Peltoavne forest. The same is

PAGE 223

213 true of P. hispida and possibly P. venosa , though sample sizes are small for the latter species. . . . e -. •• >* ' Palm trends. Oenocarpus bacaba , rare and patchy in the east, becomes relatively more common in the center east and west. Only three O. bacaba patches have been found on the eastern trail system (Fragoso unpublished data) , and all are characterized by high local density and discrete boundaries. The sites differ in topography (one is a flat upland area, one is on a slope, and one on a bottomland) . Attalea marioa is common in the east, but becomes less common in the center east, more easterly than any of the Sapotaceae. Attalea patches appear to be larger but more infrequent than Pradosia patches, since adjacent plots tend to have similar numbers of this palm (Table 6-2). Attalea does not co-occur with Peltoqyne. Astrocarvum aculeatum appears to be less aggregated, or to be aggregated at lower densities than Attalea maripa. Like Attalea marioa . it does not co-occur with Peltoqyne, and it drops out even sooner (more easterly) . In general, palms are much less common in the western half of the island; Jessenia (Oenocarpus) bataua replaces the eastern species, but never becomes as abundant in individual plots as Attalea or Astrocarvum . Other edible species trends . Tetraaastri s and Licania, important to primates as well as agoutis and peccaries, are common and apparently clumped in the east, but absent in the

PAGE 224

. 214 ,: ^ . ^ west and in Peltoavne forest (Table 6-2). Two species of Enterolobium and Hvmenea are rare everywhere, but more common in the east than west. In the Lecythidaceae , distinct understory species are common in the east and west respectively, with only Gustavia superba occurring in both areas. Lecythis corruaata . common in all eastern plots, drops out in Peltoayne and in the west. It co-occurs with both palms and sapotaceaous species, and does not appear to be patchy to the same degree the other species considered here are, based on similar abundances in all plots and absence from none . The above descriptions refer to interior terra firme forests. Floristic plots have never been established in riverine forest on Maraca. Milliken and Ratter (1989, 1998) describe the composition of riverine forest on Maraca as follows. On the eastern half of the island, to the second Santa Rosa hump (Fig. 6-2), the most common species are Attalea maripa, Chochlosoermum orinocence . and Spondias mombim . Of these, only Attalea is common in terra firme forest as well, based on my transect results. Spondias occurs only in two small patches on the main trail system (pers. obs.). The large emergents Enterolobi iim spp., Ceiba pentandra and Andira inermis are also more common on the river than on terra firme, though they are still widely spaced. In low, flooded areas of the river bank, Triolaris IS a colonist. On the western river edges, which are much steeper than the eastern ones, the terra firme forest tends

PAGE 225

215 to extend down to the river bank. Contrary to what occurs in the central terra firme forest, Attalea remains common in Peltocpy^ne forest when this forest type reaches the river bank. Peccary Ranges Overlain on Floristics The two white-lipped peccary herds tracked in 1995-97 and the larger herd tracked for six months in 1992 would have encountered the full floristic variability of the eastern and center eastern areas, including the riverine forest (Fig. 63). Any white-lipped peccary herds that used the areas to the west would have encountered very different floristics, and may have had diets lower in fruit than the eastern animals (no herds other than the radiotracked herds ever visited the main trail system, though untracked herds may have overlapped with the study herds in the center east; pers. obs.). The largest white-lip herd was never radio located within the central Peltoavne forest, even though it ' did use use Peltoqyne forests on the south east--the area where Pradosia is most common (plots 20-23). During the transect study, I encountered white-lips 7 times. Twice when white-lips where encountered on the river forest they were feeding on Attalea endocarp contents and Phenakospermiim stems. At least one herd repeatedly visited the river forest near the station during the 1995-96 and 1996-97 dry seasons, the time when Attalea endocarps are '

PAGE 226

216 available (Chapter 3) . Two times when white-lips were encountered foraging rather than traveling on the main trail system they were feeding on Pouteria surumuensis . once on Pradosia, and once on Enterolobium . In the center east, they were encountered foraging in a Mauri tia palm strand, feeding on the seeds of three palm species ( Euteroe precatoria . Socratea exorrhiza, and Mauritia f lexuosa ) , and once feeding on Attalea endocarps and Hvmenea seeds. With the exception of Phenakospermum stems, all these food items had been previously recorded by Fragoso (1994, 1998), pointing to a high temporal consistency in the diets of white-lipped peccaries in the study area. The collared peccary herds tracked in 1995-97 all partitioned the eastern tip of the island in such a way that each herd had a portion of the Sapotaceae-rich terra firme forest, and a portion of the palm and Spondias-rich riverine forest. The white-lipped herd tracked in 1992 (Fig. 6-3b) saw a similar habitat diversity, but it used all the terra fime area and the river banks on both sides of the island. It avoided the treeless savanna/wetlands on the very eastern part of the island, although it used a small savanna /wet land on the southern border to feed on fish. None of the collared peccary herds included savannas in their home range.

PAGE 227

Fruit Transects Overall fruit availability Over an 11 month monitoring period, I sampled the River Trail and Trail 2 (eastern terra firme forest) 18 times, and the Km 7 Trail (central terra firme) 15 times. Km 7 had the highest total fruit species diversity, but the lowest number of fruit patches and highest number of empty segments per sample (Table 6-3, and Fig. 6-4a-d) . The River Trail, despite its low species diversity, had the highest number of patches per transect. This was due in large part to the high density of Attalea maripa palms, which was the dominant species in 12 out of 18 transects and contributed to the higher number of patches per species on the River Trail compared to the other two transects. Attalea dominated only in 5 transects on Trail 2 and Km 7, whereas sapotaceous species dominated in 9/18 and 8/15 transects in Trail 2 and Km 7 respectively. No Sapotaceae occurred on the River Trail (a small clump of Pradosia trees produced no fruit during the study) . No other families or species overtook Sapotaceae and Attalea in frequency of dominance. All trails showed a short peak in total fruit availability in January 1996 (the mid dry season; repeated in January 1997 for Trail 2; Chapter 3) and a broader peak in May August (the early and mid rainy season).

PAGE 228

218 Edible fruit availabilit.v All species known or expected to be eaten by peccaries and agoutis were grouped as "edible species" and analyzed separately from the remaining fruits (Appendix 1). The River Trail had 7 edible species, while Trail 2 and Km 7 had 15 and 17 edible species respectively. When the contribution to the temporal pattern of fruit and seed availability of palms, Sapotaceae, other edible species and all edible species is plotted against the total fruit pattern for each trail (Fig. 6-5a-c) the following patterns emerge: On the River Trail, temporal pattern in edible species availability closely follows total species availability, and the pattern is set by Attalea maripa (Fig. 6-5a) . On Trail 2, edible species make up a smaller proportion of total availability and peak at different times; the pattern of edible species availability is set by the Sapotaceae (Fig. 6-5b) . Furthermore, palm fruit and seed availability peaks when Sapotaceae availability is low, and vice versa. The same is true on Km 7 (Fig. 6-5c). Because the river trail is palm driven and the other two trails Sapotaceae driven, and because the time of avaialbility of the two classes of items are displaced from each other, food is available on the River Trail at the time when it is scarce on the terra firme forest, and vice versa. White-lipped peccaries and collared peccaries including both habitats in their home ranges would be able to use both peaks

PAGE 229

219 (but note that this is more true for white-lips, which can use the Attalea seeds, than for collared peccaries, which only use the pulp; Chapters 3, 4). Agouti scale: within-habitat patterns of edible species availability To examine aggregation in space rather than time within each habitat, for each segment of each trail I extracted the number of patches of each edible species that occurred in the transect sample with the maximum number of patches for that species during the study period, and summed across all edible species. This estimates (slight underestimate, if trees of the same species were not in synchrony) the number of trees • in each segment that produced edible fruit during that seasonal cycle. Trail 2 showed a pattern of peaks at 50 to 100 meter intervals, overlain by a larger-scale bimodal pattern at 100 meters (Fig. 6-6a) . This pattern was again driven by the location of palm and Sapotaceae patches, with the additional contribution of several Couepia sp. and Egchwgleira ggrrugata trees at 300 500 meters along the transect. Km 7 also shows a pattern of peaks at 50 100 meter intervals, with an overlain pattern at 500600 m intervals (Fig. 6-6b) . The pattern is driven almost exclusively by Sapotaceae. The River Trail pattern differs from the other trails in having more closely spaced peaks with greater variation in the spacing between peaks (Fig. 6-

PAGE 230

; 220 6c) . The pattern is driven by Attalea with some contribution from Spondias mombin at 1200-1300 meters. These results are summarized by the trend in indices of dispersion calculated at different aggregated plot sizes (Fig. 6-7). An index of 1 reflects an even or dispersed distribution, an index of less than 1 reflects a random distribution, while and index greater than 1 reflects an aggregated distribution (Krebs 1989) . I cannot test the deviation from 1 statistically because I aggregated plot data and the samples are therefore not independent. The River Trail shows the pattern typical of a random distribution, with slightly fluctuating variance-to-mean ratios (Ludwig 1988) . Trail 2 and Km 7 both show peaks in varianceto-mean ratios at 300 to 400 meters, indicating that there are 150200 meter segments of the trail with high fruit availability interspersed by similar length segments with low fruit availability. In other words, the habitat is patchy at the scale of 150 to 200 meters. The similarity between the two habitats is probably due to the dominance of Sapotaceae in ' both. The bimodal pattern detected for Trail 2 is revealed by the near 0 ratio at the 1000 m interval, whereas the ratio remains similar to that of other plot sizes for the River Trail and near peak for Km 7 . Agouti home ranges averaged just over 0.05 km2 in size. Core areas were located within 150 by 150 meter areas (Chapter 5). Thus at the scale of activity of individual agoutis, the terra firme forests are patchy due to species

PAGE 231

221 aggregations (palm and Sapotaceae driven pattern) . The riverine habitat, on the other hand, has a random pattern of availability at a scale smaller than what the agoutis use, and a random or self-similar availability at larger scales (within the 2 km sampled) . Individual agoutis living in the two habitat types thus encounter different patterns of availability, although they are separated by only a few kilometers . Discussion Species -Specific Perspectives Focusing on the tree species actually used by the study animals rather than on community wide-patterns of availability helps elucidate and simplify spatio-temporal patterns of food availability. At the landscape scale, the food species monitored in this study followed the overall east to west species turnover noted by Milliken and Ratter (1989, 1998). Although many of the species still occur on the western end of the island, and even in the foothills of the Parima mountains, there they become very rare, in contrast to their dominance and extensive patch size in the east.

PAGE 232

222 White-lipped oeccarv scale " When the trend in vegetation turnover across the landscape is truncated at the white-lipped peccary home range scale, some species become a constant while others remain patchy. Thus Pradosia surinamensi .q and Pouteria surumuensis are found in all the terra firme forest visited by white lipped peccaries, while Attalea marioa is concentrated on the easternmost portion and becomes rare in the center east. The small scale patchiness in palm and Sapotaceae distributions detected with the index of dispersion occurs at too small a scale to be noted by animals like white-lips, which operate at a large grain size and move one or two kilometers along a single patch of food. The fact that in both 1995 and 1996, white-lipped peccaries started visiting the river trail during the same month (October) to feed on Attalea endocarps, and continued to visit it until the late dry season in February, suggests that individual white-lip herds make foraging decisions at the habitat mosaic scale (habitats clearly distinguishable to the human eye based on edaphic as well as floristic ivariables). The riverine forest is a seasonally food rich habitat; Attalea endocarp contents are maximally available between September and April (Chapter 4). '. White-lipped peccaries use other seasonal habitats (Mauritia wetlands, ponds, and savannas, which flood for various amounts of time) more than expected based on their

PAGE 233

. 223 v-:.-,availability in the environment, and terra firme forests less than expected (Fragoso 1994, in press). This suggests they use terra fime forest as one more habitat patch, to be visited when food, primarily Sapotaceae fruits and seeds, are available there. The Mauritia wetlands they rely on are seasonal with respect to the availability of fish and invertebrates (high at the end of the wet season) but aseasonal with respect to the availability of fruits. Mauritia is an irregular aseasonal fruiter; within one wetland trees fruit synchronously, but among wetlands they do not (Moskovits 1985, Fragoso unpublished data, Silvius unpublished data) . From the point of view of animals that can visit several wetlands in one foraging bout, they provide food continuously. Collared oeccarv scale . On Maraca, collared peccaries seldom use seasonal habitats such as riverine forest, Mauritia wetlands and savannas (Fragoso 1994, in press). They are confined mostly to the terra firme forest. During this study, three herds used the eastern portion of the island (east of the two inland lakes. Fig. 6-3b) . Two used the central Sapotaceae rich area, while the third used the eastern tip of the island. That tip area has an Astrocaryum-rich version of the terra firme forest (unpublished transect data), and adjacent to it a very large Pradosia dominated forest which seems to

PAGE 234

224 be related to the Peltoavne / Pradosia community found on the southeastern shore of the island (floristic plots 20-25; pers. obs.). Thus, all three herds had access to a Pradosia rich area (terra firme forest) and a palm rich area (river forest), which they used seasonally (Fragoso unpublished data) . They remained within one forest type for several weeks, and stayed at single tree patches for several days (pers. obs.). They used Astrocarvnm pulp more frequently than white-lipped peccaries. Collared peccaries may be responding to the landscape at the tree patch scale, focusing on monodominant tree species areas within terra firme forest and moving among them. Sapotaceae and palms would still provide pulp at different times of year, since Attalea . Astrocaryum and Pouteria spp. are all temporally displaced from each other in peak fruiting time. Pradosia may be a key food for collared peccaries, because like Mauritia different patches fruit asynchronously (pers. obs., Nunes 1992). Agouti scale Agouti home range sizes match the tree-patch scale, while agouti grain size is at the individual tree scale. What is a Sapotaceae-rich area to white-lipped peccaries is a high diversity area to the agoutis and provides them with food year round. Diversity enters into the agouti scale from two sources. First, 50 m segments in the terra firme habitat had more than one fruiting species per segment (Table 6-3),

PAGE 235

indicating that despite dominance by one species, individual trees still provided variation in the species available. Agoutis could therefore select home ranges on the individual tree scale, choosing enough inter-tree variation to provide food year round (c.f. Chapter 5). Second, in the area where two or more species patches come together, diversity and abundance will be high. Home ranges centered around these boundaries would have the highest availability of food. Nevertheless, individual agoutis may be at a higher risk than either peccary species of experiencing seasonally decreasing food availability, due to the clumping of food rich and food poor areas at the 150 meter scale. Food hoarding may be a necessary behavior to mitigate this spatial unpredictability. Conversely, hoarding may ensure the availability of high quality foods year round, even when lower quality foods are available . . • , other fruit-e ating species on Maraca Spider monkeys use different fruit species than the terrestrial mammals in this study, and consequently should perceive a different pattern of resource distribution. The core areas in the home range of one group (Nunes 1992), whose range overlapped the Trail 2 transect in this study, corresponds to concentrations of Spondias and Pradosia trees in the center of their range and Tetraaastri ^ and Brosimum trees on the edge (pers. obs . ) . During Nunes (1992) study

PAGE 236

v6" 226 the group was confined to terra firme forest, and did not use Mauritia wetlands, although I observed the same group three years later feeding on Euteroe precatoria fruit in a Mauritia strand; Euterpe seeds were common at spider monkey defecation sites on the trail system (sites under sleeping trees where feces become concentrated) . A group with 24 animals had a • 316 ha home range, smaller than the home range of a similarly sized group of collared peccaries. In her 1981-82 study of tortoises, Moskovits (1985, 1998) documented an aggregated pattern of fruiting trees " similar to that found for the two terra firme transects in this study. Her study site also overlapped with those for the agouti and spider monkey studies. She monitored one trail parallel to my Trail 2 and one perpendicular to it, both 2 km long, and documented fruiting tree aggregations at approximately 150 meter intervals. Tortoises used Sapotaceae, Mauritia, and Annonaceae fruits. Annonaceae trees co-occur with Attalea and Astrocarvnm in my study sites, providing an alternative food for tortoises, which did not feed on those two palm species. . Implications for Habitat Par t itioning. Comnet -i t ion . and Communi ty Structure Because the five species discussed above either use different species or perceive availability of the same

PAGE 237

227 species differently, fluctuations in food availability should differ for them. Their own populations could therefore fluctuate out of synchrony with each other, except when massive fruit failure affects all of them, as occurs during El Nino years and other extreme droughts. Out of phase fluctuations, and fluctuations at different temporal scales, would tend to reduce competition among fruit-eating species.. This study also suggests that in Neotropical forests, it is the abundant tree species, rather than keystone species sensu Power et al . (1995) that set carrying capacity and determine food availability patterns. Across the Amazon, common trees in the Sapotaceae and Palmae appear to be staple foods for fruit-eaters, especially large terrestrial mammals (Bodmer 1989, Fragoso 1994). An understanding of the distribution of species in these families may add to our understanding of animal diversity patterns in the region.

PAGE 238

228 Table 6-1. Scale of habitat use by the tree different animal species considered in this study. All home range sizes are derived from on-site studies. Grain Size Scale Species Food Patch Size Daily Range (area and linear distance ) Home Range Size Re f s . Agouti One tree 1-2 ha 150-200 m .03 .08 km2 (3-8.5 ha) This study Chpt . 5 Collared peccary Large tree or group of fruiting trees 10 ha 1 km 10-15 km2 (1000-1, 500 ha) Fragoso 1998 Whitelipped peccaries One habitat type 200 ha 5-10 km 20-150 km2 (2000 15, 000 Fragoso 1998 grain size = a) what is perceived as a food patch; b) area assessed over the short term (daily or near daily) for foraging decisions, or distance moved in one-two days between two food patches. scale = area which the individual or herd assesses over the long term (one seasonal cycle or other relevant subor supra-annual cycle for the study area) .

PAGE 239

Table 6-2. Abundances of key food species for peccaries and agoutis in floristic inventories carried out on Maraca Island Ecological Reserve by Milliken and Ratter (1989) and Nascimento (1994). A (-) signifies no data available. From left to right, species are: Pradosia surinamensis . Pouteria gurumu^nsig, Pouteri a hispida . Pouteria venosa, Ecclinusa quianensis, Astrocarvum aculeatum . Attalea marioa . Lecvthis corruqata , Tetraaastris panamensis . Brosimum sp . , Hymenea sp (courbaril and oarvifolia) , Licania kunthiana . Oenocarpus bacaba, Enterolobium sp . ( schomburaki i and cvclocarpum ) . One key species discussed in the text did not have data in any of the plots: Couepia sp . Plot numbers correspond to those shown on Figure 6-3a. Quad # Pr s Po s Po h Po V Ec c As n At m Lc c 1 c p — tJr s J Tu Hy s|! Lit K ue D En sp 1 11 1. \ ( ) ( ) 5 ( \ 5 7 "1 J I 1 1. ? ' ) If ) 2 5 { ! 34 1 r 1 u U A 3 ( J i ( ) 1 ! ( ) \ 2 c t \ u U n vJ 4 ( ) ( ) 1 c ) 1 4 3 1 1 1 1 1 1 1 1 U A V 5 ! t c 3 c 7 7 r L J u A u 6 A ( 1 0 0 1 3 1 0 A u A (J 7 it 1 2 0 24 ( 7 13 0 0 1 1 A u u A 8 6 ( 0 19 ( 3 I J 0 1 1 n I U A 0 0 9 IC 4 1 8 2 10 fi O 1 1 L U y 0 10 8 1 2 1 1 2 1 g I 1 J) U 1 o 0 1 1 7 I 0 15 2 7 4 1 \f n U lU 0 12 12 4 1 12 (J 1 7 1 Q 1 7 1 U 2 0 13 5 1 ) 6 0 22 10 4 0 0 I 0 14 8 0 13 16 4 1 20 14 10 0 20 0 15 9 1 0 1 0 17 0 0 0 0 0 16 19 1 0 4 0 7 5 0 0 0 0 0 17 8 1 7 41 i 4 22 0 0 0 3 0 18 2 9 5 4 0 0 2 0 4 2 0 2 2 0 19 1 0 0 0 3 0 2 0 14 1 0 5 21 1 20 12 7 0 6 0 0 0 0 0 0 0 0 0 21 1 1 1 0 1 0 0 0 0 0 0 0 0 0 22 24 1 0 1 0 0 0 0 0 0 0 0 0 23 20 5 0 2 1 0 0 0 0 0 0 0 0 0 24 0 2 0 0 0 0 0 0 0 0 0 0 0 0 25 2 6 1 0 1 0 3 0 3 0 0 0 0 0 26 1 3 1 1 0 0 5 3 5 1 0 0 0 0 27 0 2 -> 1 1 0 2 0 1 3 0 0 0 0 28 1 1 0 0 0 0 0 0 0 0 0 0 0 0 29 0 0 0 0 0 1 2 0 0 5 0 0 8 1 30 0 0 (J 0 0 0 0 0 0 0 1 0

PAGE 240

230 Table 6-3. Total fruit availability in three different habitat types (including all species, not just "edible" species ) . Habitat Variable River Trail Trail 2 Km 7 Total Species 42 50 63 Mean (range) # species/run 8.6 (6-15) 14.4 (9-20) 13.9 (11-19) Mean (range) # fruit patches /run 36.7 (14-65) 37.1 (18-63) 32.7 (16-56) Mean (range) # patches/ species 4.5 (2 .4-9 . 8 ) 2.6 (1.7-4.2) 2.4 (1.3-4.10 Mean (range) # empty 50 m segment s / run 18.7 (7-28) 17.4 (7-29) 19 (11-30) Mean (range) edible species per segment 1.68 (0-3) 3 (0-6) 3.23 (1-7) rj

PAGE 241

Figure 6-1. Location of Maraca Island Ecological Reserve Adapted from Milliken and Ratter (1998).

PAGE 242

252 Figure 6-2. Overview of Maraca Island Ecological Reserve showing the general location of floristic plots in the eas and center east, and the specific location of the plots in the west and center west. -. ' v #

PAGE 243

233 Metres 10000 Furo Santa Rosa N A River Trail Furo Maraca wip herd 1 wip herd 2 a) • ' \ Figure 6-3. Floristic plots and peccary home ranges on the east and center east portions of Maraca Island Ecological Reserve, a) White-lipped peccary home range (1995-97) phenology transect and floristic plot locations. b) Collared peccary (995-98) and white-lipped (1991-93) peccary home ranges

PAGE 244

234 Figure 6-3, continued

PAGE 245

235 o E z 25-1 20Riverine Forest "O Central Terra Firme Forest • O — Eastern Terra Firme Forest O 0--0. 9 Date a) Figure 6-4. Patterns of total fruit availability on the ground on each of three transects from November 4 1995 through October 13 1996. a) number of species of fruits and seeds, b) number of patches of fruits and seeds, c) mean number of patches per species, and d) number of empty segments encountered on each run. Note that three runs are missing from Km 7 in central terra firme forest compared to Trail 2 in eastern terra firme and the River Trail.

PAGE 246

236 Date b) Figure 6-4, continued. Dates as in a)

PAGE 247

237 Date Fig. 6-4, continued. Dates as in a).

PAGE 248

238 d) Figure 6-4 continued. Dates as in a)

PAGE 249

Figure 6-5. Comparison of temporal availability of all fruit species, all edible fruit species, and the contribution of all edible species, palm species, and sapotaceae species to overall fruit availability on each of three transects from November 4 1995 to October 13 1996. a) Trail 2. b) Km 7. c) River Trail

PAGE 250

240 Date b) Figure 6-5, continued. Dates as in a).

PAGE 251

Date c) Figure 6-5, continued. Dates as in a)

PAGE 252

242 12.5 -I Ou a) Edible Species O Palms • O — Sapotaceae SS8|8SS8S8S88S88S8S Segment Figure 6-6. Spatial pattern of availability of edible fruit species on three 2-km long trails divided into 50 meter segments. a) Trail 2, b) Km 7 and c) River Trail. Presented is the maximum number of patches of each edible species available during any one sample of the transect, summed across species. See text for explanation.

PAGE 253

243 Segment b) Figure 6-6, continued.

PAGE 254

244 Segment c) Figure 6-6, continued.

PAGE 255

2.i5 I 8 5 -1 43 2 riverine forest O central terra finme forest O — eastern terra f'irme forest T — I — I — I — r "I — I — I — I — I — I — I — I — I — I — I — 1— oc oooooooooooooooo T— r-9 o o o O m O 0^ Ov O Aggregated Segment Size (meters) Figure 6-7. Index of dispersion (variance/mean) for the original 50 x 3 meter segment sizes and subsequent aggregations of segments to yield 100, 150, 200, 250, 300, 350, 400, 500 and 1000 meter segments.

PAGE 256

CHAPTER 7 ; . CONCLUSIONS AND IMPLICATIONS Phenology: Methods and Implications . There has been discussion in the literature (Zhang and Wang 1995) and at recent meetings (e.g., phenology symposium at the 1998 meeting of the Association for Tropical Biology) about the quality of information yielded by different methods of describing fruit production patterns. This study relied on the use of a patch-based index for the evaluation of food availability and patterns of fruit production. Two-kilometer long un-raked transects were walked every two to three weeks and the identity and number of fruits, seeds, and germinating seed patches counted. The method proved successful in detecting the overall seasonal pattern of fruit production at the study site, as shown by comparison with two studies that used a traditional canopy count method to describe phenology. It was also successful in detecting differences among sites in the diversity of fruit species available to consumers, and in spatial (aggregated vs. dispersed) and temporal (seasonal and supra-annual) patterns of fruit availability. " ' Analysis in this study concentrated on the spatial distribution of fruit patches and the relative levels of 246 . '

PAGE 257

• : ; ' • . 247 : / fruit productions compared among different habitats. For this type of data, it was an effective, low disturbance method, which can be used when more detailed information is not required or already available from previous studies. It would also serve as a rapid way of monitoring supra-annual changes in fruit fall patterns and availability levels. Given the frequency of supra-annual variation in fruit production and diet, the current study was limited by its one and a half year time frame. However, the same major items-Sapotaceae and palm fruits-were used by collared and whitelipped peccaries at the same site during studies separated by three years (Fragoso 1994) . Additionally, studies at other sites, especially in Central America, show a remarkable similarity in the diets of at least agouties (Smythe et al . 1996, Naranjo unpublished data) . The importance of Sapotaceae and Palmae at these sites, and their importance also to ungulates in the Peruvian Amazon (Bodmer 1989) suggest that both families have a long history of ecological and evolutionary interactions with terrestrial fruit and seed eaters, and that the Sapotaceae in particular require more careful study in terms of their distribution and role in mammalian community structure. Palm-Mediated B ruchid-Vertebrate Interactions .->,'. 'tf : * _ . . ,r \ Although Janzen (1971) suggested early on that palm seed consumers may be feeding on bruchid beetle larvae rather than

PAGE 258

248 endosperm, subsequent researchers did not follow up on the inplications of this observation for f rugivore-granivore diets. This study indicates that white-lipped peccaries on Maraca rely almost exclusively on beetle larvae and seldom ingest endosperm. Agoutis and squirrels, on the other hand, ingest varying amounts of both kinds of endocarp content. The spatial distribution of endocarp piles used by each species suggests that white-lipped peccaries visit palm-rich habitats as part of their seasonal use of a large scale mosaic, whereas squirrels used the palm seeds and larvae as a staple year round. Agoutis, in contrast, used the seeds and larvae as a non-preferred food when preferred food items are not available. Variation in availability was spatial during this study, but may also be temporal due to years of low forest-wide fruit availability. Complexity in Fruaivorv Systems Figure 7-1 illustrates the participants in the Attalea ' maripa frugiyory system described to date on Maraca. At the largest scale, environmental factors act both on the palm (affecting the level of fruit production) and on the " : interactions between fruits and animals (consumers are more likely to use palm pulp and/or endocarp contents in some habitats and seasons than in others) (Chapter 4) . Seed dispersal, a consequence of animal feeding, afects habitat type by providing an input of seeds that may create new palm

PAGE 259

249 . patches or maintain existing populations (Fragoso 1997b) . At the smallest scale, pulp removal by different consumers affects oviposition by bruchid beetles (Chapter 3). In some cases, different oviposition patterns will translate to different infestation rates. The presence of different consumer species in different habitat types where the palm occurs could translate to different oviposition patterns in different environments. That is, since the distribution of consumers and consumer feeding behavior is patchily distributed over the landscape, this patchiness may superimpose patterns of seed mortality and result in ^. . aggregated tree distributions at the scale of hundreds of meters to tens of kilometers. The activities of different dispersers also affect pattern due to variation in the degree of protection afforded to the dispersed seeds (in this case protection from bruchid beetles) . In the Attalea system, agoutis and primates do not remove seeds from the area of heaviest bruchid activity ' (Chapter 3, Fragoso 1997). Long-distance seed dispersal by tapirs, on the other hand, reduces predation by bruchids, both by placing palm seeds in a palm and beetle poor habitat and by protecting them with fecal matter (Fragoso 1997b, Fragoso unpublished data). Referring back to Figure 7-1, it is possible that long distance dispersal also affects parasitism by wasps on bruchids, if parasitism rates by the Hymenopteran parasites are also habitat specific.

PAGE 260

With respect to the evolution of fruit trais, it is plausible that the thick endocarp of Attalea palms is a response to predation pressure by vertebrates. Fragoso (1994) suggests that bruchid beetles exert selection on seed number rather than endocarp thickness, given the beetles' ability to penetrate all endocarp thicknesses for the Maraca population. In fact, it is plausible that Attalea did not become potential hosts for bruchid beetles until their endocarps were thick enough to provide protection for the developing larvae. Such protection is necessary to reduce predation by vertebrates and possibly parasitism by Braconid wasps, which must also penetrate the endocarp with their ovipositor to lay their eggs on bruchid larvae. A comparison of fruit size, endocarp thicknes, and seed number for the approximately 27 species currently placed in the Attalea palm group might shed some light on the evolutionary response of the palms to different selection pressures on fruit traits. One member of the Attalea genus ("babassu", Attalea speciosa ) has large fruits with extremely thick endocarps and four to six rather than one to three seeds per endocarp (Henderson 19 95) . Attalea tessmannii produces even thicker endocarps (based on diagrams and comments in Henderson 1995), but only three seeds per endocarp. One study in Para, Brazil, suggests that rodents cannot penetrate the Attalea soeciosa endocarps (Anderson et al. 1991). It is also unlikely, based on the data from . Maraca, that white-lipped peccaries would ever crack the

PAGE 261

2 51 endocarps for A. tessmanni and A. speciosa . A systematic comparison of consumer sets, environmental conditions, relative palm abundances, and intensity of predation by bruchid beetles could help tease out some of these complex interrelations. At this point in time, however, the required information is available for only three or four species in the Attalea group. ' .., Current ecological processes can account for some of the dispersal-and survival-related features of Attalea maripa . Thus, the endocarp allows escape from bruchids; the pulp attracts tapirs (Attalea maripa has a high protein content and low fiber content relative to other palm species in the same habitat type; Castellanos and Chanin 1996), which feed on uninfested fruits; the seeds can survive predation by white-lips; and the seedling can survive fire and grazing (Anderson et al . 1991, Fragoso and Silvius unpublished data). From the palm's "point of view," rodents and primates at present do not appear to play a biologically significant role in the system, although they may at times increase infestation rates of seeds by bruchid beetles through pulp removal (Chapter 3). One possible exception is seed dispersal by spider monkeys ( Ateles sp . ) . In the Orinoco basin of southern Venezuela, Castellanos and Chanin (1996) noted that Ateles belzebuth dispersed Attalea maripa seeds to defecation sites under sleeping trees in areas of low adult Attalea density, and that seedling density was high at those

PAGE 262

252 sites. This could be a situation similar to that described by Fragoso (1997b) in northern Brazil. Our interpretation of this and any other frugivory systems, however, is limited by the processes that we currently observe. Selective pressure by currently unimportant consumers or environmental phenomena may have affected the palm in the past, and created the phenomena we see today. We could therefore also expect that current ecological processes are affecting future fruit traits and " palm distribution patterns. Recent work in French Guiana (Brunner et al . 1998) with the palm Astrocarvum sciophillum indicates that a juvenile palm may be nearly 200 years old before its stipe (meristem) emerges from the ground. This means that by the time an individual palm of this species reaches reproductive age, the processes and even some environmental conditions may have changed drastically from the time when it was first dispersed as a seed. While it is unlikely that many paim species have such a drastically • delayed growth pattern, the example of Astrocarvum . .-. , ' sciophillum serves as an important caution that we need to expand our historical perspective when interpreting ecological patterns. Mosaics of Fruit Availability: Scale and Grain Size The understanding gained from walking through a forest in which one knows the habitat preferences of certain animal

PAGE 263

species is very different from the understanding gained from walking through a forest in which one knows the home ranges of individually distinguishable animals. The first kind of information provides averages and eliminates exceptions, thus emphasizing pattern. The second kind of information points to the ways animals make choices, thus emphasizing process. Both types of information may be key to our understanding and management of tropical fruitand seed-eating animal communities. The work of Terborgh et al . (1990) clearly demonstrates the value of the second kind of knowledge described above. The authors created a layered map of the home ranges of individuals of several bird species in 97 hectares of lowland forest in Amazonian Peru. The study yielded important insights into the nature of scientific measurements of species richness. In particular it showed that small sample plots underestimate richness because they miss individuals or species with large territories (note that these are not necessarily rare species, they just move a lot with respect to the observer) . • The study also described how patchiness in forest habitat structure contributes to animal species richness, when richness is measured at a large enough scale. The maximum number of species overlapping at any one point in the study site was 160, yet a total of 319 species were identified in the entire study site. There are two possible explanations for this result. First, some species may have

PAGE 264

used the habitat at a larger spatio-temporal scale than measured in the three month study, rotating between habitats on an annual or supra-annual basis. Second, different ' species may have been restricted to different sites or habitats within the plot for either environmental reasons (food / shelter availability) or behavioral reasons (aggression / exclusion between species). Fruit and seed production by individual tree species or within habitat types or species communities is not constant year round in Neotropical forests (van Schaik et al . 1993, Leigh et al . 1996). Fruit and seed eating animals therefore need an environmental mosaic in order to survive. The nature of the mosaic will depend on the scale of activity of the animal species. As shown for the same Peruvian study site described above, a territorial ground-dwelling frugivorous bird, the white-winged trumpeter ( Psoohia leucoptera ) , established home ranges in areas larger than required to provide food during the season of high fruit availability, but large enough to provide food during the season of fruit shortage. At the trumpeters' scale of activity (approximately 70 hectare territory, or 0.7 km^ ) , either individual trees, tree patches or small-scale edaphic gradients (in this case the old Manu river flood plain) could provide the mosaic of food availability, depending on the grain size used by the animal. A smaller animal, with a smaller home range size, might perceive or encounter only the tree patch or individual tree level of variability, while a

PAGE 265

: 2 55 much larger animal might encounter large-scale edaphic gradients and a sharply segregated habitat mosaic. Such a differential perception of the environmental mosaic could play a role in the coexistence of species with similar diets, such as white-lipped peccaries, squirrels and agoutis. Temporally shifting use of small areas by the largest, more mobile species, would reduce their period of overlap with the home ranges of individual, smaller animals. The nature of the horizontal habitat mosaic (as opposed to vertical forest structure) could thus influence species diversity, when it is measured at a large enough scale. This brings us back to the conclusions of Terborgh et al . ' s (1990) study, which hints at the irrelevance of alpha diversity measures for patchy habitats. If communities are spatially structured by means of body size/home range size segregation, then small-scale measures of community structure, such as alpha diversity, will say little about the processes that structure such communities. " . Plant-Animal I nteractions and the Aggregated Distribution of Tropical Forest Plants The high floristic heterogeneity on Maraca, though undoubtedly influenced by the island's location near the forest-savanna boundary, is probably typical of other Amazonian sites, as indicated by studies of habitat heterogeneity in Peruvian rain forest (Tuomisto et al . 1995), and habitat descriptions of other transitional Amazonian

PAGE 266

256 forests (Peres 1994, Peres and Baider 1997). For example, the decrease in palm abundances from east to west may be related to the increasing distance from the forest-savanna boundary, an active disturbance edge that probably favors palm establishment (see below), Lieberman and Lieberman (1993) reviewed the literature on plant distribution data from floristic plots until 1985. They cite only one plot-based floristic inventory in Amazonian forest, but describe nine other sites in wet and dry tropical forests. All studies, except the one in Amazonian forest, showed a predominance of clumped distributions for abundant tree species. Most plot studies to date have been carried out in small plots of less than 4 ha in extent; only three covered areas of between 10 and 20 ha. Since then, larger floristic plots (50 ha) have been established at several Neo and Paleotropical forest sites (Condit 1995, Foster and Hubbell 1990) Data from one of these sites, in Borneo, analyzed at several quadrat sizes, shows clumping at the largest scale studied (He et al. 1996). Interestingly, it shows an aggregated distribution of species diversity and richness across the 50 ha site, as a result of the non-random distribution of individual species. At a smaller scale, early work on Barro Colorado Island, Panama, showed aggregated patterms at the 5 hectare, 1 hectare, and quarter hectare quadrat sizes (Thorington et al . 1996). Clumping at such scales (50 hectares and below) is biologically interesting from the

PAGE 267

point of view of plant dynamics and small animal responses. In Neotropical forests, however, the home ranges of most fruit-eating species larger than a paca ( Agouti paca ) are much larger than 50 ha, and studies at larger scales are required for understanding of food availability patterns for such animals. . ' Much of the clumping or patchiness in tropical forest plant distribution patterns can be explained by edaphic : • factors (soil type, drainage conditions and topography) . In several studies the distribution of particular plant species or plant associations have shown strong correlations with such factors (Tuomisto and Ruokolainen 1993, Tuomisto et al. 1995, Clark et al . 1995, 1996, Steege 1993). These correlations transcend taxonomic boundaries, and have been examined in detail for legumes (Steege 1993), palms (Kahn and Castro 1985, Kahn and Mejia 1990, Clark et al . 1996), and melastomes (Tuomisto and Ruokolainen 1993). These correlations also occur at several different scales-in transitions over a few meters between sand and clay soils (Tuomisto et al . 1995), along slopes with changing drainage qualities (Steege 1993), and across landscapes at the tens to thousands of kilometers scale (Tuomisto et al . 1995). On Maraca itself, the transition in species identity from east to west depends to a large extent on the shift in substrate type (Milliken and Ratter 1998). Having said all this, several of the above studies have found variation in plant distribution that cannot be

PAGE 268

258 explained by measured edaphic factors. In all cases, the unexplained variation occurs within clay soils. Both Tuomisto et al . (1995; and unpublished data in Condit 1996), and Steeger (1993) found that although white sand soils ' supported characteristic plant associations, with predictable species dominance, and although plant associations on sandy soils were easily distinguishable from those on "brown sand" or "clay" soils, much variation occurred within the clay soils that was not predictable based on measured factors. In Guayana, Steege (1993) found that species dominance varied greatly in different areas of "brown sand" soils. In Peru, species composition on clay soils was also variable (Condit 1996) . He et al . (1996) found that on the 50 hectare plot in Malayasia, much variation in diversity and distribution patterns at the smallest scales could not be explained by edaphic factors, and the authors felt that biological processes occurring at these scales were most likely responsible for the resulting patterns. It is in these soil types and at these scales that the effect of seed dispersal and seed predation processes may be most visible in plant distribution (e.g., the clumped distributions of Attalea . Ecclinusa and Oenocarpus in this study) . The short distances characteristic of seed dispersal by agoutis may create clumped patterns (c.f. Peres and Balder 1997) or maintain a clumped pattern initially created by disturbance or other historical accident. Plant-animal interactions may also affect plant distribution along edaphic

PAGE 269

...^ 2 59 gradients, as animals react to food availability on different soil types, but it is more difficult to tease out such processes against the background of correlated edaphic factors. The following hypothetical examples illustrate how such processes could be operating on Maraca. Several botanical workers have noted that on Maraca the Attalea marioa palm frequently occurs in aggregations around wetlands within the forest, or on slopes leading down to seasonal streams. They proposed that this is due to soil and drainage conditions favored by the palms (Milliken and Ratter 1989) . However, Attalea fruits are a preferred food for tapirs, which ingest the fruits intact and defecate thousands of viable seeds in latrines (Fragoso 1997b) . Since tapirs preferentially defecate in wetland sites even in the seasons when these are dry, many Attalea seeds are carried to the very sites where Attalea is commonly found. How to distinguish between the hypotheses that dispersal by tapirs or edaphic conditions control the final distribution of Attalea ? On Maraca, tapirs exhibit the unusual behavior of defecating in upland latrine sites, at the base of buttressed emergent trees (Fragoso 1997b) . Dense aggregations of seedlings and saplings occur around these latrine sites, indicating that there is higher seedling survival here than around parent trees or at random sites in the forest (Fragoso 1997b) . Thus the same process--dispersal by tapirs--can account for the location of Attalea in both upland and wetland sites. Soil and drainage conditions

PAGE 270

differ between upland and wetland sites, and therefore do not have the same explanatory power as dispersal.; A second, scale-dependent example from the same frugivory system involves a comparison of two frugivores with different home range sizes, acting on the same tree species. Tapirs are the primary dispersers of Attalea seeds on Maraca (based on number and survival rate of seeds moved) . Contrary to paradigm-based expectations (Foster 1986, Foster and Janson 1985), this large-seeded, slow to mature tree appears to be a colonist or edge and disturbance specialist. When the forest is cleared, due to a combination of high dispersal and high seedling survival, Attalea establishes monodominant stands. Attalea and other large-seeded palms are resistant to fire and grazing--their hypocotyl grows downward first before turning to the surface, ensuring that the meristem is protected underground (Anderson et al . 1991, Brunner 1998, Tomlinson 1990). Attalea is also commonly found along natural forest-savanna boundaries. Its competitive ecology and habitat preferences cause it to be densely concentrated in a narrow strip along habitat edges. On Maraca, this corresponds to the riverine forest. Tapirs come to the riverine forest specifically to feed on Attalea fruits, ingesting hundreds of ripe fruits from the ground from several trees during one foraging bout. Because tapirs have long gut retention times (Bodmer 1989) and large home ranges, they carry seeds from the river edge into the terra firme forest several kilometers away. Due to high survival of tapir defecated seeds, this

PAGE 271

261 could broaden the strip along which Attalea occurs. On Maraca, in places where there is little Attalea near the river, there is correspondingly little Attalea in the interior forest. Agoutis, frequently cited as the best dispersers for palm seeds (Chapter 5), operate at too small a scale to account for Attalea dispersal into the forest. With home ranges of about 4 hectares (Chapter 5), they cannot create a bridge between river forest and terra firme forest the way tapirs do. In yet another example from Maraca, home range size interacts with pre-existing vegetation patterns to intensify a vegetation gradient. Several fruit species ( Licania kunthiana, Pradosia surinamensis . Ecclinusa guianensis . Tetraqastris panamensis and other species in the Burseraceae) used by spider monkeys co-occur with Peltocrvne in the central monodominant forests. During a six month transect study (Mendes-Pontes 1995) spider monkeys were the only primates seen with any frequency in the Peltogxnedominated habitat. Peltoqyne itself has no known seed or seedling predators on Maraca, other than leaf cutter ants (Atta sp . ) , and no known dispersers (Nascimento 1994). In other words, with the exception of spider monkeys, the Peltoavne forest is a fairly hostile environment to fruit and seed eating vertebrates. The habitat is extensive enough that only white-lipped peccaries can easily move across it in a day and include the area in their home range. All other fruit eating species would face severe seasonal and probably sub-seasonal food

PAGE 272

262 shortages in this habitat. In the absence of fruit eating species, the Peltoavne forest also becomes a barrier to seed dispersal, and may contribute to the turnover in species composition from east to west on the island. The vegetation transition is sharpest at precisely the location of the Peltoavne area (Milliken and Ratter 1989, 1998); since it al so coincides with a geomorphological transition, it is difficult to tease apart the two alternative explanations for the vegetation turnover.

PAGE 273

263 ENVIRONMENTAL FACTORS Habitat Type^ ^ Soil Type Drainage Weather Seasonal bupra-annual White-lipped Peccary " Kodents SEED DISPERSAL Short Distance Mid Distance Long Distance Small ungulates Primates (Cebus, Saimiri, Alouata) Rodents (Dasyprocta , Agouti, Sciurus) Dung beetles Ateles Braconid Wasps Invirtebrste ljuj.p-eate rs rungi Tapir (.oviposition) Bruchid Beetles Trichogrammat ids Ants Endocarp-based Interactions Pulp-based Interactions Figure 7-1. Complex interactions among biotic and abiotic factors in the Attalea marioa system on Maraca. See text for explanation.

PAGE 274

APPENDIX I • • "EDIBLE*" PLANT SPECIES ON PHENOLOGY TRANSECTS Family Species Anacardiaceae Spondias mombin Annonaceae Duguetia quianensis Burseraceae Tetracfastris oanamensis Caryocaraceae Caryocar sp. Chrysobalanaceae Licania kunthiana Chrysobalanaceae Couepia sp. Icaniaceae Leretia sp. Lecythidaceae Augusta suberba Lecythidaceae Lecythis corrugata Lecythidaceae Couratari multi flora Leguminosae Hymenea parvi flora Leguminosae Hymenea courbaril Leguminosae Entero labium sp. Leguminosae Swartzia dipetala Leguminosae Andira inermis Moraceae Bagassa surinawensis Moraceae Brosimum sp. Moraceae Ficus sp. Myristicaceae Virola surinamensis Myristicaceae Virol a sp. Palmae Oenocarpus bacaba Palmae Attalea maripa Palmae Astrocaryum aculeatum Palmae Astrocaryum mumbaca Palmae Euterpe precatoria Palmae Mauri tia flexuosa W \ ^ 2 64

PAGE 275

265 Fami ly Species Pas si f loraceae Passi flora sp. Rubiaceae Duroia eriopila Rubiaceae Genipa americana Sapotaceae Chrysophyllum sp. Sapotaceae Pouteria venosa Sapotaceae Pouteria hispida Sapotaceae Pouteria surumuensis Sapotaceae Pradosia surinamensis Sapotaceae Pouteria sp. Sapotaceae Pouteria sp . "abiu" Sapotaceae Sapotaceae sp. *" Edible" species are those known or expected to be be eaten by agoutis, white-lipped peccaries and collared peccaries, based on studies at Maraca (Fragoso 1994, Silvius this study) and reports in the literature for other Neotropical sites.

PAGE 276

APPENDIX II FLORISTIC PLOT DATA SUMMARIZED FROM MILLIKEN AND RATTER (1989) AND NASCIMENTO (1994) Plot # Ref . Habitat Area (ha) # Spp # Trees Trees/ ha H' J 1 J M+R Tall mixec terra f irme forest , eastern plateau 0.369 54 20C 504 .61 3 . 567 0.894 M-HK Tall mixec terra f irme forest , eastern plateau 0.519 51 200 385.09 3.271 0.832 3,4,5 M-L D iXi + K iaii mixec terra f irme forest , eastern plateau 0.75 80 597 408 . 9 3.61/3 .37+/0.14 0.824/ 0.9+/0 . 02 3 N Tall mixed terra f irme forest , eastern plateau 0.25 4 1 117 -L J/ 4 N Tall mixed terra f irme forest , eastern plateau 0.25 41 107 5 N Tall mixed terra f irme forest , eastern plateau 0.25 45 117 266

PAGE 277

267 Plot Ref . Habitat Area # # Trees / H ' J # (ha) Spp Trees ha 6 M+R Angico forest (marginal form of typical terra f irme forest on the eastern end, on a steep slope 0 . 103 31 56 541.42 3 .266 0.951 leading down to campo ) 7 M+R Tall mixed terra f irme foreste , with campina influence 0.129 22 80 622.3 2 .447 0.792 8 M+R Mixed terra f irme forest on sandy soil , depauparate with marginal and colonizing speices 0.25 30 149 596 2 .706 0 .796 9, 10, N Terra 0.75 31 309 3 . 02 0 . 88 11 f irme forest T.T T +~ lit— wi unouc Peltoqyne + /2.89 + /0 . 08 + /0.01 9 N Terra f irme forest without Peltoqyne 0.25 33 106

PAGE 278

268 Plot # Ref . Habitat Area (ha) # Spp # Trees Trees / ha H' J IC N Terra f irme forest without Peltoqyne 0.25 33 101 11 N Terra f irme forest wi cnouu Peltoqyne 0.25 28 102 12 13 14 N Terra f irme forest with Io tr Peltoqyne 0.75 2 . 96 + /0.27 0 . 84 + /0.02 12 N Terra f irme forest with C 3 O d bparsG Peltoqyne 0.25 40 115 13 N Terra f irme forest with Peltoqyne 0.25 31 95 14 N Terra f irme forest with Peltoqyne 0.25 24 132 15 16 17 N Terra f irme forest dominated by Peltoqyne 0.75 2 . 58 + /0.18 0.79 + /0.04 15 N Terra f irme forest dominated by Peltoqyne 0.25 26 118

PAGE 279

269 Plot # Ref . Habitat Area (ha) # Spp # Trees Trees / ha H' J 16 N Terra f irme forest dominated by Peltoqyne 0.25 29 129 17 N Terra f irme forest dominated by Peltoqyne 0.25 24 137 18 M + R Tall mixec, terra f irme forest 0.2 40 100 499 3 .358 0.91 19 M + R Tall mixed terra f irme forest 0 . 192 41 104 542.64 3 . 15 0.848 20 21 22 N Terra f irme forest dominated by Peltoqyne 0.75 2.27 + /0 . 43 0.74 + /0.09 20 N Terra f irme forest dominated by Peltoqyne 0.25 19 150 21 N Terra f irme forest dominated Peltoqyne 0.25 27 110 22 N Terra f irme forest dominated by Peltoqyne 0.25 18 175

PAGE 280

270 Plot # Ref . Habitat Area (ha) # Spp # Trees Trees / ha H • J 23 M+R Tall forest with Peltogyne and Pradosia dominance 0 . 177 35 80 452.9 2 .988 0.841 24 M+R Tall forest with Peltogyne dominance , on flooded soil 0 .083 18 40 481 . 97 2.42 0.837 25 M+R Tall mixed terra f irme forest 0 . 115 20 40 348.74 2 .777 0.927 26 M+R Tall mixed terra f irme forest 0.329 46 100 303 . 9 3 .615 0.944 27 M+R Tall mixed terra f irme forest 0.093 27 40 429.14 3.169 0.962 28 M+R Tall mixed terra f irme forest with Peltogyne dominance 0 . 118 15 80 667 .78 1.816 0.67 29 M+R Mixed terra f irme forest 0.378 73 199 526.86 3.811 0.888 30 M+R Mixed terra f irme forest 0.061 24 40 660.8 2.934 0.923

PAGE 281

REFERENCES Abacus Concepts. 1986. Statview 512"^. Abacus Concepts, Berkeley, California, USA. Adler, G. A. 1994. Tropical forest fragmentation and isolation promote asynchrony among populations of a frugivorous rodent. Journal of Animal Ecology 63: 903-911. Adler, G. A. 1995. Fruit and seed exploitation by Central American spiny rats, Proechimvs semisoinosus . Studies on Neotropical Fauna and the Environment 30: 237-244. Adler, G. A. 1998. Impacts of resource abundance on populations of a tropical forest rodent. Ecology 79: 242254. Allen, C. M. 1997. Frugivores, palms, and conservation in Amazonia. M.S. thesis. University of Florida, Gainesville, Florida, USA. , • > -ta . Anderson, A. B. 1983. The biology of Orbianva martiana (Palmae) , a tropical dry forest dominant in Brazil. Ph.D. dissertation. University of Florida, Gainesville, Florida, USA. Anderson, A. B., P. H. May, and M. J. Balick. 1991. The subsidy from nature. Palm forests, peasantry, and development on an Amazon frontier. Columbia University Press, New York, USA. Baker, R. H. 1983. Siamodon hispidus (rata algodonera hispida, hispid cotton rat). Pp. 490-492 in D. H. Janzen, ed. Costa Rican Natural History. University of Chicago Press, Chicago, Illinois, USA. Balee, W. 1988. Indigenous adaptations to Amazonian palm forests. Principes 32: 47-54 Balick, M. J. 1988. The palm--tree of life. Advances in Economic Botany, Vol. 6. Barreto, G. R. , andO. E. Hernandez. 1988. Aspectos bioecologicos de los baquiros ( Tavassu taiacu y T. pecari ) en el Estado Cojedes: estudio comparative. Trabajo especial de grado, Universidad Central de Venezuela, Caracas, Venezuela. 271

PAGE 282

272 Bellehumeur, C, and P. Legendre. 1998. Multiscale sources of variation in ecological variables: modeling spatial dispersion, elaborating sampling designs. Landscape Ecology 13: 15-25. V. . , Bodmer, R. E. 1989a. Frugivory in Amazonian Artiodactyla : evidence for the evolution of the ruminant stomach. Journal of Zoology London 219: 457-467. Bodmer, R. E. 1989b. Frugivory in Amazon ungulates. Ph.D. dissertation. University of Cambridge, Cambridge, England. Bodmer, R. E. 1990. Responses of ungulates to seasonal inundations in the Amazon floodplain. Journal of Tropical Ecology 6: 191-201. Bodmer, R. E. 1991. Strategies of seed dispersal and seed predation in Amazonian ungulates. Biotropica 23: 255-261 Bodmer, R. E., J. F. Eisenberg, and K. H. Redford. 1997. Hunting and the likelihood of extinction of Amazonian mammals. Conservation Biology 11: 460-466. Borowicz, V. A. 1988. Do vertebrates reject decaying fruit? An experimental test with Cornus amomum fruits. Oikos 53: 74-78. . , , , Bradford, D.F., and C. C. Smith. 1977. Seed predation and seed number in Scheelea palm fruits. Ecology 58: 667-673. Brewer, S. W. , and M. Rejmanek. 1998. Small rodents are significant dispersers of tree seeds in a Neotropical forest. Paper presented at the Annual Meeting of the Association for Tropical Biology, Baltimore, Maryland, USA. Brunner, A., P. Charles-Dominique, J. Chave, B. Di-Pietro, M. A. Dubois, and B. Riera. 1998. Astrocarvum scioohilum as indicator of long term dynamics of Guianese forest. Paper presented at the Annual Meeting of the Association for Tropical Biology, Baltimore, Maryland, USA. Bucholz, R., and D. J. Levey. 1990. The evolutionary triad of microbes, fruits and seed dispersers: an experiment in fruit choice by cedar waxwings, Bombvcilla . Oikos 59: 200204 . : . _ . Buckley, J. S. 1983. The feeding behavior, social behavior, and ecology of the white-faced monkey, Cebus caoucinus . at Trujillo, Northern Honduras, Central America. Ph.D. dissertation, University of Texas, Austin, Texas, USA.

PAGE 283

273 Bullock, S. H. 1980. Dispersal of a desert palm by " opportunistic frugivores. Principes 24: 29-32. Caldecott J. 1988. Hunting and wildlife management in Sarawak. lUCN, Gland, Switzerland.. Caldecott, J. 1992. Ecology of the bearded pig in Sarawak. Pp. 117-129 in R. H, Barrett and F. Spitz, eds . Biology of Suidae. Imprimerie des Escartons, Briangon, France. Caldecott, J.O., R. A. Blouch, and A. A. MacDonald. 1993. The Bearded Pig (Sus barbatus ) . Pp. 136-144 in W. L. R. Oliver, ed. Pigs, peccaries and hippos--status survey and conservation plan. lUCN, Gland, Switzerland. Campbell, J. E., S. B. Franklin, D. J. Gibson, and J. A. Newman. 1998. Permutation of two-term local quadrat variance analysis: general concepts for interpretation of peaks. Journal of Vegetation Science 9: 41-44. Cant, J. G. H. 1977. A census of the agouti ( Dasvorocta punctata) in seasonally dry forest at Tikal, Guatemala, with some comments on strip censusing. Journal of Mammalogy 58: 688-690. Castellanos, H. G., and P. Chanin. 1996. Seasonal differences in food choice and patch preference of longhaired spider monkeys (Ateles belzebuth ) . Pp. 451-466 in Norconk, M. A., A. L. Rosenberger, and P. A. Garber, eds. Adaptive radiations of Neotropical primates. Plenum Press, New York, New York, USA. Chapman, C. A. 1988. Patterns of foraging and range use by three species of Neotropical primates. Primates 29: 177-194, Chapman, C. A., L. J. Chapman, and L. Lefebvre. 1989. Variability in parrot flock size: possible functions of communal roosts. The Condor 91: 842-847. Chapman, C. A,, L. J. Chapman, R. Wrangham, G. IsabiryeBasuta and K. Ben-David. 1997. Spatial and temporal variability in the structure of a tropical forest. African Journal of Ecology 35: 287-302. ^ Charles-Dominique, P. 1986. Inter-relations between frugivorous vertebrates and pioneer plants: Cecropia, birds and bats in French Guyana. Pp. 120-135 in A. Estrada and T. H. Fleming, eds. Frugivores and seed dispersal. Dr W. Junk Publishers, Dordrecht, Netherlands. Charles-Dominique, P., M. Atramentowicz , M. CharlesDominique, H. Gerard, A. Hladik, C. M. Hladik, and M. F. Prevost. 1981. Les mammiferes frugivores arboricles

PAGE 284

274 nocturnes d ' une foret guyanaise: inter-relations plantesanimaux. Rev. Ecol . (Terre et Vie) 35: 341-435. Cintra, R. , and V. Horna. 1997. Seed and seedling survival of the palm Astrocarvum murumuru and the legume tree Dipteryx micrantha in gaps in the Amazonian forest. Journal of Tropical Ecology: 257-277. Cipollini, M. L., and E. W. Stiles. 1993. Fruit rot, antifungal defense, and palatability of fleshy fruits for frugivorous birds. Ecology 74: 751-762. Cipollini, M. L., and D. J. Levey. 1997. Secondary metabolites of fleshy vertebrate-dispersed fruits: adaptive hypotheses and implications for seed dispersal. American Naturalist 150: 346-372. Clark, D. A., D. B. Clark, R. Sandoval, and M. V. Castro. 1995. Edaphic and human effects on landscape-scale distributions of tropical rain forest palms. Ecology 76: 2581-2594. , Clark, D. B., D. A. Clark, P. M. Rich, S. Weiss, and S. F. Oberhauer. 1996. Landscape-scale evaluation of understory light and canopy in a Neotropical lowland rain forest. Canadian Journal of Forest Research 26: 747-757. CoatesEstrada, R., and A. Estrada. 1986. Manual de identif icacion de campo de los mamiferos de la estacion de biologia "Los Tuxtlas." Universidad Nacional Autonoma de Mexico, Mexico. Condit, R. 1995. Research in large, long-term tropical forest plots. Trends in Ecology and Evolution 10: 18-22. Condit, R. 1996. Defining and mapping vegetation types in mega-diverse tropical forests. Trends in Ecology and Evolution 11: 4-5. Balling, J, W. , K. E. Harms, and R. Aizprua. 1997. Seed damage tolerance and seedling resprouting ability of Prioria copaiff in Panama. Journal of Tropical Ecology 13481490. Davis, W. M. T. 1907. Insects as the food of squirrels Canadian Entomology 39: 16. De Steven, D. , D. M. Windsor, F. E. Putz. 1987. Vegetative and reproductive phenologies of a palm assemblage in Panama, Biotropica 19: 342-356.

PAGE 285

275 Defler, T. R. 1979. On the ecology and behavior of Cebus albif rons in eastern Colombia: I. Ecology. Primates 20: 475490. Delgado, C, G. Couturier, and A. Delobel. 1997. Oviposition of seed-beetle Carvoborus serripes (Sturm) (Coloptera: Bruchidae) on palm ( Astrocarvum chambira) fruits under natural conditions in Peru. Annals of the Society of Entomology, France. (N. S.) 33: 405-409. Delobel, A., G. Couturier, F. Khan & J. A. Nilsson. 1995. Trophic relationships between palms and bruchids (Coleoptera: Bruchidae: Pachymerini) in Peruvian Amazonia. Amazoniana 8: 209-219. Dubost, G. 1988. Ecology and social life of the red acouchy, Mvoprocta exilis : comparison with the orange-rumped agouti, Dasvprocta leporina . Journal of Zoology, London. 214: 107-123. Eisenberg, J. F. 1980. The density and biomass of tropical mammals. Pp. 35-55 in M. E. Soule and B. A. Wilcox, eds . Conservation biology, an evolutionary-ecological perspective. Sinauer, Sunderland, Massachusetts, USA. Eisenberg, J. F. 1990. Neotropical mammal communities. Pp. 358-370 in A. H. Gentry, ed. Four Neotropical rainforests. Yale University Press, New Haven, Connecticut, USA Eisenberg, J. F., M. A. O'Connell, and P. V. August. 1979. Density, productivity, and distribution of mammals in two Venezuelan habitats. Pp. 187-207 in J. F. Eisenberg, ed. Vertebrate ecology in the northern Neotropics. Smithsonian Institution Press, Washington, DC, USA. Emmons, L. H. 1980. Ecology and resource partitioning among nine species of African rain forest squirrels. Ecological Monographs 50: 31-54. Emmons, L. H. 1982. Ecology of Proechimvs (Rodentia, Echimydae) in south-eastern Peru. Tropical Ecology 23: 280290. Emmons, L. H. 1984. Geographic variation in densities and diversities of non-flying mammals in Amazonia. Biotropica 16: 210-222. Emmons, L. H. , and F. Feer . 1997. Neotropical rainforest mammals. A field guide. Second edition. University of Chicago Press, Chicago, Illinois, USA. Estrada, A., and R. Coates-Estrada . 1991. Howler monkeys ( Alouatt;<^ p^lliata ) , dung beetles ( Scarabaeidae) and seed

PAGE 286

276 dispersal: ecological interactions in the tropical rain forest of Los Tuxtlas, Mexico. Journal of Tropical Ecology 7 : 459-474 . Estrada, A., and R. Coates-Estrada . 1986. Frugivory by howling monkeys (Alouatta palliata ) at Los Tuxtlas, Mexico: dispersal and fate of seeds. Pp. 93-105 in A. Estrada and T. H. Fleming, eds . Frugivores and seed dispersal. Dr W. Junk Publishers, Dordrecht, Netherlands. Estrada, A. and T. H. Fleming. 1986. Frugivores and seed dispersal. Dr W. Junk Publishers, Dordrecht, Netherlands. Fleming, T. H. 1970. Notes on the rodent faunas of two Panamanian forests. Journal of Mammalogy 51: 473-490. Fleming, T. H. 1974. The population ecology of two species of Costa Rican heteromyid rodents. Ecology 55: 493-510. Fleming, T. H. 1988. The short-tailed fruit bat. A study in plant-animal interactions. University of Chicago Press, Chicago, Illinois, USA. Fleming, T. H. 1992. How do fruitand nectarfeeding birds and mammals track their food resources. Pp. 355-391 in M. D. Hunter, T. Ohgushi , and P. W. Price, eds. Effects of resource distribution on animal-plant interactions. Academic Press, Inc., San Diego, California, USA. Fleming, T. H. and A. Estrada. 1993. Frugivory and seed dispersal: ecological and evolutionary aspects. General introduction. Vegetatio 107/108: xi-xii. Forget, P. 1990. Seed-dispersal of Vouacaooua americana (Caesalpinaceae) by caviomorph rodents in French Guiana. Journal of Tropical Ecology 6: 459-468. Forget, P. 1991. Scatterhoarding of Astrocarvum paramaca by Proechymis in French Guiana: comparison with Mvoorocta exilis . Tropical Ecology 32: 155-167. i ; • • Forget, P. 1993. Post-dispersal predation and scatterhoarding of Dipteryx panamensis (Papilionaceae) seeds by rodents in Panama. Oecologia 94: 255-261. Forget, P. ^ 1994. Recruitment pattern of Voucapoua americana (Caesalpiniaceae) , a rodent dispersed tree species in French Guiana. Biotropica 26: 408-419. Forget, P., and T. Milleron. 1991. Evidence for secondary seed dispersal by rodents in Panama. Oecologia 87: 596-599.

PAGE 287

277 Forget, P., E. Munoz, and E. G. Leigh, Jr. 1994. Predation by rodents and bruchid beetles on seeds of Scheelea palms on Barro Colorado Island, Panama. Biotropica 26: 420-426. Foster, R. 1996. Famine on Barro Colorado Island. Pp. 201212 in E. G. Leigh, Jr., A. S. Rand, and D. M. Windsor, eds . The ecology of a tropical forest: seasonal rhythms and long term changes. Second edition. Smithsonian Institution Press, Washington, D. C, USA. Foster, R. B., and N. V. L. Brokaw. 1996. Structure and history of the vegetation of Barro Colorado Island. Pp. 6782 in E. G. Leigh, Jr., A. S. Rand, and D. M. Windsor, eds. The ecology of a tropical forest: seasonal rhythms and longterm changes. Second Edition. Smithsonian Institution, Washington, D. C. Foster, R. B. and S. P. Hubbell. 1990. The floristic composition of the Barro Colorado Island forest. Pp. 85-98 in A. H. Gentry. Four Neotropical rainforests. Yale University Press, New Haven, Connecticut, USA. Foster, S. A. 1986. On the adaptive value of large seeds for tropical moist forest trees: a review and synthesis. Botanical Review 52: 260-299. Foster, S. A., and C. H. Janson. 1985. The relationship between seed size and establishment conditions in tropical woody plant. Ecology 66: 773-780. Fragoso, J, M. 1989. White-lipped peccaries on Maraca Island: Preliminary report. Pp. 34-42 in J. A. Ratter and W. Milliken, eds. Maraca Rainforest Project-Preliminary report-mammals (Part 1). Institute Nacional de Pesquisas da Amazonia; Royal Geographical Society; Secretaria Especial do Meio Ambiente Fragoso, J. M. 1991. White-lipped peccaries and palms on Maraca Island, Brazil. Project report, 1988-1991. Wildlife Conservation Society. Fragoso, J. M. 1993. The ecology of white-lipped peccaries: ecological interactions between seed predators and palms in an Amazonian forest. Project report to WWF and NYZS/WCS. Fragoso, J. M. 1994. Large mammals and the community dynamics of an Amazonian rain forest. Ph.D. dissertation. University of Florida, Gainesville, Florida, USA. Fragoso, J. M. V. 1997a. Desapariciones locales del baquiro labiado (Tayassu pecari ) en la Amazonia: migracion, sobrecosecha, o epidemia? Pp. 309-312 in T. G. Fang, R. E. Bodmer, R. Aquino, and M. H. Valqui, eds. Mane jo de fauna

PAGE 288

278 sivestre en la Amazonia. Institute de Ecologia, La Paz, Bolivia . . Fragoso, J. M. V. 1997b. Tapir-generated seed shadows: scaledependent patchiness in the Amazon rain forest. Journal of Ecology 85: 519-529. Fragoso, J. M. V. 1998. Home range and movement patterns of white-lipped peccary ( Tavassu pecari ) herds in the northern Brazilian .Amazon. Biotropica 30: 458-469. Fragoso, J. M. V. In press. Habitat partitioning by rain forest peccary herds and species: behavioral causes and ecological implications. Journal of Mammalogy. Galetti, M. , M. Paschoal, and F. Pedroni . 1992. Predation on palm nuts ( Svaarus romanzof f iana ) by squirrels ( Sciurus inarami ) in south-east Brazil. Journal of Tropical Ecology 8: 121-123. Gardner, A. L. 1977. Feeding habits. Pp. 293-350 in R. J. Baker, J. K. Jones, Jr., and D. C. Carter, eds . Biology of the New World family Phyllostomatidae, Part 2. Special Publications of the Museum, No. 16. Texas Tech University Press, Lubbock, Texas, USA. Gautier-Hion, A., R. Duplantier, R. Quris, F. Feer, C. Sourd, J. P. Decoux, G. Dubost, L. Emmons, C. Erard, B. Hecketsweiler , A. Moungazi, C. Roussilhon, and J. M. Thiollay. 1985. Fruit characters as a basis of fruit choice and seed dispersal in a tropical forest vertebrate community. Oecologia 65: 324-337. Gilbert, L. E. 1980. Food web organization and conservation of Neotropical diversity. Pp. 11-34 in M. E. Soule, ed. Conservation biology. The science of scarcity and diversity. Sinauer Associates, Inc., Sunderland, Massachusetts, USA. Gilwicz, J. -1984. Population dynamics of the spiny rat Proechimys semisoinosus on Orchid Island (Panama) . Biotropica 16: 73-78. Glanz, W. E., R. W. Thorington, Jr., J. Giacalome-Madden, and L. R. Heaney. 1996. Seasonal food use and demographic trends in Sciurus aranatensi s . Pp. 239-252 in E. G. Leigh, Jr., A. S. Rand, and D. M. Windsor, eds. The ecology of a tropical forest: seasonal rhythms and long term changes. Second edition. Smithsonian Institution Press, Washington, DC, USA. Greenhall, A. M. 1957. Food preferences of Trinidad fruit bats. Journal of Mammalogy 38: 409-410.

PAGE 289

279 Halffter, G. and W. D. Edmonds. 1982. The nesting behavior of dung beetles ( Scarabaeinae ) . An ecological and evolutive approach. Institute de Ecologia, Mexico City, DF, Mexico. Hallwachs, W. 1986. Agoutis ( Dasvorocta punctata ) : the inheritors of guapinol ( Hvmenea courbaril : Leguminosae) . Pp. 285-304 in A. Estrada and T. H. Fleming, eds . Frugivores and seed dispersal. Dr W. Junk Publishers, Dordrecht, Netherlands . . _ ... Hanski, I., and Y. Cambefort. Dung beetle ecology. Princeton University Press, Princeton, New Jersey, USA Harms, K., and J. W. Dalling. 1997. Damage and herbivory tolerance through resprouting as an advantage of large seed size in tropical trees and lianas. Journal of Tropical Ecology 13: 617-621. He, F. 1993. Ecology of a Tropical Rain Forest: spatial Structure and Ecological Processes. Ph.D. dissertation. University of Montreal, Montreal, Canada. He, F., P. Legendre, and J. V. LaFrankie. 1996. Spatial pattern of diversity in a tropical rain forest in Malaysia. Journal of Biogeography 23: 57-74. Heaney, L. R., and R. W. Thorington. 1978. Ecology of Neotropical red-tailed squirrels, Sciurus aranatensi s . in the Panama Canal zone. Journal of Mammalogy 59: 846-851. Henderson, A. 1995. The palms of the Amazon. Oxford University Press. New York, New York, USA. Hendrichs, H. 1977. Untersuchungen zur Saugetierf aua in einem palaotropischen und einem neotropischen Bergregenwaldgebiet . Saugetierkd. Mitt. 25: 214-225. Herrera, C. M. 1982. Defense of ripe fruit from pests: its significance in relation to plant-disperser interactions. American Naturalist 120: 218-247. Herrera, C. M. 1984. Avian interference of insect frugivory: an exploration into the plant-birdfruit pest evolutionary triad. Oikos 42: 203-210. Herrera, C. M. 1989. Vertebrate frugivores and their interaction with invertebrate fruit predators: supporting evidence from a Costa Rican dry forest. Oikos 54: 185-188. Hladik, A., and C. M. Hladik. 1969. Rapports trophiques entre vegetation et primates dans la foret de Barro Colorado (Panama). Terre et Vie 116: 343-413.

PAGE 290

280 Hoch, G. A., and G. H. Adler. 1997. Removal of black palm ( Astrocarvum standlevanum ) by spiny rats ( Proechimys semispinosus ) . Journal of Tropical Ecology 13: 51-58. Howe, H. F. 1990. Survival and growth of juvenile Virola surinamensis in Panama: effects of herbivory and canopy closure. Journal of Tropical Ecology 6: 259-280. Howe, H. F. 1993a. Specialized and generalized dispersal systems: where does the paradigm stand? Vegetatio 107/108: 3-13 . Howe, H. F. 1993b. Aspects of variation in a Neotropical seed dispersal system. Vegetatio 107/108: 149-162. Howe, H. F., E. W. Schupp, and L. C. Wet ley. 19 85. Early consequences of seed dispersal for a Neotropical tree. Ecology 66: 781-791. Hubbell, S. P. 1979. Tree dispersion, abundance and diversity in a tropical dry forest. Science 203: 1299-1309. Ims, R. A. 1995. Movement patterns related to spatial structures. Pp. 85-109 in L. Hansson, L. Fahrig, and G. Merriam, eds . Mosaic landscapes and ecological processes. Chapman and Hall, London, UK. Izawa, K. 1979. Foods and feeding behavior of wild blackcapped capuchins ( Cebus aoella ) . Primates 20: 57-76. Janson, C. H. , and L. G. Emmons. 1990. Ecological structure of the non-flying mammal community at the Cocha Cashu Biological Station, Manu National Park, Peru. Pp. 314-338 in A. Gentry, ed. Four Neotropical forests. Yale University Press, New Haven, Connecticut, USA. Janzen, D. H. 1970. Herbivores and the number of tree species in tropical forest. American Naturalist 104: 501-528. Janzen, D. H. 1971. The fate of Scheelea rostrata fruits beneath the parent tree: predispersal attack by bruchids. Principes 15: 89-101. Janzen, D. H. 1972. Predation on Scheelea palm seeds by bruchid beetles: seed density and distance from the parent palm. Ecology 53 (5): 954959. Janzen, D. H. 1977. Why fruits rot, seeds mold, and meat spoils. American Naturalist 111: 691-713. Janzen, D. H. 1982a. Simulation of Andira fruit pulp by bats reduces seed predation by Cleoaonus weevils. Brenesia 20: 165-170. removal

PAGE 291

281 Janzen, D. H. 1982b. Cleoaonus weevil seed predation can be predicted by fruit punctures. Brenesia 19/20: 591-593. Janzen, D. H. 1983. Larval biology of Ectomveloi s muriscis (Pyralidae: Phycitinae) , a Costa Rican fruit parasite of Hymenaea courbaril ( Leguminosae : Caesalpinoidae ) . Brenesia 21: 387-393. Janzen, D. H. 1986. Chihuahuan desert nopaleras: defaunated big mairanal vegetation. Annual Review of Ecology and Systematics 17: 595-636. Janzen, D. H., and P. S. Martin. 1982. Neotropical anachronisms: the fruits the Gomphoteres ate. Science 21519-27. Johnson, C. D. , S. Zona, and J. A. Nilsson. 1995. Bruchid beetles and palm seeds: recorded relationships. Principes 39 (1) :25-35. ,. Jordano, P. 1987. Avian fruit removal: effects of fruit variation, crop size, and insect damage. Ecology 6817111723. Kahn, F., and A. de Castro. 1985. The palm community in a forest of central Amazonia, Brazil. Biotropica 17: 210-216. Kahn, F., and J. de Granville. 1992. Palms in forest ecosystems of Amazonia. Springer-Verlag, New York, New York USA. ' Kahn, F., and K. Mejia. 1990. Palm communities in wetland forest ecosystems of Peruvian Amazonia. Forest Ecology and Management 33/ 44: 169-179. Kiltie, R. A. 1981. Distribution of palm fruits on a rain forest floor: why peccaries forage near objects. Biotropica 13: 141-145. Kiltie, R. A., and J. Terborgh. 1983. Observations of rain forest peccaries in Peru: why do white-lipped peccaries form herds? Z. Tierpsychol. 62: 241-255. ... Klein, L. L. 1972. The ecology and social organization of the spider monkey, Ateles belzebuth . Ph.D. dissertation. University of California, Berkeley, California, USA. Kozakiewicz, M. 1995. Resource tracking in space and time, pp. 136-148 in L. Hansson, L. Fahrig, and G. Merriam, eds . Mosaic landscapes and ecological processes. Chapman and Hall, London, UK. .....

PAGE 292

282 Krebs, C. J. 1989. Ecological methodology. HarperCollins Publishers, New York, New York, USA. Krischik, V., E. S. McCloud, and J. A. Davidson. 1989. Selective avoidance by vertebrate frugivores of green holly berries infested with a Cecidomyiid fly (Diptera: Cecidomyiidae) . Aerican Midland Naturalist 121: 30-354. Larson, D., and Howe, H. F. 1987. Dispersal and destruction of yirola surinamensis seeds by agoutis: Appearance and reality. Journal of Mammalogy 68: 859-860. Leigh, E. G., Jr., A. S. Rand, and D. M. Windsor, eds . 1996. The ecology of a tropical forest: Seasonal rhythms and long term changes. Second edition. Smithsonian Institution Press, Washington, DC, USA. Leighton, M. , and D. R. Leighton. 1983. Vertebrate responses to fruiting seasonality within a Bornean rain forest. Pp. 181-196 in S. L. Sutton, T. C. Whitmore, and A. C. Chadwick, eds. Tropical rain forests: Ecology and management. Blackwell Scientific Publications, London, UK. Levey, D. J., T. C. Moermond, and J. S. Dens low. 19 93. Frugivory: an overview. Pp. 2 82-2 94 in L. A. McDade, K. S. Bawa, H. A. Hespenheide, and G. S. Hartshorn, eds. La Selva. Ecology and natural history of a Neotropical rain forest. University of Chicago Press, Chicago, Illinois, USA. Lieberman, M. , and D. Lieberman. 1993. Patterns of density and dispersion of forest trees. Pp. 106-119 in L. A. McDade, K. S. Bawa, H. A. Hespenheide, and G. S. Hartshorn, eds. La' Selva. Ecology and natural history of a Neotropical rain forest. University of Chicago Press, Chicago, Illinois, USA. Lieberman, D. , M. Lieberman, R. Peralta, andG. S. Hartshorn. 1996. Tropical forest structure and composition on a largescale altitudinal gradient in Costa Rica. Journal of Ecology 84: 137-152. Losos, E. 1995. Habitat specificity of two palm species: experimental transplantation in Amazonian succesional forests. Ecology 76: 2595-2606. Ludwig, J. A. and J. F. Reynolds. 1988. Statistical Ecology. John Wiley and Sons, London, UK. Mack, A. L. 1998. An advantage of large seed size: tolerating rather than succumbing to seed predators Biotropica 30: 604-608. Malcolm, J. R. 1990. Estimation of mammalian densities in continuous forest north of Manaus . Pp. 339-357 in A. H.

PAGE 293

283 Gentry. Four Neotropical rainforests. Yale University Press, New Haven, Connecticut, USA Miller, R. P. 1991. Dinamica da regeneragao natural de jutai-mirim ( Hvmenea parvifolia Huber Leguminosae Caesalpinoidae) na Ilha de Maraca-RR. Masters thesis. Institute Nacional de Pesquisas da Amazonia, Fundagao Universidade do Amazonas, Manaus, Amazonas , Brazil. Milliken, W., and J. A. Ratter. 1989. The vegetation of the Ilha de Maraca. Field report of the vegetation survey of the Maraca Rain Forest Project. Royal Botanical Garden, Edinburgh, UK. Milliken, W. , and J. A. Ratter. 1998. Maraca: The biodiversity and environment of an Amazonian rainforest. John Wiley and Sons, Chichester, West Sussex, UK. Milton, K. 1980. The foraging strategy of howler monkeys. Columbia University Press, New York, New York, USA. Monteith, G. H., and R. I. Storey. 1981. The biology of CgphalQd??miy$, a genus of dung beetles which synthesizes "dung" from plant material (Coleoptera: Scarabaeidae : Scarabaeinae) . Memoirs of the Queensland Museum 20: 253-71. Moskovits, D. K. 1985. The behavior and ecology of two Amazonian tortoises, Geochelone carbonaria and Geochelone denticulata, in northwestern Brazil. Ph.D. dissertation. University of Chicago, Chicago, Illinois, USA. Moskovits, D. K. 1998. Population and ecology of the tortoises G?Qghglone carbonaria and £. denticulata on the Ilha de Maraca. Pp. 263-284 in W. Milliken and J. Ratter, eds. Maraca: The biodiversity and environment of an Amazonian rainforest. John Wiley and Sons, Chichester, West. Sussex, UK. Munn, C. A. •1988. Macaw biology in Manu National Park, Peru. Parrotletter 1: 18-21. Munn, C. A., J. B. Thomsen, and C . Yamashita. 1990. Audubon Wildlife Report, 1989-1990: 404-419. Nascimento, M. T. 1994. A monodominant rain forest on Maraca Island, Roraima, Brazil: forest structure and dynamics. Ph.D. thesis. University of Stirling, Stirling, Scotland. Nilsson, J. A., and CD. Johnson. 1993. A taxonomic revision of the palm bruchids ( Pachymerini ) and a description of the world genera of Pachymerinae (Coleoptera: Bruchidae) . Memoirs of the American Entomological Society 41: 1-104,

PAGE 294

284 Norconk, M. A., B. W. Grafton, and N. L. Conklin-Brittain . 1998. Seed dispersal by Neotropical seed predators. American Journal of Primatology 45: 103-126. , — • Nunes, A. P. 1992. Uso do habitat, comportamento alimentar organizagao social de Ateles belzebuth belzebuth (Primates: Cebidae) . MS thesis, Universidade Federal do Para, Belem, Para, Brazil . Nunes, A. P. 1995. Foraging and ranging patterns in whitebellied spider monkeys. Folia Primatologica 65: 85-99. Olmos, F. 1993. Diet of sympatric Brazilian caatinga peccaries ( Tavassu taiacu and T. pecari ) . Journal of Tropical Ecology 9: 255-258. .' ' Oppenheimer, J. R. 1968. Behavior and ecology of the white faced monkey, Cebus capucinus . on Barro Colorado Island, C.Z Ph.D. dissertation. University of Illinois, Urbana, Illinois USA ... ... , ; Oppenheimer, J. R. 1996. Cebus capucinus : Home range, population dynamics, and interspecific relationships. Pp. 253-289 in E. G. Leigh, Jr., A. S. Rand, and D. M. Windsor, eds. The ecology of a tropical forest: seasonal rhythms and long term changes. Second edition. Smithsonian Institution Press, Washington, DC, USA. Oyama, K. 1991. Seed predation by a curculionid beetle on the dioecious palm Chamaedorea teoei ilote . Principes 35156-160. Peres, C. A. 1993a. Diet and feeding ecology of the saddle back (Saquinus fuscicoll i s^ and moustached (S. mvstax ) tamarins in an Amazonian terra firme forest. Journal of Zoology, London 230: 567-592. Peres, C. A. 1993b. Notes on the ecology of buffy saki monkeys (Pithecia albicans, Gray 1869): a canopy seed predator. American Journal of Primatology 31: 129-140. Peres, C. A. 1994a. Primate responses to phenological changes in an Amazonian terra firme forest. Biotropica 26: 98-112 . Peres, C. A. 1994b. Composition, density and fruiting phenology of arborescent palms in an Amazonian terra firme forest. Biotropica 26: 285-294. Peres, C, and C . Balder. 1997. Seed dispersal, spatial distribution and population structure of Brazilnut trees

PAGE 295

285 (Bertholletia excelsa ) in southeastern Amazonia. Journal of Tropical Ecology 13: 595-616. Peres, C, L. C. Schiesari, and C. L. Dias-Leme. 1997. Vertebrate predation of Brazil-nuts ( Bertholletia excelsa . Lecythidaceae) , an agouti-dispersed Amazonian seed crop: a test of the escape hypothesis. Journal of Tropical Ecology 13:69-79. Pielou, E. C. 1977. Mathematical Ecology. John Wiley and Sons, New York, New York, USA. Pijl, L. van der. 1982. Principles of dispersal in higher plants, 3rd ed. Springer-Verlag , Berlin, Germany. Pinero, D. , and J. Sarukhan. 1982. Reproductive behaviour and its individual variability in a tropical palm, Astrocarvum mexicanum . Journal of Ecology 70: 461-472. Pinero, D. , M. Martinez-Ramos, and J. Sarukhan. 1984. A population model of Astrocarvum mexicanum and a sensitivity analysis of its finite rate of increase. Journal of Ecologv 72: 977-991. Piper, J. 1986. Effects of habitat and size of fruit display on removal of Smilacina stellata (Liliaceae) . Canadian Journal of Botany. 64: 1050-1054. Power, M. E., D. Tilman, J. A. Estes, B. A. Menge, W. J. Bond., L. S. Mills, G. Daily, J. C. Castilla, J. Lubchenco, and R. T. Paine. 1996. Challenges in the quest for keystones. Bioscience 4: 609-619. Prada Villalobos, M. 1994. Guilda de frugivoros associada com o buriti (Mauritia f lexuosa : Palmae) numa vereda no Brasil central. Masters thesis. Universidade de Brasilia, Brasilia, DF, Brasil. Pritchard, G. 1969. The ecology of a natural population of Queensland fruit fly, Dacus trvoni . II. The distribution of eggs and its relation to behavior. Australian Journal of Zoology 17: 293-311. Proctor, J., and R. P. Miller. 1998. Human occupation on the Ilha de Maraca : preliminary notes. Pp. 431-442 in VI. Milliken and J. Ratter, eds . Maraca: ecology and diversity of an Amazonian rainforest. John Wiley and Sons, Chichester West Sussex, UK. Quicke, D. L. J., and A. Delobel . 1995. A new Neotropical Braconme (Hym. , Braconidae) parasitic on Bruchidae (Col ) Entomologist's Monthly Magazine 131: 215-221.

PAGE 296

286 Quiroga, D. 1997. Ecologia y dispersion de semillas de Attalea phalerata en el Beni de Bolivia. Paper presented at the III International Conference on Wildlife Management and Conservation in Amazonia, Santa Cruz, Bolivia. Redford, K. H. , G. A. Bouchardet da Fonseca, and T. E. Lacher, Jr. 1984. The relationship between frugivory and insectivory in primates. Primates 25: 433-440. Robinson, J. G. 1986. Seasonal variation in use of time and space by the wedge-capped capuchin monkey, Cebus olivaceus : Implications for foraging theory. Smithsonian Contributions to Zoology No. 431. Smithsonian Institution Press, Washington, DC, USA. Rodgerson, L. 1998. Mechanical defense in seeds adapted for ant dispersal. Ecology 79: 1669-1677. Roosmalen, M. G. M. van. 1980. Habitat preferences, diet, feeding strategy and social organization of the black spider monkey (Ateles paniscus ) in Surinam. Ph.D. dissertation. University of Wageningen, Netherlands. Roosmalen, M. G. M. van. 1985. Fruits of the Guianan Flora. Institute of Systematic Botany, Utrecht University, Netherlands . Sallabanks, R. , and S. P. Courtney. 1992. Frugivory, seed predation, and insect-vertebrate interactions. Annual Review of Entomology 37: 377-400. .; ,. SAS Institute. 1989. JMP. SAS Institute, Gary, North Carolina, USA. Schaik, C. P. van., J. W. Terborgh, and S. J. Wright. 1993. The phenology of tropical forests: adaptive significance and consequences for primary consumers. Annual Review of Ecology and Systematics 24: 353-377. Scott, J. K., and R. Black. 1981. Selective predation by white-tailed black cockatoos on fruit of Banksia attenuata containing the seed-eating weevil Alohitopis nivea . Australian Wildlife Research 8: 421-430. Sherman, P. T., and P. K. Eason. 1998. Size determinants in territories with inflexible boundaries: manipulation experiments on white-winged trumpeter's territories. Ecoloov 79: 1147-1159. Sist, P. 1989. Peuplement et phenologie des palmiers en foret guyanaise (Piste de Saint Elie) . Revue d'Ecologie (Terre Vie) 44: 113-151.

PAGE 297

287 Smythe, N. 1978. The natural history of the Central American agouti ( Dasvprocta punctata ) . Smithsonian Contributions to Zoology No. 157 . , , Smythe, N. 1986. Competition and resource partitioning in the guild of Neotropical terrestrial mammals. Annual Review of Ecology and Systematics 17: 169-188. Smythe, N. 1989. Seed survival in the palm Astrocaryum standlevanum : evidence for dependence upon its seed dispersers. Biotropica 21: 50-56. Smythe, N. , W. E. Glanz, and E. G. Leigh, Jr. 1996. Population regulation in some terrestrial f rugivores . Pp. 227-238 in E. G. Leigh, Jr., A. S. Rand, and D. M. Windsor, eds. The ecology of a tropical forest: seasonal rhythms and long-term changes. Second edition. Smithsonian Institution, Washington, DC, USA. Spironelo, W. R. 1991. Importancia dos frutos de palmeiras (Palmae) na dieta de um grupo de Cebus aoella (Cebidae, Primates) na Amazonia central. Pp. 285-296 in A. B. Rylands and A. T. Bernardes, eds. A primatologia no Brasil-3. Fundagao Biodiversitas , Belo Horizonte, Minas Gerais, Brazil. Steege, H. ter. 1993. Patterns in tropical rain forest in Guyana. Trobenbos Series No. 3. Stichting Tropenbos, Utrecht, Netherlands. Steele, M. A., L. Z. Hadj-Chikh, and J. Hazeltine. 1996. Caching and feeding decisions by Sciurus carolinensis : responses to weevil-infested acorns. Journal of Mammalogy 77: 305-314. Stevenson, P. R. , M. J. Quihones, and J. A. Ahumada. 1998. Annual variation in fruiting pattern using two different methods in a lowland tropical forest, Tinigua National Park, Colombia. Biotropica 30: 129-134. Struhsaker, T. T. and L. Leland. 1977. Palm-nut smashing by Cebus a. apella in Colombia. Biotropica 9: 124-126. Symington, M. McFarland. 1987. Ecological and social correlates of party size in the black spider monkey, Ateles paniggug gh^pigk. Ph.D. dissertation, Princeton University, Princeton, New Jersey, USA. Terborgh, J. 1983. Five New World primates. Princeton University Press, Princeton, New Jersey, USA. Terborgh, J. 1986a. Community aspects of frugivory in tropical forests. Pp. 371-384 in A. Estrada and T. H.

PAGE 298

288 Fleming, eds . Frugivores and seed dispersal. Dr W. Junk Publishers, Dordrecht, Netherlands. Terborgh, J. 1986b. Keystone plant resources in the tropical forest. Pp. 330-344 in M. E. Soule, ed. Conservation biology. The science of scarcity and diversity. Sinauer Associates, Inc., Sunderland, Massachusetts, USA. Terbogh, J., S. K. Robinson, T. A. Parker, C. A. Munn, and N. Pierpont. 1990. Structure and organization of an Amazonian forest bird community. Ecological Monographs 60: 213-238. Terborgh, J., E. Losos, M. P. Riley & M. Bolanos Riley. 1993. Predation by vertebrates and invertebrates on the seeds of five canopy tree species of an Amazonian forest. Vegetatio 107/108: 375-386. Thompson, J., J. Proctor, V. Viana, W. Milliken, J. A. Ratter, and D. A. Scott. 1992. Ecological studies on a lowland evergreen rain forest on Maraca Island, Roraima, Brazil. I. Physical environment, forest structure and leaf chemistry. Journal of Ecology 80: 689-703. Thorington Jr., R. W. , B. Tannenbaum, A. Tarak, and R. Rudran. 1996. Distribution of trees on Barro Colorado Island: a five hectare sample. Pp. 83-94 in E. G. Leigh, Jr., A. S. Rand, and D. M. Windsor, eds. The ecology of a tropical forest: seasonal rhythms and long-term changes. Second edition. Smithsonian Institution, Washington, D. C. Tomlinson, P. B. 1990. The structural biology of palms. Clarendon Press, New York, New York, USA. Traveset, A. 1992. Effect of vertebrate frugivores on bruchid beetles that prey on Acacia farnesiana seeds. Oikos 63: 200-206. Tuomisto, H., and K. Ruokolainen. 1993. Distribution of Pteridophyta -and Melastomataceae along an edaphic gradient in an Amazonian rain forest. Journal of Vegetation Science 425-34. Tuomisto, H., K. Ruokolainen, R. Kalliola, A. Lina, W. Danjoy, and Z. Rodriguez. 1995. Dissecting Amazonian biodiversity. Science 269: 63-66. Uhl, N. W., and J. Dransfield. 1987. Genera Palmarum. Allen Press, Lawrence, Kansas, USA. Valburg, L. K. 1992a. Eating infested fruits: interactions in a plant-disperser-pest triad. Oikos 65: 25-28.

PAGE 299

289 Valburg, L. K. 1992b. Feeding preferences of common bushtanagers for insect-infested fruits: avoidance or attraction. Oikos 65: 29-33. .. .. , Vander Wall, S. B. 1990. Food hoarding in animals. University of Chicago Press. Chicago, Illinois, USA. '^-^^ Vandermeer, J. H. 1983. Pejibayepalm (pejibaye). Pp. 98100 in D. H. Janzen, ed. Costa Rican natural history. University of Chicago Press, Chicago, Illinois, USA. Vandermeer, J. H., J. Stout, and S. Risch. 1979. Seed dispersal of a common Costa Rican rain forest palm ( Welf ia aeoraii ) . Tropical Ecology 20: 17-26. Wenny, D. G. 1998. Seed dispersal and post-dispersal seed fate of four tree species in a Neotropical cloud forest. Ph.D. dissertation. University of Florida, Gainesville, Florida, USA. Wiens, J. A. 1995. Landscape mosaics and ecological theory. Pp. 1-26 in L. Hansson, L. Fahrig, and G. Merriam, eds . Mosaic landscapes and ecological processes. Chapman and Hall, London, UK. Wilson, D. E., and D. H. Janzen. 1972. Predation on Scheelea palm seeds by bruchid beetles: seed density and distance from the parent palm. Ecology 53: 954 -959. Wright, S. J. 1983. The dispersion of eggs by a bruchid beetle among Scheelea palm seeds and the effect of distance to the parent palm. Ecology 64:1116-1021. Wright, S. J. 1990. Cumulative satiation of seed predator over the fruiting season of its host. Oikos 58: 272-276. Wright, S. J., M. E. Gompper, and B. DeLeon. 1994. Are large predators keystone species in Neotropical forests? The evidence from Barro Colorado Island. Oikos 71: 279-294. Zhang, S., and L. Wang. 1995. Comparison of three fruit census methods in French Guiana. Journal of Tropical Ecoloav 11: 281-294. f Zona, S., and A. Henderson. 1989. A review of animal -mediated seed dispersal of Palms. Selbyana 11: 6-21.

PAGE 300

BIOGRAPHICAL SKETCH Kirsten Silvius was born in Judibana, Venezuela, to German and Dutch parents. She moved to the United States with her family in 1981 and completed her high school and college education in Maine. After graduating with a B.S. degree in biology and Romance languages from Bowdoin College, she worked as a research assistant on several field biology and conservation projects in Venezuela and the Bahamas. In 1989 she developed and obtained funding for a project on the ecology and conservation of the endangered yellow-shouldered amazon parrot in Venezuela. She started graduate studies at the University of Florida partway through that project, and obtained her master's degree from the Zoology Department in 1992. She was accepted into the Ph.D. program at UF that same year, and in 1995 went to the Brazilian Amazon to carry out research for her doctoral degree. She lived and worked m Brazil for two years and hopes to return to that wonderful country soon after completing her Ph.D. 290

PAGE 301

I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. Jo&xF . Eisenberg, Chair/ KatharT-ne Ordway pf of S"s«or of Ecosystem Conservation I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. Crawford S. Hollin£ Arthur Marshall Jr. Professor in Ecological Sciences I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. Colin A. Chapman Associate Professor of Zoology I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. Walter S. Professor Judd /'"I of Botany

PAGE 302

I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. This dissertation was submitted to the Graduate Faculty of the Department of Zoology in the College of Liberal Arts and Sciences and to the Graduate School and was accepted as partial fulfillment of the requirements for the degree of Doctor of Philosophy. May 1999 ' ; . • ; f\ Richard E. Bodmer Assistant Professor of Wildlife Ecology and Conservation Dean, Graduate School


xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID E99NFQD0Z_A2OXS3 INGEST_TIME 2015-04-16T22:08:28Z PACKAGE AA00030023_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES