The Effects of Cattle Ranching on a Primate Community in the Central

Abstract

According to the International Union for Conservation of Nature and Natural Resources

(IUCN), 64% of primate species are declining in the wild and 49% face a significant risk of extinction. This crisis is largely the result of human activity, including logging, ranching, and hunting. In this study I examine the impacts of the anthropogenic habitat disturbances associated with cattle ranching on a primate community.

Research was conducted at the preeminent site for the study of rain forest fragmentation, the Biological Dynamics of Forest Fragments Project, Manaus, Brazil. I surveyed 8 line transects totaling 13 km that sampled: 1) secondary forest on abandoned cattle pasture,2) selectively logged primary forest, and 3) undisturbed primary forest. Specifically, I tested for edge effects, niche partitioning, and interhabitat differences in population density.

Primate presence in edge habitats was negatively related to the amount of fruit consumed, a relationship that was also apparent in the densities of individual species.

Four species were more abundant in edge habitats: Alouatta macconnelli (folivore- frugivore), Chiropotes chiropotes (seed predator), Saguinus midas (generalist), and

Sapajus apella (generalist); one was less abundant: Ateles paniscus (frugivore); and the last showed no edge-related pattern: Pithecia chrysocephala (seed predator).

Niche partitioning was evident in diet and macrohabitat use. In addition, in primary forest there was partitioning along several microhabitat variables. In secondary forest, however, microhabitat partitioning was absent, possibly due to habitat constraints or low encounter rates. Body size was positively related to use of vertical strata in both habitats, hence a combination of body size and competition may drive vertical niche partitioning.

Primate characteristics were not related to their presence in selectively logged or undisturbed primary forest, though body size was inversely related to presence in secondary forest. Only the two generalists (Saguinus and Sapajus) heavily utilized secondary forest. The two largest species, Ateles (frugivore) and Alouatta (folivore- frugivore), showed an equal preference for all primary forest (logged and undisturbed) over secondary forest. Chiropotes (seed predator) also preferred undisturbed primary forest while Pithecia (seed predator) was relatively uniformly distributed.

Acknowledgments

When a Ph.D. dissertation is published, the name of a single author is placed on the title page. This action belies the multitude of people who were crucial in helping the author reach his or her goals, and I am certainly no different. First and foremost I would like to say a heartfelt thank you to my lovely wife, Amaliya, for staying strong, dealing with my being away for so long, and for listening when I needed a friend. I also need to say thank you to my little guy Vaughn for always being ready to cheer me up and for putting up with an exhausted, dissertation-writing dad for the first year of his life. I would also like to thank my parents, Richard and Janet Lenz, my brother and sister-in-law, Tom and

Bridget Lenz, and my aunt, Jane Lenz, for taking an interest in my interests - without your support this would not have been possible.

I am especially appreciative for the guidance of my advisor, Katharine Jack, who taught me how to forage for myself and who has been a great friend in addition to being a great mentor. I would also like to thank the rest of the faculty, in particular my committee members Trenton Holliday and William Balée, and the graduate students in the Department of Anthropology for all you have taught me and for being my partners in crime in the Big Easy (you know who you are). Special thanks go to Valerie Schoof for her friendship and willingness to talk shop.

I am forever indebted to Sarah Boyle for her assistance. Even though we were strangers when I decided that I was going to work at the Biological Dynamics of Forest

Fragments Project, she took time out of her busy schedule to help me find my way in

Brazil. Your assistance throughout this study was immensely valuable to me and I thank you for it.

I also owe a great deal to the people of Brazil. They embraced me with open arms, taught me an incredible amount, and were almost universally kind and generous. I will be forever grateful for the people who helped me collect my data, in particular my friends Alaercio (Léo) Marajó dos Reis and Luciana Frazão, who put up with long hours of relentless work while always maintaining good spirits. I would also like to thank the others that helped me in the field: Paulo Apóstolo de Assunção, Cicero Lopes da Silva,

Fabiana Galland, Jairo Miranda Lopes, Luiz Raimundo de Queiroz, and José Francisco

Tenaçol Andres Junior.

I would like to thank my Brazilian counterpart on this project, Wilson Spironello, and my colleagues at the Biological Dynamics of Forest Fragments Project, José Luís

Camargo, Regina Luizão, Gonçalo Ferraz, and Ana Andrade, for their advice and friendship. My thanks also go to Ary Jorge Ferreira and Rosely Hipólito for their patience and excellent work organizing all of the entradas and saídas, to the motoristas, Josimar

Facundes Menezes, Antonio José Moraes Pereira, Luiz Raimundo de Queiroz, and Paulo

Lopes Pereira, for getting me safely (and often harrowingly) from the city to the campsites and back, and to Adriane Gomes and Cleucilene Nery for helping me with a

iii

wide array of issues. Thank you also to Dona Deuza Oliveira and her family for taking me into their home and for being my surrogate family in Manaus.

I also greatly appreciate the many doctors who helped me overcome the cascade of maladies that befell me while I was in Brazil and made me wonder if I should amend my project to include data collection on tropical diseases: Susan Mclellan, Bridget Lenz,

Tony Lenz, and especially meu amigo Jorge Guerra who made my life in Manaus at least

50% easier than it would have been without his friendship.

Finally, I would like to thank the groups and individuals who provided the financial support that made this project possible: Primate Conservation, Inc., The Stone

Center for Latin American Studies, The Middle American Research Institute, The Tulane

University Department of Anthropology Graduate Student Aid Fund, The Tulane

University School of Liberal Arts Summer Merit Fellowship, and a big thank you to

Richard and Janet Lenz, Jane Lenz, and Tom and Bridget Lenz.

Table of Contents

Acknowledgments...... ii Table of Contents ...... v List of Tables ...... vii List of Figures ...... ix Chapter 1: Introduction ...... 1 Chapter 2: The History, Forests, and Primates of the BDFFP ...... 11 The History of the BDFFP ...... 11 The Forests of the BDFFP ...... 18 The Primate Community ...... 24 Chapter 3: Methodology ...... 40 Data Collection ...... 40 Primate Population Density ...... 45 Sampling the Floral Community ...... 55 Chapter 4: Edge Effects in a Community of Neotropical Primates ...... 63 Introduction...... 63 Methods ...... 72 Results ...... 83 Discussion ...... 90 Chapter 5: Niche Partitioning in a Community of Neotropical Primates in an Anthropogenic Mosaic Landscape ...... 101 Introduction...... 101 Methods ...... 105 Results ...... 114 Discussion ...... 123

Chapter 6: Primate Abundance in an Anthropogenic Mosaic Landscape in the Central Amazon ...... 132 Introduction...... 132 Methods ...... 137 Results ...... 150 Discussion ...... 156 Chapter 7: Conclusions ...... 167 Appendix ...... 178 List of References ...... 197

List of Tables

Table 2.1: Population growth in Manaus (Butler, 2013) ...... 14 Table 2.2: Characteristics of the BDFFP primates (24) ...... 27 Table 2.3: The percentages of adult trees and tree species for which I estimated reproductive phenology ...... 35 Table 2.4: Annual fruit availability by habitat ...... 35 Table 3.1: Transect information (each transect sampled two habitats) ...... 42 Table 3.2: Total area sampled by habitat ...... 43 Table 4.1: Examples of abiotic & floral edge effects in tropical forests ...... 66 Table 4.2: Characteristics of the BDFFP primates (Boyle & Smith, 2010b) ...... 69 Table 4.3: Density differences (# individuals km2-1) by edge zone (, ) ...... 85 Table 4.4: Predator encounters/evidence at the BDFFP ...... 90 Table 5.1: Characteristics of the BDFFP primates (Boyle & Smith, 2010b) ...... 107 Table 5.2: Transect information...... 109 Table 5.3: Total area sampled by habitat ...... 109 Table 5.4: Intraspecies macrohabitat use patterns ...... 116 Table 5.5: Differences in microhabitat distributions in mature forest (K-S test) (e.g. Ateles vs. Alouatta, distance to stream: Ateles (A) > Alouatta (B)) (p<0.05)* ...... 117 Table 5.6: Microhabitat use descriptive statistics ( ())* ...... 119 Table 6.1: Transect information...... 140 Table 6.2: Total area sampled by habitat ...... 140 Table 6.3: Summary of statistical tests ...... 147 Table 6.4: Characteristics of the BDFFP primates (Boyle & Smith, 2010b) ...... 148

vii

Table 6.5: The percentages of adult trees and tree species for which I estimated reproductive phenology ...... 151

Table 6.6: Primate density by habitat (# individuals/km2()) ...... 153

viii

List of Figures

Figure 2.1: The Biological Dynamics of Forest Fragments Project (Key: dark green=primary forest; light green=secondary forest or active pasture; black grids=forest control areas; red lines= a) the north-south line is the paved highway BR 174 and b) the east-west lines are dirt access roads) ...... 12 Figure 2.2: The Manaus region in 2001. Note that Manaus has grown to surround the 100 km2 Reserva Ducke. The Rios Negro and Solimões meet at Manaus and are referred to as the Amazon River east of this point (Key: pink=city, active pasture, or bare ground; dark green=primary forest; light green=young secondary forest) ...... 14 Figure 2.3: Land use around the BDFFP. Note the forest conversion in the NW and SW. (Key: pink=active pasture or bare ground; dark green=primary forest; green=riparian vegetation; light green=young secondary forest) ...... 15 Figure 2.4: Fruit availability by habitat with (A) and without (B) Cecropia sciadophylla (Alouatta, Ateles) ...... 37 Figure 2.5: Fruit availability by habitat with (A) and without (B) Cecropia sciadophylla (Saguinus, Sapajus) ...... 38 Figure 2.6: Fruit availability by habitat for Chiropotes and Pithecia and overall fruit availability with (A) and without (B) Cecropia sciadophylla ...... 39 Figure 3.1: Transect Locations ...... 41 Figure 3.2: Illustration of WDFP (Whitesides, et al., 1988) ...... 52 Figure 3.3: 50% fall-off distance (Whitesides, et al., 1988) ...... 53 Figure 4.1: Transect locations ...... 74 Figure 4.2: Types of edge responses (Reis, et al., 2003) ...... 75 Figure 5.1: Transect locations ...... 108

Figure 5.2: Vertical distribution by species in mature forest ...... 118 Figure 5.3: The relationship between body weight and use of vertical strata in mature and secondary forests ...... 123 Figure 6.1: Transect locations ...... 140

Chapter 1: Introduction

According to the International Union for Conservation of Nature and Natural Resources

(IUCN), which has been assessing the global conservation status of animal and plant taxa for over 40 years, 64% of the world’s primate species are declining in the wild. A total of

49% of species are either critically endangered (10%), endangered (20%), or vulnerable to extinction (19%), and we lack sufficient data to even assess population trends for a further 26% (IUCN, 2012). This conservation crisis is largely the result of non-mutually exclusive human activities including, but not limited to, agriculture, logging, and hunting.

Given the global nature of today's society, the amount of habitat conversion that takes place, and the omnipresent pollution that humans produce, it is safe to say that the entirety of the earth's surface has been influenced by anthropogenic activity (Balée,

1998). When considering a system like the Amazon Rain Forest, the most visible disturbances today are those that are large-scale and clearly evident on the ground and in satellite imagery (e.g. clearcut logging, intensive agriculture, and hydroelectric dams).

It is commonly believed that the dark green portions of those same satellite images hold expanses of pristine, untouched Amazon Rain Forest. This belief, however, is inaccurate

- for both the present and the past. In the present day, many of these "pristine" forests

2 are subject to anthropogenic extraction, for example hunting, selective logging, or the collection of natural forest products, that leaves much of the canopy intact and from a distance gives the incorrect impression of untouched wilderness. In the past, the activities of formerly populous indigenous groups impacted significant areas of the

Amazon and certainly influenced the current composition of many Amazonian forests

(Heckenberger, et al., 2007). In fact, there is archaeological evidence of these indigenous groups along the banks of the Rio Negro near Manaus, for example Açutuba

(Heckenberger, et al., 1999; Neves, et al., 2004), a site from which the present study is

80 km inland. As many of these Amazonian population centers were along the floodplains of major rivers (Heckenberger, et al., 2007), this distance decreases the chances of a significant past indigenous presence at the study site, the Biological

Dynamics of Forest Fragments Project. While charcoal is common in soil samples from the BDFFP (Bassini & Becker, 1990), there is little additional evidence of past human presence and the charcoal may therefore be the result of fires during recurrent El Niño- driven ultra-dry periods over the last 1,000 years (Gascon & Bierregaard Jr., 2001).

In the present study, while I acknowledge the important role that indigenous populations have played in the accumulation of the "pristine" and "undisturbed" reaches of some areas of today's Amazon Rain Forest, not to mention their current role in its protection (e.g. Nepstad, et al., 2006; Schwartzman, et al., 2000), I do not examine whether or not archaeological evidence suggests that indigenous people, especially in pre-Columbian times, had an impact on the distribution of the primate community at the study site. Instead, I consider all forests that have been free from significant

3 anthropogenic disturbance over the last several hundred years to be what I refer to hereafter as undisturbed primary forest, which is tall terra firme humid rain forest. This term is similarly applied by the other researchers that I have cited throughout this work and is not an attempt to marginalize the influences that past anthropogenic activity on today's forests. My primary goal in this dissertation is to determine the impacts on primates of one of the significant present-day anthropogenic activities that is responsible for the current primate conservation crisis: cattle ranches that are the result of market-driven forest conversion that is divorced from concern for long-term ecosystem health and is instead focused on the maximization of short-term profit. This type of land use often results in anthropogenic mosaic landscapes, which are landscapes that contain multiple habitats that experience, or have experienced, varying levels of anthropogenic disturbance. In the case of cattle ranching these habitats can include: undisturbed primary forest, selectively logged primary forest, isolated fragments of primary forest, secondary forest regenerating on abandoned pasture, riparian corridors that cross active and abandoned pastures, and active pasture itself. Similarly, I use the term secondary forest to describe the forests that regenerate naturally on abandoned agricultural clearcuts and the term selectively logged forest to refer to tall terra firme forest that has been recently disturbed by the removal of a portion of its timber.

In comparisons between these three habitats, I consider undisturbed primary forest as my baseline and examine primate behavior and ecology in the disturbed habitats (secondary and selectively logged forests) relative to the undisturbed primary forest. Underlying this framework are two assumptions. First, I assume that interhabitat

4 differences in primate behavior and ecology are the result of habitat disturbance rather than due to pre-existing differences between the areas sampled. This assumption is somewhat problematic given the spatial heterogeneity of tropical forests (Grieser Johns,

2001). However, this type of comparison is standard practice due to the amount of time and rigorous sampling necessary for a researcher to extensively sample an area before its disturbance as well as to repeat surveys in those exact locations over the decades that follow.

Second, I assume that forest succession will occur over time as clearcut pastures regenerate from bare ground to secondary forest to a climax ecosystem (Odum, 1969), in this case old-growth/mature/primary forest. Disturbance and succession are fundamental features of ecosystems, and whether those disturbances are anthropogenic or not they are essential to an ecosystem's long-term stability, function, and complexity (Kimmins, 2004). I assume that disturbed forests will regenerate to be identical to neither their pre-disturbance state nor to the neighboring undisturbed primary forests that I use as a baseline. Indeed, I am comfortable saying that there is almost no chance of this happening. Though it was the prevailing view among ecologists until the 1970s (Sprugel, 1991), I do not assume that there is a successional pathway leading to a particular climax state to which all forests in this area will return. The constant barrage of a wide range of anthropogenic and non-anthropogenic disturbances, in concert with temporal climatic and nutrient variability, result in ecosystems that are ever-changing in response to disturbance regimes that themselves change over time. This does not, however, mean that forest succession does not take

5 place following habitat disturbance, it simply means that there is not a predetermined climax state towards which that succession proceeds. It also does not imply that the study of the impacts of these disturbances on individual species should not be undertaken. In fact, studies of the ways in which biota respond to the types of anthropogenically disturbed habitats that I sample are growing in importance as the area that is impacted by these severe anthropogenic disturbances continues to grow rapidly (see below). Indeed, studies such as this are important if there is to be any hope of designing land management plans that will protect primate populations while also incorporating the anthropogenic and non-anthropogenic disturbances that have always1 impacted Amazonian forests.

Following these definitions, it is not difficult to detect disturbing trends in the way that humans utilize undisturbed primary forests. Habitat destruction and degradation, and the anthropogenic mosaic landscapes that they create, are pantropical phenomena with 50% of tropical forests now consisting of degraded old-growth forest and areas of secondary forest regenerating after clearcuts (ITTO, 2002). In the Amazon

Basin, selective logging (1.2-2.0 million ha yr-1, Asner, et al., 2005) and secondary forest regenerating on abandoned clearcuts (currently 30-37% of deforested area and projected to reach 63%, Fearnside & Guimarães, 1996) are widespread. Over recent decades, the extent of primary forest in the region where this study was conducted, the

Manaus region of the central Amazon, has dropped from 91% to 77% of the total area

1 The earliest human occupation in Amazonia is dated to 11,000-10,000 BP (Roosevelt, et al., 1996; Roosevelt, et al., 2002). The raising of crops may have begun by 4,000-5,000 BP but it does not appear to have been widespread until 500-1000 AD (Neves, et al., 2003).

(Prates-Clark, et al., 2009) while 75-80% of the deforested land is now occupied by secondary forest (Fearnside, 2006; Prates-Clark, et al., 2009). Between 1980 and 2000 alone, South American cattle ranching expanded by 35 million ha, resulting in the clearance of ~19 million ha of mature forest and ~5 million ha of disturbed forest (Gibbs, et al., 2010).

Studies in anthropogenic mosaic landscapes like those that result from ranching operations are crucial for the future survival and management of primate populations because, as the trends discussed above illustrate, it is likely that anthropogenically degraded habitats will comprise the majority of available primate habitat in the not-too- distant future. Indeed, the forests of many regions, like Brazil's Atlantic Coastal Forest

(Ribeiro, et al., 2009), have been almost entirely cleared while the remnant areas consist of highly disturbed forest. As a result, studies of the ways that primates respond to these disturbances will allow conservationists to better plan reserves and create more effective reforestation and corridor projects.

While anthropogenic disturbance often reduces primate abundance and/or the diversity of primate communities, with proper management the negative impacts of disturbance may be reduced because many disturbed habitats are not entirely inhospitable to primates. For example, cautiously managed selective logging, as long as it is not accompanied by hunting, is an economic use of forests that may permit the survival of complete, or nearly complete, primate communities (Skorupa, 1986), though logging intensity and methodology also greatly influence primate persistence (Johns &

Skorupa, 1987). On the other hand, the completely deforested fields necessary for ranching and most farming are much less viable as primate habitat than are selectively logged primary forests. However, abandoned agricultural areas do return to secondary forest. The ability of primates to use these forests is often related to both the extent of primary forest in the immediate area and to the age of the secondary growth. While these regenerating secondary forests are unsuitable habitat for some species and cannot fully support others, they also frequently supplement primary forest primate habitats, serve as dispersal corridors between mature forests in anthropogenic mosaic landscapes (Fimbel, 1994; Sorensen & Fedigan, 2000), and they even appear to be preferred habitats for some disturbance-adapted primates (Parry, et al., 2007; Vulinec, et al., 2006).

In this dissertation, I examine primate use of the anthropogenic habitats that result from cattle ranching on each of the members of a Neotropical primate community. Research was conducted at a tropical rain forest cattle ranching site, the

Biological Dynamics of Forest Fragments Project (BDFFP), which is located 80 km north of Manaus, Brazil in the central Amazon (2°25´S, 59° 50´ W) (Didham, 1997). More specifically, I examine primate distribution and ecology in secondary forests regenerating on abandoned cattle pastures, selectively logged primary forest, and undisturbed primary forest. The BDFFP's primate community is diverse, which allows for an examination of the impacts of anthropogenic disturbances on a wide range of primate evolutionary strategies. The community consists of a large-bodied specialist frugivore (Ateles paniscus), a large-bodied folivore-frugivore (Alouatta macconnelli), two

8 small- to medium-sized seed predators (Chiropotes chiropotes and Pithecia chrysocephala), and two species with generalist diets, one that is small (Saguinus midas) and one that is medium-sized (Sapajus apella apella).

The BDFFP is one of the preeminent sites in the world for research on habitat fragmentation and the impacts of cattle ranching on tropical forest biota. The project's initial intent was to experimentally determine the size of the smallest isolated primary forest habitat fragment, defined as a remnant patch of primary forest that is completely enclosed by clearcuts, that would support the majority of the species found in undisturbed continuous primary forest (Bierregaard Jr. & Gascon, 2001; Gascon &

Bierregaard Jr., 2001). Habitat fragments in three size classes (1, 10, and 100 ha) were completely isolated from undisturbed primary forest by large expanses of cattle pasture in the early 1980s, many of which have since been abandoned and now feature the secondary forests that I sampled. Today the ranches remain surrounded by large expanses of primary forest. However, human settlers are becoming more common in the area, resulting in an increase in land clearance and hunting surrounding the site

(Laurance & Luizão, 2007). Overall, the pressure from this human influx remains somewhat low as the animal community in the area is still complete. In other words, it includes the large animals that are usually the first to disappear in the face of extensive land conversion and hunting. For example, during the course of this study I encountered the six primate species that were the focus of my research, jaguars (Panthera onca), pumas (Puma concolor), tapirs (Tapirus terrestris), collared peccaries (Pecari tajacu), and harpy eagles (Harpia harpyja).

The BDFFP is perfect for this study because the history of the disturbances at the site is known, making it unlike most disturbed tropical habitats which have incomplete land use records - if records exist at all. This well-documented history is an invaluable tool that provides researchers at the BDFFP a greater ability to examine cause-and- effect relationships relative to studies conducted at poorly-documented sites where similar causal relationships may be posited though they are confounded by historical uncertainty. In addition, hunting pressure at the BDFFP is low, the site is encompassed by huge tracts of largely undisturbed primary forest, and it contains a wide range of disturbed habitats. These factors work in concert to create a landscape where primates are "free" to choose from a wide range of habitats without significant hunting pressure or the limited availability of undisturbed habitat driving their habitat selection. As a result, primate behavior at the BDFFP should more accurately reflect each species' true preferences as opposed to other disturbed sites where primates frequently have only a limited ability to move between habitats, if there are even multiple habitats from which to choose. Findings from the study site are important because voluntary use of degraded habitats suggests an increased probability of long-term survival in anthropogenic mosaic landscapes while the same cannot be said for species that are forced into degraded areas in the short-term because they have no other choice.

Including the present chapter, this dissertation consists of seven chapters.

Chapter 2 describes the history, forests, and primate community of the BDFFP. Chapter

3 contains an in-depth description of data collection methodology and analyses.

Chapters 4-6 contain the main results of this study. These chapters examine the way

10 that primates respond to a variety of human land uses that are associated with ranching in tropical forests. Chapter 4 deals with edge effects in the distributions of each of the primate species. Chapter 5 focuses on primate niche partitioning in primary and secondary forests. Chapter 6 examines primate density differences across three of the major forest classes: secondary forest, selectively logged primary forest, and undisturbed primary forest. The final chapter is a discussion of my conclusions and their significance for primate conservation.

Chapter 2: The History, Forests, and Primates of the BDFFP

This section contains a detailed history of the study site, a description of the principal habitats that I sampled, and a discussion of each of the six primates species that call the

BDFFP home.

The History of the BDFFP

I conducted field research at the Biological Dynamics of Forest Fragments Project

(BDFFP), which is located 80 km north of Manaus, Brazil in the central Amazon (2°25´S,

59° 50´ W) (Didham, 1997). The BDFFP is spread out from east to west over 3 main fazendas, or ranches, where habitat fragments are present (Dimona, Porto Alegre, and

Esteio) (Figure 2.1), each of which had thousands of hectares clearcut for cattle pasture

(Gascon & Bierregaard Jr., 2001). There are also four continuous forest sites with permanent trail networks that serve as controls (Cabo Frio, Gavião, Km37, and Km41).

At Dimona, Porto Alegre, KM37, and KM41 there is one research camp, while Esteio has four (Cidade e Powell, Cabo Frio, Gavião, and Colosso). Data for the current study were collected at Dimona and Porto Alegre, though I sampled multiple habitats at each of these sites. The age of secondary forest cannot be simply broken down by fazenda due to different areas of each ranch operating at different periods, so as a result I used vegetation surveys to characterize each forest.

While the BDFFP is located 80 km north of the city of Manaus (Figure 2.2), the rapid expansion of the city (Table 2.1) is an ever-growing threat to the forests in this region. Settlers are increasingly moving into the area and clearing forests for agriculture, movements that have been greatly facilitated by the late 1990s paving of the highway that traverses the site, the 1070 km BR 174 from Manaus to Venezuela (Figure 2.3). This influx of people increases hunting pressure in the area (Laurance & Luizão, 2007), an issue that is exacerbated by people who use the paved highway to drive to the dirt access roads (Figure 2.1) to hunt near the BDFFP (B. Lenz, personal observation). It also increases the frequency of runaway fires that escape the settlers' plots and burn the surrounding forest, something that happened along one of the dirt roads that links several of the camps (ZF-3) following an unusually arid dry season in 2009 (average monthly rainfall for August-November, 2009=43 mm; monthly average for August-

November, 2005-2010: 123 mm, (Instituto Nacional de Pesquisas Espaciais, 2012)) (B.

Lenz, personal observation). Fortunately, these anthropogenic stressors have yet to seriously impact the animal community as the large animals that usually disappear first in the face of extensive land conversion and hunting are still present. For example, in addition to primates I also encountered jaguars (Panthera onca), pumas (Puma concolor), tapirs (Tapirus terrestris), collared peccaries (Pecari tajacu), and harpy eagles

(Harpia harpyja).

Figure 2.2: The Manaus region in 2001. Note that Manaus has grown to surround the 100 km2 Reserva Ducke. The Rios Negro and Solimões meet at Manaus and are referred to as the Amazon River east of this point (Key: pink=city, active pasture, or bare ground; dark green=primary forest; light green=young secondary forest)

Table 2.1: Population growth in Manaus (Butler, 2013)

Year Population1 % Change2 1960 173,000 - 1970 309,000 +79 1980 611,000 +253 1990 959,000 +454 2000 1,436,000 +730 2010 1,785,000 +932 1 City limits; the metropolitan area had 2,178,495 inhabitants in 2008 2 Relative to 1960

Figure 2.3: Land use around the BDFFP. Note the forest conversion in the NW and SW. (Key: pink=active pasture or bare ground; dark green=primary forest; green=riparian vegetation; light green=young secondary forest)

The BDFFP is unique among fragmentation research sites in that it is an experimentally fragmented forest: the deforestation was planned and the entire history of the site is known. Through the 1970s, the area was covered in extensive primary forest, much of which remains today. In 1976, Thomas Lovejoy came up with the idea

16 for a project to determine the minimum area of tropical forest necessary to maintain the majority of the biotic diversity found in an intact ecosystem (Lovejoy & Oren, 1981).

In 1979, Lovejoy founded what at the time was known as the Minimum Critical

Size of Ecosystems Project (MCSE), a project that was possible due to a Brazilian law stating that half of each forest concession must be left standing when land was developed (Bierregaard Jr. & Gascon, 2001). Lovejoy found a group of enthusiastic cattle ranchers and together they arranged for the establishment of a series of experimental fragments between 0.1 and 10,000 ha as well as reserves of the same size in the surrounding continuous forests to serve as controls. In the end, fragments larger than

100 ha were not isolated because the government subsidies for ranching in the area ended in the mid-1980s (Bierregaard Jr. & Gascon, 2001). Some of the isolated fragments also ended up being slightly larger than called for in the initial plans, for example 1 ha fragments range from 1.1-2.8 ha (Boyle, 2008).

Baseline natural history data were collected prior to, during, and immediately following the isolation of the fragments by the ranchers’ clearcut of all trees in the areas that were to be pastures. Data were collected on canopy trees, leaf litter insects, butterflies, amphibians, non-volant mammals, and birds, among other things

(Bierregaard Jr. & Stouffer, 1997). Each of the isolated fragments was fenced with barbed wire to keep the cattle out, and as a result it is still possible to see the exact locations of the original fragment edges even where secondary forests have regenerated on abandoned pastures. Fences were not built where much of the present study occurred, at the border between the abandoned pasture and the primary forest

17 that surrounds the site, so fence lines have not been an impediment to animal movements across these borders.

The project's initial intent was to burn all of the pastures after felling the forest.

This plan was successful at two of the ranches, Fazenda Dimona and Fazenda Esteio. An unusually early rainy season in the year that the pastures were cleared at Fazenda Porto

Alegre, however, resulted in the fires being extinguished before much of the downed vegetation was consumed (Gascon & Bierregaard Jr., 2001). As a result, the forests that regenerated on the clearcut areas of the BDFFP have followed two different successional pathways depending on whether the cleared area was burned or not

(Mesquita, et al., 2001). Dimona and Esteio were burned and served as active cattle pastures for varying periods prior to their abandonment (there are still portions of

Esteio that contain active pastures). This more intensive land-use history has resulted in regenerating forests dominated by the pioneer tree genus Vismia (family

Hyperiacaceae) (Mesquita, et al., 2001). However, of my two transects in these areas, only one was dominated by Vismia (see below). At Porto Alegre, approximately half of the area that was cleared in the initial clearcut never burned. By the following dry season, the regrowth was too thick to burn so the majority of this habitat was not subjected to fire; this is the bulk of the secondary forest that I have sampled. The other half of the initial clearcut area at Porto Alegre was successfully converted to cattle pasture, much of which is still active today (I did not sample this area). This resulted in the areas of Porto Alegre that never fully burned having less soil and seed bank damage while also leading to secondary forest tree communities dominated by the pioneer

18 genus Cecropia (family Urticaceae) (Mesquita, et al., 2001). I focus on these unburned secondary forests as they are typical of regenerating forests in the Manaus region

(Prates-Clark, et al., 2009).

One of the problems that plagues most fragmentation studies is that a considerable number of years may go by before primate populations respond numerically to fragmentation, and without knowing the disturbance history researchers cannot know if the focal population has experienced disturbance-related abundance changes (Chapman, et al., 2003). The history of the BDFFP is carefully documented and, importantly, it begins before the site was disturbed. Research at the BDFFP has resulted in over 600 publications and master’s and doctoral theses, many of which address the ways in which flora, fauna, and abiotic conditions are impacted by habitat fragmentation. The site's detailed history and large and varied dataset make the BDFFP unique in its potential to produce meaningful research.

The Forests of the BDFFP

The primary forest is tall terra firme rain forest with a canopy generally between 30-37 m with emergents to 55 m (Bierregaard Jr. & Stouffer, 1997). Annual rainfall at the site ranges from 1,900-2,500 mm (Stouffer & Bierregaard Jr., 1993) and averaged 2194 mm during this study (Instituto Nacional de Pesquisas Espaciais, 2012). Over the last 106 years, the city of Manaus just to the south has received an average of 2115 mm

(Cochrane, 2011). There is a pronounced dry season from June to September (Didham,

1997). Average monthly temperatures range from 25.8⁰C in February-April to 27.9⁰C in

September (Salati, 1985). The Manaus region, and much of the Amazon Basin in general, lack significant topography. There is likewise no significant relief at the site as elevation ranges from 50-150 m (Bruna & Kress, 2002), though the landscape contains numerous steep hills cut by current and past streams (igarapés). The soil is predominantly nutrient-poor yellow latosol (Rylands & Keuroghlian, 1988).

The site has a very diverse tree community. A long-term phytodemographic study has inventoried >60,000 stems with a diameter at breast height (dbh) ≥10 cm in

66 permanent one-hectare plots (Gascon & Bierregaard Jr., 2001) and documented more than 1,300 species and morpho-species (i.e. individuals that are distinct but thus far unreliably identified) (S.G. Laurance, personal communication, cited by Laurance,

2001). The species composition of the tree community of Manaus is unusual because: 1) it is dominated by trees belonging to just a few genera, 2) many important genera from other sites are not well-represented, and 3) some of the genera that are limited elsewhere are extremely diverse near Manaus (Laurance, 2001). The 1,300 taxa represent 61 families and 288 genera, with Leguminosae the most species-rich family

(195 species and 12.1% of all trees). Several other families also make significant contributions: Lecythidaceae (12.4%), Sapotaceae (11.9%), Burseraceae (9.3%),

Euphorbiaceae (5.9%), Chrysobalanaceae (5.3%), Lauraceae (4.3%), Moraceae (4.1%),

Annonaceae (3.5%), and Violaceae (2.5%) (Gascon & Bierregaard Jr., 2001). The alpha diversity tree species richness near Manaus (280-285 species ha-1, Oliveira & Mori, 1999) is actually just as high as it is in the diverse western Amazon that was thought to be the principal center for Neotropical tree diversity (Laurance, 2001).

The BDFFP contains a number of different habitat types, including: undisturbed primary forest, selectively logged primary forest, secondary forest regenerating on abandoned cattle pastures that were or were not burned (leading to different dominant plant communities, Mesquita, et al., 2001), primary forest habitat fragments in the middle of secondary forests, riparian corridors through secondary forest, and active pasture. In this study I focused on, and will now describe in detail, undisturbed primary forest, selectively logged primary forest, and secondary forest regenerating on abandoned pastures.

Undisturbed Primary Forest

In total I sampled 2.002 ha of undisturbed primary forest spread over six transects that contained 1631 trees with a dbh ≥ 10 cm (=21.3, =12.5, range=10.0-125.1). I encountered a total of 51 families (25.5 families ha-1) with the five most prevalent being:

Sapotaceae (15.6%), Lecythidaceae (14.3%), Leguminosae (10.0%), Burseraceae (8.7%), and Chrysobalanaceae (8.3%) (Appendix 1). The 51 families were represented by 165 genera (82.4 genera ha-1) dominated by: Eschweilera (12.0%), Protium (8.1%), Pouteria

(7.5%), Licania (6.2%), and Micrandropsis (3.7%) (Appendix 2). Making up those 51 families were 424 species (211.8 species ha-1) with the most common being: Eschweilera truncata (3.8%), Micrandropsis scleroxylon (3.7%), Eschweilera wachenheimii (2.4%),

Protium hebetatum (2.3%), and Eschweilera coriacea (2.0%) (Appendix 3).

Selectively Logged Primary Forest

Trees along the two transects in selectively logged forest were likely harvested in 1983

(B. Lenz, personal observation), 25 years before the start of this study, when the forest at the Porto Alegre Ranch was cleared (Gascon & Bierregaard Jr., 2001). I use the word

"likely" because I am the first to sample this area and none of the BDFFP's founders or early researchers recall selective logging, though it clearly occurred given the stumps I encountered. It is very likely that logging occurred during the clearance of the ranch, a common practice when clearing pastures in the region, because the area was largely undisturbed prior to that time. Though I did not survey this habitat beyond my transects, I believe it is likely that the logged area does not cover significantly more than

250 ha. I recorded the impacts of logging to a depth of 10 m on each side of my transects for a total area sampled of 4.18 ha, noting the number of stumps (N=2, 0.48 stumps/ha) and the number of abandoned logging paths (~10 m wide) with relatively open canopy above them that crossed the transects (N=5, coverage=0.1 ha or 2.4% of the area sampled). According to tree expert Paulo Apóstolo Assunção (da S. Ribeiro, et al., 1999), logged trees were most likely Manilkara huberi, part of the diet of at least four of the study species: Alouatta macconnelli, Ateles paniscus, Chiropotes chiropotes, and Sapajus apella apella (Boyle, 2008; Fragazy, et al., 2004; Neves & Rylands, 1991;

Simmen & Sabatier, 1996). It is possible that this species was cut either for its timber or for its latex as there are unconfirmed reports of the latter in the 1970s at the Floresta

Nacional do Tapajós (Michael Keller, personal communication).

I sampled 0.8036 ha of selectively logged primary forest over two transects that contained 278 trees with a dbh ≥ 10 cm (=21.5, =13.9, range=10.0-110.8). I encountered a total of 39 families (48.5 families ha-1) with the five most prevalent being:

Lecythidaceae (11.2%), Sapotaceae (10.0%), Burseraceae (9.4%), Euphorbiaceae (9.0%), and Leguminosae (8.1%) (Appendix 1). The 39 families were represented by 126 genera,

(156.8 genera ha-1) dominated by: Eschweilera (9.8%), Protium (9.0%), Pouteria (5.8%),

Micrandropsis (4.5%), and Licania (3.6%) (Appendix 2). Making up those 126 genera were 278 species (345.9 species ha-1) with the most common being: Micrandropsis scleroxylon (4.5%), Protium hebetatum (2.7%), Byrsonima duckeana (2.3%), Eschweilera coriacea (2.3%), and Eschweilera truncata (2.3%) (Appendix 3).

Secondary Growth on Unburned Pastures

I had three replicates in this habitat, which is where I conducted the bulk of my secondary forest sampling. The three transects that I use to sample 25-year-old

Cecropia-dominated secondary growth regenerating on largely unburned clearcuts at the Fazenda Porto Alegre are in a sea of secondary growth of approximately 1340 ha.

These forests are regenerating on areas that were designed to be burned after being clearcut for pasture, but an early rainy season meant that most areas did not burn

(Gascon & Bierregaard Jr., 2001).

I sampled 0.896 ha of unburned pasture over three transects that contained 751 trees with a dbh ≥ 10 cm (=18.2, =7.2, range=10.0-44.8). I encountered a total of 37 families (41.3 families ha-1) with the five most prevalent being: Urticaceae (24.2%),

Euphorbiaceae (15.3%), Leguminosae (10.7%), Annonaceae (9.1%), and Malpighiaceae

(5.2%) (Appendix 1). The 39 families were represented by 87 genera (97.1 genera ha-1) dominated by: Cecropia (14.0%), Pourouma (10.3%), Inga (7.9%), Guatteria (7.1%), and

Croton (6.5%) (Appendix 2). Making up those 87 genera were 167 species (186.4 species ha-1) with the most common being: Cecropia sciadophylla (13.8%), Guatteria olivacea

(6.7%), Pourouma tomentosa tomentosa (4.3%), Byrsonima duckeana (3.9%), and

Croton matourensis (3.7%) (Appendix 3).

Secondary Growth on Burned Pastures

The single transects at Fazenda Dimona that sampled 19-year-old and 25-30-year-old secondary growth on burned pastures are in stands of those vegetation types of approximately 75 ha and 1325 ha, respectively. The latter is likely an over-estimate because it includes several riparian corridors and the area has a complex land use history.

I sampled 0.256 ha of 19-year-old secondary forest that contained 273 trees with a dbh ≥ 10 cm (=15.9, =5.2, range=10.0-49.1). I encountered a total of 12 families,

(46.9 families ha-1) with the five most prevalent being: Melastomataceae (39.6%),

Hypericaceae (34.4%), Goupiaceae (8.8%), Salicaceae (5.1%), and Euphorbiaceae (3.3%)

(Appendix 1). The 12 families were represented by 23 genera (89.8 genera ha-1) dominated by: Vismia (34.4%), Bellucia (27.8%), Goupia (8.8%), Miconia (6.6%), and

Acinodendron (4.8%) (Appendix 2). Making up those 23 genera were 33 species (128.9 species ha-1) with the most common being: Vismia cayennensis (31.5%), Bellucia

24 grossularioides (27.8%), Goupia glabra (8.8%), Acinodendron burchelli (4.8%), and Laetia procera (4.4%) (Appendix 3).

I sampled 0.264 ha of 25-30-year-old secondary forest that contained 252 trees with a dbh ≥ 10 cm (=17.5, =7.8, range=10.0-72.0). I encountered a total of 25 families (94.5 families ha-1) with the five most prevalent being: Euphorbiaceae (27.4%),

Annonaceae (9.9%), Melastomataceae (9.9%), Leguminosae (6.7%), and Hypericaceae

(6.3%) (Appendix 1). The 25 families were represented by 47 genera (178.0 genera ha-1) dominated by: Croton (25.58%), Vismia (6.3%), Laetia (6.0%), Palicourea (6.0%), and

Guatteria (5.6%) (Appendix 2). Making up those 47 genera were 70 species (265.2 species ha-1) with the most common being: Croton matourensis (25.8%), Laetia procera

(6.0%), Palicourea guianensis (6.0%), Guatteria olivacea (5.6%), Bellucia grossularioides

(5.2%) (also of note are Vismia cayennensis (3.2%) and V. japurensis (2.4%)) (Appendix

3).

The Primate Community

The primate community is composed of six species: 1) three frugivores, two of which are seed predators, the black-bearded saki (Chiropotes chiropotes) (Veiga, et al.,

2008)1 and golden-faced saki (Pithecia chrysocephala) (Marsh, In press)2, and one that specializes on the ripe fruit of large primary forest trees, the Guiana spider monkey

(Ateles paniscus) (Mittermeier, et al., 2008a); 2) two generalists, the golden-handed

1 The taxonomy of Chiropotes is unsettled. The form found at the BDFFP has also been referred to as Chiropotes satanas chiropotes (Hershkovitz, 1985) and Chiropotes sagulatus (Silva Jr. & Figueiredo, 2002). 2 The IUCN Red List (Veiga & Marsh, 2008) refers to this taxon as Pithecia pithecia chrysocephala but notes that a taxonomic revision is currently underway by Marsh (In press), whose classification I have followed.

25 tamarin (Saguinus midas) (Mittermeier, et al., 2008b) and Guianan brown capuchin

(Sapajus apella apella) (Rylands, et al., 2008); and 3) one folivore-frugivore, the Guianan red howler monkey (Alouatta macconnelli) (Boubli, et al., 2008). I follow the taxonomy used by the IUCN Redlist (IUCN, 2012), though I place the brown capuchin in the genus

Sapajus rather than Cebus (Lynch Alfaro, et al., 2012a; Lynch Alfaro, et al., 2012b; Silva

Jr., 2001; Silva Jr., 2002).

Two works provide summaries of the history of primate research at the BDFFP

(Boyle, et al., In Press; Gilbert, 2003). Early work included pre- and post-deforestation surveys of many of the habitat fragments that found that the largest species at the site, the black spider monkey, and another species thought to prefer undisturbed forest, the black-bearded saki, abandoned the fragments almost immediately (Rylands &

Keuroghlian, 1988). The next study compared forest characteristics and primate distribution in four 10-ha reserves and found more red howler monkeys, white-faced sakis, and golden-handed tamarins in the fragments with the highest structural complexity but no relationship between the distance of the fragments from continuous forest and the number of primate species present (Schwarzkopf & Rylands, 1989).

Primate presence in the fragments has also been related to ecological variables and primate characteristics, with fragment size (positive relationship) and distance to the nearest forested area (negative relationship) explaining patterns that generally show that frugivores and species with large home ranges tend to fare poorly in the fragments

(Boyle & Smith, 2010b). In general, primate use of fragments has increased over time, a trend that is likely influenced by the appearance of secondary forests on much of the

26 pasture between the fragments and the nearest continuous primary forest (Boyle, et al.,

In Press). The behavior and ecology of individual primate species have been examined in the continuous forest (Alouatta: Andresen, 2000; Andresen, 2002; Santamaría-Gómez,

1999; Santamaría & Rylands, 2003; Chiropotes: Boyle, 2008; Boyle, et al., 2009; Boyle &

Smith, 2010a; Boyle, et al., 2012; Frazão, 1991; Frazão, 1992; and Sapajus: Spironello,

1987; Spironello, 1991; Spironello, 2001) as well as in the isolated reserves (Alouatta:

Gilbert, 1994a; Gilbert, 1994b; Gilbert, 1997; Neves, 1985; Neves & Rylands, 1991;

Santamaria-Gómez, 2004; Chiropotes: Ayres, 1981; Boyle, 2008; Boyle, et al., 2009;

Boyle & Smith, 2010a; Boyle, et al., 2012; and Pithecia: Setz, 1993; Setz, 1994; Setz, et al., 1999; Setz & Gaspar, 1997). Predation of primates by raptors has also been observed on several occasions over the history of the BDFFP, both in the habitat fragments and in the undisturbed primary forest that encompasses the ranches (Gilbert, 2000; Lenz & dos

Reis, 2011).

Table 2.2 contains a summary of some of the behavioral and ecological characteristics of each of the primates in the community. The paragraphs that follow the table provide additional background on their natural history.

Table 2.2: Characteristics of the BDFFP primates (24) % of Diet Primate Weight (kg)a Group size Home range (ha) Fruit Flower Leaf Otherb Alouattac,d,e,f,g 6.8 8.2 53 40.4 6.4 48.1 5.1 Atelesh,I,j,k,l,m,n,o 7.8 14.3 224 88.7 3.8 6.1 1.4 Chiropotesi,p,q,r,s 2.8 21.8 336 87.9 5.7 5.8 0.6 Pitheciap,t,u,v,w,x 1.7 3.4 103 86.4 2.1 8.7 2.8 Saguinusj,p,y,z,A 0.5 5.7 33 66.0 1.9 0.8 31.3 Sapajush,j,m,p,B,C,D,E 2.6 14.3 429 65.0 1.4 2.7 30.9 a (Ford & Davis, 1992); b Consists mainly of insects, exudates, and small vertebrates; c (Gaulin & Gaulin, 1982); d (Julliot & Sabatier, 1993); e (Julliot, 1996a); f (Palacios & Rodriguez, 2001); g (Sekulic, 1982); h (Guillotin, et al., 1994); i (Kinzey & Norconk, 1990); j (Mittermeier & van Roosmalen, 1981); k (Norconk & Kinzey, 1994); l (Simmen, 1992); m (Simmen & Sabatier, 1996); n (Symington, 1988); o (van Roosmalen, 1985); p (Norconk, et al., 2003); q (Ayres, 1981); r (Boyle, et al., 2009); s (van Roosmalen, et al., 1981); t (Kinzey & Norconk, 1993); u (Lehman, et al., 2001); v (Norconk, 1996); w (Oliveira, et al., 1985); x (Vié, et al., 2001); y (Day & Elwood, 1999); z (Kessler, 1995); A (Oliveira & Ferrari, 2000); B (Izawa, 1980); C (Peres, 1993a); D (Spironello, 2001); E (Zhang, 1995b)

Alouatta macconnelli

The genus Alouatta is one of the largest Neotropical primates (average weight=6.4 kg) and it is the most widespread and sexually dimorphic (Ford & Davis, 1992; Kinzey,

1997a). Two of the most significant morphological features of howler monkeys are their prehensile tails and their enlarged hyoid bone/larynx complex that allows them to produce the loud calls, also known as howls or roars, for which they are best known

(Neville, et al., 1988). Howlers often consume unripe fruit as a means to reduce competition with other primates (Gaulin & Gaulin, 1982).

Red howler monkeys (Alouatta macconnelli) are found in a wide range of habitat types across their range, including gallery forest, mangrove swamps, semi-deciduous forest, savanna, woodlands, and secondary forest (Napier, 1976). At the BDFFP, the densest populations of howler monkeys have previously been recorded in the

28 undisturbed primary forest (Rylands & Keuroghlian, 1988). This species’ average weight is 6.8 kg (Table 2.2) with males weighing 5.4-9.0 kg (11.9-19.8 lbs) and females weighing less, 4.2-7.0 kg (9.3-15.4 lbs) (Ford & Davis, 1992). The howler monkey’s diet is composed of leaves (48.1%), fruit (40.4%), flowers (6.4%), and other resources (5.1%) and they have an average group size of 8.2 individuals and an average home range of 53 ha (Table 2.2). Groups are either unimale multifemale or multimale multifemale

(Sekulic, 1982).

Ateles paniscus

Spider monkeys (Ateles), which are found in both Central and South America, are among the heaviest New World primates and are characterized by their arboreality, long- slender limbs, specialization on mature fruits and mature forests, and prehensile tails

(van Roosmalen & Klein, 1988). This genus is unique among New World primates due to its fission-fusion social organization consisting of large multimale multifemale communities that forage in smaller subgroups that vary in both their number of individuals and age-sex composition (Kinzey, 1997b).

Ateles paniscus is the largest species at the BDFFP, averaging 7.8 kg (Table 2.2) with males weighing 5.5-9.2 kg (12.2-20.3 lbs) and females weighing 6.5-11.0 kg (14.2-

24.2 lbs) (Ford & Davis, 1992). Black spider monkeys are found in mature rainforest and occasionally in swamp forest (van Roosmalen & Klein, 1988). The spider monkey’s diet is composed of fruit (88.7%), leaves (6.1%), flowers (3.8%), and other resources (1.4%), with average group sizes of 14.3 individuals and average home ranges of 224 ha (Table

2.2). The study species, Ateles paniscus, spends much of its time in foraging parties of approximately three individuals (Wrangham, et al., 1993).

Chiropotes chiropotes

The distribution of the genus Chiropotes is limited to the lower Amazon and Orinoco basins and the Guianas (Kinzey, 1997d). Bearded saki monkeys are typified by their distinct beards, bulbous temporal swellings, long, thickly-haired non-prehensile tails, short body hair, and generally black pelage (van Roosmalen, et al., 1981). Bearded sakis live in multimale multifemale groups that may range from 8-30 individuals (though smaller groups were encountered during this study) and that travel as a cohesive group but spread out considerably when they reach a feeding tree (Kinzey, 1997d; Robinson, et al., 1987).

Bearded sakis average 2.8 kg (Table 2.2), with males weighing 2.2-4.0 kg (4.8-8.8 lbs) and females weighing 1.9-3.3 kg (4.2-7.3 lbs) (Ford & Davis, 1992). Chiropotes prefers high rain forest, mountain savanna forest, black-water swamp forest (igapó), and terra firme forest (van Roosmalen, et al., 1981). The bearded saki monkey’s diet is composed of fruit (87.9%), leaves (5.8%), flowers (5.7%), and other resources (0.6%) and they have average group sizes of 21.8 individuals and average home ranges of 336 ha

(Table 2.2). Chiropotes is one of two seed predators (Buchanan, et al., 1981; Kinzey &

Norconk, 1993) found at the BDFFP and was one of the first species to disappear from the habitat fragments following the clearcutting of the pastures (Rylands & Keuroghlian,

1988). It was hypothesized that it would take decades for them to return to the

30 fragments, but they surprised researchers by returning after approximately a decade

(Gilbert, 2003).

Pithecia chrysocephala

The genus Pithecia¸ which is found in the upper Amazon and from the Guianas south to the Amazon River (Buchanan, et al., 1981), is the smallest member of the closely-related pithecines (Pithecia, Chiropotes, and Cacajao) with females weighing 1.763 kg and males weighing 2.384 kg (Ford & Davis, 1992). They have long, bushy, non-prehensile tails and long, coarse, fluffy hair that makes them appear much larger than they actually are

(Buchanan, et al., 1981). Members of this genus live in small family groups and have sexually dichromatic facial pelage, a feature that is most pronounced in the study species, Pithecia chrysocephala (Kinzey, 1997e).

The white-faced saki (Pithecia chrysocephala) is one of the rare New World monkeys that is dichromatic: males are all black with a white face and black nose and females are brownish with a thin vertical white-brown stripe flanking their nose

(Eisenberg, 1989). They weigh an average of 1.7 kg (Table 2.2), with males weighing 1.0-

2.5 kg (2.1-5.5 lbs) and females weighing 0.8-1.8 kg (1.7-3.9 lbs) (Ford & Davis, 1992).

The white-faced saki monkey’s diet is composed of fruit (86.4%), leaves (8.7%), other resources (2.8%), and flowers (2.1%) with average group sizes of 3.4 individuals and average home ranges of 103 ha (Table 2.2). White-faced sakis are the second species of primate seed predator at the BDFFP (Chiropotes) (Buchanan, et al., 1981; Kinzey &

Norconk, 1993). These monkeys prefer to forage in the lower to middle canopy from 3-

15 m (Peres, 1993b) and unhabituated individuals are very skittish around human observers (B. Lenz, personal observation).

Saguinus midas

The genus Saguinus is found primarily in the Amazon Basin, though it also extends into

Central America (Kinzey, 1997f). Tamarins are small monkeys that are known as callitrichids (marmosets and tamarins), a group of Neotropical primates with claw-like nails that are used in their vertical clinging postures on large tree trunks (Kinzey, 1997f).

The callitrichids are characterized by monogamous (and perhaps polyandrous) groups that are cooperative breeders, meaning that fathers and older juveniles aid in the care of offspring (Snowdon & Soini, 1988).

The golden handed-tamarin (Saguinus midas) is the smallest primate at the site, with an average weight of 0.5 kg (Table 2.2) and males weighing 586 g (20.7 oz) with females weighing 432 g (15.2 oz) (Ford, 1994). Golden-handed tamarins are found in primary, secondary, and swamp forests as well as edge habitats, though at some sites they may prefer open high canopy (Hershkovitz, 1977). However, at the BDFFP they have been found at low densities in the continuous forest, perhaps due to their preferences for edges and secondary growth that are abundant at the study site

(Rylands & Keuroghlian, 1988). The golden-handed tamarin’s diet is composed of fruit

(66.0%), other resources including vertebrates and insects (31.3%), flowers (1.9%), and leaves (0.8%) with average group sizes of 5.7 individuals and average home ranges of 33

32 ha (Table 2.2). The study species has a variable social system with groups being either unimale multifemale or multimale multifemale (Ferrari & Ferrari, 1989).

Sapajus apella apella

Capuchins, which are classified in two genera, Cebus and Sapajus, are the second most widespread Neotropical primate genus (Alouatta) being found from Honduras to

Argentina (Kinzey, 1997c). Capuchins have semi-prehensile tails and inhabit nearly all types of Neotropical forest (Freese & Oppenheimer, 1981). They also show greatly variable behavior between species, between groups, and even within groups, perhaps explaining their success in so many different habitats (Kinzey, 1997c). Capuchins typically live in multimale multifemale groups containing an approximately equal number of each sex (Terborgh, 1983) and have linear male and female dominance hierarchies (Robinson & Janson, 1987).

The average weight of Sapajus apella is 2.6 kg (Table 2.2), with males weighing between 1.3-4.8 kg (2.9-10.6 lbs) and females weighing between 1.4-3.4 kg (3.0-7.5 lbs)

(Ford & Davis, 1992). Brown capuchins are found in primary and secondary rainforest and they also range into semi-deciduous lowland and montane forest (Eisenberg, 1989).

The brown capuchin's diet is composed of fruit (65.0%) and other resources including vertebrates and insects (30.9%), leaves (2.7%), and flowers (1.4%), with average group sizes of 14.3 individuals and average home ranges of 429 ha (Table 2.2). At the BDFFP, capuchins were recorded at lower population densities than reported in any other study

(Rylands & Keuroghlian, 1988). In Peru, 3 to 4 groups are found in 100-ha of forest, but at the BDFFP single groups have home ranges of around 900 ha (Spironello, 2001).

PRIMATE DIETS

In a study of primate habitat use it is important to derive the best possible estimate of food availability, in particular the fruit that is crucial in the diets of most primates. To collect data on the primates I used transect sampling rather than follows of focal groups, so the opportunities to record detailed diet data were extremely limited due to the small amount of time actually spent with groups. As a result, I was forced to estimate both the fruits consumed by the study species as well as fruit availability in the habitats that I sampled. To accomplish this I took several steps, all of which are discussed in greater detail in Chapter 3. Briefly, I conducted a literature review and compiled a database of all the tree species whose fruit was consumed by any one of the six study species. I carried this database with me when I identified the trees along my transects with a technician and used it, along with the technician's knowledge, to tag the trees that were important sources of fruit for primates. I then recorded monthly data on the reproductive phenology of the tagged trees in conjunction with my transect sampling. I also conducted another literature search for studies that had reproductive phenology data from sites in the Manaus region, and combined those data with the data that I collected. I then combined my database on reproductive phenology with my tree identifications to estimate the total fruit available along my transects, using my diet

34 database to identify subsets of the overall fruit data important for each individual primate species.

The primate diet database yielded data on the tree species along my transects that were most likely to be important to each primate species, whether or not I had a phenology estimate for them. I have included appendices that contain the top tree families, genera, and species for each primate species in the community (Alouatta:

Appendix 4, Ateles: Appendix 5, Chiropotes: Appendix 6, Pithecia: Appendix 7, Saguinus:

Appendix 8, Sapajus: Appendix 9). When looking at these appendices it is important to keep in mind that tree species entered my diet database on an eaten/not eaten basis, in other words, my database gives equal footing to tree species that are highly important for individual primates with those that are only rarely consumed (the reasons for this methodology are also discussed in Chapter 3).

Of the overall tree species that were important, I was able to use my reproductive phenology database to estimate fruit availability for a portion of the total tree community. The percentage of the total adult trees whose phenology I was able to estimate are summarized in Table 2.3, which provides data for both overall fruit availability as well as for each primate species. I estimated fruit availability for 65-69% of the individual trees sampled, though I was able to estimate far smaller percentages for each individual primate. Of note are the high estimates for Chiropotes, thanks to detailed diet data from the BDFFP (Boyle, 2008), as well as the low estimate for

Saguinus, which was due to the paucity of dietary studies on this species.

Table 2.3: The percentages of adult trees and tree species for which I estimated reproductive phenology # Estimated % of Estimated in Primate Diets Habitat (% of Total) Alouatta Ateles Chiropotes Pithecia Saguinus Sapajus Unburned secondary forest 780 (65) 20 25 28 12 19 28 # Individuals Logged primary forest 643( 66) 16 12 37 15 6 17 Undisturbed primary forest 1632 (68) 19 13 45 18 5 15 Unburned secondary forest 168 (47) 13 14 23 9 8 16 # Species Logged primary forest 278 (50) 13 9 22 10 6 13 Undisturbed primary forest 424 (50) 12 10 25 11 4 11

The goal of these steps was to estimate fruit availability in each habitat to determine if it explained changes in primate abundance. Table 2.4 summarizes the total amount of fruit that I estimated for each habitat, both overall and for each individual primate species. When looking at this table it is again important to remember that the fruits of commonly and infrequently eaten tree were considered to be of equal importance, so rather than summarizing the important fruiting resources for each primate species this table instead summarizes the availability of fruits that one could realistically expect to be consumed.

Table 2.4: Annual fruit availability by habitat

Species Forest Type Total Fruit (kg/ha) Species Forest Type Total Fruit (kg/ha) Total Secondary (unburned) 57,726 Pithecia Secondary (unburned) 5,183 Total Logged primary 50,218 Pithecia Logged primary 10,947 Total Undisturbed primary 45,932 Pithecia Undisturbed primary 11,873 Alouatta Secondary (unburned) 31,000 Saguinus Secondary (unburned) 26,458 Alouatta Logged primary 20,125 Saguinus Logged primary 5,438 Alouatta Undisturbed primary 22,578 Saguinus Undisturbed primary 3,177 Ateles Secondary (unburned) 32,161 Sapajus Secondary (unburned) 32,766 Ateles Logged primary 9,289 Sapajus Logged primary 15,169 Ateles Undisturbed primary 9,647 Sapajus Undisturbed primary 12,713 Chiropotes Secondary (unburned) 13,982 Chiropotes Logged primary 33,628 Chiropotes Undisturbed primary 39,586

I have made much of the way in which fruiting tree species entered my diet database, and the reason for this is clear following an examination of graphs of monthly fruit availability in secondary forests on unburned clearcuts (Figure 2.4, Figure 2.5,Figure

2.6). In this series of graphs I show the fruit estimates (overall and for each primate species) by habitat using my complete dataset (A) and using a subset of the data where I have removed the most common tree species in secondary forest, Cecropia sciadophylla

(B), for the four primate species that consumed it (Appendix 3). The removal of the most dominant species in secondary forest significantly alters the interhabitat relationship in terms of fruit availability. Therefore, if for any of these primates consumption of C. sciadophylla is rare, which seems like a distinct possibility for at least Ateles, then the interhabitat relationship in the availability of important fruits will be incorrect. None of the studies in which species consumed C. sciadophylla contains the consumption frequency, so unfortunately I am unable to investigate this possibility further. However,

I wanted to make clear the nature of the data that compose my fruit availability estimates. The fruit availability data for the two species that did not consume C. sciadophylla, the seed predators Chiropotes and Pithecia, clearly show the lowest availability in secondary growth (Figure 2.6).

Figure 2.4: Fruit availability by habitat with (A) and without (B) Cecropia sciadophylla (Alouatta, Ateles)

Figure 2.5: Fruit availability by habitat with (A) and without (B) Cecropia sciadophylla (Saguinus, Sapajus)

Figure 2.6: Fruit availability by habitat for Chiropotes and Pithecia and overall fruit availability with (A) and without (B) Cecropia sciadophylla

Chapter 3: Methodology

This chapter is an in-depth discussion of the techniques that I used to record data and describe the primate and tree communities (including the estimation of reproductive phenology). These topics are covered in less detail in the chapters that follow, though I felt that they deserved in-depth coverage due to their complexity. I do not discuss statistical analyses in this chapter because they are discussed in sufficient detail in each of the principal chapters.

Data Collection

I used line transects to survey the landscape for primates to record their density and presence/absence, to identify and describe the dendrometrics of the tree communities that I sampled, and to record the reproductive phenology of trees that are important for primates. Line transects are a systematic, objective, and quick method for comparing variables between habitats (Buckland, et al., 1996; Cant, 1978). I collected data over a three-year period, November-December 2008, June-November 2009, and

January-June 2010, using eight transects (Figure 3.1,Table 3.1) separated by 0.6-15.0 km

(spacing was not equal due to landscape limitations) that were flagged every 10 m. In total, I conducted 571.5 km of line transect sampling and recorded ad lib observations

41 over an additional 1,163.7 km (during the collection of phenology data and while walking to and from, moving between, and cutting the line transects).

Figure 3.1: Transect Locations

Table 3.1: Transect information (each transect sampled two habitats)

Time since Transect Habitat Disturbance (yr)* Length (km) Replicates Total (km) 1 Undisturbed primary forest - 0.03 47 1.41 Secondary forest (burned) 25-30 0.66 47 31.02 2 Undisturbed primary forest - 1.07 42 44.94 10 ha fragment (#2206) 24 0.35 43 15.05 3 Undisturbed primary forest - 1.10 38 41.80 Secondary forest (burned) 19 0.64 31 19.84 4 Undisturbed primary forest - 0.85 37 31.45 100 ha fragment (#2303) 18/24 1.10 42 46.20 5 Undisturbed primary forest - 1.00 46 46.00 Secondary forest (unburned) 25 0.71 46 32.66 6 Undisturbed primary forest - 1.00 46 46.00 Secondary forest (unburned) 25 0.82 44 36.08 7 Logged primary forest 25 1.03 58 59.74 Secondary forest (unburned) 25 0.07 57 3.99 8 Logged primary forest 25 1.06 55 58.30 Secondary forest (unburned) 25 0.64 50 32.00 * Relative to 2008 (Gascon & Bierregaard Jr., 2001)

All transects ran perpendicular to the primary forest-secondary growth border to allow for an examination of edge effects. I cut transects in pairs (Figure 3.1) so that I could sample one transect and move to the second and immediately sample it. This allowed me to maximize my time in the field because I was able to sample a second transect rather than sitting inactive at the end of a single transect before conducting a return walk. As a result, I was able to allow each transect to rest for at least 6 h before resampling. Although many researchers (e.g. Sorensen & Fedigan, 2000) prefer to let transects rest for at least a day before resampling them, rest periods of 3 h have been suggested as the minimum amount of time that must pass before resampling a transect to permit the primates sufficient time to redistribute themselves in the forest (Peres,

1999).

A common transect sampling protocol to avoid time-related primate detection biases is the rotation of the end of the transect where observers begin sampling (e.g.

Lehman, et al., 2006c). Landscape limitations prevented me from using this methodology because I was unable to rotate starting points without first walking the length of a transect to reach the end that was deep within the primary forest. Doing so would then necessitate a rest period for the transect and therefore be a highly inefficient use of time. Instead, for each pair of transects I rotated the transect where I began sampling (though I always started at the same end of each transect). This helps to reduced time-related biases, though was less desirable than being able to rotate starting points.

The habitats sampled are described in Table 3.1 and Table 3.2. Data collected on transects in the habitats sampled by only a single transect, two age classes of secondary growth on burned pastures and the 10-ha (#2206) and 100-ha (#2303) habitat fragments, are used only to enlighten discussion rather than being analyzed statistically due to the lack of replication.

Table 3.2: Total area sampled by habitat

Habitat Length (km) Total (with replicates) (km) Undisturbed primary forest 5.05 211.60 Logged primary forest 2.09 118.04 Secondary forest (unburned) 2.24 104.73 100 ha fragment (#2303) 1.10 46.20 Secondary forest (burned, 25-30 y) 0.66 31.02 Secondary forest (burned, 19 y) 0.64 19.84 10 ha fragment (#2206) 0.35 15.05

I walked each transect pair beginning at 06:45 and 13:15, sampling each transect once in the morning and once in the afternoon. I avoided mid-day sampling when possible to reduce data collection during the hottest parts of the day when primates tend to be less active. Many studies recommend sampling that is even more temporally restricted, with data collection only taking place only during the early morning and late afternoon when primate activity levels tend to peak (Peres, 1999; Whitesides, et al.,

1988). However, sampling transects at all times of the day is also acceptable because reduced detection probabilities during periods when primates are less active are compensated for by narrower strip widths (Chapman, et al., 1988).

Data were collected by two observers, always 5-10 m apart and with the author always in the lead for consistency, who slowly walked transects (average speed=1.01 km hr-1) while pausing for 7-10 s every 10 m to look and listen. Similar walking speeds are commonly recommended for primate surveys: 1.0 km hr-1 (National Research Council,

1981); 1.25 km hr-1 (Peres, 1999); 1.0-2.0 km hr-1 (Ross & Reeve, 2003); 1.0-1.5 km hr-1

(Whitesides, et al., 1988). Upon locating a group of primates I recorded: species, number of individuals, location on the transect, distance between first animal observed and the researcher, height of that individual, its activity, and mode of detection (Gilbert,

2003). In addition, I also recorded the time, a sighting angle, group spread along two perpendicular axes including the longest axis, the presence or absence of a polyspecific association, and the height of the first individual sighted relative to the height of both the tree it was in and the canopy. I used a Nikon Forestry 550 laser rangefinder to measure the sighting distance to, and height of, the first individual and a Suunto MC-2G

45 global compass to measure the sighting angle. All encounters with snakes, raptors, and large mammals were also recorded in the same fashion; additionally, I made notes of the locations of all mammal tracks, scats, and claw marks to allow me to estimate the distribution of primate predators and competitors in the landscape. I attempted to spend less than 10 minutes recording data for each primate group, but did not truncate encounters at the 10-minute mark if all data had not yet been recorded. Observers also remained on the transects at all times in an effort to minimize our influence on the behavior of the largely unhabituated study subjects while also reducing the stress on the animals that resulted from our presence. If we moved further along the transect (or back-tracked) while gathering data on a group, we always returned to the exact location where the encounter began to restart the transect sample to ensure that portions of the transect were not skipped or sampled twice (Chapman, et al., 1988; Peres, 1999;

Whitesides, et al., 1988).

Primate Population Density

There are several ways to calculate primate population density, the number of individuals in a given area (usually expressed in km2), with one of the most widely used methods being line transect surveys. Estimates of population density based on transect sampling calculate the number of animals within a given survey area. There are seven assumptions that underlie density estimation based on transect sampling (Seber, 1982):

1) All groups are distributed at random with respect to the survey path,

2) A sighting of one group is independent of sightings of other groups,

3) Groups are only counted once during a sample,

4) Any behavioral response by the animals to observers does not affect sighting

distances,

5) The behavior of groups does not vary with location along the transect,

6) The behavior of groups is species typical, and

7) The probability of detecting a group located directly on or above the survey path

is 1.0.

These assumptions are fairly straight-forward, with the only potential issue for this study being the fifth assumption because many of the transects cross multiple distinct habitats (e.g. undisturbed primary forest, secondary forest, and selectively logged primary forest). However, this assumption is only violated if the data from distinct habitats are lumped and considered to come from a uniform landscape without first testing for interhabitat differences. One of the goals of the current study is to examine interhabitat differences in primate abundance, so data collected in each distinct habitat were not lumped and therefore this assumption is not violated.

Density estimation using line transect sampling requires several types of data: 1) the number of animals observed (or possibly the number of groups for clustered species); 2) transect length; and 3) an estimate of the width of the area sampled along the transect

(Whitesides, et al., 1988). Much of the variation between line transect studies of primate density arises from differences in the methods used to estimate the latter, with

47 the two most common methods to estimate the total width of the area surveyed being distance sampling and strip width estimation.

Distance sampling, which can be done using either line transects or point sampling, assumes 1) that only a portion of the animals that are present in the immediate area will be detected by researchers walking a transect, and 2) that the proportion of individuals observed declines as distance between those individuals and the transect increases.

Distance sampling uses the pattern of observations and the rate at which they decline relative to their distance from the transect to estimate the total number of animals, both detected and undetected, present between the transect and the most extreme observation (Ross & Reeve, 2003). All observations are used to create a detection function (g) which is then used to calculate the number of animals that were not detected by the observers. This method is useful because, unlike strip width estimation, it does not force researchers to discard extreme observations, something that is generally undesirable, especially when data collection is costly (Eberhardt, 1968).

Distance sampling, however, does have a drawback: it requires large sample sizes because small samples provide too little information to create a function that yields accurate estimates (Buckland, et al., 2001). Calculating an accurate estimate using the distance method requires a sample size between 60 and 80 sightings, though under certain circumstances 40 observations may suffice (Buckland, et al., 2001). For group- living animals, like the primates sampled in this study, the sample size is the number of groups observed rather than the number of individuals. Each species must also be

48 handled separately, so a sample of 50 capuchins and 40 spider monkeys yields two datasets of 50 and 40 observations rather than a single dataset of 90 observations.

Population density is then estimated using the mean number of animals per group, data that can be obtained either during the line transect sampling or from group counts in other situations (Ross & Reeve, 2003). Furthermore, when using stratified line transects that sample multiple habitats, a sample of 60-80 observations for each species must be obtained in every habitat sampled. For example, assuming that there are interhabitat differences in detectability, producing accurate estimates of capuchin density in undisturbed primary forest, selectively logged primary forest, and secondary growth would require a minimum of 60 encounters in each habitat (a total of 180 encounters).

Given the amount of time that it takes to conduct transect sampling on foot using dispersed transects in a tropical rainforest where primate abundances have previously shown to be low (Rylands & Keuroghlian, 1988), this method was not practical for the current study because the necessary sample sizes were unattainable.

The second method to calculate primate densities is the strip width estimation method. There are several ways that strip width can be estimated, and they can be broken into two general groups: those that discard extreme observations and those that do not. Methods that do not discard any observations are favored by some (Chapman, et al., 1988) because, as was mentioned earlier, it is never desirable to discard observations (Eberhardt, 1968). Chapman et al. (1988) review six of these strip width estimation methods and conclude that the mean distance from the observer to the animal (MDA) is the most accurate for their strip width estimations for white-faced

49 capuchins (Cebus capucinus) and mantled howler monkeys (Alouatta palliata). However, they also note that MDA has produced mixed results in other studies where it was found to: 1) produce less biased estimates than other methods (Robinette, et al., 1974), 2) accurately estimate strip width for two species while producing very inflated estimates for a third (Sapajus apella) (Defler & Pintor, 1985), and 3) strongly overestimate population density for three Old World monkey species (Strusaker, 1981) and several

New World monkeys (Janson & Terborgh, 1988 cited by Chapman, et al., 1988). In addition to these accuracy issues, there is also the issue of required sample size.

Chapman et al. (1988) found that their precision curve did not level off until 200-500 trials, and they note that other authors (Altmann & Altmann, 1970; Neville, 1976) found a similar need for extremely large sample size. As a result, MDA’s inconsistent performance and the large sample sizes needed also result in its unsuitability for the current study.

The second group of strip width estimation techniques calls for the most extreme observations to be discarded and therefore lessens the need for such large sample sizes. The method that is going to be used in this study, which I will refer to as the Whitesides' perpendicular distance to the path (WDFP) (Whitesides, et al., 1988), is one of the methods that calls for discarding extreme observations that has been used in several primate studies (e.g. Lehman, et al., 2006c; Whitesides, et al., 1988). A closely related technique was also previously used at the BDFFP (Rylands & Keuroghlian, 1988).

WDFP is based on a method originally set out by Kelker (Kelker, 1945), who used the perpendicular distance from the path (DFP) of the first individual seen during an encounter to determine the maximum reliable strip width within which a researcher could assume that they detected 100% of the individuals present (National Research

Council, 1981). Whitesides et al. (1988) modify this technique in two important ways.

First, they incorporate group spread by using the perpendicular distance to the group center rather than the perpendicular distance to the first individual sighted as in DFP. In the calculation of primate density one is really interested in the distance to the group center rather than the distance to the first individual sighted because the data points are really the groups themselves rather than each of the individuals that compose those groups.

The perpendicular distance between the transect and the group center/first individual can be obtained in two ways: 1) by directly measuring the perpendicular distance to the transect, or 2) by recording both the distance between the observer and the group center/first individual (also known as the sighting distance) and the angle between the sighting line and the transect, and then utilizing trigonometry to calculate the perpendicular distance. The former can be problematic for a few reasons. It is often difficult, if not impossible, to determine the location of the center of a group of dispersed primates. This is especially true when those primates are unhabituated and the researcher is under the time constraints imposed by line transect methods

(Whitesides, et al., 1988). Furthermore, when a group is initially detected, the observer- first animal/group center sighting angle is rarely 90 degrees to the transect because

51 groups are almost always detected when they are ahead of observers on the transect (B.

Lenz, personal observation). As a result, observers must move down the transect to find the perpendicular sighting angle and record the sighting distance, but during this time the group is still active and it can be very difficult to relocate a distant tree limb after moving from the spot where the initial detection occurred. This decreases the accuracy of the data and greatly increases the amount of time that observers must remain stationary on the transect while they record data. Due to these issues it is preferable, for both accuracy and the efficiency of data collection, to record the sighting distance and sighting angle and use trigonometry to calculate the true perpendicular distance to the transect.

As a solution to the issue of determining the location of the group center,

Whitesides et al. (1988) recorded the sighting data for the first individual detected, data that involve far less uncertainty than field estimates of the group center, and then estimate the location of the group center relative to the first individual sighted. The rationale for this method is based on the fact that the first individual detected is usually on the periphery of a group that is entirely ahead of the observers on the transect.

Therefore, the group center often falls along the sighting line to the first individual while being farther from the observer than that first individual sighted. To estimate the sighting distance to the center of the group under this method, 50% of the species- specific average group spread is added to the sighting distance of the first individual detected, along the same sighting angle, thereby approximating the distance to the

52 group center. As above, trigonometry can then be used to calculate the perpendicular distance between the group center and the transect (Figure 3.2).

Figure 3.2: Illustration of WDFP (Whitesides, et al., 1988)

The second alteration that Whitesides et al. (1988) made to the DFP model was to add a 50% fall-off criterion. The authors examined histograms of the perpendicular distance from the transect to the first individual sighted for each primate species and estimated a fall-off distance for each based on these histograms. The fall-off criterion is

53 the first interval in which the number of species-specific sightings is <50% of the number of sightings in the interval immediately preceding it. For example, in Figure 3.3 the fall- off distances for Cercocebus atys and Cercopithecus diana are 9 m and 39 m, respectively. The 50% criterion is somewhat arbitrary, but it works well for rain forest primates because it removes the data points with the lowest precision and excludes the distances at which groups are likely to have been missed from the area sampled (Marsh

& Wilson, 1981; Skorupa, 1988).

Figure 3.3: 50% fall-off distance (Whitesides, et al., 1988)

Following the WDFP’s two modifications results in the calculation of the area sampled using the following formula:

Area=T*Lt=2WtLt=2(0.5*S+ED)*Lt with Effective Distance(ED) = (Nt/Nf)*(FD)

where T=total width sampled, Lt= total length sampled, Wt=species-specific width on one side of the transect, S=estimate of species-typical mean group spread (collected during transect and ad libitum sightings), Nt = species-specific total number of group

sightings; Nf = species-specific number of group sightings at distances less than half the fall-off distance; FD = species specific fall-off distance (Whitesides, et al., 1988).

I also calculated density following Whitesides et al. (1988):

Density= It/A =It / (2(S/2+ED)*Lt)

where It=number of individuals of each species, A=area (as above), and the other variables are as above. I modified this formula slightly because it does not take into account the area of the half-circle that is also sampled at the end of each transect. So whereas Whitesides et al. (1988) use:

Area=2(S/2+ED)*Lt when a transect terminus samples the same habitat as the rest of the transect (as opposed to a transect that ends at the border between habitats, in which case I use

Whitesides' formula) I add the term 0.5πR2 to the formula for Area and use:

2 2 Area = 2(S/2+ED)*Lt + 0.5πR = 2(S/2+ED)*Lt + 0.5π(S/2+ED) where R (radius) is the species-specific effective distance plus one half of the average group spread.

The WDFP performs better than DFP because DFP does not provide an estimate to the center of the group and therefore the group is scored as being closer to the transect than it really is, resulting in an overestimation of density (Anderson, et al.,

1979; Burnham, et al., 1981; Quinn II, 1981). A long-term study of African primates with

55 a known population density of 6.0 groups/km2 found that WDFP estimated 6.1 groups/km2 (a 2% overestimate), a significantly more accurate estimate than using the observer-animal distance (6.6 groups/km2, a 10% overestimate) (Strusaker, 1981). The

WDFP model is also attractive because it appears to deal well with relatively small sample sizes, even down to N=19 or lower (Whitesides, et al., 1988). This is due to the fact that the 50% fall-off distance removes the most extreme, and therefore least reliable, observations from the analyses.

Sampling the Floral Community

THE TREE COMMUNITY

To characterize the plant community along each transect I sampled all trees within 2 m of either side of each transect for a transect width of 4 m. In mature forest (undisturbed primary forest, selectively logged primary forest, and habitat fragments) I identified, measured, and mapped all adult trees ≥10 cm dbh (diameter at breast height, or 1.3 m), a method commonly used to characterize primate habitats (e.g. Lehman, et al., 2006c;

Sorensen & Fedigan, 2000; Warner, 2002). In secondary forests I lowered the threshold to ≥5 cm dbh because trees are younger, smaller, and may mature earlier, though for interhabitat taxonomic comparisons (see Appendices) I report only the data for individuals ≥10 cm dbh. I recorded the following for each tree: dbh (cm), species, and location on the transect (mapped in 10-m intervals). At the edges of the transect strip, trees with 50% or more of their bole (trunk) at breast height located in the transect strip were included in the transect while those with less than 50% were excluded (Warner,

2002). Species-level identifications were done in the field by a top expert on Amazonian trees, Paulo Apóstolo Assunção (da S. Ribeiro, et al., 1999). I follow the taxonomy used by The Plant List (2010).

I sampled the tree community in several different habitats: undisturbed primary forest, selectively logged primary forest, secondary forest on unburned pastures, two ages of secondary forest on burned pastures, and habitat fragments (10 and 100 ha).

The unburned secondary forests are approximately 25-years-old while the secondary forest on the two burned pastures are 25-30 and 19-years-old (Gascon & Bierregaard Jr.,

2001). The three transects that I used to sample 25-year-old secondary forest on unburned pastures at the Fazenda Porto Alegre are in a sea of secondary growth that is approximately 1340 ha (Figure 6.1). The transects at Fazenda Dimona that sampled 25-

30-year-old and 19-year-old secondary growth on burned pastures are in stands of those vegetation types of approximately 1325 ha and 75 ha, respectively (Figure 6.1).

Trees along the two transects in selectively logged forest were likely harvested in

1983 (B. Lenz, personal observation), 25 years before the start of this study, when the forest at the Porto Alegre Ranch was cleared (Gascon & Bierregaard Jr., 2001). I use the word "likely" because I am the first to sample this area and none of the BDFFP's founders or early researchers recall selective logging, though it clearly occurred given the stumps I encountered. It is very likely that logging occurred during the clearance of the ranch, a common practice when clearing pastures in the region, because the area was largely undisturbed prior to that time. Though I did not survey this habitat beyond

57 my transects, I believe it is likely that the logged area does not cover significantly more than 250 ha. I recorded the impacts of logging to a depth of 10 m on both sides of transects for a total area sampled of 4.18 ha, noting the number of stumps (N=2, 0.48 stumps ha-1) and the number of abandoned logging paths (~10 m wide) with relatively open canopy above them that crossed the transects (N=5, coverage=0.1 ha or 2.4% of the area sampled). According to tree expert Paulo Apóstolo Assunção (da S. Ribeiro, et al., 1999), the logged trees were most likely Manilkara huberi, a species whose fruit is part of the diet of at least four of the study species: Alouatta, Ateles, Chiropotes, and

Sapajus (Boyle, 2008; Fragazy, et al., 2004; Neves & Rylands, 1991; Simmen & Sabatier,

1996). It is possible that M. huberi was cut either for its timber or for its latex as there are unconfirmed reports of the latter in the 1970s at the Floresta Nacional do Tapajós

(Michael Keller, personal communication).

REPRODUCTIVE PHENOLOGY

Actual primate food availability may be unknowable (Chapman, et al., 1994), but it is still important to derive the best possible estimate. To determine monthly fruit availability I used three tools: 1) a database of fruit consumed by each primate species to tag important fruiting trees to permit me to monitor their reproductive phenology, 2) a database of reproductive phenology data from other studies, and 3) an allometric equation to estimate the fruit crop for each tree. I will now discuss each in greater detail.

Prior to tree identification in the field, I conducted a literature review to create a database containing the species of fruiting trees used by each primate (Andresen, 2000;

Boyle, 2008; Buchanan, et al., 1981; Egler, 1986; Fragazy, et al., 2004; Frazão, 1992;

Gaulin, 1977; Gaulin & Gaulin, 1982; Guillotin, et al., 1994; Julliot & Sabatier, 1993;

Mittermeier, 1977; Neves & Rylands, 1991; Norconk, 1996; Norconk & Conklin-Brittain,

2004; Oliveira & Ferrari, 2008; Pack, et al., 1999; Port-Carvalho & Ferrari, 2004;

Santamaría-Gómez, 1999; Setz, 1993; Silva & Ferrari, 2007; van Roosmalen, et al., 1981;

Veracini, 2000; Wright, 2004) (note: diet data for Saguinus midas were so limited that I also included data from the closely related S. niger). Tree species entered the database for each primate on an eaten vs. not eaten basis. In other words, the importance of any particular fruiting species was not a factor. This was done for two reasons. First, many publications do not provide this information. Second, the majority of the diet composition data available for the study species was from research at other field sites.

As a result, I wanted to be as inclusive as possible given the intersite distances and the well-documented heterogeneity of tropical forests even within the range of a single primate species. Due to my methodology, I expect to overestimate fruit availability in secondary forests for most, if not all, primate species. For example, if a mature forest fruit specialist like Ateles was foraging at the edge of a treefall gap in primary forest and on a single occasion consumed the fruit of a common secondary forest tree species like

Cecropia sciadophylla, that species entered their diet database. My fruit estimates for secondary forest would then show that this habitat contained a great deal of fruit consumed by Ateles due to the large number of individuals of C. sciadophylla in that

59 habitat, even though Ateles only consumed that species once and was in primary rather than secondary forest when it did so. If the diet lists contained only tree species that accounted for the important species in each primate's diet this issue would not arise.

Unfortunately, this source of error is unavoidable in this study for the reasons I have just discussed, so my estimates are therefore measures of the availability of any fruit that these primates have consumed in the past rather than an estimate of only important resources.

I used the diet database, along with Assunção's knowledge, to tag the most important fruiting trees during their identification. I then collected monthly data on the reproductive phenology of the tagged individuals (presence/absence of buds, flowers, and immature/ripe fruit). I also constructed a reproductive phenology database using the data that I collected in the field during this study as well as data from other studies at the BDFFP and the Manaus region (Alencar, 1998; Boyle, 2008; Bentos, et al., 2008;

Camargo, et al., 2008; Lemos, 1998; Oliveira, 1997). I limited the data used in this database to studies conducted in the immediate vicinity of the study site because the timing of the start and end of different phenophases (e.g. fruiting or flowering) can vary among individuals of the same species as the distance between those individuals increases. In the Manaus region this can happen along an east-west axis. For example, at site 1, on the eastern end of a hypothetical range, species X begins to flower on May

1. At site 2, 20 km to the west of site 1, individuals of species X flower on May 15, while individuals of species X at site 3, 20 km to the west of site 2, flower on June 1 (Camargo, et al., 2008; J.L. Camargo, personal communication).

Fruit abundance is generally recorded in one of three ways: 1) a count of the actual number of fruits, 2) the use of an index to rank the amount of fruit, or 3) a simple presence/absence score (Chapman, et al., 1994). Actual counts are the most time consuming and require frequent examinations of each tree (Chapman, et al., 1994), making this method impractical for this study due to the large number of tree species that needed to be examined to estimate fruit availability for an entire primate community. The index method involves a visual examination of a tree’s fruit crop followed by assigning it a relative rank, often on a scale of 0 to 4. All trees must be examined in a short period to yield an accurate estimate of the crop size because every individual of each species needs to be in the same phenophase during each sampling period (Chapman, et al., 1994). This method is less time consuming than the count method, but the close temporal proximity in which data must be recorded was not practical for this study given that it was conducted at two separate study sites. The third method is straight-forward, the presence or absence of fruit is noted for each tree and the presence/absence score is then combined with the tree's dbh in an allometric equation to estimate the fruit biomass of individual trees (Chapman, et al., 1994;

Sorensen & Fedigan, 2000). A comparison of the index and presence/absence methods found that the results were closely correlated (0.970, p<0.001) (Chapman, et al., 1994), so for a study that does not have reproductive phenology as its primary focus the presence/absence method is sufficient and was therefore employed in the present study.

To estimate monthly fruit availability for each primate species I used an allometric equation (Peters, et al., 1988) for fruit weight based on tree dbh that has been used in previous studies of primate ecology (Chapman, 1990a; Sorensen &

Fedigan, 2000):

Fr=47dbh1.9 where Fr is the mass of the fruit produced by an individual tree and dbh is the diameter at breast height of that tree. Individual trees ≥10 cm dbh were assumed to be reproductive adults. I combined the reproductive phenology database with the tree identifications and dbh measurements and used the formula above to calculate monthly fruit availability for individual primate species in every 10-m section of each transect.

Estimating whether a taxon is monoecious, dioecious, or hermaphroditic is important because, while it was previously assumed that dioecy was not important in tropical forests, many studies have shown that dioecious species can make up significant proportions of the forest (e.g. 22% in Costa Rica (Bawa & Opler, 1975), 26% in

Indonesia (Ashton, 1969), and 40% in Nigeria (Jones, 1955)). I used a literature search to determine the reproductive system of each tree species and assumed that congeners shared the same reproductive system when species-level data were not available. For monoecious and hermaphroditic trees I used the allometric relationship above and assumed that each individual in the community produced fruit when the reproductive phenology database said that a species was fruiting. For dioecious species I likewise assume that all individuals produce fruit but I assigned a fruit biomass of 50% to each

62 individual to account for the fact that only approximately half of the individuals bear fruit; I also assumed a balanced sex ratio for dioecious tree species (Bawa & Opler, 1975;

Glander, 1975). This method no doubt results in an overestimate of fruit availability because even when a species is fruiting not all adults produce a fruit crop, but overestimates are consistent and they may help to compensate for the many species whose reproductive phenology I was unable to estimate.

Chapter 4: Edge Effects in a Community of Neotropical Primates

Introduction

In ecology the term edge effect traditionally refers to an increase in the number of species where two habitats intergrade (Leopold, 1933), but over the last several decades the term has taken on a new meaning. Today it is most commonly used to describe the "changes occurring in previously undisturbed forest by the abrupt creation

(due to deforestation for pasture) of a very sharp edge at a reserve margin" (Lovejoy, et al., 1986, p. 258). Over the past several decades, researchers have built a body of knowledge about the impacts of edge effects on abiotic conditions as well as on flora and fauna (Laurance, et al., 2002; Laurance, et al., 2007; Laurance, et al., 2011; Murcia,

1995). One of the interesting questions in edge effect research is the magnitude of the edge penetration distance, which is the distance from the edge over which edge-related changes are detectable. This question, which I test in a community of Neotropical primates, is important because it allows an estimation of how much habitat is essentially lost for species that are not edge tolerant. The study of edge effects is especially relevant for conservation in the Brazilian Amazon where wildfires and forest clearance for cattle ranching, soy production, swidden agriculture, and logging (Asner, et al., 2005; Fearnside, 2001; Laurance, et al., 2001b; Laurance, et al., 2004) create up to

50,000 km of new forest edge each year (Broadbent, et al., 2008).

It is only recently that scientists have begun to study the impacts of edge effects on large forest fauna. In primatological research the influence of the edge has been examined in several ways, including defining the edge as a narrow band of forest

(Mittermeier & van Roosmalen, 1981; Mittermeier & van Roosmalen, 1983) and referring to all sightings of individuals seen from clearings or large trails/roads as edge habitat (Gavazzi, et al., 2008). Other researchers have studied the edge more systematically, for example examining the impact of the ratio of a fragment's edge to its interior (Mbora & Meikle, 2004). At the Budongo Forest Reserve, Uganda, Tweheyo et al. (2004) compared chimpanzee density, behavior, and ecology in the outer 150 m of a fragment (the edge) to the interior and found chimpanzee abundance to be higher in edge areas. In Madagascar, Quéméré et al. (2010) found no edge effect in the distribution of the golden-crowned sifaka (Propithecus tattersalli). A group of studies by

Lehman et al. (Lehman, et al., 2006b; Lehman, et al., 2006c; Lehman, et al., 2006d;

Lehman, 2007), also in Madagascar, employed varied methodologies to examine lemur spatial responses to the edge and the depth to which those responses are discernible.

These authors found that individual species exhibited different edge responses and that these responses do not always vary monotonically with distance to the edge. They offer several explanations for these patterns, including: food abundance and quality, temperature gradients (which can impact torpor), predation pressure, and liana abundance. A few other systematic studies have been completed, but thus far their results have only been presented at academic conferences (McGoogan, et al., 2009;

Steffens, et al., 2009).

In this study, I test a Neotropical primate community for edge effects in their population density at the Biological Dynamics of Forest Fragments Project (BDFFP) in the

Brazilian Amazon. I consider edge effect penetration depths in 150 m intervals from the forest edge. I do not anticipate detecting edge effects beyond 300 m because in tropical forests the strongest abiotic and plant community edge effects are seen in the first 150 m (Laurance, 2000) with fewer significant changes discernible to around 300 m (Table

4.1). Changes in the plant (tree) community are of the utmost importance to arboreal primates as trees provide, among other things: food (e.g. fruit, seeds, leaves, nectar, and flowers), substrates for locomotion, habitat for vertebrate and invertebrate prey, sleeping sites, and protection from predators. As a result, the magnitude of documented edge-related changes in the plant community ought to be an effective means to estimate the depth to which primates are able to perceive and respond to edge effects. I include two covariates, monthly fruit availability and the number of large trees (dbh≥60 cm; a proxy for canopy height), in the regression models used to examine the importance of edge and interior habitats in the density of each primate species.

Table 4.1: Examples of abiotic & floral edge effects in tropical forests

Variable Edge Penetration (m) Higher (+) or Lower (-) at Edge ABIOTIC Air vapor pressure deficita 20 + Air temperaturea 15; 20 + Light (PAR) a,b 20 + Soil moisturea 40 - Relative humidityc 100 - FLORAL Tree (stem) densityd 20 + Understory foliage densitye 35 + Number of treefall gaps (severe change) f 40 + Tree damage (from wind & trees) g 60 + Canopy foliage densitye 60 - Greatest height of foliage densityh 100 - Recruitment of disturbance-adapted treesi 100 + Canopy heightj 100 - Canopy damagek 150 + Number of treefall gapsf,l 160; 190 + Tree mortalitym,n 300 + Tree mortality (≥60 cm dbh) o 300 + Wind disturbancep 400 + a (Kapos, 1989); b (Williams-Linera, 1990b); c (Lovejoy, et al., 1986); d (Williams-Linera, 1993); e (Malcolm, 1994); f (Kapos, et al., 1993); g (Lovejoy, et al., 1984); h (Camargo & Kapos, 1995); i (Laurance, et al., 1998a); j (Camargo, 1993); k (Laurance, 1991); l (Wandelli, 1991); m (Laurance, et al., 1998b); n (Laurance, 2000); o (Laurance, et al., 2000); p (Lewis, 1998)

In addition to looking at the impact of the edge on primate densities, I also consider whether there are primate traits (home range size, percentage of fruit in the diet, weight, and group size) that predict primate presence in edge and interior habitats, a question which Boyle & Smith (2010b) also examined in the habitat fragments at the study site. This question is important because it provides insight on the suite of characteristics that make primates more (or less) adaptable to forest edges. I also consider whether or not edge effects in predation pressure shape primate responses to habitat edges at the BDFFP, a possibility that has been suggested in previous primate edge effect studies (Lehman, et al., 2006b; Lehman, et al., 2006c).

The forest edge in this study is the boundary between primary terra firme forest and 19-30 year-old secondary forests that are regenerating on areas that were clear cut for cattle pasture. I do not sample any edges that abut active pastures and, as a result, many of the microclimatic edge effects that immediately follow edge creation are likely reduced or absent because forest edges are sealed by a proliferation of vines and secondary growth over time (Camargo & Kapos, 1995; Denyer, et al., 2006; Didham &

Lawton, 1999; Kapos, et al., 1993). However, edge effects like desiccation stress, wind shear, and wind turbulence increase tree mortality and damage (Laurance, et al., 1997;

Laurance, et al., 1998b) and result in significant changes in the composition of the mature tree (Laurance, et al., 2000; Laurance, et al., 2006a; Laurance, et al., 2006b) and liana communities (Laurance, et al., 2001a). Edge effects also lead to much higher rates of tree mortality and damage within 100-300 m of the edge (Ferreira & Laurance, 1997;

Laurance, et al., 1998b) while large trees over 60 cm dbh perish three times faster near edges than they do in forest interiors (Laurance, et al., 2000). These effects are long- lasting and do not immediately disappear when the edge is sealed. Indeed, they may even become more severe following edge closure (Laurance, et al., 2011) as the winds that act on the first 60-370 m (and perhaps as much as 550 m) of the forest edge at the

BDFFP (Bierregaard Jr. & Stouffer, 1997; Reville, et al., 1990; Savill, 1983; Somerville,

1980) may actually become stronger as the forest edge is sealed, weakening only when the canopy of the regenerating forest is approximately the same height as that of the mature forest (Savill, 1983). As a result, while the habitat edges in this study border regenerating forests rather than active pastures, the plant community at the forest edge

68 has been accumulating, and continues to accumulate, significant edge effects that may in turn impact the primate community.

I predict that the percentage of fruit in the diet will be negatively related to edge use. Edges are attractive to folivores due to increased leaf production and a higher percentage of poorly defended leaves (discussed below) and unattractive to frugivores, particularly those that rely on large trees, due to edge-related increases in tree mortality

(discussed below). Body weight may also be negatively related to primate presence because damaged edge forests have discontinuous canopies and denser understories

(Table 4.1) that should facilitate movement by smaller species and make it more difficult for larger species to move between the crowns of big trees. I do not think that home range size or group size will be relevant because the edge areas under consideration are part of large tracts of primary forest inhabited by all six primate species and therefore area requirements and average group sizes should not to be factors. I also predict that none of the primate characteristics will be related to primate presence in the forest interior because I believe that most, if not all, species will be encountered on each of the forest interior transects.

Both Alouatta and Pithecia should be more abundant at the forest edge. The highly folivorous Alouatta (Table 4.2) will favor edges because they have two characteristics that make them attractive to folivorous primates: a proliferation of successional tree species (e.g. Cecropia spp. and Croton spp.) with poorly defended foliage (Coley, 1980) and an increase in the production of poorly defended new leaves in

69 mature forest species due to the increase in sunlight (Chen, et al., 1992; Coley, 1980;

Sork, 1983; Williams-Linera, 1990a). Alouatta groups are also found in 1 ha fragments at the BDFFP (Boyle, et al., In Press), habitats that are so small (100 m x 100 m) that they are 100% edge-affected, further suggesting a preference for these habitats.

Table 4.2: Characteristics of the BDFFP primates (Boyle & Smith, 2010b)

% of Diet Primate Weight (kg)a Group size Home range (ha) Fruit Flower Leaf Otherb Alouattac,d,e,f,g 6.8 8.2 53 40.4 6.4 48.1 5.1 Atelesh,I,j,k,l,m,n,o 7.8 14.3 224 88.7 3.8 6.1 1.4 Chiropotesi,p,q,r,s 2.8 21.8 336 87.9 5.7 5.8 0.6 Pitheciap,t,u,v,w,x 1.7 3.4 103 86.4 2.1 8.7 2.8 Saguinusj,p,y,z,A 0.5 5.7 33 66.0 1.9 0.8 31.3 Sapajush,j,m,p,B,C,D,E 2.6 14.3 429 65.0 1.4 2.7 30.9 a (Ford & Davis, 1992); b Consists mainly of insects, exudates, and small vertebrates; c (Gaulin & Gaulin, 1982); d (Julliot & Sabatier, 1993); e (Julliot, 1996a); f (Palacios & Rodriguez, 2001); g (Sekulic, 1982); h (Guillotin, et al., 1994); i (Kinzey & Norconk, 1990); j (Mittermeier & van Roosmalen, 1981); k (Norconk & Kinzey, 1994); l (Simmen, 1992); m (Simmen & Sabatier, 1996); n (Symington, 1988); o (van Roosmalen, 1985); p (Norconk, et al., 2003); q (Ayres, 1981); r (Boyle, et al., 2009); s (van Roosmalen, et al., 1981); t (Kinzey & Norconk, 1993); u (Lehman, et al., 2001); v (Norconk, 1996); w (Oliveira, et al., 1985); x (Vié, et al., 2001); y (Day & Elwood, 1999); z (Kessler, 1995); A (Oliveira & Ferrari, 2000); B (Izawa, 1980); C (Peres, 1993a); D (Spironello, 2001); E (Zhang, 1995b)

I predict that Pithecia, a small-bodied frugivorous seed predator (Table 4.2), will also prefer the edge because previous work at the site found a correlation between their presence and increasing structural complexity of the forest fragments they occupy

(Schwarzkopf & Rylands, 1989). More specifically, they prefer areas with many small trees and lianas and that have few trees with dbh>10 cm. In essence, they select habitats similar to what is expected at the forest edge (Table 4.1). Pithecia may also be well-suited for life at the edge because they tend to use low and mid-forest vertical strata (Chapter 5), focus on habitats with few emergent trees (Walker, 1996), and they

70 are known to utilize both primary and secondary forests (Chapter 6; Kinzey, et al., 1988;

Mittermeier, 1977; Wolfheim, 1983).

It seems likely that only one species, the large-bodied, highly frugivorous Ateles

(Table 4.2), will not utilize edge habitats. I predict a negative edge response due to their preference for undisturbed primary forest and the observation that they avoid edge areas at other sites (Lehman, 2004; Mittermeier & van Roosmalen, 1981). In addition, they only rarely use even the largest (100 ha) fragments at the study site (Boyle, et al.,

In Press), suggesting that spider monkeys do not prosper in habitats under the influence of habitat edges. Edge effects also reduce the abundance of the large trees upon whose fruit spider monkeys rely (Table 4.1) (Simmen & Sabatier, 1996; van Roosmalen, 1985), thereby making edge areas significantly less attractive.

I expect that three species will have equal abundances in edge and interior habitats: Chiropotes, Saguinus, and Sapajus. The behavior of Chiropotes, a medium-sized frugivorous seed predator (Table 4.2), is difficult to predict. Bearded saki monkeys are traditionally thought to prefer primary forest habitats (Ayers, 1989; Johns & Ayers,

1987; Mittermeier, 1977; Wolfheim, 1983), but previous work at the site (Boyle, et al.,

2009) found groups frequently using edge-affected habitat fragments and sometimes leaving them to move into secondary forests. Boyle et al. (2009) found that small groups that use 10 ha fragments utilize almost the entire fragment, suggesting at least a tolerance of edge habitats, but she also found that they avoid using some edge areas in larger 100 ha fragments where the larger area available may make it easier to do so.

Chiropotes is also known to prefer feeding in the canopy, which suggests a negative edge response, but travel in the understory (van Roosmalen, et al., 1981; Walker, 1993), which suggests a possible positive edge response due to the proliferation of understory vegetation at the edge (Table 4.1).

Saguinus and Sapajus are medium- and small-bodied species with generalist diets (Table 4.2) that utilize both primary and secondary forests (Eisenberg, 1989;

Hershkovitz, 1977). As a result, I hypothesize that they will view the edge as a transitional zone between two habitats that they frequent rather than focusing on or avoiding it. The generalist diets of both species will help to guard against potential declines in fruit abundance at the edge due to the edge-related loss of mature trees

(Table 4.1) because a potential decline in fruit may be compensated for by positive edge-related changes in the abundance of some insects (Fowler, et al., 1993). In addition, several studies have found that Sapajus spp. (Aldana, et al., 2008; Chapter 6;

Parry, et al., 2007; Vulinec, et al., 2006) and several callitrichids (Branch, 1983; Chapter

6; Parry, et al., 2007; Vulinec, et al., 2006) can be found at higher abundances in secondary forest than in primary forest, suggesting that the edge effects that cause mature forest edges to be invaded by secondary forest species likely will not be a significant deterrent to either Saguinus or Sapajus. Sapajus has also been found to have huge home ranges at the BDFFP (852-915 ha) (Spironello, 2001) which suggests that they are less likely able to focus on the narrow forest edge given their area requirements.

Methods

STUDY SITE

The BDFFP is a long-term, experimental study of the impacts of habitat fragmentation in a Neotropical rain forest. It is located in the central Amazon 80 km north of Manaus,

Brazil (2°25´S, 59° 50´ W) (Didham, 1997). The forest is tall terra firme tropical rain forest with a closed canopy between 30-37 m and emergents to 55 m (Bierregaard Jr. &

Stouffer, 1997). The entire site is 1,000 km2 and includes three large, isolated cattle ranches, Fazendas Dimona, Porto Alegre, and Esteio (3,000-4,000 ha each), and the continuous forest around them (Laurance, et al., 2007). This study was conducted at

Fazendas Dimona and Porto Alegre. The site consists of undisturbed primary forest, habitat fragments in three size classes (1, 10, and 100 ha), active pasture, abandoned/regenerating pasture, and selectively logged forest. The majority of the pastures have been abandoned and now contain regenerating forests, including those which meet the edge of the primary forests sampled in this study. Annual rainfall at the site ranges from 1,900-2,500 mm (Stouffer & Bierregaard Jr., 1993) and averaged 2194 mm during this study (Instituto Nacional de Pesquisas Espaciais, 2012); over the last 106 years the city of Manaus just to the south has received an average of 2115 mm

(Cochrane, 2011). There is a pronounced dry season from June to September (Didham,

1997). Average monthly temperatures range from 25.8⁰C in February-April to 27.9⁰C in

September (Salati, 1985). While the Manaus region, and much of the Amazon Basin in general, are considered lacking in large-scale topography, there are numerous steep

73 hills cut by current and past streams (igarapés) even though there is no significant large- scale relief. Indeed, the elevation of the site ranges from 50 to 150 m (Bruna & Kress,

2002). The soil is predominantly nutrient-poor yellow latosol (Rylands & Keuroghlian,

1988).

The primate community is composed of six species: black-bearded saki

(Chiropotes chiropotes) (Veiga, et al., 2008)1, golden-faced saki (Pithecia chrysocephala)

(Marsh, In press)2, golden-handed tamarin (Saguinus midas) (Mittermeier, et al., 2008b),

Guiana spider monkey (Ateles paniscus) (Mittermeier, et al., 2008a), Guianan brown capuchin (Sapajus apella apella) (Rylands, et al., 2008), and Guianan red howler monkey

(Alouatta macconnelli) (Boubli, et al., 2008). I follow the taxonomy used by the IUCN

Redlist (IUCN, 2012), though I place the brown capuchin in the genus Sapajus rather than Cebus (Lynch Alfaro, et al., 2012a ; Lynch Alfaro, et al., 2012b; Silva Jr., 2001; Silva

Jr., 2002). Hereafter each species will be referred to by genus alone.

DATA COLLECTION

Between November-December 2008, June-November 2009, and January-June 2010 I surveyed a total of 209.35 km over five primary forest transects: 1) 42 replicates*1.05 km, 2) 38*1.1 km, 3) 37*0.85 km, 4) 46*1.0 km, 5) 46*1.0 km) and a total of 118.04 km over two very lightly (0.48 trees ha-1, Chapter 6) selectively logged primary forest

74 transects: 6) 58*1.03 km, 7) 55*1.06 km (Figure 4.1). Vegetation sampling along the primary forest transects yielded a total sample area of 2.002 ha with 1639 stems ≥10 cm dbh from 425 species while the selectively logged forest had a total area sampled of

0.836 ha with 657 stems ≥10 cm dbh from 282 species. The species that was cut from these lightly-logged forests was likely Manilkara huberi (Chapter 2).

Figure 4.1: Transect locations

Edge responses are defined as positive, negative, or neutral (Lehman, et al.,

2006c; Reis, et al., 2003). An increase in primate density with increasing edge proximity is considered a positive edge effect, a decrease in density with decreasing edge proximity is considered negative, and a density pattern that has no apparent relationship with edge proximity is considered neutral (Figure 4.2).

Figure 4.2: Types of edge responses (Reis, et al., 2003)

PRIMATE SAMPLING

I collected data by repeatedly sampling seven line transects, each of which was perpendicular to the primary forest edge (Lehman, et al., 2006b; Lehman, et al., 2006c), and between 950-1100 m in length (Figure 4.1). Transects were flagged every 10 m and were separated by 0.575-15.0 km, spacing was not equal due to landscape limitations such as the shape of the cattle ranches and the matrix type. I walked transects beginning at 06:45 and 13:15, avoiding mid-day sampling when possible because primates tend to be less active at this time. There was a minimum of 6.25 hrs between resampling, which is more than the 3 hr that have been suggested as a sufficient rest period to allow groups to redistribute themselves (Peres, 1999). Two observers, 5-10 m apart and with the author always in the lead for consistency, slowly walked transects

(average speed=1.01 km/hr) while pausing for 7-10 seconds every 10 m to look and listen. Upon sighting a group, I used a Nikon Forestry 550 laser rangefinder to measure the sighting distance and height of the first individual and a Suunto MC-2G global compass to measure the sighting angle. I also recorded our exact location on the

76 transect, counted the number of individuals, and estimated group spread along two perpendicular axes, always including the longest axis.

BOTANICAL SAMPLING

Trees ≥10 cm dbh (diameter at breast height, or 1.3 m) were identified, measured, and mapped in 10-m intervals to a depth of 2 m on both sides of each transect.

Identifications, following the taxonomy of The Plant List (2010), were done in the field by a top expert on Amazonian trees, Paulo Apóstolo Assunção (da S. Ribeiro, et al.,

1999). Prior to tree identification I used previous research to create a database of the fruiting tree species consumed by each primate (Andresen, 2000; Boyle, 2008;

Buchanan, et al., 1981; Egler, 1986; Fragazy, et al., 2004; Frazão, 1992; Gaulin, 1977;

Gaulin & Gaulin, 1982; Guillotin, et al., 1994; Julliot & Sabatier, 1993; Mittermeier, 1977;

Neves & Rylands, 1991; Norconk, 1996; Norconk & Conklin-Brittain, 2004; Oliveira &

Ferrari, 2008; Pack, et al., 1999; Port-Carvalho & Ferrari, 2004; Santamaría-Gómez,

1999; Setz, 1993; Silva & Ferrari, 2007; van Roosmalen, et al., 1981; Veracini, 2000;

Wright, 2004) (note: diet data for Saguinus midas were so limited that I also included data from the closely related S. niger). I used this database, along with Assunção's knowledge, to tag the most important fruiting trees. I then collected monthly data on the reproductive phenology of the tagged individuals (presence/absence of buds, flowers, immature fruit, and ripe fruit).

ANALYSES & STATISTICS

Fruit Availability

I constructed a reproductive phenology database using the data collected during this study as well as data from other studies at the BDFFP and the Manaus region (Alencar,

1998; Bentos, et al., 2008; Boyle, 2008; Camargo, et al., 2008; Lemos, 1998; Oliveira,

1997). To estimate the monthly fruit production of adult trees (≥10 cm dbh) I used an allometric equation (Peters, et al., 1988) that has been used in previous studies of primate ecology (Chapman, 1990a; Sorensen & Fedigan, 2000):

Fr=47dbh1.9 where Fr is the mass of fruit and dbh is the dbh of the tree. I then combined the reproductive phenology database, tree identifications and measurements, and the primate diet database to calculate monthly fruit availability for each primate species. I used the scientific literature to determine the reproductive system of each tree species

(or genus when species-level information was not available). I made the generous assumptions that 100% of the individuals in monoecious and hermaphroditic taxa produced fruit and that all individuals of dioecious taxa also produced fruit, but for the latter I assigned a fruit biomass of 50% to each individual to account for the fact that only approximately half of the individuals are capable of bearing fruit.

Primate Density Calculations

Density is calculated as the number of individuals in a given area. To determine the area sampled, I estimated species-specific transect widths using the perpendicular distance to the first individual sighted and a 50% criterion for fall-off distance (Whitesides, et al.,

1988). The fall-off distance is the maximum distance from the transect in which it is assumed that 100% of the individuals present were detected. The data for perpendicular distance estimation come transects in undisturbed primary forest (five), selectively logged forest (two), and secondary forests on unburned pastures (three), as well as from additional sampling I conducted in other mature forests (10 ha and 100 ha fragments (one each)) and single transects in two age classes of secondary forest regenerating on burned pastures (19- and 25-30-years old). This bolsters the number of sightings and allows me to obtain the best possible fall-off distance estimates, though before combining sightings from multiple habitats I tested for differences using Kruskal

Wallis and Mann Whitney tests (see Statistics).

I followed the area sampled and density formulas of Whitesides et al. (1988) except in instances where a transect terminus sampled the same habitat as the rest of the transect (as opposed to a transect that ended at the border between habitats). In this case, I modified Whitesides' area sampled to include the semi-circle sampled beyond the end of the transect by adding the 0.5πR2 term to the original equation:

2 2 Area = 2(S/2+ED)*Lt + 0.5πR = 2(S/2+ED)*Lt + 0.5π(S/2+ED)

with ED (Effective Distance) = (Nt/Nf)*(FD)

where T=total width sampled, Lt=total length sampled, Wt=species-specific transect width on one side of the transect, S=estimate of species-typical mean group spread

(collected during transect and ad libitum sightings), Nt=species-specific total number of group sightings; Nf=species-specific number of group sightings at distances less than half the fall-off distance; FD=species specific fall-off distance (Whitesides, et al., 1988); and R

(radius)=species-specific effective distance plus one half the average group spread.

Statistics

Analyses for primate presence in edges and interiors followed Boyle and Smith (2010b).

I used a binomial generalized linear model with a logit link and arcsine transformed the percent of fruit in the diet. Each transect was scored 1/0 for the presence/absence of each primate species based on whether or not each was observed over all transect samples; the primate characteristics of interest entered the model as covariates.

Covariates in the model are largely identical to those used by Boyle and Smith (2010b):

80 home range size, percentage of fruit in the diet, body weight, and group size (Table

4.2).However, I do not consider seeds as they found this variable was not a factor (Boyle

& Smith, 2010b).

I used Kruskal Wallis to test for differences in the species-specific perpendicular distance to the transect among habitats to determine if I could combine sightings from multiple habitats in transect width estimation. To be certain that I was justified in combining data, I also reclassified the forest types into two broader habitat types

(secondary growth, and mature forest, with the latter including primary forest, selectively logged forest, and habitat fragments) and conducted Mann Whitney tests for interhabitat differences.

Primate surveys yield abundance (count) data and can therefore be modeled with generalized linear models utilizing the Poisson or negative binomial distributions.

Count data are frequently over-dispersed, which was the case in this study due to an excess of zeroes that is common in ecological sampling (Zuur, et al., 2009). To deal with this issue, I used negative binomial models because they often outperform other methods of handling over-dispersion (Warton, 2005). I used negative binomial generalized estimating equations (GEE) with a log link to account for repeated measures and included the natural log of the area sampled as an offset to treat counts as densities and allow for intertransect differences in area sampled. I use QIC to determine the best correlation matrix and a model-based estimator because I had fewer than 20 clusters

(Garson, 2012). For all GEEs I report the full main effects model to avoid the many issues that arise with the use of stepwise methods (Babyak, 2004).

To test for interhabitat density differences for each primate species I first used

GEE planned comparisons, with the least significant difference test, testing only the impact of habitat. I ran the habitat-only model because it isolated the interhabitat differences (i.e. this test does not adjust for covariates). I then ran a full effects GEE model with habitat, average dbh, and fruit availability to determine the relative influences of all covariates on abundance. I calculated the standard deviation for each population density estimate as the weighted standard deviation (where the weight is the area sampled per transect in km2 because transect lengths varied) with each transect survey considered to be a separate replicate.

To determine if there are edge effects in the distribution of each primate species

I first tested to see if data from undisturbed and selectively logged primary forests could be combined for each primate species. I also examined differences in the number of stems ≥10 cm dbh/ha (Mann-Whitney U Test), the average dbh (Independent Samples

T-test), and the density of each primate species in each habitat (Chapter 6). In order for transects to be combined, all three tests had to find no significant differences. To enlighten the discussion, I also tested for edge vs. interior differences in the number of large trees ha-1 with the Mann-Whitney U Test. When I determined how to group the data, I used a GEE, as above, to test the impact of three habitats (0-150 m from the edge, 150-300 m from the edge, and the interior (meter 300 to the end of each transect,

82 ranging from 850 to 1100 m)) on the density of each primate to determine edge penetration depth. When all three habitats differed, I added successive 150 m increments from the edge and retested until the farthest increment from the edge was equal to the forest interior. When all three habitats were initially equal I stopped and considered edge effects to be neutral. When the density in 0-150 m differed from the interior but density in 150-300 m did not, I lumped the 150-300 m category with the interior and considered edge effects to penetrate to a depth of 150 m. I then proceeded to test hypotheses using 0-150 m as the edge, comparing it to the rest of the transect

(i.e. the forest interior).

After determining the edge depth for each primate species (all were 150 m or had neutral edge effects), I ran a full main effects GEE to determine whether the number of large trees or fruit availability also influenced primate distribution. I tested for multicollinearity among all predictors using a conservative variance inflation factor

(VIF) threshold of VIF≤4.0 (Chatterjee, et al., 2000); no covariates required removal.

Dummy coding can influence multicollinearity statistics when at least one variable in categorical, so I ran multicollinearity tests with different coding schemes to determine the scheme that best reduced multicollinearity (Wissmann, et al., 2007).

All analyses were conducted in SPSS version 19.0.0.1.

Results

PRIMATE PRESENCE/ABSENCE:

I considered primate presence in edge habitats (0-150 m) vs. forest interior because overall that was found to be the most common edge depth (see below). I consider the edge and interior separately. At the forest edge the overall model was not significant (2=7.883, df=4, p=0.096), but the percent of fruit in the diet significantly predicted species richness (Intercept: =5.880, df=1, p=0.023; Fruit: =-5.698, df=1, p=0.022; Weight: =-0.238, df=1, p=0.204; Group size: =0.174, df=1, p=0.134; Home range size: =-0.005, df=1, p=0.299). An edge model with fruit as the only covariate was significant overall (2=5.058, df=1, p=0.025) as was the fruit term (Intercept: =6.199, df=1, p=0.013; Fruit: =-2.781, df=1, p=0.032). In the forest interior neither the overall model (2=3.739, df=4, p=0.443) nor any of the predictors was significant (Intercept:

=5.077, df=1, p=0.131; Fruit: =-2.111, df=1, p=0.517; Weight: =-0.021 df=1, p=0.916;

Group size: =-0.111, df=1, p=0.411; Home range size: <0.001, df=1, p=0.958).

EDGE EFFECTS:

Kruskal Wallis tests on the average perpendicular distance to the transect for each primate species over all habitats detected no differences (Alouatta: N=55 in 4 habitats,

2=5.144, df=3, p=0.162; Ateles: N=22 in 3 habitats, 2=1.790, df=2, p=0.409; Sapajus:

N=59 in 6 habitats, 2=1.847, df=5, p=0.870; Chiropotes: N=40 in 5 habitats, 2=4.082, df=4, p=0.395; Pithecia: N=21 in 5 habitats, 2=5.114, df=4, p=0.276; Saguinus: N=75 in 6

habitats, 2=8.835, df=5, p=0.116). Mann Whitney tests for differences between secondary growth (SG) and mature forest (MF) revealed that data for Alouatta needed to be split but that all other species could still be combined (Alouatta: N=53 (MF) 2 (SG),

Z=-2.025, p=0.034); Ateles: N=22 (MF) 0 (SG), no test; Sapajus: N=36 (MF) 23 (SG), Z=-

0.054, p=0.975); Chiropotes: N=36 (MF) 4 (SG), Z=-1.782, p=0.074); Pithecia: N=16 (MF)

5 (SG), Z=0.186, p=0.208); Saguinus: N=37 (MF) 37 (SG), Z=-1.109, p=0.282)).

In the following regression results the covariate "Habitat" refers to the influence of edge and interior habitats on the dependent variable. The dendrometric data suggested that I could combine undisturbed primary forest and logged primary forest transects (average dbh: N=2274, t=-0.122, p=0.903; stems ha-1: N=7; mean rank: primary=4.40, logged=3.00; U=3.000; Z=-0.775; p=0.571). However, there were significant interhabitat differences in population density for all primates other than

Ateles (Chapter 6), so I examine data from only the five primary forest transects for

Alouatta, Sapajus, Chiropotes, Pithecia, and Saguinus while lumping data collected on primary forest (N=5) and logged primary forest (N=2) transects for Ateles.

Tests of the relationship between fruit availability (a GEE covariate) and the density of each primate species are listed below under the heading for each primate species, but the covariate of the number of large trees only needs to be reported once because it is the same for all species. There is a non-significant difference in the number of large trees (dbh≥60 cm) in undisturbed primary forest edge (0-150 m) vs. interior habitats (N=10; mean rank: edge=3.80, interior=7.20; U=21.000; Z=1.803; p=0.095). A

85 larger sample might show a significant difference as other work has found that edges contain fewer large trees (Laurance, et al., 2000). Indeed, the same comparison for undisturbed primary forest and logged forest combined, which is relevant for Ateles, shows significantly fewer large trees at the edge than in the interior (N=14; mean rank: edge=4.86, interior=10.14; U=43.000; Z=2.446; p=0.017).

The differences in primate density between edge and interior habitats are summarized in (Table 4.3) and reported fully after it. The table summarizes both the tests between 0-150 m vs. 150-300 m vs. forest interior as well as the tests solely between 0-150 m vs. the interior. Table 4.3 contains the true means and weighted standard deviations while in the text I report the estimated marginal means and standard errors used in the statistical analyses.

Table 4.3: Density differences (# individuals km2-1) by edge zone (, )

Habitat Alouatta Ateles Chiropotes Pithecia Saguinus Sapajus 0-150 11.1 (75.0)a 0.4 (8.5)c 15.3 (132.8)h,i 2.5 (46.0) 10.5 (89.2)k 18.6 (134.1)e,f 150-300 6.6 (59.0) 1.1 (17.5) 4.0 (65.3)h 1.6 (30.1) 4.9 (64.3)k 6.0 (68.4)e Interior 6.6 (23.3)a 1.3 (7.7)c 3.2 (18.7)i 2.4 (14.5) 5.9 (26.2) 6.4 (29.2)f 0-150 11.1 (75.0)b 0.4 (8.5)d 15.3 (132.8)j 2.5 (46.0) 10.5 (89.2)l 18.6 (134.1)g Interior 6.6 (20.7)b 1.3 (7.1)d 3.3 (18.0)j 2.2 (12.8) 5.7 (23.2)l 6.3 (26.4)g Letters denote significant differences (see text)

Alouatta

0-150 m vs. 150-300 m vs. Interior: Planned comparisons following the habitat only GEE showed density differences between 0-150 m (=11.14, SE=2.061) and the interior

(=6.55, SE=0.805) (p=0.038) but not between 0-150 m and 150-300 m (=6.57,

SE=1.537) (p=0.075) or between 150-300 m and the interior (p=0.994) (Table 4.3). I

86 therefore combined the 150-300 m and forest interior zones to examine a forest edge located at 0-150 m from the edge vs. the forest interior.

0-150 m vs. Interior: Planned comparisons following the habitat only GEE showed density differences between 0-150 m (=11.14, SE=2.061) and the interior (=6.55,

SE=0.757) (p=0.036), a positive edge effect (Table 4.3). In the full model, habitat and fruit availability were significant while the number of large trees ha-1 was not (Intercept:

=0.191, df=1, p=0.002; Habitat (edge): =1.484, df=1, p<0.001; Fruit: =0.001, df=1, p<0.001; # large trees: =0.009, df=1, p=0.448).

Ateles

0-150 m vs. 150-300 m vs. Interior: Planned comparisons following the habitat only GEE showed density differences between 0-150 m (=0.40, SE=0.246) and the interior

(=1.36, SE=0.212) (p=0.003) but not between 0-150 m and 150-300 m (=1.06,

SE=0.402) (p=0.162) or between 150-300 m and the interior (p=0.508) (Table 4.3). I therefore combined the 150-300 m and forest interior zones to examine the edge of 0-

150 m vs. the forest interior.

0-150 m vs. Interior: Planned comparisons following the habitat only GEE showed density differences between 0-150 m (=0.40, SE=0.247) and the interior (=1.33,

SE=0.194) (p=0.003), a negative edge effect (Table 4.3). In the full model, the number of large trees ha-1 and fruit availability were significant while habitat was not (Intercept:

=-3.140, df=1, p<0.001; Habitat (edge): =0.880, df=1, p=0.286; Fruit: =0.001, df=1, p=0.001; # large trees: =0.105, df=1, p<0.001).

Chiropotes

0-150 m vs. 150-300 m vs. Interior: Planned comparisons following the habitat only GEE showed density differences between 0-150 m (=15.29, SE=2.123) and the interior

(=3.18, SE=0.436) (p<0.001) and between 0-150 m and 150-300 m (=4.00, SE=0.981)

(p<0.001) but no differences between 150-300 m and the interior (p=0.446) (Table 4.3).

As a result, I combined the 150-300 m and forest interior zones to examine an edge of 0-

150 m vs. the forest interior.

0-150 m vs. Interior: Planned comparisons following the habitat only GEE showed density differences between 0-150 m (=15.29, SE=2.123) and the interior (=3.29,

SE=0.417) (p<0.001), a positive edge effect (Table 4.3). In the full model, habitat and the number of large trees ha-1 were significant while fruit availability was not (Intercept: =-

0.872, df=1, p=0.088; Habitat (edge): =2.697, df=1, p<0.001; Fruit: <0.001, df=1, p=0.053; # large trees: =0.077, df=1, p<0.001).

Pithecia

0-150 m vs. 150-300 m vs. Interior: Planned comparisons following the habitat only GEE showed no density differences between 0-150 m (=2.52, SE=0.902) and the interior

(=2.40, SE=0.420) (p=0.898), between 0-150 m and 150-300 m (=1.58, SE=0.710)

(p=0.410), or between 150-300 m and the interior (p=0.321) (Table 4.3). I did not analyze data for Pithecia further because there is a neutral edge effect.

Saguinus

0-150 m vs. 150-300 m vs. Interior: Planned comparisons following the habitat only GEE showed density differences between 0-150 m (=10.49, SE=2.212) and 150-300 m

(=4.86, SE=1.458) (p=0.034) but not between 0-150 m and the interior (=5.91,

SE=0.829) (p=0.053) or between 150-300 m and the interior (p=0.529) (Table 4.3). I therefore combined the 150-300 m and forest interior zones to examine an edge of 0-

150 m vs. the forest interior.

0-150 m vs. Interior: Planned comparisons following the habitat only GEE showed density differences between 0-150 m (=10.49, SE=2.239) and the interior (=5.72,

SE=0.768) (p=0.044), a positive edge effect (Table 4.3). In the full model, habitat, fruit availability, and the number of large trees ha-1 were all significant (Intercept: =-0.039, df=1, p<0.001; Habitat (edge): =2.023, df=1, p<0.001; Fruit: =0.003, df=1, p<0.001; # large trees: =0.027, df=1, p=0.011).

Sapajus

0-150 m vs. 150-300 m vs. Interior: Planned comparisons following the habitat only GEE showed density differences between 0-150 m (=18.63, SE=1.956) and the interior

(=6.45, SE=0.571) (p<0.001) and between 0-150 m and 150-300 m (=5.99, SE=0.987)

(p<0.001) but no differences between 150-300 m and the interior (p=0.686) (Table 4.3). I therefore combined the 150-300 m and forest interior zones to examine an edge of 0-

150 m vs. the forest interior.

0-150 m vs. Interior: Planned comparisons following the habitat only GEE show density differences between 0-150 m (=18.62, SE=2.086) and the interior (=6.38, SE=0.574)

(p<0.001), a positive edge effect (Table 4.3). In the full model all three covariates predicted Sapajus density (Intercept: =1.064, df=1, p<0.001; Habitat (edge): =1.618, df=1, p<0.001; Fruit: <0.001, df=1, p=0.022; # large trees: =0.020, df=1, p=0.016).

PREDATION:

Evidence of predator habitat use is summarized in Table 4.4. There are two published reports of eagle predation on primates at the BDFFP (Gilbert, 2000; Lenz & dos Reis, 2011) and the site has five eagle species that are large enough to be potential predators of at least one of the primate species: harpy eagle (Harpia harpyja), crested eagle (Morphnus guianensis), black-and-white hawk-eagle (Spizastur melanoleucus), black hawk-eagle (Spizaetus tyrannus), and ornate hawk-eagle (Spizaetus ornatus)

(Cohn-Haft, et al., 1997). In addition, there are two big cats, both of which were encountered during the present study, the jaguar (Panthera onca) and puma (Puma concolor), as well as several species of small cats, including the ocelot (Leopardus pardalis) and jaguarundi (Herpailurus yaguarondi), that pose a threat to some or all of the primates at the site.

Table 4.4: Predator encounters/evidence at the BDFFP

Species Evidence Regenerating Forest Primary Forest* Habitat Fragment (N=number of encounters) Harpy eagle Encounter 3(i) 4(e) 1(i) 1(e)** (Harpia harpyja) Indirect Crested eagle Encounter 1(rd)** (Morphnus guianensis) Indirect Probable eagle Encounter 1(rd) 6(i) 1(e) Indirect Jaguar Encounter 2(rd) 2(i) 1(e) (Panthera onca) Indirect 4(rd) 1(i) Puma Encounter 1 (Puma concolor) Indirect 4(rd) Jaguar/puma Direct 1(e) Indirect 1 1(i) Ocelot/jaguarondi Direct (Leopardus pardalis/ Indirect 2(1 rd) 1(i) Herpailurus yaguarondi) e=edge (0-150m); i=interior; rd=seen on a dirt road surrounded by forest Encounter=visual or audio; Indirect=tracks, scat, claw marks on trees *Includes selectively logged forest; **Probable

Discussion

I was correct in my prediction that none of the primate characteristics would be related to primate presence in the interior because all six species were observed on the majority of the forest interior transects. At the forest edge I expected both the percentage of fruit in the diet and body size to be significant, but only the former ended up being relevant. The percentage of fruit in the diet was very strongly negatively related to primate presence at the edge, showing that species that heavily rely on fruit may struggle to utilize the edge while species that are less reliant on fruit and have more diverse diets will be more likely to thrive. These patterns are driven by the positive edge effects in the three species with the lowest percentage of fruit in their diets,

Alouatta, Saguinus, and Sapajus, and the negative edge response of the species with the

91 highest percentage of fruit in its diet, Ateles (Table 4.2). I expected that forest structure at the edge, like lower canopy height, thicker understory, and increased number of treefall gaps (Table 4.1) would make use of these habitats less appealing to larger species and more appealing to smaller species, but this was not the case. While the largest species at the site, Ateles, avoided the edge, four of the remaining five species ranging from the smallest (Saguinus) to the second largest (Alouatta) experienced positive edge effects.

Five primate species displayed edge effects in their density, Alouatta, Ateles,

Chiropotes, Saguinus, and Sapajus. Edge-related density changes extended 150 m into the forest, the same edge penetration as the strongest changes in the plant community

(Table 4.1). Of these five species, all displayed positive edge effects in their distributions except for Ateles, which had a negative response. In the sixth species, Pithecia, edge response was neutral.

Alouatta showed a positive edge response, a pattern that was strongly explained by habitat type (edge vs. interior) and very weakly by fruit availability. Though it was not tested in this study, it is probable that howler monkeys prefer edge areas (and do well in habitat fragments) as a result of their highly folivorous diets (Table 4.2). Indeed, it may be for this reason that habitat type remained such a strong predictor in the full regression model. Edge areas have several characteristics that make them attractive to arboreal folivores, including higher abundances of foliage that is easier to digest and is more nutritious than interior habitats. For example, there is an elevated number of

92 pioneer trees at the edge due to the increased mortality of large trees and higher treefall rates, and the foliage of these pioneer species tends to have weak chemical defenses (Coley, 1980). In addition, elevated light levels at the forest edge (Kapos, 1989;

Lehman, et al., 2006d; Williams-Linera, 1990b) bolster the production of poorly defended new leaves by mature forest species (Chen, et al., 1992; Coley, 1980; Sork,

1983; Williams-Linera, 1990a) and spur trees to create higher-quality leaves with increased protein concentrations (Ganzhorn, 1995).

Also showing a positive edge response, contrary to my prediction of neutral edge effects, Chiropotes, Saguinus, and Sapajus all preferred the forest edge to the interior.

The positive edge effect observed in Chiropotes is difficult to explain. My findings

(Chapter 6) confirm that Chiropotes prefers undisturbed primary forest habitats (Ayers,

1989; Johns & Ayers, 1987; Mittermeier, 1977; Wolfheim, 1983) but also utilizes secondary growth forests (Boyle, et al., 2009). However, unlike the pattern seen in

Saguinus and Sapajus, the density of Chiropotes is highest in undisturbed primary forest.

Therefore, elevated edge use does not appear to be a case of groups that spend a good deal of time in regenerating forests supplementing those habitats with primary forest edges. The full regression model showed that habitat strongly predicted density while the number of large trees was weakly positively related to density, so there is an unmeasured edge characteristic that is responsible for making them so attractive. The positive edge effect in Chiropotes may instead be due to edge-related changes in the plant community (Table 4.1) and their creation of a denser understory. This could create an ideal habitat structure for Chiropotes, which is known to prefer feeding in the canopy

93 and traveling in the understory (van Roosmalen, et al., 1981; Walker, 1993). A small proportion of the diet of Chiropotes at the site is composed of insects (2.4%; Boyle,

2008) so it is also possible that higher insect abundances at the forest edge (e.g. Fowler, et al., 1993) are also partially responsible for their positive edge response.

I hypothesized that Saguinus and Sapajus would have neutral edge effects because they utilize both primary and secondary forests (Eisenberg, 1989; Hershkovitz,

1977) and as a result I thought they would not view the edge as a distinct habitat but rather as a narrow transition zone. This would especially be the case for Sapajus because they have been found to have huge home ranges at the BDFFP (852-915 ha)

(Spironello, 2001). Indeed, in a landscape where the matrix is denuded (i.e. active pasture or farmland) a prediction of positive edge effects for these two species, both of which consume a fair number of arthropods (included in "other" in Table 4.2), would make sense because some studies have shown increases in insect abundances near tropical forest edges (e.g. Fowler, et al., 1993). In this setting, the tree community at the edge would also still be a potentially attractive mix of primary and secondary species without the nearby secondary forest proving to be more appealing.

At the BDFFP, Saguinus and Sapajus are the only two primate species with their highest densities in regenerating forests (Chapter 6), a fact that goes a long way towards explaining their observed positive edge effect. For both species, habitat type (edge vs. interior) remained a strong predictor in the full regression model while the number of large trees and fruit availability were both very weak predictors, so there is something

94 about edge habitats which I did not measure that makes them exert a strong influence. I believe that this factor is a combination of their proximity to the secondary growth habitats in which Saguinus and Sapajus densities are their highest and the possibility that edges have higher insect abundances (Fowler, et al., 1993). Groups that heavily utilize secondary growth may also supplement secondary growth resources with forest edges because they contain vertebrate or invertebrate food sources not found in secondary growth or because they provide access to canopy water which may not be as readily available in the drier secondary growth.

Ateles was the only species with a negative edge response. This is not a surprise considering spider monkeys are undisturbed primary forest specialists that are thought to avoid edge areas (Lehman, 2004; Mittermeier & van Roosmalen, 1981). In the full regression model, habitat type (edge vs. interior) was no longer a significant predictor, so it seems that Ateles habitat use is mainly explained by canopy height and the number of large trees. Spider monkeys spend most of their time high in the canopy (25-30 m)

(Chapter 5; van Roosmalen & Klein, 1988) and their large body size forces them to travel great distances each day to find enough of the clumped and scarce fruit on which they depend (Collins & Durbach, 2000). This results in unsuitable edge habitats because the elevated number of treefall gaps in these areas (Kapos, et al., 1993; Wandelli, 1991) likely decrease canopy connectivity enough to hinder their rapid locomotion between the large canopy trees on which they specialize. In this study fruit availability had very weak explanatory power, so insufficient fruit production at the edge does not appear to

95 be the reason for the observed negative edge effect in this highly frugivorous species

(Table 4.2).

Pithecia was the only species to exhibit a neutral edge response, though I expected a positive edge effect in their abundance. My prediction was based on previous work at the BDFFP that demonstrated their preference for habitat fragments with high levels of floral structural complexity (many small trees and lianas and few trees with DBH >10 cm) (Schwarzkopf & Rylands, 1989), which I hypothesized would make them similar to the forest edge (Table 4.1). A study in Venezuela adds further evidence that edge habitats might be preferred by Pithecia through its finding that golden-faced sakis tended to use the lower and central portions of trees, focus on areas with few emergent trees, and that the average dbh of their feeding trees and of all trees in their range were small at 14.5 cm and 7.4 cm, respectively (Walker, 1996). The neutral edge response in Pithecia may actually be the result of interspecific competition with the BDFFP's other seed predator, Chiropotes. While the edge characteristics discussed above would seemingly make this habitat attractive to Pithecia, they may choose to under-utilize this habitat to avoid competing with Chiropotes, who displayed a positive edge response. Indeed, the abundances of Chiropotes albinasus and Pithecia irrorata were lower at sites where they were sympatric as compared to sites where each was the lone seed predator so intergeneric competition may be a serious factor (Ferrari, et al., 1999).

My encounters with primate predators (Table 4.4) were spread across regenerating forests and the edges and interiors of both primary forest and habitat fragments. While it is impossible to estimate predator habitat preferences from these data, the fact that encounters with these difficult to detect species were somewhat evenly distributed suggests that there likely is not a major edge effect on predator distribution at the BDFFP. Many of the studies of edge effects in predation pressure are examinations of rates of avian nest predation (many of which do not find edge effects)

(Lahti, 2001) and studies of temperate mesopredators (Cervinka, et al., 2011; Salek, et al., 2010). These studies might not extrapolate to the large mammal and large raptor primate predators whose enormous home ranges prohibit a focus on narrow edge habitats. However, while it may be that anthropogenic edges in and of themselves do not lead to edge effects in predation pressure, the type of matrix, degree of human pressure, and total amount of forested habitat are certainly influential. For example,

Balme et al. (2010) found lower leopard densities near reserve edges due to elevated mortality rates from accidental and intentional human off-take of individuals that spend time outside of reserves. Large African eagles also experience strong population declines at the edge of protected areas and severe declines outside of them where over- grazing likely reduces prey availability (Herremans & Herremans-Tonnoeyr, 2000).

Studies like these, in concert with anecdotal data like those presented here, suggest that large primate predators at sites with significant remaining habitat and low human hunting pressures may not face significant edge effects. The BDFFP is such a site, though hunting may be becoming a more serious issue (B. Lenz personal observation; Laurance

& Luizão, 2007). Primate predators in truly isolated fragments or in areas with high human hunting pressure, on the other hand, can be expected to experience negative edge effects. For example, an isolated 100 ha fragment that does not have any nearby expanse of primary forest is certainly too small to support breeding pairs of harpy eagles

(Harpia harpyja) which have 4,300 ha territories in largely undisturbed habitat and require even more space in disturbed areas (Bierregaard, 1994). In terms of the impacts on primate behavior and ecology, my findings, along with the studies just discussed, suggest any potential edge effect on predation pressure should be negative and result in increased edge attractiveness due to a reduction in predation pressure. However, I question whether primate researchers could easily quantify this impact because of the large scales at which the protected area edge effects on predators discussed above tend to operate (e.g. large African eagles: 0-50 km outside of the protected area and an edge area in the protected habitat of 30 km, Herremans & Herremans-Tonnoeyr, 2000).

One of the main foci of primate research at the BDFFP has been the examination of primate persistence in the habitat fragments. It appears that success in fragments is due largely to fragment size, distance to the nearest mature forest, or a combination of the two (Boyle & Smith, 2010b). It has also been posited that the age and amount of forest regenerating on abandoned pastures, and the arboreal pathways they create between continuous primary forest and the fragments, are highly influential in primate recolonizations of abandoned fragments (Boyle & Smith, 2010b). Alouatta fared the best in fragments of all sizes over all matrix phases, from active pasture to decades-old regenerating forest (see Table 2 in Boyle, et al., In Press), likely because their highly

98 folivorous diets allow them to meet their needs in relatively small areas of a variety of habitats (Gilbert, 2003; Rylands & Keuroghlian, 1988). The next four species to return to the fragments, all within 10 years of fragment creation, were Saguinus, Pithecia,

Sapajus, and Chiropotes (in approximately that order). Ateles has been the last species to return to the fragments and most recolonizations by Ateles consist of single, non- resident individuals making brief forays into the larger fragments before returning to primary forest. It is interesting to note that the general pattern of primate success in, and recolonization of, habitat fragments more or less mirrors the edge effect patterns found in this study. Alouatta, Chiropotes, Pithecia, Saguinus, and Sapajus experience positive or neutral edge effects and returned to the fragments earliest and persist best in them while Ateles, the only species to experience negative edge effects, only infrequently uses habitat fragments to this day. This overlap suggests that edge and fragmentation responses may be driven by the same suite of behavioral and ecological traits. This is logical because large areas of fragments are subjected to edge effects, whose influence is compounded where multiple edges meet at fragment corners

(Laurance, et al., 2011). As a result, in settings where data are only available for either edge or fragmentation responses, one might hypothesize that primate responses to both phenomena are similar.

I would like to close with a consideration of the conservation implications of these findings. While this study's lack of active pasture habitat edges might be viewed as a drawback, in some regards studying edge effects in forests that border regenerating forests may actually prove more useful for conservation purposes. In the conservation

99 of isolated primate habitats we already know that bigger (area), closer (to other mature forests), and more connected (to viable habitats) are better (e.g. Boyle & Smith, 2010b).

What is less understood, however, is how species will respond to the regeneration of the matrix at the edge of their habitats. I have demonstrated that edge effects in primate distribution persist beyond, and may be influenced by, the regeneration of a secondary growth matrix that is suitable habitat for some species. Positive edge responses, whether the matrix consists of regenerating forests or active pastures/farmland, result in the creation of a preferred primate habitat and allow land managers to consider an entire fragment to be habitat for that species. Additionally, edge habitats are in close proximity to, and therefore might encourage at least occasional use of, the regenerating matrix. This would bolster seed dispersal from primary forests into the matrix and aid in the regeneration of a more diverse forest. The seed input that this would provide is crucial because secondary forest tree diversity is determined by seed dispersal from surrounding areas (Chazdon, et al., 2009) because agricultural clearcuts often destroy much of the seed bank (Aide & Cavelier, 1994;

López-Toledo & Martínez-Ramos, 2011; Quintana-Ascencio, et al., 1996; Wijdeven &

Kuzee, 2000). Negative edge responses, particularly those that exist in spite of the presence of regenerating forests, suggest a species with extremely sensitive habitat requirements. These species, like Ateles, should therefore be given special consideration during reserve planning because they avoid the regenerating matrix and only infrequently utilize edge-affected areas of the remnant mature forest. Interestingly, even negative edge effects and an unfavorable matrix habitat were not enough to

100 completely prevent Ateles from traversing both habitats on rare occasions to spend brief periods in a few habitat fragments that are surrounded by a matrix of secondary forest. This is good news for conservationists because it means that in relatively short periods (25-30 yr) following the abandonment of pastures where the secondary forests were completely unmanaged to promote tree diversity, even the most ecologically sensitive primate species can disperse through the landscape to maintain gene flow in spite of rarely using edge and regenerating habitats.

101

Chapter 5: Niche Partitioning in a Community of Neotropical Primates in an Anthropogenic Mosaic Landscape

Introduction

In order to coexist, sympatric species must diverge in their behavior and ecology to reduce interspecific competition, a process commonly referred to as niche partitioning.

Niche partitioning is especially important when considering competition between closely related and/or ecologically similar sympatric species (Fleagle & Mittermeier,

1980; Mittermeier & van Roosmalen, 1981). Primates partition their niches along several lines: diet, vertical stratification, forest type, location within the forest, branch size, ranging behavior, prey capture techniques, and activity patterns (Schreier, et al.,

2009). Vertical stratification is the predominant form of niche partitioning in African and

Asian species while dietary differences are the principal form in the Neotropics and

Madagascar (Schreier, et al., 2009). Examinations of interspecies division of space and resources by primates tend to focus on two of the most prevalent lines of niche partitioning: nutritional (dietary) and physical (macro- or microhabitat). In this study I examine physical niche partitioning in a central Amazonian primate community inhabiting an anthropogenic mosaic landscape consisting of a variety of habitats with differing levels of anthropogenic disturbance.

102

Dietary differences are always important in the way that primate niches diverge, in particular the relative proportion of different food categories (e.g. fruit, leaves, insects, etc.) (Mittermeier & van Roosmalen, 1981; Porter, 2001) as well as the taxa consumed within each food category (Terborgh, 1983). However, the degree of dietary overlap is not constant throughout the year: the lowest and highest degrees of overlap occur at times of resource scarcity and abundance, respectively (e.g. Gautier-Hion,

1980; Porter, 2001). While I consider dietary niche partitioning in the study community, unfortunately my diet data are compiled from other studies as it is not practical to collect dietary data using the methodology employed in the present study, line transect sampling, because contact periods with each group are brief.

Macro- and microhabitat are terms that describe habitat patch size relative to the mobility and range size of a species (Morris, 1984). For primates, macrohabitat generally refers to large patches (e.g. terra firme or flooded forests) while microhabitat generally refers to small patches within each macrohabitat (e.g. vertical strata, structurally complex areas, or narrow bands of riparian/riverine forest). Studies of primate community ecology and distribution suggest the active avoidance of competition though both macrohabitat (Bobadilla & Ferrari, 2000; Peres, 1993a) and microhabitat preferences (Mittermeier & van Roosmalen, 1981; Peres, 1993a; Porter,

2004; Terborgh, 1983). Within primate communities, different subsets of species utilize each of the available forest habitats, with all members of the community tending to use terra firme forest to varying degrees where it is available (e.g. Bobadilla & Ferrari, 2000;

Peres, 1993a). Vertical stratification is also evident in many primate communities. A

103 positive relationship between forest height and body weight has also been observed, though when a forest stratum contains an abundant resource many apparently divergent species may simultaneously utilize that stratum (Peres, 1993a).

While the bulk of primate macrohabitat niche partitioning research has been conducted in relatively undisturbed forests, niche partitioning has also been examined at a few sites with various types of anthropogenically disturbed habitats. Typically these studies discuss differences in abundance and species richness between undisturbed and disturbed macrohabitats, with the latter commonly including selectively logged and/or regenerating forests following clear cuts for agriculture (Bicknell & Peres, 2010; Chapter

6; Johns, 1991; Parry, et al., 2007). This research shows that undisturbed and lightly disturbed sites often have the same species richness even though the densities of individual species vary based on their adaptability. Sites that are heavily disturbed, however, often experience reductions in both species richness and individual densities, though genera that are disturbance specialists, like Saguinus, may experience population increases.

It is less common for researchers to consider the influence of large swaths of anthropogenically disturbed macrohabitats in concert with diet and microhabitat variables like vertical stratification (e.g. Johns, 1991). The degree to which individual primate species are able to focus on disturbed habitats is more pronounced at sites with large areas of anthropogenic disturbance than it is at sites with only natural disturbance regimes. In anthropogenic mosaic landscapes, disturbed habitat patch sizes tend to be

104 large, for example a regenerating 2500 ha cattle ranch, while patches in areas with natural disturbance regimes are often small and widely dispersed, for example isolated treefalls or blowdowns that range from 5-10 trees to <30 ha in the Amazon (Negrón-

Juárez, et al., 2010). (Blowdowns up to 3370 ha do occur, though most blowdowns >30 ha (~74%) are <300 ha and overall they disturb only 0.04% of the forest every 2 yr,

Nelson, et al., 1994.) As a result of these differences, in naturally-disturbed areas it is more difficult for primates to specialize on disturbed habitats for a significant proportion of their needs because heavily disturbed habitats are less extensive under natural disturbance regimes than under many anthropogenic disturbance regimes

(Peres, 1993a).

Perhaps the clearest evidence of niche partitioning occurs when primates travel and forage together in polyspecific (multispecies) associations that benefit of one or more of the participant species through increases in foraging efficiency and/or reductions in predation pressure (Terborgh, 1983; Terborgh, 1990). The principle of competitive exclusion holds that multiple species competing for the same resources cannot coexist (Gause, 1934), so niche partitioning should be especially pronounced among closely related species that form these associations. Indeed, one group where niche partitioning is especially clear is the ecologically, behaviorally, and morphologically similar callitrichids. Many callitrichids are known for their long-lasting polyspecific associations (Garber, 1992; Porter, 2004; Terborgh, 1983) during which they partition their niches along many lines, including vertical stratification, foraging techniques, and diet (Buchanan-Smith, 1999; Peres, 1992; Porter, 2001).

105

In this study I use data collected during line transect sampling to examine macro- and microhabitat use in a community of six Neotropical primates at the Biological

Dynamics of Forest Fragments Project (BDFFP) in the state of Amazonas, Brazil. I begin by looking at macrohabitat partitioning by examining population density differences between mature forest and secondary forest regenerating following anthropogenic disturbance (clearcuts for cattle ranches) while also examining polyspecific associations.

Within these habitat types, I analyze several traditional microhabitat variables where primate behavior and ecology might diverge: vertical stratification, use of riparian forest

(measured as the distance to the nearest permanent stream), and habitat complexity.

Lastly, I examine the relationship between body size and vertical stratification to determine whether the use of different strata is tied to body mass or if other factors like resource availability might be at play.

Methods

STUDY SITE

The Biological Dynamics of Forest Fragments Project is a long-term, experimental study of the impacts of habitat fragmentation on Neotropical rain forest flora, fauna, and abiotic variables. The project is located in the central Amazon, 80 km north of the city of

Manaus, Brazil (2°25´S, 59° 50´ W) (Didham, 1997). The entire site is 100,000 ha and encompasses three large, isolated cattle ranches (3,000-4,000 ha each) and a portion of the vast expanse of largely undisturbed primary forest that surrounds them (Laurance, et al., 2007). The forest is tall terra firme tropical rain forest with a closed canopy

106 between 30-37 m and emergents up to 55 m (Bierregaard Jr. & Stouffer, 1997). The landscape contains undisturbed continuous primary forest, habitat fragments in three size classes (1, 10, and 100 ha), active pasture, secondary forest on abandoned pasture, and selectively logged continuous primary forest. Annual rainfall at the site ranges from

1,900-2,500 mm (Stouffer & Bierregaard Jr., 1993) and averaged 2194 mm during this study (Instituto Nacional de Pesquisas Espaciais, 2012); over the last 106 years the city of Manaus just to the south has received an average of 2115 mm (Cochrane, 2011).

There is a pronounced dry season from June to September (Didham, 1997) with average monthly temperatures ranging from 25.8⁰C in February-April to 27.9⁰C in September

(Salati, 1985). While the Manaus region, and much of the Amazon Basin in general, are considered lacking in large-scale topography, there are numerous steep hills cut by current and past streams (igarapés) even though there is no significant large-scale relief.

Indeed, the elevation of the site ranges from 50 to 150 m (Bruna & Kress, 2002). The soil is predominantly nutrient-poor yellow latosol (Rylands & Keuroghlian, 1988).

The primate community at the BDFFP (Table 5.1) is composed of six species: black-bearded saki (Chiropotes chiropotes) (Veiga, et al., 2008)1, golden-faced saki

(Pithecia chrysocephala) (Marsh, In press)2, golden-handed tamarin (Saguinus midas)

(Mittermeier, et al., 2008b), Guiana spider monkey (Ateles paniscus) (Mittermeier, et al.,

2008a), Guianan brown capuchin (Sapajus apella apella) (Rylands, et al., 2008), and

Guianan red howler monkey (Alouatta macconnelli) (Boubli, et al., 2008). I follow the

107 taxonomy used by the IUCN Redlist (IUCN, 2012), though I place the brown capuchin in the genus Sapajus rather than Cebus (Lynch Alfaro, et al., 2012a; Lynch Alfaro, et al.,

2012b; Silva Jr., 2001; Silva Jr., 2002). Hereafter each study species will be referred to by genus alone.

Table 5.1: Characteristics of the BDFFP primates (Boyle & Smith, 2010b)

DATA COLLECTION

I collected data collected over a three-year period: November-December 2008, June-

November 2009, and January-June 2010. Data were collected along eight line transects

(Figure 5.1) that sampled multiple habitats, ran perpendicular to the edges between mature and secondary growth forests, and varied in length from 690-2250 m depending on the number of habitats sampled (Table 5.2 and Table 5.3). I analyzed species-specific density differences between mature (undisturbed primary forest, lightly selectively logged forest, and primary forest habitat fragments) and secondary forests (25 year-old forest regenerating on unburned pasture and 19- and 25-30-year-old forest

108 regenerating on burned pasture). Transects were separated by 0.575-15.0 km, spacing was not equal due to landscape limitations, and were flagged every 10 m. I walked transects beginning at 06:45 and 13:15, avoiding mid-day sampling when possible. There was a minimum of 6.25 hrs between resampling, which is more than the 3 hrs that has been suggested as a sufficient rest period to allow groups to redistribute themselves

(Peres, 1999). Two observers, always 5-10 m apart and always with the author in the lead for consistency, slowly walked transects (average speed=1.01 km hr-1) while pausing for 7-10 seconds every 10 m to look and listen.

Figure 5.1: Transect locations

109

Table 5.2: Transect information

Transect Habitat Approximate Age (yr) Length (km) Replicates Total (km) 1 Undisturbed primary forest - 0.03 47 1.41 Secondary forest (burned) 25-30 0.66 47 31.02 2 Undisturbed primary forest - 1.07 42 44.94 10 ha fragment (#2206) - 0.35 43 15.05 3 Undisturbed primary forest - 1.10 38 41.80 Secondary forest (burned) 19 0.64 31 19.84 4 Undisturbed primary forest - 0.85 37 31.45 100 ha fragment (#2303) - 1.10 42 46.20 5 Undisturbed primary forest - 1.00 46 46.00 Secondary forest (unburned) 25 0.71 46 32.66 6 Undisturbed primary forest - 1.00 46 46.00 Secondary forest (unburned) 25 0.82 44 36.08 7 Logged primary forest 25 1.03 58 59.74 Secondary forest (unburned) 25 0.07 57 3.99 8 Logged primary forest 25 1.06 55 58.30 Secondary forest (unburned) 25 0.64 50 32.00

Table 5.3: Total area sampled by habitat

Habitat Length (km) Total (with replicates) (km) Undisturbed primary forest 5.05 211.60 Logged primary forest 2.09 118.04 Secondary forest (unburned, 25 y) 2.24 104.73 100 ha fragment (#2303) 1.10 46.20 Secondary forest (burned, 25-30 y) 0.66 31.02 Secondary forest (burned, 19 y) 0.64 19.84 10 ha fragment (#2206) 0.35 15.05

Upon sighting a group, I used a Nikon Forestry 550 laser rangefinder to measure the sighting distance and height of the first individual and a Suunto MC-2G global compass to measure the sighting angle. I only recorded microhabitat data on the first individual sighted, essentially a focal animal sample, because all individuals tended to be at the same level of the forest, though as in other studies capuchins (Sapajus) were often at multiple levels (Mittermeier & van Roosmalen, 1981). This method also minimized the influence of observers on the behavioral data from those groups that were unhabituated by recording data immediately, before the group had a chance to

110 react to our presence. I recorded our exact location on the transect, counted the number of individuals, and estimated group spread along two perpendicular axes, always including the longest axis. Multiple species were considered to be in a polyspecific association if individuals from each species were within 50 m (e.g.

Buchanan-Smith, 1990; Chapman & Chapman, 1996; Wachter, et al., 1997) though in most cases individuals were much closer, and if my observations suggested active maintenance of interspecies proximity rather than simply a chance meeting of multiple species as in an assemblage (e.g. Mendes Pontes, 1997). The microhabitat data that I discuss with my polyspecific association findings were recorded both during and outside of associations. Traditionally, studies of polyspecific associations only consider data recorded during active associations. However, following that protocol here would not have left enough data due to the small number of associations and the fact that each association only provides a single data point for each species because I did not conduct focal group follows.

I measured, identified, and mapped all trees ≥10 cm dbh (diameter at breast height, or 1.3 m) that fell within 2 m of each transect. Amazonian tree expert Paulo

Apóstolo Assunção (da S. Ribeiro, et al., 1999) identified all trees in the field. I calculated the distance between the estimated group center and the nearest permanent stream by first using Google Earth 6.1 to plot estimated group centers relative to my exact location on the transect in relation to GPS points taken at 50-m intervals along all transects. I then added my data on the locations of permanent streams to satellite images in Google

Earth and combined those stream data with drainage maps for the area from SUFRAMA

111

(Superintendência da Zona Franca de Manaus) in ArcGIS 9.3. Finally, I used ArcGIS 9.3 to calculate the distance from the group center to the nearest permanent stream.

DENSITY CALCULATIONS

I estimated the area sampled for each primate using species-specific transect widths where I assumed that I detected every individual within a specified distance from the transect. I determined species-specific transect widths using the perpendicular distance to the first individual, an adjustment for group spread, and a 50% criterion for fall-off distance (Whitesides, et al., 1988). The fall-off distance is the maximum distance from the transect in which it is assumed that 100% of the individuals present were detected. In this method, the perpendicular distances of all sightings for each species are displayed in a histogram in both 5- and 10-m groups and the fall-off distance is the smallest distance at which there is a drop of more than 50% between consecutive bins using the bin size that best fits the data (Whitesides, et al., 1988). I also weighed sighting distances during my evaluation of fall-off distance in cases where a single low outlier bin would have resulted in a fall-off distance that did not accurately describe the entire dataset. The data for perpendicular distance estimation are from replicates of all eight transects. Before combining sightings from multiple habitats, I used Kruskal Wallis and

Mann Whitney U tests to look for interhabitat differences in species-specific perpendicular distances and combined observations across habitats only when there were no interhabitat differences.

112

2 2 Area = 2(S/2+ED)*Lt + 0.5πR = 2(S/2+ED)*Lt + 0.5π(S/2+ED)

with ED (Effective Distance) = (Nt/Nf)*(FD)

where T=total width sampled, Lt=total length sampled, Wt=species-specific transect width on one side of the transect, S=estimate of species-typical mean group spread

(radius)=species-specific effective distance plus one half the average group spread.

I also estimated the location of group center following Whitesides et al. (1988), assuming that the first individual sighted is the closest to the observer and that the center of the group is along the sighting angle at a distance of 50% of the average group spread.

113

STATISTICS

Primate surveys yield abundance (count) data and can therefore be modeled with generalized linear models utilizing the Poisson or negative binomial distributions. Count data are frequently over-dispersed, which was the case in this study due to an excess of zeroes that is common in ecological sampling (Zuur, et al., 2009). To deal with this issue I used negative binomial models because they often outperform other methods of handling over-dispersion (Warton, 2005). Specifically, to test for density differences between mature and secondary forests I used the planned comparisons (least significant difference) associated with a negative binomial generalized estimating equation (GEE) with a log link to account for repeated measures, including the natural log of the area sampled as an offset to treat counts as densities and allow for intertransect differences in area sampled. I use QIC to determine the best correlation matrix and a model-based estimator because I have fewer than 20 clusters (Garson, 2012). I calculated the standard deviation for each population density estimate as the weighted standard deviation (where the weight is the area sampled per transect in km2 because transect lengths varied) with each transect survey considered to be a separate replicate.

I used the Kolmogorov-Smirnov test to examine interspecific differences in the distributions of all other microhabitat variables: vertical stratification, use of riparian forest, and habitat complexity (Bobadilla & Ferrari, 2000; Peres, 1993a). Vertical strata refer to the height (m) at which the first individual was observed. The use of riparian forest, which is forest adjacent to a permanent stream, is compared among species

114 through an examination of the distances between the centers of the groups encountered and the nearest permanent stream. Finally, there are two measures of habitat complexity: the average dbh and the number of stems in the area immediately surrounding the first individual. The "immediate area" consists of the 10-m section of the transect perpendicular to the first individual sighted plus the 10-m sections on either side of it (a total of 30 m*4 m or 120 m2). I examined each of these microhabitat variables for all primate species pairs in both mature and secondary forests. Finally, I used simple linear regression to examine the relationship between body size and vertical strata.

All analyses were conducted in SPSS version 19.0.0.1.

Results

Kruskal Wallis tests on the average perpendicular distance to the trail for each primate species between the 7 habitats sampled (primary forest, selectively logged forest, 100 ha fragment, 10 ha fragment, three types of secondary growth) detected no differences

(Alouatta: N=55 in 4 habitats, 2=5.144, df=3, p=0.162; Ateles: N=22 in 3 habitats,

2=1.790, df=2, p=0.409; Chiropotes: N=40 in 5 habitats, 2=4.082, df=4, p=0.395;

Pithecia: N=21 in 5 habitats, 2=5.114, df=4, p=0.276; Saguinus: N=74 in 5 habitats,

2=8.835, df=5, p=0.116; Sapajus: N=59 in 6 habitats, 2=1.847, df=5, p=0.870). To be positive that I could combine habitats, I also reclassified the forest types and conducted

Mann Whitney tests for differences between two broader habitat types: 1) secondary growth (SG), and 2) mature forest (MF) (including primary forest, logged forest, and

115 habitat fragments). This analysis revealed that data for Alouatta needed to be split but that all other species could still be combined (Alouatta: N=53 (MF) 2 (SG), Z=-2.025, p=0.034); Ateles: N=22 (MF) 0 (SG), no test; Chiropotes: N=36 (MF) 4 (SG), Z=-1.782, p=0.074); Pithecia: N=16 (MF) 5 (SG), Z=0.186, p=0.208); Saguinus: N= 37 (MF) 37 (SG),

Z=-1.109, p=0.282); Sapajus: N=36 (MF) 23 (SG), Z=-0.054, p=0.975)).

Interhabitat population density differences and polyspecific association incidences for each species are summarized in Table 5.4. Only four species (Chiropotes,

Pithecia, Saguinus, and Sapajus) were observed in polyspecific associations, occurring in two species pairings (Chiropotes-Sapajus and Pithecia-Saguinus). The table reports true mean densities and weighted standard deviations whereas the means associated with the statistical tests reported in the text are the estimated marginal means and standard errors (number of individuals km2 -1) used in the GEEs. Alouatta: mature forest

(EMM=5.68; SE=0.488), secondary forest (EMM=1.54; SE=0.570) Wald 2=30.362, df=1, p<0.001); Ateles: not tested, N=0; Chiropotes: mature forest (EMM=5.14; SE=0.491), secondary forest (EMM=1.67; SE=0.413) Wald 2=29.172, df=1, p<0.001); Pithecia: mature forest (EMM=1.51; SE=0.240), secondary forest (EMM=1.36; SE=0.353) Wald

2=0.127, df=1, p=0.721); Saguinus: mature forest (EMM=5.56; SE=0.524), secondary forest (EMM=13.43; SE=1.419) Wald 2=27.088, df=1, p<0.001); Sapajus: mature forest

(EMM=6.80; SE=0.491), secondary forest (EMM=12.02; SE=1.089) Wald 2=19.091, df=1, p<0.001).

116

Table 5.4: Intraspecies macrohabitat use patterns

Species Forest Type Density (individ./km2; ) Polyspecific Associations1 Alouatta Mature 5.7 (18.2)* 0% (0/53) Secondary 1.5 (17.9)* 0% (0/2) Ateles Mature 1.3 (9.7)** 0% (0/22) Secondary 0.0 (0)** 0% (0/0) Chiropotes Mature 5.0 (21.7)* 16.7% (6/36) Secondary 1.7 (15.1)* 50% (2/4) Pithecia Mature 1.5 (9.6) 18.8% (3/16) Secondary 1.4 (10.5) 40% (2/5) Saguinus Mature 5.6 (20.5)* 8.3% (3/36) Secondary 13.7 (41.7)* 5.4% (2/37) Sapajus Mature 6.6 (28.5)* 16.7% (6/36) Secondary 11.6 (50.4)* 8.7% (2/23) * Statistically significant difference (p<0.05); **Not tested, N=0 in secondary forest 1Number of polyspecific associations/number of encounters

I split all microhabitat variables by habitat (mature vs. secondary forest) and considered each habitat separately. Table 5.5, which excludes both measures of habitat complexity (tree size and stem density) because there were no interspecies differences, summarizes significant differences in mature forests detected by the Kolmogorov-

Smirnov test. Full test results are reported in the text below the table. There were also no differences in secondary growth, so I did not create a summary table, but the full results are again reported in the text. Figure 5.2 shows the vertical distributions of each of the primate species in mature forest, the variable with the most interspecies differences. I also report the means and standard deviations for each variable tested, though I did not statistically analyze the means as the Kolmogorov-Smirnov test uses distributions (Table 5.6).

117

Table 5.5: Differences in microhabitat distributions in mature forest (K-S test) (e.g. Ateles vs. Alouatta, distance to stream: Ateles (A) > Alouatta (B)) (p<0.05)*

Species A: Species B: Alouatta Ateles Chiropotes Pithecia Saguinus Sapajus Alouatta Vertical stratification > > > Distance to stream < < Ateles Vertical stratification > > > > Distance to stream > < Chiropotes Vertical stratification < > Distance to stream Pithecia Vertical stratification < < Distance to stream < Saguinus Vertical stratification < < < < Distance to stream > > > > Sapajus Vertical stratification < < > Distance to stream < *Empty cells denote statistical equivalence. Vertical stratification and tree height are based on the first individual sighted while distance to water is measured from the group center

118

Figure 5.2: Vertical distribution by species in mature forest

119

Table 5.6: Microhabitat use descriptive statistics ( ())*

Habitat Complexity Species (N) Height (m) Distance to Stream (m) Dbh (cm) Stem density Alouatta (54) 20.8 (5.5) 297 (213) 21.4 (4.6) 9.4 (2.8) Ateles (22) 23.3 (4.5) 352 (158) 22.8 (4.9)1 8.9 (2.4)1 Mature Chiropotes (36) 18.6 (5.4)2 382 (272) 20.6 (4.0) 10.2 (3.2) Forest Pithecia (16) 15.7 (7.4) 288 (185) 21.6 (2.7) 11.3 (3.0) Saguinus (36) 13.3 (5.8) 428 (151) 21.4 (4.0) 8.6 (2.5) Sapajus (36) 18.6 (5.9)3 281 (205) 21.6 (4.2)4 9.4 (3.4)4 Alouatta (2) 20.9 (1.0) 138 (71) 18.3 (2.8) 9.0 (2.8) Ateles (0) n/a n/a n/a n/a Secondary Chiropotes (4) 16.7 (2.0) 297 (269) 17.3 (0.6) 10.5 (5.3) Forest Pithecia (5) 14.8 (2.3) 365 (271) 18.0 (1.6) 11.8 (3.4) Saguinus (37) 13.9 (4.5) 415 (210) 17.5 (2.7) 10.6 (3.2) Sapajus (23) 13.6 (6.0)5 435 (269) 18.6 (2.5)5 10.5 (2.7)5 *Note: This table is descriptive only, it contains means while statistical tests used distributions 1N=21; 2N=35; 3N=35; 4N=34; 5N=22

VERTICAL STRATIFICATION:

Mature forest: Kolmogorov-Smirnov: Alouatta vs. (Ateles N=76, Z=1.171, p=0.129; Chiropotes N=89, Z=1.253, p=0.086; Pithecia N=70, Z=1.765, p=0.004; Saguinus

N=90, Z=2.754, p<0.001; Sapajus N=89, Z=1.475, p=0.026), Ateles vs. (Chiropotes N=58,

Z=1.995, p=0.001; Pithecia N=38, Z=2.283, p<0.001; Saguinus N=58, Z=2.977, p<0.001;

Sapajus N=57, Z=1.575, p=0.014); Sapajus vs. (Chiropotes N=70, Z=0.956, p=0.320;

Pithecia N=51, Z=1.065, p=0.207; Saguinus N=71, Z=2.090, p<0.001); Chiropotes vs.

(Pithecia N=51, Z=1.024, p=0.245; Saguinus N=71, Z=1.729, p=0.005); Pithecia vs.

(Saguinus N=52, Z=0.948, p=0.330).

Secondary forest: Kolmogorov-Smirnov: Alouatta vs. (Ateles n/a, for Ateles N=0;

Chiropotes N=6, Z=1.155, p=0.139; Pithecia N=7, Z=1.195, p=0.115; Saguinus N=39,

Z=1.266, p=0.081; Sapajus N=24, Z=1.231, p=0.097), Ateles vs. (n/a because for Ateles

120

N=0); Sapajus vs. (Chiropotes N=26, Z=1.045, p=0.225; Pithecia N=27, Z=0.642, p=0.804;

Saguinus N=59, Z=0.429, p=0.993); Chiropotes vs. (Pithecia N=9, Z=0.820, p=0.512;

Saguinus N=41, Z=0.976, p=0.297); Pithecia vs. (Saguinus N=42, Z=0.737, p=0.648).

DISTANCE TO STREAM (RIPARIAN FOREST):

Mature forest: Kolmogorov-Smirnov: Alouatta vs. (Ateles N=76, Z=1.358, p=0.05;

Chiropotes N=90, Z=1.248, p=0.089; Pithecia N=70, Z=0.667, p=0.765; Saguinus N=90,

Z=2.281, p<0.001; Sapajus N=90, Z=0.689, p=0.730), Ateles vs. (Chiropotes N=58,

Z=1.064, p=0.208; Pithecia N=38, Z=0.968, p=0.305; Saguinus N=58, Z=1.428, p=0.034;

Sapajus N=58, Z=1.269, p=0.08); Sapajus vs. (Chiropotes N=72, Z=1.061, p=0.211;

Pithecia N=52, Z=0.693, p=0.722; Saguinus N=72, Z=1.768, p=0.004); Chiropotes vs.

(Pithecia N=52, Z=0.994, p=0.277; Saguinus N=72, Z=1.296, p=0.069); Pithecia vs.

(Saguinus N=52, Z=1.433, p=0.033).

Secondary forest: Kolmogorov-Smirnov: Alouatta vs. (Ateles n/a; Chiropotes N=6,

Z=0.866, p=0.441; Pithecia N=7, Z=0.717, p=0.683; Saguinus N=39, Z=1.080, p=0.194;

Sapajus N=25, Z=0.944, p=0.335), Ateles vs. (n/a); Sapajus vs. (Chiropotes N=27,

Z=0.742, p=0.640; Pithecia N=28, Z=0.529, p=0.943; Saguinus N=60, Z=0.982, p=0.289);

Chiropotes vs. (Pithecia N=9, Z=0.522, p=0.948; Saguinus N=41, Z=0.963, p=0.312);

Pithecia vs. (Saguinus N=42, Z=0.556, p=0.917).

121

HABITAT COMPLEXITY (AVERAGE DBH):

Mature forest: Kolmogorov-Smirnov: Alouatta vs. (Ateles N=75, Z=0.802, p=0.540; Chiropotes N=90, Z=0.732, p=0.658; Pithecia N=70, Z=1.082, p=0.193; Saguinus

N=90, Z=0.602, p=0.861; Sapajus N=88, Z=0.746, p=0.633), Ateles vs. (Chiropotes N=57,

Z=0.766, p=0.600; Pithecia N=37, Z=0.726, p=0.667; Saguinus N=57, Z=0.650, p=0.791;

Sapajus N=55, Z=0.590, p=0.877); Sapajus vs. (Chiropotes N=70, Z=0.690, p=0.728;

Pithecia N=50, Z=0.861, p=0.449; Saguinus N=70, Z=0.588, p=0.880); Chiropotes vs.

(Pithecia N=52, Z=1.017, p=0.252; Saguinus N=72, Z=0.707, p=0.699); Pithecia vs.

(Saguinus N=52, Z=0.624, p=0.831).

Secondary forest: Kolmogorov-Smirnov: Alouatta vs. (Ateles n/a, for Ateles N=0;

Chiropotes N=6, Z=0.577, p=0.893; Pithecia N=7, Z=0.598, p=0.867; Saguinus N=39,

Z=0.521, p=0.949; Sapajus N=24, Z=0.554, p=0.919), Ateles vs. (n/a because for Ateles

N=0); Sapajus vs. (Chiropotes N=26, Z=1.087, p=0.188; Pithecia N=27, Z=0.826, p=0.503;

Saguinus N=59, Z=1.018, p=0.252); Chiropotes vs. (Pithecia N=9, Z=0.894, p=0.400;

Saguinus N=41, Z=0.770, p=0.593); Pithecia vs. (Saguinus N=42, Z=0.624, p=0.831).

HABITAT COMPLEXITY (STEM DENSITY):

Mature forest: Kolmogorov-Smirnov: Alouatta vs. (Ateles N=75, Z=0.483, p=0.974; Chiropotes N=90, Z=0.473, p=0.978; Pithecia N=70, Z=1.008, p=0.261; Saguinus

N=90, Z=0.689, p=0.730; Sapajus N=88, Z=0.503, p=0.962), Ateles vs. (Chiropotes N=57,

Z=0.751, p=0.625; Pithecia N=37, Z=1.166, p=0.132; Saguinus N=57, Z=0.405, p=0.997;

122

Sapajus N=58, Z=0.505, p=0.961); Sapajus vs. (Chiropotes N=70, Z=0.629, p=0.824;

Pithecia N=50, Z=1.310, p=0.065; Saguinus N=70, Z=0.519, p=0.950); Chiropotes vs.

(Pithecia N=52, Z=0.832, p=0.493; Saguinus N=72, Z=0.943, p=0.336); Pithecia vs.

(Saguinus N=52, Z=1.341, p=0.055).

Secondary forest: Kolmogorov-Smirnov: Alouatta vs. (Ateles n/a, for Ateles N=0;

Chiropotes N=6, Z=0.577, p=0.893; Pithecia N=7, Z=0.598, p=0.867; Saguinus N=39,

Z=0.633, p=0.818; Sapajus N=24, Z=0.615, p=0.843), Ateles vs. (n/a because for Ateles

N=0); Sapajus vs.: (Chiropotes N=26, Z=0.753, p=0.623; Pithecia N=27, Z=0.514, p=0.954;

Saguinus N=59, Z=0.584, p=0.885); Chiropotes vs. (Pithecia N=9, Z=0.745, p=0.635;

Saguinus N=41, Z=0.693, p=0.722); Pithecia vs. (Saguinus N=42, Z=0.420, p=0.995).

RELATIONSHIP BETWEEN BODY SIZE AND USE OF VERTICAL STRATA:

Body size (kg) strongly and significantly predicted the height of the average vertical strata (m) used in both mature forest (r2=0.884, F=30.409 (1,4), p=0.005) and secondary growth (r2=0.846, F=16.460 (1,3), p=0.027) (Figure 5.3).

123

Figure 5.3: The relationship between body weight and use of vertical strata in mature and secondary forests

Discussion

The primate community of the BDFFP consists of six species that vary in size, behavior, and ecology. There is little overlap in dietary classifications, with a large specialist frugivore (Ateles), a large folivore-frugivore (Alouatta), a small- (Pithecia) and medium-sized (Chiropotes) seed predator, and a small- (Saguinus) and medium-sized

(Sapajus) generalist. In addition to the niche partitioning revealed by these dietary differences are the macrohabitat preferences that also suggest that species segregate space to reduce competition. The species that prefer mature forest are large-bodied and/or frugivores (including seed-predators) (Alouatta, Ateles, and Chiropotes) while those that prefer secondary forest are small-to-medium-sized generalists (Saguinus and

Sapajus). The single species without a clear habitat preference, Pithecia, is a small- bodied frugivore, a combination of the characteristics of the species that prefer mature or secondary forests.

124

In addition to dietary and macrohabitat niche partitioning, microhabitat partitioning within individual macrohabitats is another means to reduce competition because even species with different diet types compete with each other within habitats

(e.g. Alouatta and Ateles, Chapman, 1987). To elucidate microhabitat partitioning I considered species dyads composed of the species that are most likely to compete with each other due to their similar diets and/or body sizes.

The first dyad that I considered was Alouatta and Ateles. These species are similar in their large body sizes, preference for mature forest, and use of vertical strata; they differ in their diets, proximity to riparian forest, and use of edge habitats with

Alouatta utilizing edges and Ateles avoiding them (Chapter 4). The large body sizes of these species necessitate their use of similar vertical strata and likely explain their preferences for mature forests because secondary forests comprised of small trees lack large enough substrates for safe locomotion (Chapter 6). Indeed, I found a positive relationship between body mass and stratum use in mature (r2=0.884) and secondary forests (r2=0.846) illustrating the limitations that short-statured forests place on larger species. Given that their body mass restricts them to the same macrohabitat and vertical strata within that habitat, niche partitioning comes down to 1) diet, Alouatta is a folivore-frugivore while Ateles is a specialized frugivore, and 2) microhabitat preferences for small-scale habitats within macrohabitats. The latter is evident in howler monkey preferences for habitat edges (Chapter 4) and riparian forest. I found Alouatta groups to be closer to permanent streams than Ateles, and the closely related Alouatta seniculus shows a clear preference for lake- and river-edge habitats in Amazonia while generally

125 not establishing home ranges more than 1600-1800 m into terra firme forests from water sources (Defler and Peres, personal communication, cited by Palacios, 1998).

Howler monkeys are more likely to use both edge and riparian habitats because their folivorous diets permit them to shift to these areas to avoid competition: habitat edges contain successional tree species with poorly defended foliage (Coley, 1980) and an increase in the production of poorly defended new leaves in mature forest species due to the increase in sunlight (Chen, et al., 1992; Coley, 1980; Sork, 1983; Williams-Linera,

1990a) while howler use of seasonally flooded forest (igapó) increases when new leaves are abundant (Palacios, 1998). Howler monkeys may also be driven into these habitats in other locations due to displacement from terra firme habitats by other large-bodied sympatric primates (e.g. Lagothrix, Palacios, 1998), a relationship that may also exist between Alouatta and Ateles at the BDFFP as Alouatta palliata tends to lose direct contests over fruit with Ateles geoffroyi in Central America (Chapman, 1987).

The small- and medium-sized species, Chiropotes, Pithecia, Saguinus, and

Sapajus can be broken into two dyads based on their diets: generalists (Saguinus and

Sapajus) and frugivores, both of which are pithecine seed predators (Chiropotes and

Pithecia) (Table 5.1). As a result of their small to medium body sizes, none of these species' weights prevents them from spending significant amounts of time in secondary forests (Chapter 6). Both Saguinus and Sapajus show preferences for edge habitats in undisturbed primary forest (Chapter 4) and for secondary forests (Chapter 6). Both species are also omnivores that consume large amounts of fruit, arthropods, and vertebrates (Table 5.1). In mature forests, niche partitioning among these two species

126 can be seen in their use of vertical strata and riparian forest. Sapajus spends more time at higher forest strata and is found closer to permanent streams than Saguinus. Sapajus may make more use of riparian areas because at the BDFFP these areas often contain palms (B. Lenz, personal observation) that are important in their diets (Fragazy, et al.,

2004) and which provide preferred sleep sites (Zhang, 1995a). Saguinus, on the other hand, might avoid riparian forests because many have somewhat open canopies relative to the surrounding forest which may therefore increase their exposure to aerial predators (B. Lenz, personal observation). Sapajus is more than four times heavier than

Saguinus (2.6 vs. 0.5 kg), a weight difference that means Saguinus is able to locomote on a wider range of substrate sizes than Sapajus, facilitating interspecies vertical stratification even though they consume many of the same food types. It has also been suggested that these two species divide fruit resources by focusing on fruits from different plant families and species while consuming different size classes and species of vertebrate and invertebrate fauna due to foraging differences (Mittermeier & van

Roosmalen, 1981).

The final dyad contains the pithecines, Chiropotes and Pithecia. Chiropotes shows a clear preference for mature forest while Pithecia is evenly distributed across both habitats. Interestingly, the only two seed predators at the site do not differ in any of the other microhabitat variables that I measured, though I did previously find that

Chiropotes and Pithecia experience positive and neutral, respectively, edge effects

(Chapter 4). The absence of significant niche differentiation at the BDFFP may be due to two factors. First, it is possible that the low encounter rate for the uncommon Pithecia

127 yielded an insufficient sample size to detect differences. For example, a graphical examination of the vertical distributions of these species (Figure 5.2) reveals that

Pithecia appears to be more common than Chiropotes at the lowest forest strata (note: removal of the outlier Pithecia point between 35-40 m more than halved the p value but still did not produce a significant finding). Indeed, a true lack of niche partitioning at the

BDFFP would be a surprise as sympatric Chiropotes and Pithecia are known to partition their niches by forest type and vertical stratification (Kinzey & Norconk, 1993;

Mittermeier & van Roosmalen, 1981; Walker, 1996). The second possible explanation is the rarity of Pithecia at the study site itself. Pithecia occurs at such low densities across all habitats that encounters and direct competition with Chiropotes are likely to be infrequent. As a result, both species may be able to utilize the forest as if the other is not present, though this may not be the case at the forest edge (Chapter 4).

Based on dietary data compiled from the literature, there is clear niche partitioning in the primate community at the BDFFP in terms of the food types that each species consumes. I also found niche partitioning among macrohabitat and mature forest microhabitats, though there was an absence of niche partitioning in secondary forest. This pattern could be the result of the smaller sample sizes due to low encounter rates in secondary forests because several species show a clear aversion to it.

Interestingly, there were no differences even between the two species that were common in secondary growth, Saguinus and Sapajus, even though they have similar diets. It is therefore also possible that species are limited in their ability to physically partition their niches in secondary forest because this habitat does not provide much

128 opportunity for vertical stratification due to the shorter, more uniform size of the trees relative to mature forest. In other words, in mature forests non-canopy strata permit vertical niche partitioning because they contain branches large enough to support primates while sufficiently large non-canopy strata are absent from secondary forests.

The other potential explanation for the lack of niche partitioning may be that the resources consumed by Saguinus and Sapajus in secondary growth are common enough that there is little need to avoid competition for them.

In terms of polyspecific associations, which I use to refer only to mixed groups rather than chance assemblages, four species of the primate community entered into associations in two pairings (Chiropotes-Sapajus and Pithecia-Saguinus; hereafter CH-

SAP and PI-SAG). Before continuing, I should again note that the data on microhabitat use that I analyzed were collected both in and outside of associations, so my results need to be interpreted cautiously as traditional discussions of niche partitioning during associations involve only data collected while multiple species are engaged in an association. Associations between both species sets occurred in both mature and secondary forests, and they are not unique to the BDFFP (Bobadilla & Ferrari, 2000;

Mittermeier & van Roosmalen, 1981) though they may not occur everywhere that these species are sympatric (Lehman, et al., 2006a). Each of the two polyspecific association dyads at the BDFFP was composed of one pithecid seed predator (Chiropotes/Pithecia) that preferred mature forest (Chiropotes) or had neutral habitat preferences (Pithecia) and one of the two generalist species at the site (Saguinus/Sapajus), both of which showed clear preferences for secondary forests.

129

The individual species in CH-SAP and PI-SAG associations did not interact with each other in the same fashion at the BDFFP. CH-SAP associations were heterogeneous, with individuals from each species mixed together in a larger group, as has been observed elsewhere (Bobadilla & Ferrari, 2000). PI-SAG associations tended to be homogeneous groups of each species side-by-side or nearby without individuals from the two species intermingling. This may be an artifact of Sapajus and Chiropotes having significantly larger group sizes than Pithecia and Saguinus, therefore making maintenance of tightly-knit groups extremely difficult, or it may demonstrate a different balance of motivations for association formation.

While each association was recorded in mature and secondary forest, pithecids were more likely to be in an association in secondary forests than they were in mature forests. Pithecids may not heavily use secondary forests outside of associations because they are dominated by pioneer trees with small, low quality seeds that have been hypothesized to negatively impact pithecid populations in logged forests (Grieser Johns,

1997; Plumptre & Grieser Johns, 2001). They may also be more likely to enter secondary forests when they are in an association if they are interested in more efficient foraging for items like insects that are part of the "other" portion of their diets (0.6% and 2.8% of their diets, Table 5.1) because prey capture rates may increase when foraging in a larger multispecies group.

It is also interesting that the heaviest (CH-SAP) and lightest (PI-SAG) pithecids and generalists associate with each other, a division that has several potential

130 explanations. First, in addition to being a division by body size, this is also essentially a division by home range and group size (Chiropotes=336 ha, 21.8 individuals;

Sapajus=429 ha, 14.3 individuals vs. Pithecia=103 ha, 3.4 individuals; Saguinus=33 ha,

5.7 individuals, Table 5.1). It is logical that species with similar home range sizes choose to associate because it makes long-lasting associations possible where they would not be between species with disparate home and day ranges. Second, body size itself may also be important. While species of similar sizes often use the same strata, potentially leading to more resource conflicts, species of similar size also face the same predation threats. As a result, associations likely lower predation risk because all participants are looking for the same group of potential predators, a population that is healthy and diverse at the BDFFP (Chapter 4; Cohn-Haft, et al., 1997; Gilbert, 2000; Lenz & dos Reis,

2011). It may be that the antipredator benefit outweighs the negatives of increased resource competition because that increase is small due to the degree of difference between the diets of the two species in the associations.

In summary, I detected niche partitioning in macrohabitat use, diet, and mature forest microhabitat use but found no evidence of microhabitat partitioning in secondary forests. The myriad ways in which primates partition their niches hints at the evolutionary importance of interspecific competition's role in shaping primate behavior and ecology. The apparent lack of microhabitat partitioning in secondary forests, even between the two species (Saguinus and Sapajus) that have their highest relative abundances in this habitat while sharing generalist diets and small-to-medium body sizes, suggests that avenues for niche partitioning in secondary forests are limited,

131 though one cannot rule out the possibility that abundant resources reduce the need for competition avoidance. The patterns of niche partitioning in this study demonstrate just how intertwined body size, diet, and micro- and macrohabitat preferences are, and they suggest that the different evolutionary pathways that species take to reduce competition involve suites of characteristics rather than divergence in a series of isolated socioecological traits.

132

Chapter 6: Primate Abundance in an Anthropogenic Mosaic Landscape in the Central Amazon

Introduction

Human exploitation of tropical forests creates ecosystems that are best characterized as mosaics of cleared land, disturbed habitats and, in some locations, residual undisturbed forest. Indeed, nearly half of all tropical forest is now a combination of regenerating secondary forest and degraded old-growth forest due to human activities like selective logging and ranching (ITTO, 2002). Such anthropogenic habitat disturbance alters ecological variables, like food availability and forest structure, that are crucial for the success of resident primate populations (e.g. Johns, 1985a; Johns, 1988; Johns &

Skorupa, 1987; Sorensen & Fedigan, 2000). While not all primates respond to anthropogenic disturbances in the same fashion, the abundances of many primary forest species are often reduced following human land use (Johns, 1991; Johns &

Skorupa, 1987; Parry, et al., 2007). Among these disturbances, clearcutting for agriculture is the most damaging to primate populations. Carefully managed selective logging, on the other hand, has less of an impact and can present an economic use of forests that is not incompatible with primate survival as long as it is not accompanied by

133 hunting (Skorupa, 1986). Though they are highly disruptive to primates, when agricultural lands are abandoned they frequently return to secondary forest that may be of some use to a few primates after only 10-20 years (e.g. Parry, et al., 2007; This Study).

Following longer periods, on the order of several decades, these forests can supplement primary forest primate habitats and play a role in animal dispersal through disturbed landscapes (Fimbel, 1994; Sorensen & Fedigan, 2000; This Study). While not equivalent to primary forest, some secondary forests may even be preferred habitat for a few primate species (Parry, et al., 2007; This Study, Vulinec, et al., 2006).

In this study I examine the presence and abundance of a community of six primate species across three habitat types: 1) 25-30-year-old secondary forest regenerating on abandoned cattle pastures, 2) primary forest that was lightly selectively logged 25 years ago, and 3) undisturbed primary forest at the Biological Dynamics of

Forest Fragments Project (BDFFP) in the central Amazon. These habitats, together with the BDFFP’s isolated habitat fragments, riparian corridors, and areas of active pasture are typical byproducts of tropical forest cattle ranching operations in the central

Amazon (B. Lenz, personal observation). The BDFFP, which was conceived in 1976 and became a reality in 1979, was designed to experimentally test the minimum size of isolated forest fragment necessary to maintain the majority of the flora and fauna of undisturbed primary forests (Bierregaard Jr. & Gascon, 2001; Gascon & Bierregaard Jr.,

2001). To achieve this goal, replicates of three size classes of square habitat fragments

(1, 10, and 100 ha) were isolated for study in the middle of large clearcut cattle pastures

operated by landowners working with the project. Pastures were in turn surrounded by

134 the vast expanses of undisturbed primary forest that dominated the area before the creation of the BDFFP. Many of the areas cleared for pasture have since been abandoned and are now covered in secondary forests that have reconnected the isolated fragments and surrounding primary forests, though the project periodically reclears a 100-m strip at the fragment borders.

Primate research at the BDFFP has focused on the habitat fragments and undisturbed primary forests (Boyle, et al., In Press; Gilbert, 2003). The present study is the first to sample the secondary forests regenerating on the abandoned pastures and the selectively logged primary forest at the study site, thereby further refining our knowledge of how primates respond to ranching and forest conversion. There is a record of all major disturbances at the BDFFP, an invaluable tool that improves our ability to discuss cause-and-effect relationships and that sets the BDFFP apart from most disturbed tropical habitats which lack detailed data on past habitat disturbances. The site also has low hunting pressure, a wide range of disturbed habitats, and abuts vast undisturbed primary forests. As a result, the primate habitat preferences displayed at the BDFFP should be an accurate reflection of each species' true preferences. This is in contrast to other anthropogenically disturbed sites where primates are restricted in their ability to move between habitat types (if there are even multiple habitats from which to choose). Findings from the BDFFP are therefore important because voluntary use of degraded habitats suggests that species are more likely to survive in them over the long-term whereas the same cannot necessarily be said of species that are forced into degraded areas because they have no alternative.

135

In this chapter I address two separate questions: 1) which primate characteristics are related to presence in secondary growth, selectively logged primary forest, and undisturbed primary forest, and 2) are there interhabitat differences in population density for each species and how influential are fruit availability and tree size on the observed patterns. For the former I follow Boyle and Smith (2010b) who examined primate presence in the BDFFP's habitat fragments. I consider the influence of the percentage of fruit in the diet, body mass, home range size, and group size. Diet is known to be influential on primate survivability in disturbed habitats (Johns & Skorupa,

1987), body size influences the size of the branches (i.e. trees) required for locomotion

(Grand, 1984), and home range size and group size were found to be the second and third most important predictors of primate presence in habitat fragments at the study site (Boyle & Smith, 2010b) so they may also impact species' ability to utilize other disturbed habitats. Given that all primate species at the BDFFP have been reported to utilize undisturbed primary forest (Boyle, 2008), I do not expect to find any primate characteristics that explain their presence there because all species should be encountered on all transects. I also predict that all species will be present in selectively logged primary forest because the logging intensity was low and the time elapsed since the disturbance is on the order of decades, two factors that suggest that no species will have disappeared even if their abundances have declined (Johns & Skorupa, 1987). In secondary forest I predict that primate presence will be explained by a generalist diet, measured as the percentage of fruit consumed, because generalist primates tend to be the most successful in disturbed (and many undisturbed) habitats (Johns, 1985b; Johns

136

& Skorupa, 1987). Body size will also be influential because the mass of arboreal animals determines the size of the branches available to them for locomotion (Grand, 1984). I therefore expect larger species to be absent from a greater percentage of secondary forest transects than smaller species because large expanses of young trees lack the substantial branches necessary to support large-bodied species (Fimbel, 1994).

For my second question on interhabitat density differences for each primate species, I predict that the density of individual species will be related to the level of habitat disturbance. I expect population density to increase with increasing disturbance for the two species with generalist diets, Saguinus midas and Sapajus apella apella. In addition to their generalist diets, these species are small-to-medium sized so their mass should not restrict their use of secondary forests. Saguinus spp. generally prefer disturbed and secondary forests (Garber, 1993; Freese, et al., 1982; Oliveira & Ferrari,

2000) and Sapajus spp. also often thrive in disturbed habitats (Johns, 1985a; Sorensen &

Fedigan, 2000) so I expect that they will heavily utilize these forests to escape competition with the other medium- and large-bodied primates at the site. The densities of the four remaining species should generally decrease with increasing disturbance due to their specialist diets and/or large body sizes. The first of the two largest species, Alouatta macconnelli, will not experience a population density decline from undisturbed to logged forest though secondary forest will have a lower density than both undisturbed and selectively logged primary forests. Howler monkeys should be resilient to some level of disturbance because they are largely folivorous (Julliot,

1996a; Julliot & Sabatier, 1993; Palacios & Rodriguez, 2001), they can shift their diets

137 based on resource availability (Grieser Johns, 1997), and they are the only species able to regularly utilize the highly disturbed 1 ha habitat fragments at the study site (Boyle, et al., In Press). However, their large size may make secondary forest dangerous, so I expect to find their lowest abundances in this habitat. I predict that Ateles paniscus, the largest species at the BDFFP, will become less abundant with increasing disturbance and that they will be absent from secondary growth. Spider monkeys are large-bodied, fast- moving brachiators, and they are therefore without sufficient substrates in secondary forest. They also specialize on undisturbed tall forest (Lehman, 2004; Mittermeier & van

Roosmalen, 1981) and ripe fruit from large mature forest trees (van Roosmalen, 1985) that are absent from secondary forests. I expect that the two seed predators, Chiropotes chiropotes and Pithecia chrysocephala, will have negative responses to disturbance.

Primate seed predators are adversely impacted by selective logging, perhaps due to the poor nutrient content of the seeds of pioneer trees (Plumptre & Grieser Johns, 2001).

While their small to medium sizes will not prohibit their use of secondary forest, the pithecines will occur there at their lowest levels because these forests are comprised primarily of the same pioneer species that produce inadequate seeds in logged forests.

Methods

STUDY SITE

The BDFFP is located in the central Amazon, 80 km north of the city of Manaus, Brazil

(2°25´S, 59° 50´ W) (Didham, 1997). The entire site spans 1,000 km2 (100,000 ha) and encompasses three large, isolated cattle ranches (each is 3,000-4,000 ha) that are

138 surrounded by a sea of largely undisturbed primary forest (Laurance, et al., 2007). I collected data at two of these ranches, Fazenda Dimona and Fazenda Porto Alegre, which are approximately 14 km apart. The primary forest is tall terra firme tropical rain forest with a closed canopy between 30-37 m and emergents to 55 m (Bierregaard Jr. &

Stouffer, 1997). Hunting pressure is low, though it still does occur (B. Lenz, personal observation) and may be becoming a more serious issue as Manaus continues to grow

(Laurance & Luizão, 2007). Annual rainfall at the site ranges from 1,900-2,500 mm

(Stouffer & Bierregaard Jr., 1993) and averaged 2194 mm during this study (Instituto

Nacional de Pesquisas Espaciais, 2012); over the last 106 years the city of Manaus just to the south has received an average of 2115 mm (Cochrane, 2011). There is a pronounced dry season from June to September (Didham, 1997), with average monthly temperatures ranging from 25.8⁰C in February-April to 27.9⁰C in September (Salati,

1985). While the Manaus region, and much of the Amazon Basin in general, are considered lacking in large-scale topography, there are numerous steep hills cut by current and past streams (igarapés) even though there is no significant large-scale relief.

Indeed, the elevation of the site ranges from 50 to 150 m (Bruna & Kress, 2002). The soil is predominantly nutrient-poor yellow latosol (Rylands & Keuroghlian, 1988).

The primate community consists of six species: black-bearded saki (Chiropotes chiropotes) (Veiga, et al., 2008)1, golden-faced saki (Pithecia chrysocephala) (Marsh, In

139 press)1, golden-handed tamarin (Saguinus midas) (Mittermeier, et al., 2008b), Guiana spider monkey (Ateles paniscus) (Mittermeier, et al., 2008a), Guianan brown capuchin

(Sapajus apella apella) (Rylands, et al., 2008), and Guianan red howler monkey (Alouatta macconnelli) (Boubli, et al., 2008). I follow the taxonomy of the IUCN Redlist (IUCN,

2012), though I place the Guianan brown capuchin in the genus Sapajus rather than

Cebus (Lynch Alfaro, et al., 2012a; Lynch Alfaro, et al., 2012b; Silva Jr., 2001; Silva Jr.,

2002). Hereafter each species is referred to by genus alone.

DATA COLLECTION

Primate Sampling

I collected data over a three-year period, November-December 2008, June-November

2009, and January-June 2010, using eight line transects (Figure 6.1) separated by 0.575-

15.0 km (spacing was not equal due to landscape limitations) that were flagged every 10 m. The habitats sampled are described in Table 6.1 and Table 6.2.

1 The IUCN Red List (Veiga & Marsh, 2008) refers to this taxon as Pithecia pithecia chrysocephala but notes that a taxonomic revision is currently underway by Marsh (In press), whose classification I have followed.

140

Figure 6.1: Transect locations

Table 6.1: Transect information

Transect Habitat Age (yr)* Length (km) Replicates Total (km) 1 Undisturbed primary forest - 0.03 47 1.41 Secondary forest (burned) 25-30 0.66 47 31.02 2 Undisturbed primary forest - 1.07 42 44.94 3 Undisturbed primary forest - 1.10 38 41.80 4 Undisturbed primary forest - 0.85 37 31.45 5 Undisturbed primary forest - 1.00 46 46.00 Secondary forest (unburned) 25 0.71 46 32.66 6 Undisturbed primary forest - 1.00 46 46.00 Secondary forest (unburned) 25 0.82 44 36.08 7 Logged primary forest 25 1.03 58 59.74 Secondary forest (unburned) 25 0.07 57 3.99 8 Logged primary forest 25 1.06 55 58.30 Secondary forest (unburned) 25 0.64 50 32.00 * Approximate age relative to 2008

Table 6.2: Total area sampled by habitat

Habitat Length (km) Total (with replicates) (km) Secondary forest (burned) 0.66 31.02 Secondary forest (unburned) 2.24 104.73 Logged primary forest 2.09 118.04 Undisturbed primary forest 5.05 211.60

141

I began sampling each transect pair at 06:45 and 13:15, avoiding mid-day sampling when possible and allowing a minimum of 6.25 hrs between resampling, more than the three hours that has been suggested as a sufficient rest period (Peres, 1999).

Transect pairs were never sampled on consecutive days. Two observers, always 5-10 m apart and with the author always in the lead, walked transects at an average speed of

1.01 km hr-1, pausing for 7-10 s every 10 m. Upon sighting a group, I used a Nikon

Forestry 550 laser rangefinder to measure the sighting distance and height of the first individual and a Suunto MC-2G global compass to measure the sighting angle. I recorded our exact location on the transect, counted the number of individuals, and estimated group spread along the longest axis and the axis perpendicular to it.

For species-specific tests of interhabitat differences in primate density, I consider density in undisturbed primary forest as a baseline and compare it to densities in disturbed habitats. Though a common methodology, confidently attributing interhabitat abundance differences to primate behavior rather than to pre-existing differences between the areas sampled is somewhat problematic given the spatial heterogeneity of tropical forests (Grieser Johns, 2001). I partially account for this limitation with line transects that sample multiple habitats to somewhat reduce these concerns, though this criticism remains valid and needs to be acknowledged.

Botanical Sampling

Trees ≥10 cm dbh (diameter at breast height, or 1.3 m) in mature forests (undisturbed and selectively logged primary forest and habitat fragments) were identified, measured,

142 and mapped in 10-m intervals to a depth of 2 m on both sides of each transect; in all secondary forests the dbh threshold was ≥5 cm because many pioneer trees mature at smaller sizes. Identifications, following the taxonomy of The Plant List (2010), were done in the field by a top expert on Amazonian trees, Paulo Apóstolo Assunção (da S. Ribeiro, et al., 1999). I tagged fruiting trees that are part of primate diets using Assunção's knowledge and a fruit database derived from the literature (Andresen, 2000; Boyle,

2008; Buchanan, et al., 1981; Egler, 1986; Fragazy, et al., 2004; Frazão, 1992; Gaulin,

1977; Gaulin & Gaulin, 1982; Guillotin, et al., 1994; Julliot & Sabatier, 1993; Mittermeier,

1977; Neves & Rylands, 1991; Norconk, 1996; Norconk & Conklin-Brittain, 2004; Oliveira

& Ferrari, 2008; Pack, et al., 1999; Port-Carvalho & Ferrari, 2004; Santamaría-Gómez,

1999; Setz, 1993; Silva & Ferrari, 2007; van Roosmalen, et al., 1981; Veracini, 2000;

Wright, 2004) (note: diet data for Saguinus midas were so limited that I also included data from the closely related S. niger). I then collected monthly data on the reproductive phenology of tagged individuals (presence/absence of buds, flowers, immature fruit, and ripe fruit).

Tree species entered the diet database on an eaten vs. not eaten basis rather than by including only the most important species for each primate. I did this for two reasons: 1) almost all of the diet data were collected at other sites so I wanted to be as inclusive as possible given the spatial heterogeneity and diversity of tropical tree communities, and 2) many published diet lists lack consumption frequencies. As a result,

I expect to overestimate fruit availability in secondary forests for most, if not all, primate species. I anticipate this because, for example, if a mature forest fruit specialist like

143

Ateles consumed a common secondary forest species like Cecropia sciadophylla on a single occasion that species entered their diet database. The result being that secondary forests were scored as having far more fruit available than if the diet list only contained species that accounted for, say, >5% of each primate's diet in which case Cecropia would not have been included. My fruit availability estimate is therefore a measure of all fruit that could be consumed rather than a list of only important resources.

The three transects that I used to sample 25-year-old secondary forest on unburned pasture at the Fazenda Porto Alegre were in a sea of secondary growth of approximately 1340 ha (Figure 6.1). The transect at Fazenda Dimona that I used to sample 25-30-year-old secondary growth on burned pasture was in a stand of approximately 1325 ha (Figure 6.1), though this is likely an over-estimate because it includes several riparian corridors and contains a mixture of secondary forest ages due to a complex land use history.

Trees along the two transects in selectively logged forest were likely harvested in

144 my transects, I also believe it is likely that the logged area does not cover significantly more than 250 ha. I recorded the impacts of logging to a depth of 10 m on both sides of transects for a total area sampled of 4.18 ha, noting the number of stumps (N=2, 0.48 stumps ha-1) and the number of abandoned logging paths (~10 m wide) with relatively open canopy above them that crossed the transects (N=5, coverage=0.1 ha or 2.4% of the area sampled). According to tree expert Paulo Apóstolo Assunção (da S. Ribeiro, et al., 1999), the logged trees were most likely Manilkara huberi, part of the diet of at least four of the study species: Alouatta, Ateles, Chiropotes, and Sapajus (Boyle, 2008;

Fragazy, et al., 2004; Neves & Rylands, 1991; Simmen & Sabatier, 1996). It is possible that this species was cut either for its timber or for its latex as there are unconfirmed reports of this practice in the 1970s at the Floresta Nacional do Tapajós (Michael Keller, personal communication).

ANALYSES & STATISTICS

Fruit Availability & Habitat Differences

I constructed a reproductive phenology database using the data collected during this study as well as data from other studies at the BDFFP and in the Manaus region

(Alencar, 1998; Bentos, et al., 2008; Boyle, 2008; Camargo, et al., 2008; Lemos, 1998;

Oliveira, 1997). To estimate the monthly fruit production of adult trees (≥10 cm dbh, unless smaller mature individuals were found at the BDFFP, Bentos, et al., 2008) I used

145 an allometric equation (Peters, et al., 1988) that has been used in previous studies of primate ecology (Chapman, 1990a; Sorensen & Fedigan, 2000):

Fr=47dbh1.9 where Fr is the mass of the fruit and dbh is the dbh of the tree. I then combined the reproductive phenology database, tree identifications and measurements, and the primate diet database to calculate monthly fruit availability for each primate species. I used the scientific literature to determine the reproductive system of each tree species

(or genus when species-level information was not available). I make the generous assumptions that 100% of the individuals in monoecious and hermaphroditic taxa produced fruit and that all individuals of dioecious taxa also produced fruit, but for the latter I assign a fruit biomass of 50% to each individual to account for the fact that only approximately half of the individuals are capable of bearing fruit.

Primate Density Calculations

I calculated density as the number of individuals in a given area. To determine the area sampled, I estimated species-specific transect widths using the perpendicular distance to the first individual sighted and a 50% criterion for fall-off distance (Whitesides, et al.,

146 forest on unburned pastures that are the focus of this study as well as from additional sampling I conducted in other mature forests (10 ha and 100 ha fragments) and two age classes of secondary growth on burned pastures (19- and 25-30-year-old). This bolsters the number of sightings and allows me to obtain the best possible fall-off distance estimates, though before combining sightings from multiple habitats I tested for differences using Kruskal Wallis and Mann Whitney tests (see Statistics).

I followed the area sampled and density formulas of Whitesides et al. (1988) except in instances where a transect terminus sampled the same habitat as the rest of the transect (as opposed to a transect that ended at the border between habitats). In this case I modified Whitesides' area sampled to include the semi-circle sampled beyond the end of the transect by adding the 0.5πR2 term to Whitesides' original equation:

2 2 Area = 2(S/2+ED)*Lt + 0.5πR = 2(S/2+ED)*Lt + 0.5π(S/2+ED)

with ED (Effective Distance) = (Nt/Nf)*(FD)

where T=total width sampled, Lt=total length sampled, Wt=species-specific transect width on one side of the transect, S=estimate of species-typical mean group spread

(radius)=species-specific effective distance plus one half the average group spread.

147

Statistics

To compare fruit availability and tree size among secondary growth, selectively logged forest, and undisturbed primary forest, I use Kruskal Wallis tests followed by pairwise comparisons (see Table 6.3 for an overview of all statistical tests).

Table 6.3: Summary of statistical tests

Question Test Variables (i=indep.; d=dep.) Do dbh & fruit production vary between habitats? Kruskal Wallis (d) Fruit production, (i) habitat (d) Dbh, (i) habitat What traits influence presence in each habitat? Firth logistic regression (d) Presence (0/1) (i) Home range size (i) % fruit in the diet (i) Body mass (i) Group size Combine habitats in transect width estimation? Kruskal Wallis; Mann Whitney (d) Perpendicular distance (i) Habitat Are there interhabitat density differences? GEE & planned comparisons (d) Species-specific density (i) Habitat Which independent variables predict density? GEE (d) Species-specific density (i) Habitat (i) Dbh (i) Fruit available

Primate presence/absence analyses using several primate traits (Table 6.4), including an arcsine transformation of percent of fruit in the diet, followed Boyle and

Smith (2010b). However, I used Firth logistic regression (Heinze & Schemper, 2002;

UCLA Academic Technology Services Statistical Consulting Group, 2012) rather than the standard generalized linear model due to quasi-complete separation. These analyses use only data from long transects (≥500 m). For secondary forests, I combined data from secondary forests on burned and unburned pastures that are the same age (25-30 years old) because here I am not interested in fine-scale differences between the two forest

148 communities. Each transect was scored 1/0 for the presence/absence of each primate species based on whether or not each species was observed over all transect samples.

Primate characteristics entered the model as covariates (Boyle & Smith, 2010b).

Table 6.4: Characteristics of the BDFFP primates (Boyle & Smith, 2010b)

To justify my combination of data from multiple habitats to improve estimation of fall-off distances, I used the Kruskal Wallis test to look for differences in species- specific perpendicular distance to the transect between all habitats. To be certain that I did not overlook differences I also reclassified the forest types into two broader habitat types, secondary growth and mature forest (primary forest, selectively logged forest, and habitat fragments), and conducted Mann Whitney tests.

Primate surveys yield abundance (count) data and can therefore be modeled with generalized linear models utilizing the Poisson or negative binomial distributions.

Count data are frequently over-dispersed, which was the case in this study due to an excess of zeroes (i.e. zero inflation) that is common in ecological sampling (Zuur, et al.,

149

2009). To deal with this issue I used negative binomial models because they often outperform other methods of handling over-dispersion (Warton, 2005). I used a negative binomial generalized estimating equations (GEE) with a log link to account for repeated measures and included the natural log of the area sampled as an offset to treat counts as densities and allow for intertransect differences in area sampled. I use

QIC to determine the best correlation matrix and a model-based estimator because I have fewer than 20 clusters (Garson, 2012). For all GEEs I report the full main effects model to avoid the many issues that arise with the use of stepwise methods (Babyak,

2004).

To test for interhabitat density differences for each primate species I used GEE planned comparisons, testing only the impact of habitat with the least significant difference test. I ran the habitat-only model because it isolated the interhabitat differences in which I am most interested (i.e. the test does not adjust the means for continuous covariates). I then ran a full effects GEE model with habitat, average dbh, and fruit availability to determine the relative influences of all covariates on abundance.

I calculated the standard deviation for each population density estimate as the weighted standard deviation where the weight is the area sampled per transect (km2); each transect survey was considered a separate replicate.

I tested for multicollinearity among all predictors using a conservative variance inflation factor (VIF) threshold of VIF≤4.0 (Chatterjee, et al., 2000); no covariates required removal. Dummy coding can influence multicollinearity statistics when at least

150 one variable in categorical, so I ran multicollinearity tests with different coding schemes to determine the scheme that best reduced multicollinearity (Wissmann, et al., 2007).

Analyses were conducted in SPSS 19.0.0.1 and SAS 9.3.

Results

My ability to estimate fruit availability for each primate depended on the number of fruiting tree species from their diet list that were found at the BDFFP, and of those species, the number for which I was able to estimate reproductive phenology (Table

6.5). Fruit availability varied across habitats for all species other than Pithecia (for all cases N=116 and df=2: Alouatta: 2=27.286, p<0.001; Ateles 2=34.996, p<0.001;

Chiropotes: 2=9.702, p=0.008; Pithecia: 2=3.751, p=0.153; Saguinus: 2=30.739, p<0.001; Sapajus: 2=26.624, p<0.001). Pairwise tests for all primates, excluding

Pithecia, showed that interhabitat fruit relationships were all the same: undisturbed primary forest and selectively logged primary forest did not differ while secondary forests had the greatest fruit availability. There were also significant interhabitat differences in dbh (N=4010, 2=1081.027, df=2, p<0.001) where undisturbed (=21.29,

=12.47) and selectively logged (=21.38, =13.93) forests did not differ (2=20.474, p=1.000), while secondary forest trees (=11.94, =7.32) were smaller than the trees of both undisturbed primary forest (2=1219.397, p=<0.001) and selectively logged forest

(2=-1198.923, p<0.001).

151

Table 6.5: The percentages of adult trees and tree species for which I estimated reproductive phenology

# Estimated % of Estimate in the Diet of Each Primate Species Habitat (% of Total) Alouatta Ateles Chiropotes Pithecia Saguinus Sapajus Secondary forest (unburned) 780 (65) 20 25 28 12 19 28 Individuals Logged primary forest 643( 66) 16 12 37 15 6 17 Undisturbed primary forest 1632 (68) 19 13 45 18 5 15 Secondary forest (unburned) 168 (47) 13 14 23 9 8 16 Species Logged primary forest 278 (50) 13 9 22 10 6 13 Undisturbed primary forest 424 (50) 12 10 25 11 4 11

SPECIES PRESENCE/ABSENCE:

Each primate species was present on the majority of the five undisturbed primary forest transects (Alouatta: 5, Ateles: 4, Chiropotes: 5, Pithecia: 5, Saguinus: 5, Sapajus: 4). None of the primate characteristics predicted undisturbed primary forest use (Intercept:

=1.4783, df=1, p=0.5851; Fruit: =0.000, df=1, p=1.000; Weight: =0.000, df=1, p=1.000; Group size: =0.000, df=1, p=1.000; Home range size: =0.000, df=1, p=1.000) and the overall model was not significant (LR: 2=0.000, df=4, p=1.000, Score:

2=2.3557, df=4, p=0.6706, Wald: 2=0.000, df=4, p=1.000). Almost every primate species was also present on the two selectively logged primary forest transects

(Alouatta: 2, Ateles: 2, Chiropotes: 2, Pithecia: 1, Saguinus: 2, Sapajus: 2). Similarly, primate characteristics were not related to presence in logged primary forest (Intercept:

=2.3979, df=1, p=0.6906; Fruit: =0.000, df=1, p=1.000; Weight: =0.000, df=1, p=1.000; Group size: =0.000, df=1, p=1.000; Home range size: =0.000, df=1, p=1.000) and the overall model was not significant (LR: 2=0.000, df=4, p=1.000, Score:

2=8.0522, df=4, p=0.0897, Wald: 2=0.000, df=4, p=1.000). Presence on each of the 4

152 secondary forest transects was more varied (Alouatta: 2, Ateles: 0, Chiropotes: 3,

Pithecia: 3, Saguinus: 4, Sapajus: 4). Body weight was the only trait that was (negatively) related to the use of secondary forests (Intercept: =5.5793, df=1, p=0.0958; Fruit: =-

3.4409, df=1, p=0.2465; Weight: =-0.5373, df=1, p=0.0393; Group size: =0.0102, df=1, p=0.9435; Home range size: =0.00215, df=1, p=0.7603) and the overall model was significant (LR: 2=11.0569, df=4, p=0.0259, Score: 2=12.2142, df=4, p=0.0158, Wald:

2=5.3315, df=4, p=0.2549).

INTERHABITAT DIFFERENCES IN PRIMATE DENSITY:

Kruskal Wallis tests on the average perpendicular distance to the trail for each primate species between the all habitats sampled, including those not used in the current study (e.g. forest fragments), detected no differences (Alouatta: N=55, 4 habitats, 2=5.144, df=3, p=0.162; Ateles: N=22, 3 habitats, 2=1.790, df=2, p=0.409;

Chiropotes: N=40, 5 habitats, 2=4.082, df=4, p=0.395; Pithecia: N=21, 5 habitats,

2=5.114, df=4, p=0.276; Saguinus: N=75, 6 habitats, 2=8.835, df=5, p=0.116; Sapajus:

N=59, 6 habitats, 2=1.847, df=5, p=0.870). Mann Whitney tests for perpendicular distance differences between two broader habitat types, secondary growth (SG) vs. mature forest (MF), revealed that data for Alouatta needed to be split but that all other species could still be combined (Alouatta: N= 53 (MF), 2 (SG), Z=-2.025, p=0.034; Ateles:

N= 22 (MF), 0 (SG), no test; Chiropotes: N= 36 (MF), 4 (SG), Z= -1.782, p=0.074; Pithecia:

N=16 (MF), 5 (SG), Z=0.186, p=0.208; Saguinus: N= 37 (MF), 37 (SG), Z= -1.109, p=0.282;

Sapajus: N= 36 (MF), 23 (SG), Z=-0.054, p=0.975).

153

In the text below I report the models for the habitat only and full GEEs but I do not report the estimated marginal means used in the pairwise interhabitat density tests.

Instead, in Table 6.6 I report the means and weighted standard deviations (number of individuals km2-1) along with the p values obtained in the GEEs. In the regression results the covariate "Habitat" refers to the influence of habitat type (secondary growth on unburned pasture, selectively logged forest, and undisturbed primary forest) on the dependent variable.

Table 6.6: Primate density by habitat (# individuals/km2())

Habitat Alouatta Ateles Chiropotes Pithecia Saguinus Sapajus Undisturbed primary forest 6.8(20.5)a,b 1.8(12.0) 4.9(22.7)c,d 2.2(11.9)e 6.3(23.2)f,g 8.3(34.0)i,j Logged primary forest 4.3(16.3)a 1.1(7.4) 1.2(9.5)c 1.0(7.5)e 7.2(21.1)f,h 4.6(17.8)i,k Secondary forest (unburned) 2.3(23.2)b 0.0* 2.1(18.2)d 1.7(12.2) 18.3(49.4)g,h 14.5(58.8)j,k a-k Letters denote significant intraspecies pairwise differences; * not included in tests a-k p values (b,c,d,g,h,k<0.001; f,i=0.001; j=0.006; a,e=0.012)

Alouatta

The habitat only GEE showed that habitat was a significant predictor of Alouatta density

(Intercept: Wald 2=102.312, df=1, p<0.001; Habitat: Wald 2=102.312, df=2, p=0.003).

The results of planned comparisons are summarized in Table 6.6. The full GEE demonstrated that habitat type and fruit availability significantly predicted Alouatta density while average dbh did not (Intercept: =2.205, df=1, p=0.376; Habitat:

(Secondary forest=-1.541, undisturbed=0.502, logged=reference), df=2, p=0.006; Fruit:

=0.018, df=1, p=0.007; dbh: =-0.065, df=1, p=0.569).

154

Ateles

Spider monkeys were not observed in secondary forests and that habitat was therefore excluded from analyses, leaving a model with a two-level habitat variable. The habitat only GEE showed that habitat was not a significant predictor of Ateles density (Intercept:

Wald 2=2.473, df=1, p=0.116; Habitat: Wald 2=0.869, df=1, p=0.351). The results of planned comparisons are summarized in Table 6.6. The full GEE, again with a two-level habitat variable, showed that fruit availability was the only significant predictor of Ateles density (Intercept: =-4.656, df=1, p=0.293; Habitat: (undisturbed=0.5287, logged=reference), df=1, p=0.131; Fruit: =0.046, df=1, p=0.002; dbh: =0.161, df=1, p=0.446).

Chiropotes

The habitat only GEE showed that habitat significantly predicted Chiropotes density

(Intercept: Wald 2=59.862, df=1, p<0.001; Habitat: Wald 2=35.801, df=2, p<0.001).

The results of planned comparisons are summarized in Table 6.6. The full GEE demonstrated that habitat type, fruit availability, and average dbh were all significant predictors of Chiropotes density (Intercept: =-17.186, df=1, p<0.001; Habitat: 

(Secondary forest=8.139, undisturbed=1.521, logged=reference), df=2, p<0.001; Fruit:

=-0.037, df=1, p<0.001; dbh: =0.839, df=1, p<0.001).

155

Pithecia

The habitat only GEE showed that habitat alone was a not significant predictor of

Pithecia density (Intercept: Wald 2=7.934, df=1, p=0.005; Habitat: Wald 2=4.912, df=2, p=0.086). The results of planned comparisons are summarized in Table 6.6. The full GEE demonstrated that habitat type and fruit availability significantly predicted Pithecia density while average dbh did not (Intercept: =-5.012, df=1, p=0.285; Habitat:

(Secondary forest=2.594, undisturbed=0.951, logged=reference), df=2, p=0.027; Fruit:

=0.022, df=1, p=0.030; dbh: =0.209, df=1, p=0.287).

Saguinus

The habitat only GEE showed that habitat was a significant predictor of Saguinus density

(Intercept: Wald 2=2143.041, df=1, p<0.001; Habitat: Wald 2=46.101, df=2, p<0.001).

The results of planned comparisons are summarized in Table 6.6. The full GEE demonstrated that habitat type and average dbh predicted Saguinus density while fruit availability did not (Intercept: =-18.309, df=1, p<0.001; Habitat: (Secondary forest=9.764, undisturbed=-0.484, logged=reference), df=2, p<0.001; Fruit: =-0.003, df=1, p=0.390; dbh: =0.953, df=1, p<0.001).

Sapajus

The habitat only GEE showed that habitat predicted Sapajus density (Intercept: Wald

2=811.911, df=1, p<0.001; Habitat: Wald 2=28.410, df=2, p<0.001). The results of

156 planned comparisons are summarized in Table 6.6. The full GEE demonstrated that habitat type and average dbh were significant predictors of Sapajus density while fruit availability was not (Intercept: =-2.527, df=1, p=0.285; Habitat: (Secondary forest=2.328, undisturbed=0.023, logged=reference), df=2, p=0.027; Fruit: =0.005, df=1, p=0.225; dbh: =0.217, df=1, p=0.015).

Discussion

In this study I examined the impacts of two types of anthropogenic disturbance, selective logging and secondary forest regeneration on abandoned cattle pastures, on primate presence and density. As predicted, none of the primate characteristics that I tested was related to presence in either undisturbed primary forests or logged forests because most primate species were detected on most transects. These findings were expected because all species at the BDFFP are known to utilize undisturbed primary forest (Boyle, 2008) and because low intensity logging and nearly three decades of recovery time have created a selectively logged forest that is similar to undisturbed primary forest in some ways (Johns & Skorupa, 1987). This does not, however, mean that logging had no impact on the primate community (see below).

In secondary growth only body weight was related to primate presence. This pattern was driven by the absence of the two largest species, Ateles and Alouatta, from

100% and 50%, respectively, of secondary growth transects. I predicted that body size would be important because secondary forests are dominated by young, small trees with small branches that cannot support the weight of large animals who would risk

157 falls, injuries, and even death by choosing substrates that are too weak. Selection pressure from falls is a significant force shaping primate habitat use. Indeed, primates seem to fracture a large number of bones in falls (Bramblett, 1967; Jurmain, 1997;

Lovell, 1990; Lovell, 1991; Schultz, 1944), leading to more cautious habitat selection by larger species whose greater masses increase the risk of falls and serious injuries from falls (B. Lenz personal observation; Cartmill & Milton, 1977; Vogel, 1988). Body size may also influence secondary forest use in that young secondary forests simply may not provide enough food per unit area to permit substantial use by large primates who require more food than smaller ecologically similar species (Wolfheim, 1983).

I predicted a positive relationship between population density and the level of disturbance for the small- to medium-sized generalists, Saguinus and Sapajus. This prediction that was accurate for Saguinus, whose population density increased from undisturbed primary forest to selectively logged primary forest to its peak in secondary forest. This species' small size, generalist diet, and widespread success in highly disturbed (Garber, 1993; Oliveira & Ferrari, 2000) and mosaic habitats (Ayres, 1983;

Freese, et al., 1982; Johns, 1991) make them excellent pioneer species and explain their high density in secondary and selectively logged forests at the BDFFP. The observed preference of Saguinus for logged over undisturbed primary forest suggests that the footprint of mild logging operations that took place decades ago is still influencing primate habitat quality and may be especially apparent in the understory and midstory, habitats that this species frequents.

158

Sapajus, which was most common in secondary forest, did not support my prediction of increasing density with increasing disturbance because they were more abundant in undisturbed primary forests than in selectively logged forest. This species' responses to logging and secondary growth are variable (Appendix 11 and Appendix 12), perhaps due to their large home ranges, which at the BDFFP are 852-915 ha (Spironello,

2001). In anthropogenic mosaic landscapes, large ranges yield groups whose core areas can include multiple distinct habitats among which they move on a daily basis ( B. Lenz personal observation; Fedigan, et al., 1996). The selectively logged area at the site is small (estimated at 250 ha) compared to the large extent of primary and secondary forests. As a result, it may be that Sapajus densities are lowest in logged forest because it is a small habitat that is a portion of the range of 1-2 groups that forage most effectively in habitats at the disturbance extremes of undisturbed and secondary forests. This is in contrast to Saguinus where the extent of selectively logged forest is sufficient to contain the ranges of multiple groups (33 ha each, Table 6.4) therefore making abundance patterns more predictable.

For four species, Alouatta, Ateles, Chiropotes, and Pithecia, I predicted that different body size/diet combinations for each species would generally result in their becoming less abundant with increasing habitat disturbance. My predictions for

Alouatta, the second largest species at the site, were partially correct: their highest density was in undisturbed primary forest but their densities in selectively logged primary and secondary forests were equivalent. It is curious that density was lower in logged primary forest than in undisturbed primary forest because howlers alter their

159 diets under conditions of habitat disturbance (Grieser Johns, 1997) and when necessary rely heavily on young leaves (Julliot, 1996b; Julliot & Sabatier, 1993) that might be more abundant in logged forests (Johns, 1988; Plumptre & Reynolds, 1994). While the characteristics that allow them to exploit logged forest also ought to permit the exploitation of secondary forest, I nonetheless correctly predicted that their lowest abundance would be lowest in secondary forest because Alouatta is the BDFFP's second heaviest primate. Howler monkeys are wary of using smaller trees, like those that dominate secondary growth. Indeed, even in mature forests Alouatta spp. frequently select habitats (Chapman, 1988) and feeding sites (Chapman, 1988; Chapman, 1990b;

Larose, 1996; Leighton & Leighton, 1982) that are comprised of larger trees. While I found no density difference between selectively logged and secondary forests, I suspect that further data collection would reveal the observed difference to be significant. The number of pioneer tree species in secondary growth with easily digestible, poorly defended leaves (Coley, 1980) might increase the attractiveness of this habitat relative to lightly logged primary forest, but the large body size of Alouatta should prevent groups from making extensive use of secondary forests given its smaller trees.

My predictions for Ateles, the largest primate species at the BDFFP, were partially correct: they were not recorded in secondary growth but their densities in undisturbed and selectively logged primary forest were equal. The absence of spider monkeys from secondary growth was not a surprise because Ateles strongly prefers undisturbed forests (Lehman, 2004; Mittermeier & van Roosmalen, 1981). They also specialize on the fruits of large mature forest trees (e.g. Simmen & Sabatier, 1996; van

160

Roosmalen, 1985) rather than the smaller nutrient-poor fruits of pioneer tree species

(Grieser Johns, 1997) that dominate secondary growth. Furthermore, the large size of

Ateles also makes use of secondary forests dangerous enough that the border between mature and secondary forest more or less filters them out of secondary forests. Indeed, all studies that include secondary growth have recorded densities of Ateles spp. of zero in that habitat (Table 6.6). However, although I never encountered them in secondary growth during sampling or ad libitum data collection, spider monkeys apparently still use secondary forests at the BDFFP on rare occasions. For example, they may use these forests as a corridor during their rare, brief forays into the BDFFP's larger habitat fragments (B. Lenz personal observation; Boyle, et al., In Press) or to move between patches of mature forest in Central America (Chapman, et al., 1989; Fedigan, et al.,

1996; Sorensen, 1999). It is interesting that a species that is sensitive to disturbance had equivalent densities in selectively logged and undisturbed primary forests. These habitats may be similar to Ateles due to the low logging levels and long period since logging. While even low intensity selective logging can increase canopy gaps from 1% to

10-20% (Crome, et al., 1992; Johns, 1992; White, 1994), the low intensity and decades of regeneration time at the BDFFP may have left a forest with an acceptable number of canopy pathways due in part to the extent of unlogged islands within the larger logged area (e.g. Johns, 1989). Indeed, I found that the number of stems ha-1 (Chapter 4) and the average dbh in the two habitats did not differ. Within these two mature forest habitats, the only factor that significantly explained Ateles density was fruit availability, which illustrates how important fruit is in the diet of this specialized frugivore.

161

My predictions of declining abundance with increasing habitat disturbance for the seed predators, Chiropotes and Pithecia, were partially correct. The density of

Chiropotes was highest in undisturbed primary forest and equal in the disturbed habitats while Pithecia was fairly uniformly distributed across all habitats with only a single small difference between undisturbed primary forest and secondary forest. Given that both species are small-to-medium sized, body size constraints cannot account for their lower densities in the disturbed forests. Primate seed predators in Africa, and especially the pithecines in the Neotropics, tend to be adversely affected by logging due to their inability to utilize the smaller seeds of pioneer species that increase in abundance following logging (Grieser Johns, 1997; Plumptre & Grieser Johns, 2001).

Chiropotes spp. are generally considered to prefer undisturbed primary forest and not do well in disturbed habitats (Ayers, 1989; Johns & Ayers, 1987; Mittermeier, 1977; This

Study; Wolfheim, 1983) so decreased use of all disturbed habitats may be tenable especially in light of the relationship between tree size and population density.

Pithecia exhibits variable responses to both logging and secondary growth

(Appendix 11 and Appendix 12). This species also prefers habitat fragments with many small trees and lianas at the BDFFP (Schwarzkopf & Rylands, 1989) and uses the lower and central portions of trees, areas with few emergent trees, and habitats dominated by small trees (Walker, 1996). In spite of this, for Pithecia there was no relationship between average dbh and density, though there was an extremely weak relationship between fruit availability and density. Tree size may not have been revealed as important due to my methodology: I consider tree size by transect on scales of

162 hundreds of meters which may obfuscate patches of forest comprised of smaller trees, areas where a species with a small home range like Pithecia (103 ha, Table 6.4) might focus. It may be that this species' relatively even distribution is due to its small size and ability to exploit the smaller trees that dominate secondary forest and are found in scattered areas of dense understory in logged and undisturbed primary forest.

Conservation Implications & Recommendations

I discuss many of the conservation implications of this study along with my recommendations, but would like to separately consider the impacts that primate habitat use patterns have on seed dispersal and habitat regeneration. Primates disperse significant numbers of seeds (Chapman & Russo, 2011) and often account for 25-40% of tropical forest frugivore biomass (Chapman, 1995). In the current study, 50% (3/6) and

67% (4/6) of primate species were found at lower densities in secondary and selectively logged forests, respectively. In selectively logged forests the residual stand may be sufficient to permit enough intrahabitat seed dispersal to ensure that long-term changes in forest composition are not drastic. However, in secondary forests on agricultural lands the seed bank is so depleted that interhabitat seed dispersal from nearby old- growth or old secondary forests is the lone avenue to the regeneration of a diverse mature forest (Chazdon, et al., 2009). The primates of the BDFFP are therefore important agents of forest regeneration because the speed at which seed dispersers return to secondary growth is an important factor in how quickly late successional tree species reach the secondary forest canopy (Steininger, 2000). The primates of the BDFFP

163 also disperse seeds long distances, averaging several hundred meters (up to 1.1 km,

Chapman & Russo, 2011), and 100% of the primate species use selectively logged primary forest while 83% use secondary forest. The true importance of primates in forest regeneration is suggested by Saguinus midas niger, a conspecific of the study species, which was found to disperse 50% of the seeds it consumes into habitats other than the habitat in which they were consumed (19.3% move from primary to secondary forest) (Oliveira & Ferrari, 2000). Such significant input from a small species with a small home range and a generalist diet illustrates just how important primates are in the regeneration of diverse forests.

I would like to close by discussing four recommendations that will help to ensure primate survival in anthropogenic mosaic landscapes. First, hunting must be curtailed because strong hunting pressure can quickly lead to empty forests even in areas of low habitat disturbance (Redford, 1992). When you add hunting pressure to disturbance- related declines, medium to large primates stand little chance of survival. Fortunately, hunting occurs at low levels at the BDFFP, though it may be becoming a more serious issue as the city of Manaus continues to grow rapidly (B. Lenz personal observation ;

Laurance & Luizão, 2007). Second, where it is possible, large patches of undisturbed primary forest should be protected from hunting and left in areas with high levels of anthropogenic disturbance. I believe the expansive forests surrounding the study site are a significant reason that primates are able to successfully utilize so many of the

BDFFP's disturbed habitats, a pattern also noted elsewhere (Fimbel, 1994; Johns, 1991).

These forests are critical because they are refuges for primates while disturbances are

164 on-going and they provide source populations that can later return to disturbed areas when habitat recovery has reached sufficient levels.

Third, government incentive programs should encourage landowners to reduce their impacts on the primate community and the forest itself. In selective logging this includes practices like reduced impact logging (RIL) (Bicknell & Peres, 2010) where the trees to be cut are carefully mapped, felled, and extracted to minimize damage to the residual stand. In areas that contain well-studied primates, it may even be possible to alter the species composition of the extracted trees to ensure that sufficient primate fruit resources remain after the disturbance. Incentives should target different goals in agricultural clearcuts. I recommend leaving important fruiting trees standing in clearcuts so that when agricultural lands are abandoned the secondary forests that regenerate will have mature forest fruits that would otherwise be absent. This will allow more primates to use these habitats, thereby increasing seed dispersal into secondary forests by enticing primates to use this habitat (Wunderle Jr, 1997). This strategy is often aimed at volant species and involves leaving large fruiting trees as perches for frugivorous birds and bats to disperse primary forest seeds into pastures with depauperate seed banks (e.g. Guevara, et al., 1992; Guevara, et al., 2004) though it has also been briefly recommended for primates (Terborgh, 1986; Sorensen & Fedigan,

2000). While large (e.g. ≥80 cm dbh, Guevara, et al., 2004) remnant trees are an effective strategy for attracting bird- and bat-dispersed seeds, these trees will be isolated from primates even when secondary forests begin to regenerate because their canopies are so far above the tops of even decades-old regenerating forests that

165 primates will be unable to reach them unless their trunks are covered in lianas that might act as a ladder (though fallen fruit is still accessible to terrestrial and semi- terrestrial species). To circumvent this, I propose leaving a mixture of large and small fruiting trees. Leaving smaller individuals, which will be more agreeable to landowners because their timber is less valuable due to their small size, will result in secondary forests that are pre-stocked with preferred fruiting tree species whose crowns will be accessible to primates not long after secondary forests take over.

Finally, there is one habitat type that I did not sample but which greatly facilitates interhabitat primate movements: riparian corridors through pastures and secondary growth. All six primate species use riparian corridors at the BDFFP and elsewhere (Estrada, et al., 1994; Green, 1978), though Ateles does not commonly utilize them at the BDFFP (B. Lenz, personal observation) and appears to avoid them at some sites (Lees & Peres, 2008). When the forests of the BDFFP were initially cleared, Brazilian law required 100-m buffer zones around most streams (Bierregaard Jr. & Gascon, 2001)

(30-500 m depending on stream size, Tollefson, 2012), though in 2012 the Congresso

Nacional do Brasil voted to reduce buffer zones from 30-500 m to 5-100 m (Tollefson,

2012). The entirety of these narrower corridors are subject to strong edge effects (Lees

& Peres, 2008) while providing less protection from aerial predators (Lees & Peres,

2009). Any reduction in the ability of riparian corridors to support primates hurts long- term primate survival by limiting colonization/dispersal pathways and by reducing gene flow and seed dispersal. Corridors of at least 400 m (200 m on each side of a stream)

166 have previously been recommended (Lees & Peres, 2008), so not only should the new law be repealed but the old law might even be made more restrictive.

167

Chapter 7: Conclusions

Summary of Findings

I examined a community of six Neotropical primates to determine the impacts of

Amazonian cattle ranching and the anthropogenic mosaic landscape that it creates.

Chapter 4 deals with edge effects in primate distribution on the margins of continuous primary forests. I found that primate presence at the edge was negatively related to the amount of fruit that a species consumes. This relationship also largely held during my examination of the edge-related densities of individual species where I found: 1) positive edge effects in the densities of four species, Alouatta (folivore-frugivore),

Chiropotes (frugivorous seed predator), Saguinus (generalist), and Sapajus (generalist),

2) a negative edge response in one species, Ateles (frugivore), and 3) a neutral response in the final species, Pithecia (frugivorous seed predator).

In Chapter 5 I examined niche partitioning in mature and secondary forests.

Niche partitioning was evident in the composition of individual species' diets as well as in their preferred habitats. In terms of microhabitat variables, I found evidence of partitioning between species dyads in mature forests but an apparent absence of

168 physical microhabitat niche partitioning in secondary forests. The latter may have been the result of a combination of small sample sizes for some species and/or the limitations on potential niche differentiation imposed by the more uniform secondary forests. In both habitats body size was positively related to the height of a species' preferred vertical stratum, suggesting that body size limits the strata that larger species can use.

In Chapter 6 I looked for relationships between primate traits and their presence in secondary forest, selectively logged primary forest, and undisturbed primary forest while also examining interhabitat differences in population density for individual species. No primate characteristics were related to presence in selectively logged or undisturbed primary forests because all species were present on almost every transect in these two habitats. However, in secondary forests body size was negatively related to primate presence. The differences in the interhabitat densities of individual species were variable, with Saguinus and Sapajus, the two generalists at the site, being the only two species with their highest densities in secondary growth.

Implications

The overall message from this study is a positive one: under the right conditions a primate community can adapt to, and rebound from, large-scale cattle ranching. While individual species responded to the presence of the multitude of available habitats in different ways, I nonetheless encountered most species in every habitat. I recorded all six species in undisturbed primary forest (and edge habitat), selectively logged primary

forest, and a 100 ha habitat fragment, though not all species were permanent residents

169 in the latter. I also recorded five of six species in secondary forests as well as in riparian corridors that cross them, while the sixth species, Ateles, seems to use both habitats only on rare occasions based on my observations and the long-term observations of my primary field assistant, Alaercio Marajó de Reis. Four of the six species were found in a

10 ha fragment, though no species were confined to it. In general, it appears that flexible diets and small-to-medium body sizes are related to primate success in this landscape, with the species that combine these two traits, Saguinus and Sapajus, generally fairing the best in disturbed habitats. These patterns suggest that conservation planning in anthropogenic landscapes should focus on the habitat requirements of large-bodied and specialist species because they are less able to adapt to disturbed habitats than species that are small and/or have flexible diets.

Data on primate responses to anthropogenic mosaic habitats are also relevant for reserve planning, expansion, and connection. Edge effect data (Chapter 4), along with patterns of primate persistence in habitat fragments (Boyle, et al., In Press; Gilbert,

2003), can be used to determine the amount of habitat available to, and needed by, species in isolated reserves so that we can ensure that reserves are large enough to support viable populations. Reserve expansion can generally happen in two ways: 1) where intact, unprotected primary forest abuts a reserve it can be legally annexed to an existing protected area, and 2) where reserves are surrounded by degraded anthropogenic habitats, the land at margins of the reserve can be protected and allowed to regenerate. In the latter setting, data on edge effects (Chapter 4) and primate use of secondary forests (Chapter 5 and Chapter 6) will be crucial in estimating

170 which primates will use these regenerating forests, how long it will take them to return to these areas, and whether there is sufficient ecological space for multiple species to heavily utilize the newly protected secondary forests. In many cases the two options for reserve expansion just discussed may not be practical, affordable, or possible, and it may make more sense to create habitat corridors to connect isolated habitats and primate populations. In order to entice primates to use habitat corridors to promote gene flow between otherwise isolated populations, primate behavior and ecology should receive strong consideration during corridor planning to ensure that corridors are sufficiently wide and contain an appropriate mixture of tree species and sizes.

Especially important to these efforts will be data like those reported in the present study on the ability of species to use riparian corridors (Chapter 6), proneness to edge effects (Chapter 4), and on general habitat preferences in a disturbed landscape

(Chapter 6).

Primate responses to anthropogenic mosaic landscapes provide information not only on primate adaptability but also on the importance of primates in the growth and maintenance of healthy forests. Perhaps the most significant influence that primates have on forest composition stems from their role as seed dispersers. Primate seed dispersal is a service that is especially critical in the regeneration of secondary forests on abandoned agricultural fields because the return of seed dispersers to these habitats strongly influences how quickly late successional trees reach the canopy (Steininger,

2000).

171

Primate seed dispersal into secondary forests is so crucial because agricultural clearcuts often destroy much of the soil seed bank (Aide & Cavelier, 1994; López-Toledo

& Martínez-Ramos, 2011; Quintana-Ascencio, et al., 1996; Wijdeven & Kuzee, 2000). To create effective pastures/farmland, mature forests must be felled, dried, and burned to release the nutrients stored in their vegetation to grow grass or crops on what are often nutrient-poor soils. This process decimates the seed bank, thereby drastically reducing the species richness of the secondary forests that follow abandonment, for several reasons: 1) the seeds of most old-growth tropical forest trees are especially susceptible to being destroyed by the high heat of fires (Brinkmann & Vieira, 1971), 2) deforestation, soil disturbance, and weeding promote the germination of seeds present at the time of deforestation that are not replaced due to the level of disturbance (Aide

& Cavelier, 1994; Garwood, 1989; Uhl, et al., 1981; Uhl, et al., 1988) and it also seems likely that these initial seedlings do not survive because they are destroyed by the burning of felled vegetation, grazing, or weeding, and 3) burning and dense grass may also impede or prevent seed germination, leading to a reduction in the number of viable seeds in the seed bank (Nepstad, et al., 1991; Uhl, et al., 1981). In the end, the composition of the seedbank in agricultural clearcuts is largely at the mercy of seed colonization from nearby mature forests or old secondary forests (Chazdon, et al.,

2009). As a result, an examination of the diets of the primate species that are most frequently found in secondary forests, overwhelmingly Saguinus and Sapajus at the

BDFFP, yields information on which mature forest tree species are likely to colonize secondary forests that are initially dominated by pioneer trees. Perhaps more

172 importantly, the diets of primates that avoid/rarely use secondary forests, like Ateles, provide us a glimpse of the mature forest seeds that are not being dispersed into these depleted seed banks without our having to wait 50-100 years to survey the tree communities in old secondary forests to learn which tree species have been dispersal- limited. Data on primate diets can therefore be used to guide reforestation programs seeking to speed the regeneration of primate-friendly secondary forests by bolstering the seed bank with un- and under-represented tree species.

I began the discussion of the implications of this study by noting that I have demonstrated that primates can survive in anthropogenic mosaic landscapes "under the right conditions," a qualifier that merits consideration. Although hunting pressure at the

BDFFP may be rising (B. Lenz personal observation; Laurance & Luizão, 2007), it can currently be considered to be low. This leaves habitat disturbance as the principle anthropogenic pressure with which the BDFFP's primates must contend. Across the

Amazon local hunting pressure is hard to quantify, varies from site to site due to local factors, and its effects on primate populations are often confounded by factors like habitat quality and heterogeneity (Wiederholt, et al., 2010). One thing that is agreed upon is that high levels of hunting of large animals that reproduce slowly, as is the case for primates, can wipe out populations even where the forests themselves appear to be free of serious anthropogenic disturbances (Redford, 1992). In the case of the BDFFP, the "right conditions" for primate survival include low hunting pressure so that primates are not forced to adapt to drastically altered habitats while also having their populations decimated by hunting. In mosaic landscapes where hunting pressure is high, I think that

173 there is little hope that medium- and large-bodied primates will escape drastic population reductions or even local extinctions until that hunting pressure is lessened because severe habitat disturbances in concert with high hunting pressure are simply too great of a stress.

The other condition that influenced the primate patterns that I observed is the amount of primary forest that surrounds the BDFFP. The study site is encompassed by an expanse of primary forest that serves as a refuge during habitat conversion, active agriculture, and initial forest regeneration. These forests also hold source populations

(as opposed to being a sink) that are able to repopulate disturbed areas when they have recovered enough to once again be suitable habitat. The likely impact of the presence of this type of extensive intact forest on primate persistence has also been noted in other studies (Fimbel, 1994; Johns, 1991), though these conditions do not exist across the

Amazon Basin. The majority of the deforestation in the Brazilian Amazon has occurred in a crescent moon-shaped belt across the southern and eastern Amazon referred to as the arc of deforestation (Margulis, 2004). This extensive habitat loss has left many forested sites in this region without large areas of undisturbed forest nearby to serve as refuges and areas for the maintenance of source populations. As a result, many primates, in particular those with large bodies and/or specialist diets, are much less likely to persist in this area than they are in regions with large tracts of undisturbed forest. This, however, is not the case in the less disturbed central and western Amazon.

For example, in 2003 in the region of the rapidly expanding city of Manaus, where this study took place, mature forests still covered 77% of the total area (down from 91% in

174

1973) (Prates-Clark, et al., 2009) with secondary forests occupying 75-80% of the area that has been cleared (Fearnside, 2006; Prates-Clark, et al., 2009). As a result, the findings of the current study are more easily extrapolated to sites in the central and western Amazon than they are to research conducted in the arc of deforestation.

The data from the BDFFP, where the conditions are "right" for primate persistence following cattle ranching, are valuable because they can be applied to other similar areas. Findings from the BDFFP are also important because the low hunting pressure and extensive primary forests mean that primate use of disturbed habitats is voluntary. This suggests that these species are likely to persist in these degraded habitats over the long-term, provided they are protected from hunting, whereas the same cannot be said of species that are forced into degraded areas because extreme habitat loss or strong hunting pressure have left them without an alternative.

Where Do We Go From Here?

Over the course of this dissertation I have made several recommendations aimed at helping ensure primate survival in anthropogenic landscapes: expand the amount of forest that must be left along the banks of streams when fields are cleared (Chapter 4,

Chapter 6), preserve undisturbed forests where possible, discourage hunting, provide land use incentives that will reduce the impacts of habitat conversion (e.g. reduced impact logging), and leave important fruiting trees, both large and small, in areas that are cleared for pastures/fields so that secondary forests will more quickly return to viable primate habitat following the abandonment of agriculture (Chapter 6).

175

In addition to these ideas, I think that a program designed to replenish the seed bank of newly abandoned lands could be extremely beneficial to the health of the forest and to its value to primates. As I mentioned above, primates that use secondary forests help to restore the diversity of seed banks decimated by clearcuts and pastures, though species that avoid or rarely use secondary forests (e.g. spider monkeys) play little role in first 50-100 years of regeneration. Because a diverse seed bank promotes a diverse forest, seed dispersal into secondary forests bolsters tree species richness thereby making these habitats more able to meet the needs of all species in the primate community. Primate seed dispersal also helps mold secondary forests for future primate use because primates disperse seeds of the fruit that they eat, thereby ensuring that the trees that upon which they rely will be a part of the canopy of secondary forests in the future.

The proposed program involves gathering the seeds of mature forest tree species that are important in primate diets and dropping them from planes over abandoned agricultural lands. A program like this is necessary because the trees that compose much of the mature forest canopy have large seeds that are naturally dispersed into abandoned agricultural areas at rates so low that animal dispersers cannot be relied upon to disperse enough of them for secondary forests to approach primary forest diversity (Parrotta, et al., 1997; Tucker & Murphy, 1997; Wunderle Jr,

1997). This may be due to the fact that the large primate species that are most likely to consume large seeds tend to make less use of secondary forests (Chapter 6). Such a program would be a cost effective way to promote the regeneration of diverse

176 secondary forests because it requires no management following the seed drop and it also poses no threat to land owners or the environment. One initial hesitation that many may have is that all of the anthropogenically dispersed seeds would simply provide a bonanza for terrestrial seed predators and therefore not bolster the seed bank. The solution to this hurdle is to treat the seeds with natural plant defensive compounds like quinolizidine alkaloids (QAs) that deter predation on both seeds that naturally contain them as well as on seeds to which QAs have been artificially applied

(Guimarães Jr., et al., 2003). In fact, QAs even deter agouti (Dasyprocta leporina) seed predation without inhibiting seed caching (i.e. dispersal) (Guimarães Jr., et al., 2003).

QAs also deter vertebrates like sheep, hares, and rabbits (Wink, 1993a; Wink, 1993b) and therefore hold real promise due to their ability to deter a wide set of seed predators and permit the success of this type of reforestation effort.

A final, and unfortunate, consideration that should be discussed in any conversation research on landscape-level processes is climate change. By the year 2100, temperatures in Amazonia are projected to have risen 3.3°C (5.94°F) (range: 1.8-5.1°C,

3.24-9.18°F) (Christensen, et al., 2007), though forest dieback could lead to changes of up to 8°C (14.4°F) (Betts, et al., 2004). Much of the Amazon Basin also faces potentially significant reductions in rainfall (Christensen, et al., 2007; Malhi, et al., 2008) that could lead to a permanently drier climate (Oyama & Nobre, 2003). A more arid climate could increase deforestation rates because a drier Amazon Basin may increase the area suitable for agriculture (Malhi, et al., 2008) because many crops, like the widely grown soya bean, do best in areas where there is a strong dry season (Nepstad, et al., 2008). In

177 my opinion, these significant climate changes could have several potential impacts on primates, including but not limited to: a) rainfall-related reductions in the availability of fruit or other resources, which could be especially significant if dry season fallback food sources were disrupted, b) large-scale shifts in the ranges of flora and fauna or even of entire habitats themselves, with the latter having a potentially devastating effect in areas where anthropogenic disturbances have created barriers to migration, c) accelerated habitat destruction due to the creation of newly desirable agricultural land, and d) increased hunting pressure as human populations move into new areas in response to the changing climate, hunting pressure that could be catastrophic in concert with shrinking forests and drastic habitat shifts. It is important to remember that I, along with all other conservation-oriented field researchers, have conducted a study of how wildlife reacts to disturbance under the current climate regime. The likely result of the looming climate changes will be an increased sensitivity of all primate species to anthropogenic habitat disturbance because shifting climates will add significant adaptive pressures to the already significant pressures exerted by habitat conversion and hunting.

178

Appendix

Appendix 1: The most common tree families by habitat UNDISTURBED N (%) LOGGED (25 y) N(%) UNBURNED (25 y) N (%) BURNED (25-30 y) N (%) BURNED (19 y) N (%) Sapotaceae 254 (15.6) Lecythidaceae 72 (11.2) Urticaceae 182 (24.2) Euphorbiaceae 69 (27.4) Melastomataceae 108 (39.6) Lecythidaceae 233 (14.3) Sapotaceae 64 (10.0) Euphorbiaceae 115 (15.3) Annonaceae 25 (9.9) Hypericaceae 94 (34.4) Leguminosae 163 (10.0) Burseraceae 60 (9.4) Leguminosae 80 (10.7) Melastomataceae 25 (9.9) Goupiaceae 24 (8.8) Burseraceae 142 (8.7) Euphorbiaceae 58 (9.0) Annonaceae 68 (9.1) Leguminosae 17 (6.7) Salicaceae 14 (5.1) Chrysobalanaceae 135 (8.3) Leguminosae 52 (8.1) Malpighiaceae 39 (5.2) Hypericaceae 16 (6.3) Euphorbiaceae 9 (3.3) Euphorbiaceae 99 (6.1) Moraceae 39 (6.1) Hypericaceae 33 (4.4) Rubiaceae 16 (6.3) Urticaceae 6 (2.2) Moraceae 61 (3.7) Chrysobalanaceae 32 (5.0) Melastomataceae 30 (4.0) Salicaceae 16 (6.3) Annonaceae 5 (1.8) Annonaceae 58 (3.6) Annonaceae 24 (3.7) Burseraceae 26 (3.5) Urticaceae 15 (6.0) Leguminosae 5 (1.8) Lauraceae 54 (3.3) Melastomataceae 23 (3.6) Salicaceae 23 (3.1) Burseraceae 10 (4.0) Rubiaceae 5 (1.8) Meliaceae 41 (2.5) Lauraceae 21 (3.3) Boraginaceae 22 (2.9) Malpighiaceae 10 (4.0) Boraginaceae 1 (0.4) Malvaceae 40 (2.5) Arecaceae 19 (3.0) Lauraceae 21 (2.8) Goupiaceae 5 (2.0) Malvaceae 1 (0.4) Arecaceae 39 (2.4) Meliaceae 17 (2.7) Moraceae 15 (2.0) Anacardiaceae 4 (1.6) Moraceae 1 (0.4) Myristicaceae 35 (2.1) Myristicaceae 17 (2.7) Lecythidaceae 12 (1.6) Lecythidaceae 4 (1.6)

Melastomataceae 22 (1.3) Urticaceae 16 (2.5) Simaroubaceae 11 (1.5) Malvaceae 4 (1.6)

Olacaceae 22 (1.3) Malpighiaceae 15 (2.3) Myristicaceae 9 (1.2) Lauraceae 3 (1.2)

Violaceae 20 (1.2) Hypericaceae 12 (1.9) Malvaceae 8 (1.1) Arecaceae 2 (0.8)

Myrtaceae 19 (1.2) Malvaceae 11 (1.7) Arecaceae 6 (0.8) Boraginaceae 2 (0.8)

Humiriaceae 18 (1.1) Olacaceae 11 (1.7) Anacardiaceae 5 (0.7) Caryocaraceae 2 (0.8)

Rubiaceae 15 (0.9) Elaeocarpaceae 10 (1.6) Caryocaraceae 5 (0.7) Apocynaceae 1 (0.4)

Apocynaceae 13 (0.8) Violaceae 9 (1.4) Nyctaginaceae 5 (0.7) Lacistemataceae 1 (0.4)

Vochysiaceae 13 (0.8) Apocynaceae 8 (1.2) Vochysiaceae 5 (0.7) Moraceae 1 (0.4)

Ochnaceae 12 (0.7) Humiriaceae 7 (1.1) Rubiaceae 4 (0.5) Nyctaginaceae 1 (0.4)

Anacardiaceae 11 (0.7) Myrtaceae 7 (1.1) Apocynaceae 3 (0.4) Sapindaceae 1 (0.4)

Elaeocarpaceae 10 (0.6) Vochysiaceae 6 (0.9) Chrysobalanaceae 3 (0.4) Sapotaceae 1 (0.4)

Hypericaceae 10 (0.6) Anacardiaceae 4 (0.6) Goupiaceae 3 (0.4) Vochysiaceae 1 (0.4)

Urticaceae 10 (0.6) Ochnaceae 4 (0.6) Myrtaceae 3 (0.4)

Combretaceae 9 (0.6) Bignoniaceae 3 (0.5) Sapotaceae 3 (0.4)

Salicaceae 7 (0.4) Combretaceae 3 (0.5) Violaceae 3 (0.4)

Bignoniaceae 6 (0.4) Rubiaceae 3 (0.5) Bombacaceae 1 (0.1)

Caryocaraceae 5 (0.3) Sapindaceae 3 (0.5) Combretaceae 1 (0.1)

Dichapetalaceae 5 (0.3) Boraginaceae 2 (0.3) Dichapetalaceae 1 (0.1)

Linaceae 4 (0.2) Caryocaraceae 2 (0.3) Erythroxylaceae 1 (0.1)

REMAINDER (19) 46 (2.8) REMAINDER (7) 7 (1.1) REMAINDER (5) 5 (0.7)

179

Appendix 2: Most common tree genera by habitat UNDISTURBED N(%) LOGGED (25 y) N(%) UNBURNED (25 y) N(%) BURNED (25-30 y) N(%) BURNED (19 y) N(%) Eschweilera 196 (12.0) Eschweilera 63 (9.8) Cecropia 105 (14.0) Croton 65 (25.8) Vismia 94 (34.4) Protium 132 (8.1) Protium 58 (9.0) Pourouma 77 (10.3) Vismia 16 (6.3) Bellucia 76 (27.8) Pouteria 122 (7.5) Pouteria 37 (5.8) Inga 59 (7.9) Laetia 15 (6.0) Goupia 24 (8.8) Licania 101 (6.2) Micrandropsis 29 (4.5) Guatteria 53 (7.1) Palicourea 15 (6.0) Miconia 18 (6.6) Micrandropsis 60 (3.7) Licania 23 (3.6) Croton 49 (6.5) Guatteria 14 (5.6) Acinodendron 13 (4.8) Micropholis 54 (3.3) Byrsonima 15 (2.3) Byrsonima 39 (5.2) Pourouma 14 (5.6) Laetia 12 (4.4) Chrysophyllum 38 (2.3) Micropholis 13 (2.0) Vismia 33 (4.4) Bellucia 13 (5.2) Croton 7 (2.6) Oenocarpus 31 (1.9) Oenocarpus 13 (2.0) Mabea 25 (3.3) Inga 12 (4.8) Cecropia 6 (2.2) Couepia 28 (1.7) Swartzia 13 (2.0) Cordia 22 (2.9) Miconia 11 (4.4) Palicourea 3 (1.1) Ocotea 26 (1.6) Brosimum 12 (1.9) Laetia 22 (2.9) Byrsonima 10 (4.0) Casearia 2 (0.7) Swartzia 25 (1.5) Virola 12 (1.9) Protium 22 (2.9) Protium 8 (3.2) Isertia 2 (0.7) Inga 23 (1.4) Vismia 12 (1.9) Micrandropsis 20 (2.7) Bocageopsis 6 (2.4) Maprounea 2 (0.7) Mabea 19 (1.2) Guarea 11 (1.7) Bellucia 17 (2.3) Goupia 5 (2.0) Rollinia 2 (0.7) Rinorea 18 (1.1) Helianthostylis 11 (1.7) Ocotea 17 (2.3) Eschweilera 4 (1.6) Xylopia 2 (0.7) Carapa 17 (1.0) Miconia 11 (1.7) Rollinia 14 (1.9) Rollinia 4 (1.6) Zygia 2 (0.7) Unonopsis 17 (1.0) Ocotea 11 (1.7) Glycydendron 13 (1.7) Tapirira 4 (1.6) Apeiba 1 (0.4) Brosimum 16 (1.0) Croton 10 (1.6) Simarouba 11 (1.5) Caryocar 2 (0.8) Bocageopsis 1 (0.4) Scleronema 16 (1.0) Inga 10 (1.6) Eschweilera 8 (1.1) Cordia 2 (0.8) Cordia 1 (0.4) Virola 16 (1.0) Sloanea 10 (1.6) Acinodendron 7 (0.9) Mabea 2 (0.8) Helicostylis 1 (0.4) Eugenia 15 (0.9) Cecropia 9 (1.4) Virola 7 (0.9) Sterculia 2 (0.8) Inga 1 (0.4) Iryanthera 15 (0.9) Hevea 9 (1.4) Helicostylis 6 (0.8) Trattinnickia 2 (0.8) Loreya 1 (0.4) Chromolucuma 14 (0.9) Minquartia 7 (1.1) Miconia 6 (0.8) Acinodendron 1 (0.4) Ormosia 1 (0.4) Duguetia 14 (0.9) Pourouma 7 (1.1) Caryocar 5 (0.7) Aniba 1 (0.4) Stryphnodendron 1 (0.4) Ecclinusa 14 (0.9) Rinorea 7 (1.1) Neea 5 (0.7) Apeiba 1 (0.4)

Guarea 14 (0.9) Couepia 6 (0.9) Oenocarpus 5 (0.7) Casearia 1 (0.4)

Minquartia 14 (0.9) Duguetia 6 (0.9) Parkia 5 (0.7) Cecropia 1 (0.4)

Sclerolobium 14 (0.9) Mabea 6 (0.9) Aparisthmium 4 (0.5) Conceveiba 1 (0.4)

Theobroma 14 (0.9) Bellucia 5 (0.8) Brosimum 4 (0.5) Dinizia 1 (0.4)

Bocoa 13 (0.8) Chrysophyllum 5 (0.8) Sterculia 4 (0.5) Endlicheria 1 (0.4)

Zygia 13 (0.8) Corythophora 5 (0.8) Trattinnickia 4 (0.5) Eperua 1 (0.4)

Lecythis 12 (0.7) Erisma 5 (0.8) Couratari 3 (0.4) Euterpe 1 (0.4)

Mouriri 12 (0.7) Eugenia 5 (0.8) Erisma 3 (0.4) Faramea 1 (0.4)

REMAINDER (133) 498 (30.5) REMAINDER (94) 185 (28.9) REMAINDER (55) 77 (10.3) REMAINDER (15) 15 (6.0)

180

Appendix 3: Most common tree species by habitat (CONTINUED ON NEXT PAGE)

UNDISTURBED N(%) LOGGED (25 y) N(%) UNBURNED (25 y) N(%) Eschweilera truncata 62 (3.8) Micrandropsis scleroxylon 29 (4.5) Cecropia sciadophylla 104 (13.8) Micrandropsis scleroxylon 60 (3.7) Protium hebetatum 17 (2.7) Guatteria olivacea 50 (6.7) Eschweilera wachenheimii 39 (2.4) Byrsonima duckeana 15 (2.3) Pourouma tomentosa tomentosa 32 (4.3) Protium hebetatum 37 (2.3) Eschweilera coriacea 15 (2.3) Byrsonima duckeana 29 (3.9) Eschweilera coriacea 33 (2.0) Eschweilera truncata 15 (2.0) Croton matourensis 28 (2.0) Oenocarpus bacaba 27 (1.7) Eschweilera wachenheimii 14 (2.2) Vismia cayennensis 28 (3.7) Chrysophyllum sanguinolentum 23 (1.4) Helianthostylis sprucei 11 (1.7) Mabea speciosa 24 (3.2) Licania impressa 21 (1.3) Croton matourensis 10 (1.6) Laetia procera 22 (2.9) Micropholis guyanensis guyanensis 21 (1.3) Hevea guianensis 9 (1.4) Croton draconoides 21 (2.8) Carapa guianensis 17 (1.0) Oenocarpus bacaba 9 (1.0) Micrandropsis scleroxylon 20 (1.0) Pouteria vernicosa 16 (1.0) Brosimum rubescens 7 (1.0) Pourouma tomentosa essequiboensis 18 (1.0) Scleronema micranthum 16 (1.0) Cecropia sciadophylla 7 (1.0) Bellucia grossularioides 17 (1.0) Eschweilera cyathiformis 15 (0.9) Minquartia guianensis 7 (1.1) Cordia exaltata 15 (2.0) Pouteria anomala 15 (0.9) Vismia cayennensis 7 (1.1) Inga rubiginosa 14 (1.9) Chromolucuma rubriflora 14 (0.9) Swartzia reticulata 6 (0.9) Pourouma villosa 14 (1.9) Minquartia guianensis 14 (0.9) Bellucia grossularioides 5 (0.8) Rollinia insignis 14 (1.9) Bocoa viridiflora 13 (0.8) Corythophora alta 5 (0.8) Glycydendron amazonicum 13 (1.7) Licania heteromorpha 13 (0.8) Mabea speciosa 5 (0.8) Simarouba amara 11 (1.5) Zygia racemosa 13 (0.8) Micropholis casiquiarensis 5 (0.8) Byrsonima spicata 9 (1.2) Protium robustum 12 (0.7) Ocotea amazonica 5 (0.8) Inga thibaudiana 9 (1.2) Brosimum rubescens 11 (0.7) Pouteria ambelaniifolia 5 (0.8) Acinodendron burchelli 7 (0.9) Pouteria erythrochrysa 11 (0.7) Rinorea guianensis 5 (0.8) Inga huberi 7 (0.9) Swartzia reticulata 11 (0.7) Virola calophylla 5 (0.8) Helicostylis tomentosa 6 (0.8) Theobroma sylvestre 11 (0.7) Acinodendron burchelli 4 (0.6) Caryocar villosum 5 (0.7) Unonopsis duckei 11 (0.7) Duguetia surinamensis 4 (0.6) Cordia fallax 5 (0.7) Andira micrantha 10 (0.6) Eschweilera collina 4 (0.6) Eschweilera coriacea 5 (0.7) Eschweilera grandiflora 10 (0.6) Eschweilera grandiflora 4 (0.6) Inga paraensis 5 (0.7) Hevea guianensis 10 (0.6) Euterpe precatoria 4 (0.6) Oenocarpus bacaba 5 (0.7) Licania coriacea 10 (0.6) Guarea humaitensis 4 (0.6) Pourouma cecropiifolia 5 (0.7) Pouteria bilocularis 10 (0.6) Guatteria olivacea 4 (0.6) Aparisthmium cordatum 4 (0.5) Rinorea guianensis 10 (0.6) Licania heteromorpha 4 (0.6) Inga alba 4 (0.5)

Gustavia elliptica 9 (0.6) Licania micrantha 4 (0.6) Inga crassiflora 4 (0.5) 181

REMAINDER (392) 1026 (62.9) REMAINDER (246) 387 (60.4) REMAINDER (135) 197 (26.2%)

Appendix 3 (CONTINUED)

BURNED (25-30 y) N(%) BURNED (19 y) N(%) Croton matourensis 65 (25.8) Vismia cayennensis 86 (31.5) Laetia procera 15 (6.0) Bellucia grossularioides 76 (27.8) Palicourea guianensis 15 (6.0) Goupia glabra 24 (8.8) Guatteria olivacea 14 (5.6) Acinodendron burchelli 13 (4.8) Bellucia grossularioides 13 (5.2) Laetia procera 12 (4.4) Byrsonima duckeana 10 (4.0) Miconia cuspidata 10 (3.7) Vismia cayennensis 8 (3.2) Croton matourensis 7 (2.6) Bocageopsis multiflora 6 (2.4) Vismia japurensis 6 (2.2) Pourouma tomentosa tomentosa 6 (2.4) Miconia biglandulosa 4 (1.5) Vismia japurensis 6 (2.4) Cecropia sciadophylla 3 (1.1) Goupia glabra 5 (2.0) Palicourea guianensis 3 (1.1) Pourouma villosa 5 (2.0) Cecropia purpurascens 2 (0.7) Inga huberi 4 (1.6) Isertia hypoleuca 2 (0.7) Miconia pyrifolia 4 (1.6) Maprounea guianensis 2 (0.7) Rollinia insignis 4 (1.6) Miconia holosericea 2 (0.7) Eschweilera coriacea 3 (1.2) Miconia pyrifolia 2 (0.7) Inga paraensis 3 (1.2) Vismia guianensis 2 (0.7) Miconia biglandulosa 3 (1.2) Zygia latifolia var. latifolia 2 (0.7) Tapirira guianensis 3 (1.2) Apeiba petoumo 1 (0.4) Caryocar villosum 2 (0.8) Bocageopsis multiflora 1 (0.4) Cordia exaltata 2 (0.8) Casearia arborea 1 (0.4) Inga alba 2 (0.8) Casearia manausensis 1 (0.4) Inga thibaudiana 2 (0.8) Cecropia ulei 1 (0.4) Miconia argyrophylla 2 (0.8) Cordia nodosa 1 (0.4) Miconia cuspidata 2 (0.8) Helicostylis tomentosa 1 (0.4) Pourouma guianensis guianensis 2 (0.8) Inga alba 1 (0.4) Protium grandifolium 2 (0.8) Loreya spruceana 1 (0.4) Trattinnickia burserifolia 2 (0.8) Ormosia paraensis 1 (0.4) Acinodendron burchelli 1 (0.4) Rollinia impressivena 1 (0.4) Aniba canellila 1 (0.4) Rollinia insignis 1 (0.4) Apeiba petoumo 1 (0.4) Stryphnodendron pulcherrimum 1 (0.4) Casearia arborea 1 (0.4) Xylopia amazonica 1 (0.4) REMAINDER (38) 38 (15.1) REMAINDER (1) 1 (0.4)

182

Appendix 4: The trees on my transects whose fruit is consumed by Alouatta (Note: lists include species with and without phenology estimates)

Family N (%) Genus N (%) Species N (%) Melastomataceae 215 (17.2) Bellucia 211 (16.9) Bellucia grossularioides 211 (16.9) Sapotaceae 194 (15.5) Cecropia 147 (11.8) Cecropia sciadophylla 147 (11.8) Urticaceae 164 (13.1) Goupia 135 (10.8) Goupia glabra 135 (10.8) Goupiaceae 135 (10.8) Protium 92 (7.4) Bocageopsis multiflora 45 (3.6) Burseraceae 95 (7.6) Pouteria 56 (4.5) Scleronema micranthum 35 (2.8) Moraceae 61 (4.9) Micropholis 50 (4.0) Eschweilera grandiflora 34 (2.7) Leguminosae 51 (4.1) Bocageopsis 45 (3.6) Chrysophyllum sanguinolentum 32 (2.5) Annonaceae 45 (3.6) Chrysophyllum 45 (3.6) Micropholis guyanensis guyanensis 28 (2.2) Lecythidaceae 45 (3.6) Eschweilera 41 (3.3) Protium grandifolium 28 (2.2) Myristicaceae 42 (3.4) Scleronema 35 (2.8) Minquartia guianensis 27 (2.2) Malvaceae 41 (3.3) Virola 34 (2.7) Hevea guianensis 26 (2.1) Chrysobalanaceae 36 (2.9) Licania 32 (2.6) Protium tenuifolium 24 (1.9) Olacaceae 27 (2.2) Brosimum 31 (2.5) Swartzia reticulata 24 (1.9) Euphorbiaceae 26 (2.1) Minquartia 27 (2.2) Virola calophylla 23 (1.8) Anacardiaceae 24 (1.9) Hevea 26 (2.1) Protium apiculatum 22 (1.8) Apocynaceae 11 (0.9) Helicostylis 25 (2.0) Licania heteromorpha 21 (1.7) Humiriaceae 8 (0.6) Swartzia 24 (1.9) Brosimum rubescens 21 (1.7) Combretaceae 6 (0.5) Ecclinusa 22 (1.8) Pouteria anomala 20 (1.6) Ochnaceae 6 (0.5) Manilkara 19 (1.5) Helicostylis tomentosa 19 (1.5) Solanaceae 4 (0.3) Pourouma 17 (1.4) Ecclinusa guianensis 16 (1.3) Arecaceae 3 (0.2) Anacardium 12 (1.0) Chrysophyllum pomiferum 13 (1.0) Lauraceae 3 (0.2) Tapirira 12 (1.0) Protium altsonii 13 (1.0) Malpighiaceae 3 (0.2) Geissospermum 11 (0.9) Micropholis williamii 12 (1.0) Polygalaceae 2 (0.2) Eperua 10 (0.8) Tapirira guianensis 12 (1.0) Myrtaceae 1 (0.1) Inga 10 (0.8) Geissospermum argenteum 11 (0.9) TOTAL 1247 (100) Osteophloeum 8 (0.6) Pourouma minor 11 (0.9) Vantanea 8 (0.6) Virola michelii 11 (0.9) Buchenavia 6 (0.5) Eperua glabriflora 10 (0.8) Quiina 6 (0.5) Inga alba 10 (0.8) Sterculia 6 (0.5) Licania oblongifolia 10 (0.8) Clarisia 5 (0.4) Manilkara bidentata 9 (0.7) Dipteryx 5 (0.4) Brosimum parinarioides 9 (0.7) REMAINDER (13) 35 (2.8) REMAINDER (40) 187 (15.0)

183

Appendix 5: The trees on my transects whose fruit is consumed by Ateles (Note: lists include species with and without phenology estimates)

Family N (%) Genus N (%) Species N (%) Melastomataceae 211 (17.3) Bellucia 211 (17.3) Bellucia grossularioides 211 (17.3) Urticaceae 164 (13.4) Cecropia 147 (12.0) Cecropia sciadophylla 147 (12.0) Goupiaceae 135 (11.0) Goupia 135 (11.0) Goupia glabra 135 (11.0) Sapotaceae 131 (10.7) Laetia 130 (10.6) Laetia procera 130 (10.6) Salicaceae 130 (10.6) Inga 84 (6.9) Oenocarpus bacaba 48 (3.9) Leguminosae 121 (9.9) Oenocarpus 48 (3.9) Inga thibaudiana 36 (2.9) Arecaceae 70 (5.7) Protium 47 (3.8) Chrysophyllum sanguinolentum 32 (2.6) Burseraceae 50 (4.1) Chrysophyllum 45 (3.7) Micropholis guyanensis 32 (2.7) Moraceae 50 (4.1) Micropholis 33 (2.7) Minquartia guianensis 27 (2.2) Olacaceae 27 (2.2) Minquartia 27 (2.2) Inga rubiginosa 24 (2.0) Malvaceae 22 (1.8) Helicostylis 19 (1.6) Protium tenuifolium 24 (2.0) Lecythidaceae 21 (1.7) Manilkara 19 (1.6) Protium apiculatum 22 (1.8) Myristicaceae 19 (1.6) Virola 19 (1.6) Helicostylis tomentosa 19 (1.6) Chrysobalanaceae 12 (1.0) Euterpe 18 (1.5) Euterpe precatoria 18 (1.5) Humiriaceae 12 (1.0) Pouteria 18 (1.5) Ecclinusa guianensis 16 (1.3) Meliaceae 10 (0.8) Pourouma 17 (1.4) Chrysophyllum pomiferum 13 (1.1) Annonaceae 7 (0.6) Ecclinusa 16 (1.3) Couratari stellata 12 (1.0) Anacardiaceae 6 (0.5) Couratari 13 (1.1) Pterocarpus officinalis 12 (1.0) Boraginaceae 5 (0.4) Brosimum 12 (1.0) Sacoglottis guianensis 12 (1.0) Clusiaceae 4 (0.3) Pterocarpus 12 (1.0) Licania micrantha 11 (0.9) Putranjivaceae 4 (0.3) Sacoglottis 12 (1.0) Pourouma minor 11 (0.9) Lauraceae 3 (0.2) Licania 11 (0.9) Virola michelii 11 (0.9) Bignoniaceae 2 (0.2) Theobroma 9 (0.7) Inga alba 10 (0.8) Polygalaceae 2 (0.2) Trymatococcus 9 (0.7) Brosimum parinarioides 9 (0.7) Ulmaceae 2 (0.2) Apeiba 7 (0.6) Manilkara bidentata 9 (0.7) Apocynaceae 1 (0.1) Duguetia 7 (0.6) Theobroma subincanum 9 (0.7) Violaceae 1 (0.1) Lecythis 7 (0.6) Trymatococcus amazonicus 9 (0.7) TOTAL 1222 (100) Anacardium 6 (0.5) Inga stipularis 8 (0.7) Ormosia 6 (0.5) Pouteria guianensis 8 (0.7)

Perebea 6 (0.5) Virola surinamensis 8 (0.7)

Sterculia 6 (0.5) Apeiba petoumo 7 (0.6)

Trichilia 6 (0.5) Duguetia surinamensis 7 (0.6)

REMAINDER (22) 60 (4.9) REMAINDER (38) 134 (11.0)

184

Appendix 6: The trees on my transects whose fruit is consumed by Chiropotes (Note: lists include species with and without phenology estimates)

Family N (%) Genus N (%) Species N (%) Lecythidaceae 425 (15.6) Eschweilera 388 (14.3) Bellucia grossularioides 211 (7.8) Euphorbiaceae 344 (12.7) Bellucia 212 (7.8) Micrandropsis scleroxylon 179 (6.6) Sapotaceae 317 (11.7) Micrandropsis 179 (6.6) Croton matourensis 139 (5.1) Melastomataceae 258 (9.5) Pouteria 173 (6.4) Goupia glabra 135 (5.0) Leguminosae 182 (6.7) Protium 162 (6.0) Laetia procera 130 (4.8) Burseraceae 165 (6.1) Croton 139 (5.1) Eschweilera truncata 128 (4.7) Goupiaceae 135 (5.0) Goupia 135 (5.0) Eschweilera coriacea 93 (3.4) Salicaceae 130 (4.8) Inga 130 (4.8) Eschweilera wachenheimii 75 (2.8) Urticaceae 125 (4.6) Laetia 130 (4.8) Protium hebetatum 67 (2.5) Chrysobalanaceae 115 (4.2) Pourouma 125 (4.6) Pourouma tomentosa tomentosa 53 (1.9) Moraceae 108 (4.0) Licania 99 (3.6) Oenocarpus bacaba 48 (1.8) Annonaceae 73 (2.7) Micropholis 52 (1.9) Bocageopsis multiflora 45 (1.7) Arecaceae 52 (1.9) Chrysophyllum 50 (1.8) Acinodendron burchelli 44 (1.6) Malvaceae 35 (1.3) Oenocarpus 48 (1.8) Inga thibaudiana 36 (1.3) Olacaceae 29 (1.1) Bocageopsis 45 (1.7) Scleronema micranthum 35 (1.3) Meliaceae 27 (1.0) Brosimum 45 (1.7) Eschweilera grandiflora 34 (1.3) Myristicaceae 25 (0.9) Acinodendron 44 (1.6) Inga huberi 33 (1.2) Anacardiaceae 24 (0.9) Scleronema 35 (1.3) Chrysophyllum sanguinolentum 29 (1.1) Humiriaceae 24 (0.9) Minquartia 27 (1.0) Micropholis guyanensis guyanensis 28 (1.0) Caryocaraceae 18 (0.7) Hevea 26 (1.0) Minquartia guianensis 27 (1.0) Simaroubaceae 18 (0.7) Helicostylis 24 (0.9) Pourouma villosa 27 (1.0) Vochysiaceae 18 (0.7) Carapa 23 (0.8) Hevea guianensis 26 (1.0) Apocynaceae 16 (0.6) Ecclinusa 22 (0.8) Pouteria vernicosa 26 (1.0) Malpighiaceae 13 (0.5) Pseudolmedia 20 (0.7) Protium tenuifolium 25 (0.9) Combretaceae 11 (0.4) Manilkara 19 (0.7) Inga rubiginosa 24 (0.9) Elaeocarpaceae 7 (0.3) Caryocar 18 (0.7) Carapa guianensis 23 (0.8) Putranjivaceae 4 (0.1) Iryanthera 17 (0.6) Licania impressa 23 (0.8) Solanaceae 4 (0.1) Pterocarpus 16 (0.6) Eschweilera cyathiformis 22 (0.8) Ebenaceae 3 (0.1) Unonopsis 16 (0.6) Protium apiculatum 22 (0.8) Nyctaginaceae 3 (0.1) Naucleopsis 15 (0.6) Protium nitidifolium 22 (0.8) Sapindaceae 3 (0.1) Lecythis 14 (0.5) Brosimum rubescens 21 (0.8) Bignoniaceae 2 (0.1) Simarouba 14 (0.5) Licania heteromorpha 21 (0.8) REMAINDER (4) 6 (0.2) REMAINDER (47) 257 (9.5) REMAINDER (143) 868 (31.9)

185

Appendix 7: The trees on my transects whose fruit is consumed by Pithecia (Note: lists include species with and without phenology estimates)

Family N (%) Genus N (%) Species N (%) Euphorbiaceae 179 (18.0) Micrandropsis 179 (18.0) Micrandropsis scleroxylon 179 (18.0) Sapotaceae 129 (13.0) Palicourea 102 (10.3) Pourouma guianensis guianensis 108 (10.9) Rubiaceae 103 (10.4) Pourouma 88 (8.9) Pourouma tomentosa tomentosa 53 (5.3) Moraceae 88 (8.9) Oenocarpus 57 (5.7) Oenocarpus bacaba 48 (4.8) Urticaceae 88 (8.9) Inga 40 (4.0) Micropholis guyanensis guyanensis 28 (2.8) Leguminosae 71 (7.1) Micropholis 40 (4.0) Protium grandifolium 28 (2.8) Arecaceae 57 (5.7) Pouteria 39 (3.9) Minquartia guianensis 27 (2.7) Burseraceae 35 (3.5) Protium 35 (3.5) Miconia biglandulosa 25 (2.5) Chrysobalanaceae 35 (3.5) Brosimum 33 (3.3) Inga rubiginosa 24 (2.4) Malvaceae 31 (3.1) Theobroma 31 (3.1) Swartzia reticulata 24 (2.4) Melastomataceae 29 (2.9) Minquartia 27 (2.7) Pouteria anomala 22 (2.2) Olacaceae 27 (2.7) Miconia 25 (2.5) Theobroma sylvestre 22 (2.2) Annonaceae 17 (1.7) Licania 24 (2.4) Brosimum rubescens 21 (2.1) Myristicaceae 17 (1.7) Swartzia 24 (2.4) Helicostylis tomentosa 19 (1.9) Nyctaginaceae 17 (1.7) Chrysophyllum 19 (1.9) Pourouma tomentosa essequiboensis 18 (1.8) Meliaceae 16 (1.6) Helicostylis 19 (1.9) Neea floribunda 17 (1.7) Lecythidaceae 12 (1.2) Iryanthera 17 (1.7) Ecclinusa guianensis 15 (1.5) Salicaceae 9 (0.9) Neea 17 (1.7) Naucleopsis caloneura 15 (1.5) Anacardiaceae 6 (0.6) Ecclinusa 15 (1.5) Chrysophyllum pomiferum 13 (1.3) Lacistemataceae 6 (0.6) Manilkara 15 (1.5) Trichilia micrantha 12 (1.2) Violaceae 5 (0.5) Naucleopsis 15 (1.5) Brosimum guianense 11 (1.1) Simaroubaceae 4 (0.4) Eschweilera 12 (1.2) Pourouma minor 11 (1.1) Siparunaceae 4 (0.4) Trichilia 12 (1.2) Pseudolmedia laevis 11 (1.1) Bignoniaceae 2 (0.2) Pseudolmedia 11 (1.1) Couepia longipendula 10 (1.0) Polygalaceae 2 (0.2) Couepia 10 (1.0) Fusaea longifolia 10 (1.0) Sapindaceae 2 (0.2) Fusaea 10 (1.0) Inga alba 10 (1.0) Ulmaceae 2 (0.2) Casearia 9 (0.9) Licania sprucei 10 (1.0) Celastraceae 1 (0.1) Trymatococcus 9 (0.9) Casearia javitensis 9 (0.9) TOTAL 994 (100) Stryphnodendron 7 (0.7) Eschweilera tessmannii 9 (0.9) Anacardium 6 (0.6) Manilkara bidentata 9 (0.9)

Lacistema 6 (0.6) Palicourea guianensis 9 (0.9)

Leonia 5 (0.5) Theobroma subincanum 9 (0.9)

REMAINDER (15) 36 (3.6) REMAINDER (44) 158 (15.9)

186

Appendix 8: The trees on my transects whose fruit is consumed by Saguinus (Note: lists include species with and without phenology estimates and species consumed by S. niger)

Family N (%) Genus N (%) Species N (%) Urticaceae 180 (27.9) Cecropia 147 (22.8) Cecropia sciadophylla 147 (22.8) Goupiaceae 135 (20.9) Goupia 135 (20.9) Goupia glabra 135 (20.9) Leguminosae 114 (17.7) Inga 101 (15.7) Oenocarpus bacaba 48 (7.4) Burseraceae 83 (12.9) Protium 80 (12.4) Inga thibaudiana 36 (5.6) Arecaceae 51 (7.9) Oenocarpus 48 (7.4) Pourouma villosa 27 (4.2) Sapotaceae 37 (5.7) Pourouma 33 (5.1) Inga rubiginosa 24 (3.7) Melastomataceae 13 (2.0) Chrysophyllum 17 (2.6) Protium tenuifolium 24 (3.7) Anacardiaceae 12 (1.9) Manilkara 15 (2.3) Protium apiculatum 22 (3.4) Ochnaceae 6 (0.9) Miconia 12 (1.9) Inga paraensis 20 (3.1) Boraginaceae 5 (0.8) Tapirira 12 (1.9) Chrysophyllum pomiferum 13 (2.0) Annonaceae 3 (0.5) Parkia 10 (1.6) Protium decandrum 13 (2.0) Sapindaceae 2 (0.3) Quiina 6 (0.9) Miconia holosericea 12 (1.9) Ulmaceae 2 (0.3) Cordia 5 (0.8) Tapirira guianensis 12 (1.9) Moraceae 1 (0.2) Micropholis 5 (0.8) Protium aracouchini 11 (1.7) Myrtaceae 1 (0.2) Astrocaryum 3 (0.5) Inga alba 10 (1.6) TOTAL 645 (100) Dialium 3 (0.5) Manilkara bidentata 9 (1.4) Guatteria 3 (0.5) Inga stipularis 8 (1.2) Tetragastris 3 (0.5) Parkia multijuga 7 (1.1) Ampelocera 2 (0.3) Manilkara bidentata surinamensis 6 (0.9) Talisia 2 (0.3) Pourouma guianensis guianensis 6 (0.9) Bellucia 1 (0.2) Protium trifoliolatum 6 (0.9) Clarisia 1 (0.2) Quiina amazonica 6 (0.9) Myrcia 1 (0.2) Cordia sagotii 5 (0.8) TOTAL 645 (100) Micropholis venulosa 5 (0.8) Chrysophyllum prieurii 4 (0.6) Protium guianense pilosissimum 4 (0.6) Astrocaryum gynacanthum 3 (0.5) Dialium guianense 3 (0.5) Guatteria scytophylla 3 (0.5) Inga lateriflora 3 (0.5) Parkia nitida 3 (0.5) Tetragastris panamensis 3 (0.5) REMAINDER (5) 7 (1.1)

1 87

Appendix 9: The trees on my transects whose fruit is consumed by Sapajus (Note: lists include species with and without phenology estimates)

Family N (%) Genus N (%) Species N (%) Melastomataceae 219 (14.0) Bellucia 211 (13.5) Bellucia grossularioides 211 (13.5) Leguminosae 205 (13.1) Inga 159 (10.2) Cecropia sciadophylla 147 (9.4) Urticaceae 191 (12.2) Cecropia 147 (9.4) Goupia glabra 135 (8.6) Goupiaceae 135 (8.6) Goupia 135 (8.6) Laetia procera 130 (8.3) Salicaceae 130 (8.3) Laetia 130 (8.3) Eschweilera coriacea 93 (5.9) Sapotaceae 106 (6.8) Eschweilera 99 (6.3) Oenocarpus bacaba 48 (3.1) Lecythidaceae 104 (6.6) Oenocarpus 57 (3.6) Inga thibaudiana 36 (2.3) Arecaceae 99 (6.3) Pourouma 44 (2.8) Inga huberi 33 (2.1) Moraceae 81 (5.2) Brosimum 41 (2.6) Micropholis guyanensis guyanensis 28 (1.8) Meliaceae 43 (2.7) Micropholis 38 (2.4) Minquartia guianensis 27 (1.7) Burseraceae 31 (2.0) Zygia 38 (2.4) Pourouma villosa 27 (1.7) Malvaceae 30 (1.9) Protium 28 (1.8) Zygia racemosa 26 (1.7) Olacaceae 29 (1.9) Minquartia 27 (1.7) Inga rubiginosa 24 (1.5) Myristicaceae 27 (1.7) Carapa 23 (1.5) Protium tenuifolium 24 (1.5) Chrysobalanaceae 21 (1.3) Licania 21 (1.3) Carapa guianensis 23 (1.5) Boraginaceae 19 (1.2) Pouteria 20 (1.3) Brosimum rubescens 21 (1.3) Nyctaginaceae 17 (1.1) Cordia 19 (1.2) Licania heteromorpha 21 (1.3) Bignoniaceae 13 (0.8) Helicostylis 19 (1.2) Inga paraensis 20 (1.3) Anacardiaceae 12 (0.8) Manilkara 19 (1.2) Helicostylis tomentosa 19 (1.2) Annonaceae 10 (0.6) Virola 19 (1.2) Euterpe precatoria 18 (1.2) Siparunaceae 8 (0.5) Euterpe 18 (1.2) Neea floribunda 17 (1.1) Apocynaceae 6 (0.4) Neea 17 (1.1) Syagrus inajai 17 (1.1) Myrtaceae 6 (0.4) Syagrus 17 (1.1) Ecclinusa guianensis 16 (1.0) Caryocaraceae 5 (0.3) Ecclinusa 16 (1.0) Cordia nodosa 14 (0.9) Combretaceae 5 (0.3) Trichilia 16 (1.0) Chrysophyllum pomiferum 13 (0.8) Sapindaceae 5 (0.3) Sterculia 14 (0.9) Tapirira guianensis 12 (0.8) Polygalaceae 2 (0.1) Chrysophyllum 13 (0.8) Trichilia micrantha 12 (0.8) Rubiaceae 2 (0.1) Tapirira 12 (0.8) Zygia latifolia var. latifolia 12 (0.8) Clusiaceae 1 (0.1) Jacaranda 11 (0.7) Brosimum guianense 11 (0.7) Connaraceae 1 (0.1) Pseudolmedia 11 (0.7) Jacaranda copaia 11 (0.7) Peraceae 1 (0.1) Theobroma 9 (0.6) Pourouma minor 11 (0.7) Violaceae 1 (0.1) Iryanthera 8 (0.5) Pseudolmedia laevis 11 (0.7) TOTAL 1565 (100) REMAINDER (31) 109 (7.0) REMAINDER (62) 297 (19.0)

188

Appendix 10: Population densities of the study species at sites with only undisturbed habitat (see Appendices 11 & 12 for additional sites)

Species Location Forest Type Density Source Alouatta macconnelli Brazil TF 6.8 THIS STUDY TF 10.5 (Rylands & Keuroghlian, 1988) TF 3.8-6.1 " " French Guiana TF 18.9 (Guillotin, et al., 1994) TF 11.0-15.0 (Kessler, 1998) TF 19.9-20.0 (Simmen, et al., 2001) TF 17-22 (Julliot, 1992) Guyana TF? 7.5 (locally 4.7-63.8) (Muckenhirn, et al., 1976) Suriname PF 17.0 (Mittermeier, 1977) Venezuela Is 0.0-860.0 (Feeley & Terborgh, 2006) Ateles paniscus Brazil TF 1.8 THIS STUDY TF 1.0 (Rylands & Keuroghlian, 1988) TF 0.0 " " French Guiana TF 8.57 (7-10) (Kessler, 1998) TF 14.0 (Guillotin, et al., 1994) TF 8.4 (Simmen, et al., 1998; Simmen, et al., 2001) Guyana TF? 2.4-6.2 (Muckenhirn, et al., 1976) Suriname PF 7.1 (van Roosmalen, 1980) Chiropotes chiropotes Brazil TF 4.9 THIS STUDY TF 0.07 (Boyle & Smith, 2010a) TF 0.04 " " TF 5.5 (Rylands & Keuroghlian, 1988) TF 0.0 " " Guyana TF? 7.6 (Muckenhirn, et al., 1976) TF? 12.7 " " Suriname PF 7.0-8.0 (Mittermeier, 1977) Pithecia chrysocephala Brazil TF 2.2 THIS STUDY (incl. P. pithecia) TF 0.7 (Rylands & Keuroghlian, 1988) TF 0.0-4.8 " " French Guiana TF 0.64 (Vié, et al., 2001) TF 3.0 (Guillotin, et al., 1994) Guyana TF? 2.6-12.0 (Muckenhirn, et al., 1976)

Suriname PF 3.6 (Mittermeier, 1977) 189 PF 0.8-7.0 (Buchanan, 1978)

(Continued) Location Forest Type Density Source Saguinus midas Brazil TF 6.3 THIS STUDY Mix 16.4-32.9 (Thorington, 1968) TF 3.9 (Rylands & Keuroghlian, 1988 French Guiana TF 16.5; 22.9 (Kessler, 1995; Kessler, 1998) TF 11.3-14.0 (Simmen, et al., 1998; Simmen, et al., 2001) Guyana TF? 2.3-13.9 (Muckenhirn, et al., 1976) Suriname PF 23.5 (Mittermeier, 1977) Sapajus apella Brazil TF 8.3 THIS STUDY TF 38.3 (Haugaasen & Peres, 2005) Vz 57.6 " " Ig 45.8 " " TF 1.6 (Spironello, 2001) TF 2.9-49.6 (Peres, 1997) Vz 17.4-55.7 " " TF 6.7-24.1 (Peres, 1998) TF 32.3 (Peres, 1993a) TF 2.2 (Rylands & Keuroghlian, 1988) TF 0.0-17.1 " " Bolivia TF 14.1 (Wallace, et al., 1998) Colombia TF 6.1 (Izawa, 1980) French Guiana TF 3.66 (Zhang, 1995a) TF 15.0 (Kessler, 1998) TF 8.6-12.9 (Simmen, et al., 1998; Simmen, et al., 2001) Guyana TF? 10.6-17.1 (Muckenhirn, et al., 1976) Peru TF/Ig? 12.5 (Terborgh, 1983) Ig igapó, Is islands; PF primary forest; TF terra firme; Vz várzea

190

Appendix 11: Population densities of the study species and their close relatives at sites that sampled both undisturbed and secondary forest habitats

Species Location Forest Type Density (ind/km2) Change (%)* Notes Alouatta A. belzebul Brazil TF 0.7 a SG 0.0a Decrease 1 A. macconnelli** Brazil TF 6.8 SG 2.3 -66.2 2 Age: 25 yr; Cecropia-dominated TF 1.12 SG 0.13 -88.43 Age: 13-18 y; Disturbance: tree plantations A. seniculus Brazil TF 0.9 SG 0.2 -77.8 4 Agriculture, re-growth, corridors TF/Vz 0.0 SG 1.28 Increase 5 Age: 50 y; Disturbance: clear-cut TF 1.11 SG 0.69 -37.8 5 Age: 30 y; Disturbance: tree plantation Colombia TF (logged - low) 3.0 n/a No logging data TF (logged - med) 10.0 n/a No logging data SG 0.0 Decrease6 No data on age Ateles A. belzebuth Brazil TF (wet 0, dry 0)b Low forest TF (wet 0, dry 1.36)b High forest SG (wet 0, dry 0)b Decrease 1 A. chamek Brazil TF 1.30 SG 0.0 Decrease 4 Agriculture, re-growth, corridors TF (wet 0, dry 0)b Low forest TF (wet 0, dry 1.13)b High forest SG (wet 0, dry 0)b Decrease 1 A. hybridus Colombia TF (logged - low) 29.0 n/a No logging data TF (logged - med) 38.0 n/a No logging data SG 0.0 Decrease6 No data on age A. paniscus** Brazil TF 1.8 SG 0.0 Decrease 2 Age: 25 yr; Cecropia-dominated TF 1.39 SG 0.0 Decrease3 Age: 13-18 y; Disturbance: tree plantations

191

(Continued) Location Forest Type Density (ind/km2) Change (%)* Notes Chiropotes C. albinasus Brazil TF 1.11 SG 0.0 Decrease5 Age: 30 y; Disturbance: tree plantation TF 0.9a SG 0.0a Decrease 1 C. chiropotes Brazil TF 4.9 SG 2.1 -57.1 2 Age: 25 yr; Cecropia-dominated Pithecia P. albicans Brazil TF 9.0 SG 0.9 -90.04 Agriculture, re-growth, corridors P. chrysocephala** Brazil TF 2.2 SG 1.7 -22.7 2 Age: 25 yr; Cecropia-dominated P. hirsuta Brazil TF 0.0a SG 0.7a Increase 1 P. irrorata Brazil TF/Vz 2.29 SG 4.18 +82.5 5 Age: 50 y; Disturbance: clear-cut TF/Vz 1.0 SG 2.73 +173.0 5 Age: 50 y; Disturbance: clear-cut P. pithecia Brazil TF 0.21 SG 0.13 -38.13 Age: 13-18 y; Disturbance: tree plantations Saguinus S. bicolor Brazil TF 0.17 SG 4.58 +2594.1 5 Age: 30 y; Disturbance: tree plantation S. fuscicollis Brazil TF 0.0a SG 0.05a Increase 7 Agriculture, re-growth, corridors TF/Vz 3.33 SG 3.25 -2.4 5 Age: 50 y; Disturbance: clear-cut S. labiatus Brazil TF/Vz 2.13 SG 0.0 Decrease 5 Age: 50 y; Disturbance: clear-cut S. midas** Brazil TF 6.3 SG 18.3 +190.5 2 Age: 25 yr; Cecropia-dominated TF 0.0 SG 0.91 Increase 3 Age: 13-18 y; Disturbance: tree plantations S. m. mystax Brazil TF 78.0 SG 23.0 -70.5 4 Agriculture, re-growth, corridors 192

(Continued) Location Forest Type Density (ind/km2) Change (%)* Notes Sapajus (Cebus) albifrons Brazil TF 14.0 SG 5.2 -62.9 4 Agriculture, re-growth, corridors Colombia TF (logged - low) 34.0 n/a No logging data TF (logged - med) 83.0 n/a No logging data SG 136.0 Increase6 No data on age S. apella** Brazil TF 8.3 SG 14.5 +74.7 2 Age: 25 yr; Cecropia-dominated TF 11.5 SG 5.6 -51.34 Agriculture, re-growth, corridors TF 1.9a SG 1.1a -42.1 1 TF 0.78 SG 1.55 +98.7 3 Age: 13-18 y; Disturbance: tree plantations TF/Vz 17.08 SG 6.61 -61.3 5 Age: 50 y; Disturbance: clear-cut TF/Vz 1.38 SG 8.18 +492.8 5 Age: 50 y; Disturbance: clear-cut TF 0.44 SG 0.0 Decrease 5 Age: 30 y; Disturbance: tree plantation * Relative to primary forest at the same site; ** Study species; Ig igapó; SG secondary growth; TF terra firme; Vz várzea a Groups/km2; b Groups/10km; c Modified by (Johns & Skorupa, 1987) 1 (Branch, 1983); 2(This Study), 3 (Parry, et al., 2007), 4 (Johns, 1991), 5 (Vulinec, et al., 2006) ,6 (Aldana, et al., 2008), 7 (Johns, 1986)

193

Appendix 12: Populations densities of the study species and their close relatives at sites that sampled both undisturbed and selectively logged habitats

Species Location Forest Type (Intensity) Density (ind/km2) Change (%)* Notes Alouatta A. belzebul Brazil TF 0.7a TF (logged - low) 0.4a -42.91 <1 tree ha-1; elapsed >10 y A. macconnelli** Brazil TF 6.8 TF (logged - low) 4.3 -37.82 0.48 trees ha-1; skid area 2.4%; elapsed: 25 y Guyana TF 4.6 (2.4-9.0) TF (logged - low) 5.4 (2.6-11.1) 17.43 1.12 trees ha-1; 3.0% damage; elapsed: 11 y A. seniculus Bolivia TF 0.40 (0.19)d TF (logged) 0.0d Decrease 15? m3ha-1; canopy cover: -17%; elapsed: 1 y TF (logged) 0.0d Decrease4 14.5 m3ha-1; canopy cover: -28%; elapsed: 2 y Brazil TF 0.9 TF (logged - low) 0.2 -77.85 3-5 trees ha-1; 61% damage; elapsed: 11 y Brazil Vz (logged - low) 3.2a n/a <0.1 trees ha-1; <1% damage; elapsed: 0 Vz (logged - medium) 10.7a n/a6 4.6 trees ha-1; 5% damage; elapsed: 0 Colombia TF (logged - low) 3 n/a No logging data TF (logged - medium) 10 n/a7 No logging data Peru TF/Ig 4.8a TF/Ig (logged - low?) 6.1a 27.18 Elapsed >35 years Ateles A. belzebuth Brazil TF (wet 0, dry 0)b Low forest TF (wet 0, dry 1.36)b n/a1 High forest A. chamek Bolivia TF 1.68 (0.47)d TF (logged) 0.43 (0.25)d -74.4 15? m3ha-1; canopy cover: -17%; elapsed: 1 y TF (logged) 0.42 (0.30)d -754 14.5 m3ha-1; canopy cover: -28%; elapsed: 2 y A. chamek Brazil TF (wet 0, dry 0)b Low forest TF (wet 0, dry 1.13)b High forest TF (logged - low) (wet 0, dry 0)b Decrease <1 tree ha-1; elapsed >10 y TF (logged - low) (wet 0, dry 0)b Decrease1 <1 tree ha-1; elapsed >10 y Brazil TF 1.3 TF (logged - low) 1.1 -15.45 3-5 trees ha-1; 61% damage; elapsed: 11 y Peru TF/Ig 3.2a Reported A. paniscus TF (logged - low) 0.0a Decrease8 Elapsed >35 years; reported A. paniscus

A. paniscus** Brazil TF 1.8 194 TF (logged - low) 1.1 -38.92 0.48 trees ha-1; skid area 2.4%; elapsed: 25 y

(Continued) Location Forest Type (Intensity) Density (ind/km2) Change (%)* Notes A. paniscus** Guyana TF 6.90 (3.9-12.2) TF (logged - low) 3.60 (1.5-8.0) -47.83 1.12 trees ha-1; 3.0% damage; elapsed: 11 y A. hybridus Colombia TF (logged - low) 29 n/a No logging data TF (logged - medium) 38 n/a7 No logging data Chiropotes C. albinasus Brazil TF 0.9a 55.51 TF (logged - low) 1.4a <1 tree ha-1; time elapsed >10 y C. chiropotes** Brazil TF 4.9 TF (logged - low) 1.2 -75.52 0.48 trees ha-1; skid area 2.4%; elapsed: 25 y Brazil TF&Ig 0.8a TF/Ig (logged - high) 0.0a Decrease6 ~10 trees ha-1; 60% damage; elapsed: 1 y Brazil TF 1.0a TF (logged - low) 0.5a -50.0 1-2 trees ha-1; 10% damage; elapsed: 2 y TF (logged - high) Solitaries Decrease6 ~10 trees ha-1; 50% damage; elapsed: 2 y Pithecia P. albicans Brazil TF 9 TF (logged - low) 18 100 .05 3-5 trees ha-1; 61% damage; elapsed: 11 y P. chrysocephala** TF 2.2 TF (logged - low) 1.0 -54.52 0.48 trees ha-1; skid area 2.4%; elapsed: 25 y P. hirsuta Brazil TF 0.0a TF (logged - medium) 0.4a Increase1 <1 tree ha-1; elapsed >10 y P. irrorata Peru TF/Ig 0.0a Reported P. monachus TF/Ig (logged - low?) 1.8a Decrease8 Elapsed >35 years Saguinus S. fuscicollis Brazil TF 0.0a TF (logged - low) 0.0a No change6 3-5 trees ha-1; 61% damage; elapsed: 11 y Peru TF/Ig 1.8a TF/Ig (logged - low?) 2.5a 38.98 Elapsed >35 years S. midas** Brazil TF 6.3 TF (logged - low) 7.2 14.32 0.48 trees ha-1; skid area 2.4%; elapsed: 25 y TF&Ig 3.1a TF/Ig (logged - high) 3.6a 16.16 ~10 trees ha-1; 60% damage; elapsed: 1 y TF 3.0a TF (logged - low) 1.5a -50 1-2 trees ha-1; 10% damage; elapsed: 2 y a 6 -1 TF (logged - high) 0.4 -86.7 ~10 trees ha ; 50% damage; elapsed: 2 y 195

(Continued) Location Forest Type (Intensity) Density (ind/km2) Change (%)* Notes S. m. mystax Brazil TF 78 TF (logged - low) 88 12.85 3-5 trees ha-1; 61% damage; elapsed: 11 y Sapajus (Cebus) albifrons Brazil TF 14 TF (logged - low) 31 121.4 5 3-5 trees ha-1; 61% damage; elapsed: 11 y Colombia TF (logged - low) 34 n/a No logging data TF (logged - medium) 83 n/a7 No logging data (Cebus) olivaceous Guyana TF 0.42c TF (logged - low) 0.05c -88.13 1.12 trees ha-1; 3.0% damage; elapsed: 11 y S. apella** Bolivia TF 0.31 (0.11)d TF (logged) 0.48 (0.20)d 15? m3ha-1; canopy cover: -17%; elapsed: 1 y TF (logged) 0.17 (0.12)d n/a 4 14.5 m3ha-1; canopy cover: -28%; elapsed: 2 y Brazil TF 8.3 TF (logged - low) 4.6 -46.52 0.48 trees ha-1; skid area 2.4%; elapsed: 25 y TF 11.5 TF (logged - low) 32 178.35 3-5 trees ha-1; 61% damage; elapsed: 11 y TF 1.9a TF (logged - low) 4.3a 126.31 <1 tree ha-1; elapsed >10 y TF&Ig 2.3a TF/Ig (log) 2.4a 4.36 ~10 trees ha-1; 60% damage; elapsed: 1 y TF 1.8a TF (logged - low) 0.5a -72.2 1-2 trees ha-1; 10% damage; elapsed: 2 y TF (logged - high) 0.0a Decrease6 ~10 trees ha-1; 50% damage; elapsed: 2 y Vz (logged - low) 1.1a n/a <0.1 trees ha-1; <1% damage; elapsed: 0 Vz (logged - medium) 0.6a n/a6 4.6 trees ha-1; 5% damage; elapsed: 0 Peru TF/Ig 3.6a TF/Ig (logged - low?) 2.5a -30.68 Elapsed >35 years * Relative to primary forest at the same site; ** Study species; Ig igapó; TF terra firme; Vz várzea a Groups/km2; b Groups/10km, c Encounter rate; d # animals/day; e Modified by (Johns & Skorupa, 1987) 1 (Branch, 1983),2(This Study),3 (Bicknell & Peres, 2010),4 (Fredericksen, et al., 2007),5 (Johns, 1991),6 (Johns, 1986),7 (Aldana, et al., 2008), 8 (Freese, et al., 1982)

196

197

List of References

Aide, T. & Cavelier, J., 1994. Barriers to lowland tropical forest restoration in the Sierra Nevada de Santa Marta, Colombia. Restoration Ecology, Volume 2, pp. 219-229.

Aldana, A., Beltran, M., Torres-Neira, J. & Stevenson, P., 2008. Habitat characterization and population density of brown spider monkeys (Ateles hybridus) in Magdalena Valley, Colombia. Neotropical Primates, 15(2), pp. 46-50.

Alencar, J., 1998. Fenologia de espécies arbóreas tropicais na Amazônia central. In: Floresta Amazônica: Dinâmica, regeneração, e manejo. Manaus: Ministerio da Ciência Tecnologia INPA, pp. 25-40.

Altmann, S. & Altmann, J., 1970. Baboon ecology: African field research. Chicago: University of Chicago Press.

Anderson, D., Laake, J., Crain, B. & Burnham, K., 1979. Guidelines for line transect sampling of biological populations. Journal of Wildlife Management, Volume 43, pp. 70-78.

Andresen, E., 2000. The role of dung beetles in the regeneration of rainforest plants in central Amazonia (Ph.D. dissertation). Gainesville: University of Florida.

Andresen, E., 2002. Primary seed dispersal by red howler monkeys and the effect of defecation pattern on the fate of dispersed seeds. Biotropica, 34(2), pp. 261-272.

Ashton, P., 1969. Speciation among tropical forest trees: Some deductions in the light of recent evidence. Biological Journal of the Linnean Society, Volume 1, pp. 155- 196.

198

Asner, G., Knapp, D., Broadbent, E., Oliveira, P., Keller, M., Silva, J., 2005. Selective logging in the Brazilian Amazon. Science, 310(5747), pp. 480-482.

Ayres, J., 1981. Observações sobre a ecologia e o comportamento dos cuxiús (Chiropotes albinasus e Chriopotes satanas, Cebidae: Primates) (M.S. thesis). Manaus: Instituto Nacional de Pesquisas da Amazônia and Fundação Universidade do Amazonas.

Ayres, J., 1983. Conservation of primates in Brazililan Amazonia: Problems and strategies. In: D. Harper, ed. Conservation of primates and their habitats, Vaughan Paper No. 31. Leicester: University of Leicester, pp. 2-33.

Ayers, J., 1989. Comparative feeding ecology of the uakari and bearded saki, Cacajao and Chiropotes. Journal of Human Evolution, Volume 18, pp. 697-716.

Babyak, M., 2004. What you see many not be what you get: A brief, nontechnical introduction to overfitting in regression-type models. Psychosomatic Medicine, Volume 66, pp. 411-421.

Balée, W., 1998. Historical ecology: Premises and postulates. In: W. Balée, ed. Advances in historical ecology. New York: Columbia University Press, pp. 13-29.

Balme, G., Slotow, R. & Hunter, L., 2010. Edge effects and the impact of non-protected areas in carnivore conservation: Leopards in the Phinda-Mkhuze Complex, South Africa. Animal Conservation, Volume 13, pp. 315-323.

Bassini, F. & Becker, P., 1990. Charcoal's occurrence in soil depends on topography in terra firme forest near Manaus, Brazil. Biotropica, 22(4), pp. 420-422.

Bawa, K. & Opler, P., 1975. Dioecism in tropical forest trees. Evolution, 29(1), pp. 167- 179.

Bentos, T., Mesquita, R. & Williamson, G., 2008. Reproductive phenology of central Amazon pioneer trees. Tropical Conservation Science, 1(3), pp. 186-203.

199

Betts, R., Cox, P., Collins, M., Harris, P., Huntingford, C., Jones, C.D., 2004. The role of ecosystem-atmosphere interactons in simulated Amazonian precipitation decrease and forest dieback under global climate warming. Theoretical and Applied Climatology, 78(1-3), pp. 157-175.

Bicknell, J. & Peres, C., 2010. Vertebrate population responses to reduced-impact logging in a neotropical forest. Forest Ecology and Management, Volume 259, pp. 2267-2275.

Bierregaard, R., 1994. Harpy Eagle. In: J. del Hoyo, A. Elliot & J. Sargatal, eds. Handbook of birds of the world. Volume 2. Barcelona: Lynx Edicions, p. 191.

Bierregaard Jr., R. & Gascon, C., 2001. The Biological Dynamics of Forest Fragments Project: Overview and history of a long-term conservation project. In: J. R. Bierregaard, C. Gascon & T. M. R. Lovejoy, eds. Lessons from Amazonia: The ecology and conservation of a fragmented forest. New Haven & London: Yale University Press, pp. 5-12.

Bierregaard Jr., R. & Stouffer, P., 1997. Understory birds and dynamic habitat mosaics in Amazonian rainforests. In: W. Laurance & R. Bierregaard, eds. Tropical forest remnants: Ecology, management, and conservation of fragmented communities. Chicago: University of Chicago Press, pp. 138-155.

Bobadilla, U. & Ferrari, S., 2000. Habitat use by Chiropotes satanas utahicki and syntopic platyrrhines in eastern Amazonia. American Journal of Primatology, Volume 50, pp. 215-224.

Boubli, J.-P., Di Fiore, A. & Mittermeier, R., 2008. Alouatta macconnelli. [Online] Available at: http://www.iucnredlist.org/details/40642/0 [Accessed 15 10 2012].

Boyle, S., 2008. The effects of forest fragmentation on primates in the Brazilian Amazon (Ph.D. dissertation). Phoenix: Arizona State University.

200

Boyle, S., Lenz, B., Gilbert, K., Spironello, W., Santamaria-Gómez, M., Setz, E., dos Reis, A., da Silva, O., Keuroghlian, A., Pinto, F., In Press. The primates of the Biological Dynamics of Forest Fragments Project: A history. In: L. Marsh, ed. Primates in Fragments. pp. xxx-xxx.

Boyle, S., Lourenço, W., da Silva, L. & Smith, A., 2009. Travel and spatial patterns change when Chiropotes satanas chiropotes inhabit forest fragments. International Journal of Primatology, Volume 30, pp. 515-531.

Boyle, S. & Smith, A., 2010a. Behavioral modifications in northern bearded saki monkeys (Chiropotes satanas chiropotes) in forest fragments of central Amazonia. Primates, Volume 51, pp. 43-51.

Boyle, S. & Smith, A., 2010b. Can landscape and species characteristics predict primate presence in forest fragments in the Brazilian Amazon? Biological Conservation, Volume 143, pp. 1134-1143.

Boyle, S., Zartman, C., Spironello, W. & Smith, A., 2012. Implications of habitat fragmentation on the diet of bearded saki monkeys in central Amazonia forest. Journal of Mammalogy, 93(4), pp. 959-976.

Bramblett, C., 1967. Pathology in the Daranjani baboon. American Journal of Physical Anthropology, Volume 26, pp. 331-340.

Branch, L., 1983. Seasonal and habitat difference in the abundance of primates in the Amazon (Tapajós) National Park, Brazil. Primates, Volume 24, pp. 424-431.

Brinkmann, W. & Vieira, A., 1971. The effect of burning on germination of seeds at different soil depths of various tropical tree species. Turrialba, Volume 21, pp. 77-82.

Broadbent, E, Anser, G., Keller, M., Knapp, D., Oliveira, P., Silva, J., 2008. Forest fragmentation and edge effects from deforestation and selective logging in the Brazilian Amazon. Biological Conservation, 141(7), pp. 1745-1757.

201

Bruna, E. & Kress, W., 2002. Habitat fragmentation and the demographic structure of an Amazonian understory herb (Heliconia acuminata). Conservation Biology, 16(5), pp. 1256-1266.

Buchanan, D., 1978. Communication ecology of pithecine monkeys, with special reference to Pithecia pithecia. Ph.D. dissertation. Detroit: Wayne State University.

Buchanan, D., Mittermeier, R. & van Roosmalen, M., 1981. The saki monkeys, genus Pithecia. In: A. Coimbra-Filho & R. Mittermeier, eds. Ecology and behavior of Neotropical primates. Rio de Janeiro: Academia Brasileira de Ciências, pp. 391- 417.

Buchanan-Smith, H., 1990. Polyspecific association of two tamarin species, Saguinus labiatus and Saguinus fuscicollis, in Bolivia. American Journal of Primatology, Volume 22, pp. 205-214.

Buchanan-Smith, H., 1999. Tamarin polyspecific associations: Forest utilization and stability of mixed-species groups. Primates, Volume 40, pp. 233-247.

Buckland, S., Anderson, D., Burnham, K. & Laake, J., 1996. Distance sampling: Estimating abundances of biological populations. London: Chapman and Hall.

Buckland, S., Anderson, D., Burnham, K., Laake, J., Borchers, D., Thomas, L., 2001. Introduction to distance sampling: Estimating abundance of biological populations. Oxford: Oxford University Press.

Burnham, K., Anderson, D. & Laake, J., 1981. Line transect estimation of bird population density using a Fourier Series. Studies in Avian Biology, Volume 6, pp. 466-482.

Butler, R., 2013. Mongabay.com. [Online] Available at: http://books.mongabay.com/population_estimates/full/Manaus- Brazil.html [Accessed 15 January 2013].

202

Camargo, J., 1993. Variation in soil moisture and air vapour pressure deficit relative to tropical rain forest edges near Manaus, Brazil (M.Phil. thesis). Cambridge: Cambridge University.

Camargo, J., Ferraz, I., Mesquita, M., Santos, B., Brum, H., 2008. Guia de propágulos plântulas da Amazônia. Manaus: INPA.

Camargo, J. & Kapos, V., 1995. Complex edge effects on soil moisture and microclimate in central Amazonian forest. Journal of Tropical Ecology, Volume 11, pp. 205-211.

Cant, J., 1978. Population survey of the spider monkey Ateles geoffroyi at Tikal, Guatemala. Primates, 19(3), pp. 525-535.

Cartmill, M. & Milton, K., 1977. The lorisiform wrist joint and the evolution of "brachiating" adaptations in the Hominoidea. American Journal of Physical Anthropology, Volume 47, pp. 249-272.

Cervinka, J., Salek, M., Pavluvcik, P. & Kreisinger, J., 2011. The fine-scale utilization of forest edges by mammalian mesopredators related to patch size and conservation issues in central European farmland. Biodiversity Conservation, Volume 20, pp. 3459-3475.

Chapman, C., 1987. Flexibility in diets of three species of Costa Rican primates. Folia Primatologica, Volume 49, pp. 90-105.

Chapman, C., 1988. Patch use and patch depletion by the spider and howling monkeys of Santa Rosa National Park. Behaviour, Volume 105, pp. 99-116.

Chapman, C., 1990a. Association patterns of spider monkeys: The influence of ecology and sex on social organization. Behavioral Ecology and Sociobiology, Volume 26, pp. 409-414.

Chapman, C., 1990b. Ecological constraints on group size in three species of Neotropical primates. Folia Primatologica, Volume 55, pp. 1-9.

203

Chapman, C., 1995. Primate seed dispersal: Coevolution and conservation implications. Evolutionary Anthropology, Volume 4, pp. 74-82.

Chapman, C. & Chapman, L., 1996. Mixed-species primate groups in the Kibale Forest: Ecological constraints on association. International Journal of Primatology, Volume 17, pp. 31-50.

Chapman, C., Chapman, L. & Glander, K., 1989. Primate populations in northwestern Costa Rica: Potential for recovery. Primates, Volume 10, pp. 37-44.

Chapman, C., Fedigan, L. & Fedigan, L., 1988. A comparison of transect methods of estimating population densities of Costa Rican primates. Brenesia, Volume 30, pp. 67-80.

Chapman, C., Lawes, M., Naughton-Treves, L. & Gillespie, T., 2003. Primate survival in community-owned forest fragments: Are metapopulation models useful amidst intensive use? In: L. Marsh, ed. Primates in fragments: Ecology and conservation. New York: Kluwer Academic/Plenum Publishers, pp. 63-78.

Chapman, C. & Russo, S., 2011. Primate seed dispersal: Linking behavioral ecology with forest community structure. In: C. Campbell, A. Fuentes, K. MacKinnon, S. Bearder, R. Stumpf eds. Primates in Perspective. Oxford: Oxford University Press, pp. 523-534.

Chapman, C., Wrangham, R. & Chapman, L., 1994. Indicies of habitat-wide fruit abundance in tropical forests. Biotropica, 26(2), pp. 160-171.

Chatterjee, S., Hadi, A. & Price, B., 2000. Regression analysis by example. New York: Wiley.

Chazdon, R., Peres, C., Dent, D., Sheil, D., Lugo, A., Lamb, D., Stork, N., Miller, S., 2009. The potential for species conservation in tropical secondary forests. Conservation Biology, 23(6), pp. 1406-1417.

204

Chen, J., Franklin, J. & Spies, T., 1992. Vegetation response to edge environments in old- growth Douglas-fir forests. Ecological Applications, 2(4), pp. 387-396.

Christensen, J., Hewitson, B., Busuioc, A., Chen, A., Gao, X., Held, R., Jones, R., Kolli, R., Kwon, W., Laprise, R., Magana Rueda, V., Mearns, L., Menendez, C., Räisänen, J., Rinke, A., Sarr, A., Whetton, P., Arritt, R., Benestad, R., 2007. Regional climate projections, climate change, 2007: The physical science basis. Contribution of working group I to the fourth assessment report of the Intergovernmental Panel on Climate Change. Chapter 11., Cambridge: University Press.

Cochrane, T., 2011. Agteca. [Online] Available at: http://www.agteca.com/climate.htm (using data formerly available from NOAA at: http://lwf.ncdc.noaa.gov/oa/pub/data/ghcn/v2/ghcnftp_zipd.html) [Accessed 15 June 2011].

Cohn-Haft, M., Whittaker, A. & Stouffer, P., 1997. A new look at the "species-poor" central Amazon: The avifauna north of Manaus, Brazil. Ornithological Monographs, Volume 48, pp. 205-235.

Coley, P., 1980. Effects of leaf age and plant life history patterns on herbivory. Nature, 484(7395), pp. 545-546.

Collins, A. & Durbach, J., 2000. Biogeographic and ecological forces responsible for speciation in Ateles. International Journal of Primatology, 21(3), pp. 421-444.

Crome, F., Moore, L. & Richards, G., 1992. A study of logging damage in upland rainforest in north Queensland. Forest Ecology and Management, Volume 49, pp. 1-29. da S. Ribeiro, J., Hopkins, M., Vicentini, A., Sothers, C., da S. Costa, M., de Brito, J., de Souza, M., Martins, L., Lohmann, L., Assunção, P., 1999. Flora da Reserva Ducke. Manaus: INPA/DFID.

205

Day, R. & Elwood, R., 1999. Sleeping site selection by the golden-handed tamarin Saguinus midas midas: The role of predation risk, proximity to feeding sites, and terriotrial defense. Ethology, Volume 105, pp. 1035-1051.

Defler, T. & Pintor, D., 1985. Censusing primates by transect in a forest of known primate density. International Journal of Primatology, Volume 6, pp. 243-249.

Denyer, K., Burns, B. & Ogden, J., 2006. Buffering of native forest edge microclimate by adjoining tree plantations. Austral Ecology, Volume 31, pp. 478-489.

Didham, R., 1997. The influence of edge effects and forest fragmentation on leaf litter invertebrates in central Amazonia. In: W. Laurance & R. Bierregaard, eds. Tropical forest remnants: Ecology, management, and conservation of fragmented communities. Chicago: University of Chicago Press, pp. 55-70.

Didham, R. & Lawton, J., 1999. Edge structure determines the magnitude of changes in microclimate and vegetation structure in tropical forest fragments. Biotropica, Volume 31, pp. 17-30.

Eberhardt, L., 1968. A preliminary appraisal of line transects. Journal of Wildlife Management, Volume 32.

Egler, S., 1986. Estudos bionomicos de Saguinus bicolor em uma mata alterada em Manaus, AM (Master's thesis). Campinas: Universidade Estadual de Campinas.

Eisenberg, J., 1989. Mammals of the Neotropics: The northern Neotropics, Vol. 1. Chicago: University of Chicago Press.

Estrada, A., Coates-Estrada, R. & Meritt, D. J., 1994. Non flying mammals and landscape changes in the tropical rain forest region of Los Tuxtlas, Mexico. Ecography, Volume 17, pp. 229-241.

Fearnside, P., 2001. Soybean cultivation as a threat to the environment in Brazil. Environmental Conservation, Volume 28, pp. 23-38.

206

Fearnside, P., 2006. Brazil's Manaus-Porto Velho Highway and the potential impacts of linking the arc of deforestation in central Amazonia. Environmental Management, Volume 38, pp. 705-716.

Fearnside, P. & Guimarães, W., 1996. Carbon uptake by secondary forests in Brazilian Amazonia. Forest Ecology and Management, Volume 80, pp. 35-46.

Fedigan, L., Rose, L. & Avila, R., 1996. See how they grow: Tracking capuchin monkey populations in a regenerating Costa Rican dry forest. In: M. Norconk, A. Rosenberger & P. Garber, eds. Adaptive radiations of Neotropical primates. New York: Plenum Press, pp. 289-307.

Feeley, K. & Terborgh, J., 2006. Habitat fragmentation and effects of herbivore (howler monkey) abundances on bird species richness. Ecology, 87(1), pp. 144-150.

Ferrari, S. & Ferrari, M., 1989. A re-evaluation of the social organization of the Callitrichidae, with reference to the ecological differences between genera. Folia Primatologica, Volume 52, pp. 132-147.

Ferrari, S., Iwanaga, S., Coutinho, P., Messias, M., Cruz Neto, E., Ramos, E., Ramos, P., 1999. Zoogeography of Chiropotes albinasus (Platyrrhini, Atelidae) in southwestern Amazonia. International Journal of Primatology, 20(6), pp. 995- 1004.

Ferreira, L. & Laurance, W., 1997. Effects of forest fragmentation on mortality and damage of selected trees in central Amazonia. Conservation Biology, Volume 11, pp. 797-801.

Fimbel, C., 1994. Ecological correlates of species success in modified habitats may be disturbance and site-specific: The primates of Tiwai Island. Conservation Biology, Volume 1, pp. 106-113.

207

Fleagle, J. & Mittermeier, R., 1980. Locomotor behavior, body size, and comparative ecology of seven Surinam monkeys. American Journal of Physical Anthropology, Volume 52, pp. 301-314.

Ford, S., 1994. Evolution of sexual dimorphism in body weight in platyrrhines. American Journal of Primatology, Volume 34, pp. 221-244.

Ford, S. & Davis, L., 1992. Systematics and body size: Implications for feeding adaptations in New World monkeys. American Journal of Physical Anthropology, Volume 88, pp. 415-468.

Fowler, H., Silva, C. & Venticinque, E., 1993. Size, taxonomic and biomass distributions of flying insects in central Amazonia: Forest edge vs. understory. Revista de Biologica Tropical, 41(3), pp. 755-760.

Fragazy, D., Visalberghi, E. & Fedigan, L., 2004. The complete capuchin. Cambridge: Cambridge University Press.

Frazão, E., 1991. Insectivory in free-ranging bearded sakis. Primates, Volume 32, pp. 245-247.

Frazão, E., 1992. Comportamento alimentar de Chiropotes satanas chiropotes (CEBIDAE: PRIMATAS) na Amazônia central Brazileira (M.A. thesis). Manaus: INPA.

Fredericksen, N., Terceros, C., Chubinu, J., Karamians, R., Fredericksen, T., 2007. Impacts of silvicultural treatments on wildlife populations in a Bolivian humid forest: Determining the impacts of logging on wildlife using indicator species, Santa Cruz Bolivia: Proyecto BOLFOR.

Freese, C., Heltne, P., Castro, R. & Whitesides, G., 1982. Patterns and determinants of monkey densities in Peru and Bolivia, with notes on distributions. International Journal of Primatology, Volume 3, pp. 53-90.

208

Freese, C. & Oppenheimer, J., 1981. The capuchin monkeys, genus Cebus. In: A. Coimbra-Filho & R. Mittermeier, eds. Ecology and behavior of Neotropical primates. Volume 1. Rio de Janeiro: Academia Brasileira de Ciências, pp. 331-390.

Ganzhorn, J., 1995. Low-level forest disturbance effects on primary production, leaf chemistry, and lemur populations. Ecology, Volume 76, pp. 2084-2096.

Garber, P., 1992. Vertical clinging, small body size, and the evolution of feeding adaptations in the Callitrichinae. American Journal of Physical Anthropology, Volume 88, pp. 469-482.

Garber, P., 1993. Feeding ecology and behaviour of the genus Saguinus. In: Marmosets and tamarins: Systematics, ecology, and behaviour. Oxford: Oxford University Press, pp. 273-295.

Garson, G., 2012. Generalized linear models & generalized estimating equations. Asheboro(NC): Statistical Associates Publishing.

Garwood, N., 1989. Tropical soil seed banks: A review. In: M. Leck, V. Parker & L. Simpson, eds. Ecology of soil seed banks. San Diego: Academic Press, Inc., pp. 149-209.

Gascon, C. & Bierregaard Jr., R., 2001. The Biological Dynamics of Forest Fragments Project: The study site, experimental design, and research activity. In: R. Bierregaard Jr., C. Gascon, T. Lovejoy & R. Mesquita, eds. Lessons from Amazonia: The ecology and conservation of a fragmented forest. New Haven & London: Yale University Press, pp. 31-42.

Gaulin, S., 1977. The ecology of Alouatta seniculus in Andean cloud forest (Ph.D. dissertation). Cambridge(MA): Harvard University.

Gaulin, S. & Gaulin, C., 1982. Behavioral ecology of Alouatta seniculus in Andean cloud forest. International Journal of Primatology, Volume 3, pp. 1-32.

209

Gause, G., 1934. The struggle for existence. Baltimore: Williams and Wilkins.

Gautier-Hion, A., 1980. Seasonal variations of diet related to species and sex in a community of Cercopithecus monkeys. Journal of Animal Ecology, Volume 49, pp. 237-269.

Gavazzi, A., Cornick, L., Markowitz, T., Green, D., Markowitz, H., 2008. Density, distribution, and home range of the black howler monkey (Alouatta pigra) at Lamanai, Belize. Journal of Mammalogy, 89(5), pp. 1105-1112.

Gibbs, H., Ruesch, A., Achard, F., Clayton, M., Holmgren, P., Ramankutty, N., Foley, J.A., 2010. Tropical forests were the primary sources of new agricultural land in the 1980s and 1990s. Proceedings of the National Academy of Sciences of the United States of America, 107(38), pp. 16732-16737.

Gilbert, K., 1994a. Endoparasitic infection in red howling monkeys (Alouatta seniculus) in the central Amazonian Basin: A cost of sociality? (Ph.D. dissertation). New Brunswick: Rutgers University.

Gilbert, K., 1994b. Parasitic infection in red howling monkeys in forest fragments. Neotropical Primates, Volume 2, pp. 10-12.

Gilbert, K., 1997. Red howling monkey use of defecation sites as a parasite avoidance strategy. Animal Behaviour, 54(2), pp. 451-455.

Gilbert, K., 2000. Attempted predation on a white-faced saki in the central Amazon. Neotropical Primates, 8(3), pp. 103-104.

Gilbert, K., 2003. Primates and fragmentation of the Amazon forest. In: L. Marsh, ed. Primates in fragments. New York: Kluwer Academic, pp. 145-157.

Glander, K., 1975. Habitat description and resource utilization: A preliminary report on mantled howler monkey ecology. In: R. Tuttle, ed. Socioecology and psychology of primates. The Hague: Mouton, pp. 37-57.

210

Grand, T., 1984. Motion economy withi the canopy: Four strategies for mobility. In: P. Rodman & J. Cant, eds. Adaptations for foraging in nonhuman primates. New York: Columbia University Press, pp. 54-72.

Green, K., 1978. Primate censusing in northern Colombia: A comparison of two techniques. Primates, 19(3), pp. 537-550.

Grieser Johns, A., 1997. Timber production and biodiversity conservation in tropical rain forests. Cambridge: Cambridge University Press.

Grieser Johns, A., 2001. Natural forest management and biodiversity conservation. In: R. Fimbel, A. Grajal & J. Robinson, eds. The cutting edge: Conserving wildlife in logged tropical forest. New York: Columbia University Press, pp. 405-422.

Guevara, S., Laborde, J. & Sánchez-Rios, G., 2004. Rain forest regeneration beneath the canopy of fig trees isolated in pastures of Los Tuxtlas, Mexico. Biotropica, 36(1), pp. 99-108.

Guevara, S., Meave, J., Moreno-Casasola, P. & Laborde, J., 1992. Floristic composition and structure of vegetation under isolated trees in Neotropical pastures. Journal of Vegetative Science, Volume 3, pp. 655-664.

Guillotin, M., Dubost, G. & Sabatier, D., 1994. Food choice and food competition among three major primate species of French Guiana. Journal of Zoology, 233(3), pp. 551-579.

Guimarães Jr., P., José, J., Galetti, M. & Trigo, J., 2003. Quinolizidine alkaloids in Ormosia arborea seeds inhibit predation but not hoarding by agoutis (Dasyprocta leporina). Journal of Chemical Ecology, 29(5), pp. 1065-1072.

Haugaasen, T. & Peres, C., 2005. Primate assemblage structure in Amazonian flooded and unflooded forests. American Journal of Primatology, Volume 67, pp. 243- 258.

211

Heckenberger, M., Petersen, B. & Neves, E., 1999. Village size and permanence in Amazonia: Two archaeological case examples from Brazil. Latin American Antiquity, Volume 10, pp. 353-376.

Heckenberger, M., Russell, J., Toney, J. & Schmidt, M., 2007. The legacy of cultural landscapes in the Brazilian Amazon: Implications for biodiversity. Philosophical Transactions of the Royal Society B, Volume 362, pp. 197-208.

Heinze, G. & Schemper, M., 2002. A solution to the problem of separation in logistic regression. Statistics in Medicine, Volume 21, pp. 2409-2419.

Herremans, M. & Herremans-Tonnoeyr, D., 2000. Land use and the conservation status of raptors in Botswana. Biological Conservation, Volume 94, pp. 31-41.

Hershkovitz, P., 1977. Living New World monkeys (platyrrhini): With an introduction to primates. Chicago: University of Chicago Press.

Hershkovitz, P., 1985. A preliminary taxonomic review of the South American bearded saki monkeys genus Chiropotes (Cebidae: Platyrrhini), with the description of a new species. Zoology, Volume 27, pp. 1-46.

Instituto Nacional de Pesquisas Espaciais, 2012. SONDA (Sistema de Organização de Dados Ambientais). [Online] Available at: http://sonda.ccst.inpe.br/ [Accessed 2011 June 2012].

ITTO, 2002. ITTO (International Tropical Timber Organization) guidelines for the restoration, management and rehabilitation of degraded and secondary tropical forests. ITTO policy development series 13., Yokohama, Japan: ITTO.

IUCN, 2012. IUCN Red List of threatened species. Version 2012.2. [Online] Available at: www.iucnredlist.org [Accessed 15 10 2012].

212

Izawa, K., 1980. Social behavior of the wild black-capped capuchin (Cebus apella). Primates, Volume 21, pp. 443-467.

Janson, C. & Terborgh, J., 1988. Censusing primates in rainforest, with reference to the primate community of Cocha Cashu Biological Station, Manu National Park, Peru, Unpublished report.

Johns, A., 1985a. Primates and forest exploitation at Tefé, Brazilian Amazonia. Primate Conservation, Volume 6, pp. 27-29.

Johns, A., 1985b. Selective logging and wildlife conservation in tropical rain forest: Problems and recommendations. Biological Conservation, Volume 31, pp. 355- 375.

Johns, A., 1986. Effects of habitat disturbance on rain forest wildlife in Brazilian Amazonia. Unpublished report, Washington D.C.: World Wildlife Fund US.

Johns, A., 1988. Effects of "selective" timber extraction on rain forest structure and composition and some consequences for frugivores and folivores. Biotropica, Volume 20, pp. 31-37.

Johns, A., 1989. Timber, the environment and wildlife in Malaysian rain forests. Report to ODA (Overseas Development Administration)/NERC (Natural Environment Resource Council) project F3CR26/G1/05, Aberdeen, Scotland: University of Aberdeen Institute of South-east Asian Biology.

Johns, A., 1991. Forest disturbance and Amazonian primates. In: Primate responses to environmental change. London: Chapman & Hall, pp. 115-135.

Johns, A., 1992. Species conservation in managed tropical forests. In: T. Whitmore & J. Sayer, eds. Tropical deforestation and species extinction. London: Chapman and Hall, pp. 15-53.

213

Johns, A. & Ayers, J., 1987. Conservation of white uacaris in Amazonian varzea. Oryx, 21(2), pp. 74-80.

Johns, A. & Skorupa, J., 1987. Responses of rain-forest primates to habitat disturbance: A review. International Journal of Primatology, 8(2), pp. 157-191.

Jones, E., 1955. Ecological studies on the rain forest of Southern Nigeria IV. The plateau forest of the Okomu Forest Reserve. Journal of Ecology, Volume 43, pp. 564-594.

Julliot, C., 1992. Utilisation des resources alimentaires par le singe hurleur roux, Alouatta seniculus (Atelidae, Primates), en Guyane: Impact de la dissemmination des graines sur la regeneration forestiere. These de doctorat. Tours: Universite de Tours.

Julliot, C., 1996a. Fruit choice by red howlers monkeys (Alouatta seniculus) in a tropical forest. American Journal of Primatology, Volume 40, pp. 261-282.

Julliot, C., 1996b. Seed dispersal by red howling monkeys (Alouatta seniculus) in the tropical rain forest of Guiana. International Journal of Primatology, Volume 17, pp. 239-258.

Julliot, C. & Sabatier, D., 1993. Diet of the red howler monkey (Alouatta seniculus) in French Guiana. International Journal of Primatology, 14(4), pp. 527-550.

Jurmain, R., 1997. Skeletal evidence of trauma in African apes, with special reference to the Gombe chimpanzees. Primates, 38(1), pp. 1-14.

Kapos, V., 1989. Effects of isolation on the water status of forest patches in the Brazilian Amazon. Journal of Tropical Ecology, Volume 5, pp. 173-185.

Kapos, V., Ganade, G., Matsui, E. & Victoria, R., 1993. ∂ 13C as an indicator of edge effects in tropical rainforest reserves. Journal of Ecology, Volume 81, pp. 425- 432.

214

Kelker, G., 1945. Measurement and interpretation of forces that determine populations of managed deer (Ph.D. dissertation). Ann Arbor: University of Michigan.

Kessler, P., 1995. Preliminary filed study of the red-handed tamarin, Saguinus midas, in French Guiana. Neotropical Primates, Volume 3, p. 184.

Kessler, P., 1998. Primate densities in the Natural Reserve of Nourages, French Guiana. Neotropical Primates, 6(2), pp. 45-46.

Kimmins, J., 2004. Forest ecology. In: T. Northcote & G. Hartman, eds. Fishes and forestry: Worldwide watershed interactions and management. Oxford: Blackwell Science Ltd., pp. 19-43.

Kinzey, W., 1997a. Alouatta. In: W. Kinzey, ed. New world primates: Ecology, evolution, and behavior. New York: Aldine de Gruyter, pp. 174-185.

Kinzey, W., 1997b. Ateles. In: W. Kinzey, ed. New World primates: Ecology, evolution, and behavior. New York: Aldine de Gruyter, pp. 192-199.

Kinzey, W., 1997c. Cebus. In: W. Kinzey, ed. New World primates: Ecology, evolution, and behavior. New York: Aldine de Gruyter, pp. 248-257.

Kinzey, W., 1997d. Chiropotes. In: W. Kinzey, ed. New World primates: Ecology, evolution, and behavior. New York: Aldine de Gruyter, pp. 258-263.

Kinzey, W., 1997e. Pithecia. In: W. Kinzey, ed. New World primates: Ecology, evolution, and behavior. New York: Aldine de Gruyter, pp. 281-288.

Kinzey, W., 1997f. Saguinus. In: W. Kinzey, ed. New World primates: Ecology, evolution, and behavior. New York: Aldine de Gruyter, pp. 289-296.

Kinzey, W. & Norconk, M., 1990. Hardness as a basis of friut choice in two sympatric primates. American Journal of Physical Anthropoology, 81(1), pp. 5-15.

215

Kinzey, W. & Norconk, M., 1993. Physical and chemical properties of fruit and seeds eaten by Pithecia and Chiropotes in Surinam and Venezuela. International Journal or Primatology, Volume 14, pp. 207-227.

Kinzey, W., Norconk, M. & Alvarez-Cordero, E., 1988. Primate survey of eastern Bolivia. Primate Conservation, Volume 9, pp. 66-70.

Lahti, D., 2001. The "edge effect on nest predation" hypothesis after twenty years. Biological Conservation, Volume 99, pp. 365-374.

Larose, F., 1996. Foraging strategies, group size, and food competition in the mantled howler monkey, Alouatta palliata. Ph.D. thesis. Alberta: University of Alberta.

Laurance, W., 1991. Edge effects in tropical forest fragments: Application of a model for the design of nature reserves. Biological Conservation, 57(2), pp. 205-219.

Laurance, W., 2000. Do edge effects occur over large spatial scales? Trends in Ecology and Evolution, Volume 15, pp. 134-135.

Laurance, W., 2001. The hyper-diverse flora of the central Amazon. In: R. Bierregaard Jr., C. Gascon, T. Lovejoy & R. Mesquita, eds. Lessons from Amazonia: The ecology and conservation of a fragmented forest. New Haven: Yale University Press, pp. 47-53.

Laurance, W., Albernaz, A., Fearnside, P., Vasconcelos, H., Ferreira, L., 2004. Deforestation in Amazonia. Science, 304(5674), p. 1109.

Laurance, W., Camargo, J., Luizão, R., Laurance, S., Pimm, S., Bruna, E., Stouffer, P., Williamson, G., Benítez-Malvido, J., Vasconcelos, H., van Houtan, K., Zartman, C., Boyle, S., Didham, R., Andrade, A., Lovejoy, T., 2011. The fate of Amazonian forest fragments: A 32-year investigation. Biological Conservation, Volume 144, pp. 56-67.

216

Laurance, W., Cochrane, M., Bergen, S., Fearnside, P., Delamonica, P., Barber, C., D'Angelo, S., Fernandes, T., 2001b. The future of the Brazilian Amazon. Science, 291(5503), pp. 438-439.

Laurance, W., Delamonica, P., Laurance, S., Vasconcelos, H., Lovejoy, T., 2000. Rainforest fragmentation kills big trees. Nature, Volume 404, p. 836.

Laurance, W., Ferreira, L., Rankin de Merona, J., Laurance, S., Hutchings, R., Lovejoy, T., 1998a. Effects of forest fragmentation on recruitment patterns in Amazonian tree communities. Conservation Biology, Volume 12, pp. 460-464.

Laurance, W., Ferreira, L., Rankin-de Merona, J. & Laurance, S., 1998b. Rain forest fragmentation and the dynamics of Amazonian tree commmunities. Ecology, Volume 79, pp. 2032-2040.

Laurance, W., Laurance, S., Ferreira, L., Rankin de Merona, J., Gascon, C., Lovejoy, T., 1997. Biomass collapse in Amazonian forest fragments. Science, Volume 278, pp. 1117-1118.

Laurance, W., Lovejoy, T., Vasconcelos, H., Bruna, E., Didham, R., Stouffer, P., Gascon, C., Bierregaard, R., Laurance, S., Sampaio, E., 2002. Ecosystem decay of Amazonian forest fragments: A 22-year investigation. Conservation Biology, pp. 605-618.

Laurance, W. & Luizão, R., 2007. Driving a wedge into the Amazon. Nature, Volume 448, pp. 409-410.

Laurance, W., Nascimento, H., Laurance, S., Andrade, A., Fearnside, P., Ribeiro, J., 2006a. Rain forest fragmentation and the proliferation of successional trees. Ecology, Issue 87, pp. 469-482.

Laurance, W., Nascimento, H., Laurance, S., Andrade, A., Ribeiro, J., Giraldo, J., Lovejoy, T., Condit, R., Chave, J., D'Angelo, S., 2006b. Rapid decay of tree-community composition in Amazonian forest fragments. Proceedings of the National Academy of Sciences USA, Volume 103, pp. 19010-19014.

217

Laurance, W., Nascimento, H., Laurance, S., Andrade, A., Ewers, R., Harms, K., Luizão, R., Ribeiro, J., Bennett, P., 2007. Habitat fragmentation, variable edge effects, and the landscape-divergence hypothesis. PLoS ONE, 2(10).

Laurance, W., Perez-Salicrup, D., Delaminica, P., Fearnside, P., D'Angelo, S., Jerozolinski, A., Pohl, L., Lovejoy, T., 2001a. Rain forest fragmentation and the structure of Amazonian liana communities. Ecology, Volume 82, pp. 105-116.

Lees, A. & Peres, C., 2008. Conservation value of remnant riparian forest corridors of varying quality for Amazonian birds and mammals. Conservation Biology, 22(2), pp. 439-449.

Lees, A. & Peres, C., 2009. Gap-crossing movements predict species occupancy in Amazonian forest fragments. Oikos, 118(2), pp. 280-290.

Lehman, S., 2004. Biogeography of the primates of Guyana: Effects of habitat use and diet on geographic distribution. International Journal of Primatology, 25(6), pp. 1225-1242.

Lehman, S., 2007. Spatial variations in Eulemur fulvus rufus and Lepilemur mustelinus densities in Madagascar. Folia Primatologica, 78(1), pp. 46-55.

Lehman, S., Prince, W. & Mayor, M., 2001. Variations in group size in white-faced sakis (Pithecia pithecia): Evidence for monogamy or seasonal congregation? Netropical Primates, Volume 9, pp. 96-101.

Lehman, S., Rajaonson, A. & Day, S., 2006b. Edge effects and their influence on lemur density and distribution in southeast Madagascar. American Journal of Physical Anthropology, 129(2), pp. 232-241.

Lehman, S., Rajaonson, A. & Day, S., 2006c. Edge effects on the density of Chierogaleus major. International Journal of Primatology, 27(6), pp. 1569-1588.

218

Lehman, S., Rajaonson, A. & Day, S., 2006d. Lemur responses to edge effects in the Vohibola III Classified Forest, Madagascar. American Journal of Primatology, pp. 293-299.

Lehman, S., Sussman, R., Phillips-Conroy, J. & Prince, W., 2006a. Ecological biogeography of primates in Guyana. In: S. Lehman & J. Fleagle, eds. Primate biogeography: Progress and prospects. New York: Springer, pp. 105-130.

Leighton, M. & Leighton, D., 1982. The relationship of size of feeding aggregate to size of food patch: Howler monkeys (Alouatta palliata) feeding in Trichilia cipo fruit trees on Barro Colorado Island. Biotropica, 14(2), pp. 81-90.

Lemos, M., 1998. Padrões de frutificao e efeito do corte seletivo sobre a produção de frutos em uma floresta de terra-firme da Amazônia central (M.Sc. thesis). Manaus: INPA.

Lenz, B. & dos Reis, A., 2011. Harpy eagle-primate interactions in the central Amazon. The Wilson Journal of Ornithology, 123(2), pp. 404-408.

Leopold, A., 1933. Game management. New York: Scribners.

Lewis, S., 1998. Treefall gaps and regeneration: A comparison of continuous forest and fragmented forest in central Amazonia (Ph.D. dissertation). Cambridge: University of Cambridge.

López-Toledo, L. & Martínez-Ramos, M., 2011. The soil seed bank in abandoned tropical pastures: Source of regeneration or invasion? Revista Mexicana de Bioversidad, Volume 82, pp. 663-678.

Lovejoy, T., Bierregaard Jr., R., Rylands, A., Malcolm, J., Quintela, C., Harper, L., Brown, Jr., K., Powell, A., Powell, G., Schubart, H., Hays, M., 1986. Edge and other effects of isolation on Amazon forest fragments. In: M. Soule, ed. Conservation biology: The science of scarcity and diversity. Sunderland: Sinauer, pp. 257-285.

219

Lovejoy, T. & Oren, D., 1981. Minimum critical size of ecosystems. In: R. Burgess & D. Sharp, eds. Forest island dynamics in man-dominated landscapes. New York: Springer-Verlag, pp. 7-12.

Lovejoy, T., Rankin, J., Bierregaard Jr., R., Brown Jr., K., Emmons, L., van der Voort, M., 1984. Ecosystem decay of Amazon forest remnants. In: M. Nitecki, ed. Extinctions. Chicago: University of Chicago Press, pp. 295-325.

Lovell, N., 1990. Patterns of injury and illness in great apes: A skeletal analysis. Washington DC: Smithsonian Institution Press.

Lovell, N., 1991. An evolutionary framework for assessing illness and injury in nonhuman primates. Yearbook of Physical Anthropology, Volume 34, pp. 117-155.

Lynch Alfaro, J., Boubli, J., Olson, L., Di Fiore, A., Wilson, B., Gutiérres-Espeleta, G., Chiou, K., Schulte, M., Neitzel, S., Ross, V., Schwochow, D., Nguyen, M., Farias, I., Janson, C., Alfaro, M., 2012a. Explosive Pleistocene range expansion leads to widespread Amazonian sympatry between robust and gracile capuchin monkeys. Journal of Biogeography, 39(2), pp. 272-288.

Lynch Alfaro, J., Silva Jr., J. & Rylands, A., 2012b. How different are robust and gracile capuchin monkeys? An argument for the use of Sapajus and Cebus. American Journal of Primatology, 74(4), pp. 273-286.

Malcolm, J., 1994. Edge effects in central Amazonian forest fragments. Ecology, Volume 75, pp. 2438-2445.

Malhi, Y., Roberts, J., Betts, R., Killeen, T., Li, W., Nobre, C., 2008. Climate change, deforestation, and the fate of the Amazon. Science, Volume 319, pp. 169-172.

Margulis, S., 2004. Causes of deforestation of the Brazilian Amazon. World Bank working paper no. 22, Washington DC: World Bank Publications.

220

Marsh, C. & Wilson, W., 1981. A survey of primates in peninsular Malaysian forests. Cambridge: Universiti Kebangsaan Malaysia and University of Cambridge.

Marsh, L., In press. A taxonomic revision of the saki monkeys, group Pithecia (Desmarest 1804). Neotropical Primates, Volume Special monograph, p. 235.

Mbora, D. & Meikle, D., 2004. Forest fragmentation and the distribution, abundance and conservation of the Tana River red colobus (Procolobus rufomitratus). Biological Conservation, Volume 118, pp. 67-77.

McGoogan, K., Steffens, T. & Lehman, S., 2009. Edge effects on home range size of Propithecus coquereli in northwest Madagascar [Abstract]. 71(Suppl 1), p. 71.

Mendes Pontes, A., 1997. Habitat partitioning among primate in Maracá Island, Roraima, northern Brazilian Amazonia. International Journal of Primatology, 18(2), pp. 131-157.

Mesquita, R., Ickes, K., Ganade, G. & Williamson, G., 2001. Alternative successional pathways in the Amazon Basin. Journal of Ecology, Volume 89, pp. 528-537.

Mittermeier, R., 1977. Distribution, synecology and conservation of Surinam monkeys (Ph.D. dissertation). Cambridge(MA): Harvard University.

Mittermeier, R., Rylands, A. & Boubli, J.-P., 2008a. Ateles paniscus. [Online] Available at: http://www.iucnredlist.org/details/2283/0 [Accessed 15 10 2012].

Mittermeier, R., Rylands, A. & Boubli, J.-P., 2008b. Saguinus midas. [Online] Available at: http://www.iucnredlist.org/details/41525/0 [Accessed 15 10 2012].

Mittermeier, R. & van Roosmalen, M., 1981. Preliminary observations on habitat utilization and diet in eight Surinam monkeys. Folia Primatologica, Volume 36, pp. 1-39.

221

Mittermeier, R. & van Roosmalen, M., 1983. A synecological study of Surinam monkeys. In: A. Rhodin & K. Miyata, eds. Advances in herpetology and evolutionary biology. Cambridge(MA): Museum of Comparative Zoology, pp. 521-534.

Morris, D., 1984. Patterns and scale of habitat use in two temperate-zone, small mammal faunas. Candaian Journal of Zoology, Volume 62, pp. 1540-1547.

Muckenhirn, N., Mortensen, S., Vessey, S., Fraser, C., Singh, B., 1976. Report on a primate survey in Guyana, Washington D.C.: Pan American Health Organization.

Murcia, C., 1995. Edge effects in fragmented forests: Implications for conservation. Trends in Ecology and Evolution, Volume 10, pp. 58-62.

Napier, P., 1976. Catalogue of primates in the British Museum (Natural History), Part 1: Family Callitrichidae and Cebidae. London: British Museum (Natural History).

National Research Council, 1981. Techniques for the study of primate population ecology. Washington D.C.: National Academy Press.

Negrón-Juárez, R., Chambers, J., Guimaraes, G., Zeng, H., Raupp, C., Marra, D., Ribeiro, G., Saatchi, S., Nelson, B., Higuchi, N., 2010. Widespread Amazon forest tree mortality from a single cross-basin squall line event. Geophysical Research Letters, 37(16), p. L16701.

Nelson, B., Kapos, V., Adams, J., Oliveira, W., Braun, O., do Amaral, I., 1994. Forest disturbance by large blowdowns in the Brazilian Amazon. Ecology, 75(3), pp. 853- 858.

Nepstad, D., Schwartzman, S., Bamberger, B., Santilli, M., Ray, D., Schlesinger, P., Lefebvre, P., Alencar, A., Prinz, E., Fiske, G., Rolla, A., 2006. Inhibition of Amazon deforestation and fire by parks and indigenous lands. Conservation Biology, 20(1), pp. 65-73.

222

Nepstad, D., Stickler, C., Soares-Filho, B. & Merry, F., 2008. Interactions among Amazon land use, forests and climate: Prospects for a near-term forest tipping point. Philosophical Transactions of the Royal Society of London B: Biological Sciences, 363(1498), pp. 1737-1746.

Nepstad, D., Uhl, C. & Serrão, E., 1991. Recuperation of a degraded Amazonian landscape: Forest recovery and agricultural restoration. Ambio, Volume 20, pp. 248-255.

Neves, A., 1985. Alguns aspectos da ecologia de Alouatta seniculus em reserva isolada na Amazônia Central (Monograph). São Paulo: Universidade de São Paulo, Ribeirão Preto.

Neves, A. & Rylands, A., 1991. Diet of a group of howling monkeys, Alouatta seniculus, in an isolated forest patch in central Amazonia. A Primatologia no Brasil, Volume 3, pp. 263-274.

Neves, E., Petersen, J., Bartone, R. & da Silva, C., 2003. Historical and socio-cultural origins of Amazonian dark earths. In: J. Lehmann, D. Kern, B. Glaser & W. Woods, eds. Amazonian dark earths: Origin, properties, management. Dordrecht: Kluwer Academic, pp. 29-50.

Neves, E., Petersen, J., Bartone, R. & Heckenberger, M., 2004. The timing of terra preta formation in the Central Amazon. In: B. Glaser & W. Woods, eds. Amazonian dark earths: Explorations in space and time. Berlin: Springer-Verlag, pp. 125-134.

Neville, M., 1976. The population and conservation of howler monkeys in Venezuela and Trinidad. In: R. Thorington & P. Heltne, eds. Neotropical primates: Field studies and conservation. Washington D.C.: National Academy of Sciences, pp. 101-109.

223

Neville, M., Glander, K., Braza, F. & Rylands, A., 1988. The howling monkeys, genus Alouatta. In: R. Mittermeier, A. Rylands, A. Coimbra-Filho & A. da Fonseca, eds. Ecology and behavior of Neotropical primates. Volume 2. Washington D.C.: World Wildlife Fund, pp. 349-453.

Norconk, M., 1996. Seasonal variations in the diets of white-faced and bearded sakis (Pithecia pithecia and Chiropotes satanas) in Guri Lake, Venezuela. In: M. Norconk, A. Rosenberger & P. Garber, eds. Adaptive Radiations of Neotropical Primates. New York: Plenum, pp. 403-423.

Norconk, M. & Conklin-Brittain, N., 2004. Variation on frugivory: The diet of Venezuelan white-faced sakis. International Journal of Primatology, 25(1), pp. 1-26.

Norconk, M. & Kinzey, W., 1994. Challenge of Neotropical frugivory: Travel patterns of spider monkeys and bearded sakis. American Journal of Primatology, Volume 34, pp. 171-183.

Norconk, M., Raghanti, M., Martin, S., Grafton, B., Gregory, L., De Dijn, B., 2003. Primates of Brownsberg Natuurpark, Suriname, with particular attention to the pitheciins. Neotropical Primates, Volume 11, pp. 94-100.

Odum, E., 1969. The strategy of ecosystem development: An understanding of ecological succession provides a basis for resolving man's conflict with nature. Science, Volume 870, pp. 262-270.

Oliveira, A., 1997. Diversidade estrutura e dinâmica do componente arbóreo de uma floresta de terra firme de Manaus, Amazonas (Ph.D. dissertation). São Paulo: Universidade de São Paulo.

Oliveira, A. & Ferrari, S., 2000. Seed dispersal by black-handed tamarins, Saguinus midas niger (Callitrichinae, Primates): Implications for the regeneration of degraded forest habitats in eastern Amazonia. Journal of Tropical Ecology, Volume 16, pp. 709-716.

224

Oliveira, A. & Ferrari, S., 2008. Habitat exploitation by free-ranging Saguinus niger in eastern Amazonia. International Journal of Primatology, Volume 29, pp. 1499- 1510.

Oliveira, A. & Mori, S., 1999. A central Amazonian terra firme forest. I. High tree species richness on poor soils. Biodiversity and Conservation, Volume 8, pp. 1219-1244.

Oliveira, J., Lima, M., Bonvincino, C., Ayres, J., Fleagle, J., 1985. Preliminary notes on the ecology and behavior of the Guianan saki (Pithecia pithecia, Linnaeus 1766; Cebidae, Primates). Acta Amazônica, Volume 15, pp. 249-263.

Oyama, M. & Nobre, C., 2003. A new climate-vegetation equilibrium state for tropical South America. Geophysical Research Letters, 30(23, 2199), pp. CLM 5-1 - CLM 5- 4.

Pack, K., Henry, O. & Sabatier, D., 1999. The insectivorous-frugivorous diet of the golden-handed tamarin (Saguinus midas midas) in French Guiana. Folia Primatologica, Volume 70, pp. 1-7.

Palacios, E., 1998. Ecological bases for lake- and river-side habitat use of Alouatta seniculus in Colombian Amazonia. ASP Bulletin, 22(3), p. 8.

Palacios, E. & Rodriguez, A., 2001. Ranging pattern and use of space in a group of red howler monkeys (Alouatta seniculus) in a southeastern Colombian rainforest. American Journal of Primatology, 55(4), pp. 233-251.

Parrotta, J., Knowles, O. & Wunderle Jr., J., 1997. Development of floristic diversity in 10-year-old restoration forests on a bauxite mined site in Amazonia. Forest Ecology and Management, 99(1-2), pp. 21-42.

Parry, L., Barlow, J. & Peres, C., 2007. Large-vertebrate assemblages of primary and secondary forests in the Brazilian Amazon. Journal of Tropical Ecology, Volume 23, pp. 653-662.

225

Peres, C., 1992. Prey-capture benefits in a mixed-species group of Amazonian tamarins, Saguinus fuscicollis and S. mystax. Behavioral Ecology and Sociobiology, Volume 31, pp. 339-347.

Peres, C., 1993a. Structure and spatial organization of an Amazonian terra firme forest primate community. Journal of Tropical Ecology, Volume 9, pp. 259-276.

Peres, C., 1993b. The ecology of buffy saki monkeys (Pithecia albicans, Gray 1860): A canopy seed predator. American Journal of Primatology, Volume 31, pp. 129- 140.

Peres, C., 1997. Primate community structure at twenty western Amazonian flooded and unflooded forests. Journal of Tropical Ecology, Volume 13, pp. 381-405.

Peres, C., 1998. Primate community structure in western Brazilian Amazonia. Primate Conservation, Volume 9, pp. 83-86.

Peres, C., 1999. General guidelines for standardizing line-transect surveys of tropical forest primates. Neotropical Primates, 7(1), pp. 11-16.

Peters, R., Cloutier, S., Dube, D., Evans, A., Hastings, P., Kaiser, H., Kohn, D., Sarwer- Foner, B., 1988. The allometry of the weight of fruit on trees and shrubs in Barbados. Oecologia, Volume 74, pp. 612-616.

Plumptre, A. & Grieser Johns, A., 2001. Changes in primate communitiess following logging disturbance. In: R. Fimbel, A. Grajal & J. Robinson, eds. The cutting edge: Conserving wildlife in logged tropical forest. New York: Columbia University Press, pp. 71-92.

Plumptre, A. & Reynolds, V., 1994. The effect of selective logging on the primate populations in the Budongo Forest Reserve. Journal of Applied Ecology, Volume 31, pp. 631-641.

226

Port-Carvalho, M. & Ferrari, S., 2004. Occurrence and diet of the black bearded saki (Chiropotes satanas satanas) in the fragmented landscape of western Maranhão, Brazil. Neotropical Primates, 12(1), pp. 17-21.

Porter, L., 2001. Callimico goeldii and Saguinus: Dietary differences between sympatric callitrichines in northern Bolivia. International Journal of Primatology, Volume 22, pp. 961-992.

Porter, L., 2004. Differences in forest utilization and activity patterns among three sympatric callitrichines: Callimico goeldii, Saguinus fuscicollis and S. labiatus. American Journal of Physical Anthropology, Volume 124, pp. 139-153.

Prates-Clark, C., Lucas, R. & dos Santos, J., 2009. Implications of land-use history for forest regeneration in the Brazilian Amazon. Canadian Journal of Remote Sensing, 35(6), pp. 534-553.

Quéméré, E., Champeau, J., Besolo, A., Rasolondraibe, E., Rabarivola, C., Crouau-Roy, B., Chikhi, L., 2010. Spatial variation in density and total size estimates in fragmented primate populations: The golden-crowned sikfaka (Propithecus tattersalli). American Journal of Primatology, Volume 72, pp. 72-80.

Quinn II, T., 1981. The effect of group size on line transect estimators of abundance. Studies in Avian Biology, Volume 6, pp. 502-508.

Quintana-Ascencio, P., González-Espinosa, M., Ramírez-Marcial, N., Domínguez- Vázquez, G., Martínez-Icó, M., 1996. Soil seed banks and regeneration of tropical rain forest from milpa fields at the selva Lacandona, Chiapas, Mexico. Biotropica, Volume 28, pp. 192-209.

Redford, K., 1992. The empty forest. Bioscience, 42(6), pp. 412-422.

Reis, L., Fletcher, R., Battin, J. & Sisk, T., 2003. Ecological responses to habitat edges: Mechanisms, models, and variability explained. Annual Review of Ecology and Systematics, Volume 28, pp. 567-572.

227

Reville, B., Tranter, J. & Yorkston, H., 1990. Impacts of forest clearing on the endangered seabird Sula abbottii. Biological Conservation, Volume 51, pp. 23-38.

Ribeiro, M., Metzger, J., Martensen, A., Ponzoni, F., Hirota, M., 2009. The Brazilian Atlantic Forest: How much is left, and how is the remaining forest distributed? Implications for conservation. Biological Conservation, Volume 142, pp. 1141- 1153.

Robinette, W., Loveless, C. & Jones, D., 1974. Field tests of strip census methods. Journal of Wildlife Management, Volume 38, pp. 81-96.

Robinson, J. & Janson, C., 1987. Capuchins, squirrel monkeys, and atelines: Socioecological convergence with Old World primates. In: B. Smuts, D. Cheney, R. Seyfarth, R. Wrangham, T, Struhsaker eds. Primate societies. Chicago: University of Chicago Press, pp. 69-82.

Robinson, J., Wright, P. & Kinzey, W., 1987. Monogamous cebids and their relatives: Inter-group calls and spacing. In: B. Smuts, D. Cheney, R. Seyfarth, R. Wrangham, T, Struhsaker eds. Primate societies. Chicago: University of Chicago Press, pp. 44- 53.

Roosevelt, A., da Costa, M., Lopes Machado, C., Michab, M., Mercier, N., Valladas, H., Feathers, J., Barnett, W., da Silveira, M., Henderson, A., Silva, J., Chernoff, B., Reese, D., Holman, J., Toth, N., Schick, K., 1996. Paleoindian cave dwellers in the Amazon: The peopling of the Americas. Science, Volume 272, pp. 373-384.

Roosevelt, A., Douglas, J. & Brown, L., 2002. The migrations and adaptations of the first Americans: Clovis and pre-Clovis viewed from South America. In: N. Jablonski, ed. The first Americans: The Pleistocene colonization of the New World. San Francisco: Memoirs of the California Academy of Sciences Number 27, pp. 159- 235.

228

Ross, C. & Reeve, N., 2003. Survey and census methods: Population distribution and density. In: J. Setchell & D. Curtis, eds. Field and laboratory methods in primatology: A practical guide. Cambridge: Cambridge University Press, pp. 90- 109.

Rylands, A., Boubli, J., Mittermeier, R., Wallace, A., Ceballos-Mago, N., 2008. Cebus apella. [Online] Available at: http://www.iucnredlist.org/details/39949/0 [Accessed 15 10 2012].

Rylands, A. & Keuroghlian, A., 1988. Primate populations in continuous forest and forest fragments in central Amazonia. Acta Amazônica, Volume 18, pp. 291-307.

Salati, E., 1985. The climatology and hydrology of Amazonia. In: G. Prance & T. Lovejoy, eds. Key environments: Amazonia. Oxford: Pergamon Press, pp. 18-48.

Salek, M., Kreisinger, J., Sedlacek, F. & Albrecht, T., 2010. Do prey densities determine preferences of mammalian predators for habitat edges in an agricultural landscape? Landscape and Urban Planning, Volume 98, pp. 86-91.

Santamaría-Gómez, A., 1999. Ecologia e comportamento de Alouatta seniculus em uma mata de terra firme na Amazônia central (M.A. thesis). Minas Gerais: Universidade Federal de Minas Gerais.

Santamaria-Gómez, M., 2004. The effect of home range reduction on the ecology of red howler monkeys in central Amazonia (Ph.D. dissertation). Cambridge: University of Cambridge.

Santamaría, M. & Rylands, A., 2003. Ecología básica de un grupo de Alouatta seniculus durante una estación seca en la Amazonía central brasileira. In: V. Pereira- Benoga, F. Montoya & A. Savage, eds. Primatología del nuevo mundo: Biología, medicína, manejo y conservación. Bogotá: Centro de Primatologia Araguiatos Ltda, pp. 96-111.

229

Savill, P., 1983. Silviculture in windy climates. Forestry Abstracts, Volume 44, pp. 473- 488.

Schreier, B., Harcourt, A., Coppeto, S. & Somi, M., 2009. Interspecific competition and niche separation in primates: A global analysis. Biotropica, 41(3), pp. 283-291.

Schultz, A., 1944. Age changes and variability in gibbons. American Journal of Physical Anthropology, Volume 2, pp. 1-129.

Schwartzman, S., Moreira, A. & Nepstad, D., 2000. Rethinking tropical forest conservation: Perils in parks. Conservation Biology, 14(5), pp. 1351-1357.

Schwarzkopf, L. & Rylands, A., 1989. Primate species richness in relation to habitat structure in Amazonian rainforest fragments. Biological Conservation, Volume 48, pp. 1-12.

Seber, G., 1982. The estimation of animal abundance and related parameters, 2nd edition. New York: MacMillan.

Sekulic, R., 1982. Daily and seasonal patterns of roaring and spacing in four red howler Alouatta seniculus troops. Folia Primatologica, Volume 39, pp. 22-48.

Setz, E., 1993. Ecologia alimentar de um grupo de parauacus (Pithecia pithecia chrysocephala) em um fragmento florestal na Amazônia central (Ph.D. dissertation). São Paulo: Universidade Estadual de Campinas.

Setz, E., 1994. Feeding ecology of golden-faced sakis. Neotropical Primates, Volume 2, pp. 13-14.

Setz, E, Enzweiler, J., Solferini, V., Amêndola, M., Berton, R., 1999. Geophagy in the golden-faced saki monkey (Pithecia pithecia chrysocephala) in the central Amazon. Journal of the Zoological Society of London, Volume 247, pp. 91-103.

230

Setz, E. & Gaspar, D., 1997. Scent-marking behavior in free-ranging golden-faced saki monkeys, Pithecia pithecia chrysocephala in the central Amazon. Journal of Zoology, Volume 241, pp. 603-611.

Silva Jr., J., 2001. Especiação nos macacos-pregos e caiarara, gênero Cebus Erxleben 1777 (Primates, Cebidae). Ph.D. thesis. Rio de Janeiro: Universidade Federal do Rio de Janeiro.

Silva Jr., J., 2002. Sistemática dos macacos-pregos e caiararas, gênero Cebus Erxleben 1777 (Primates, Cebidae). In: Livro de resumos, X Congresso Brasileiro de Primatologia: Amazônia - A Última Fronteira. Belém: s.n., p. 35.

Silva Jr., J. & Figueiredo, W., 2002. Revisão sistemática dos cuxiús, gênero Chiropotes Lesson, 1840 (Primates Pithecidae). In: Livro de resumos do do X° congresso da Sociedade Brazileira de Primatologia, Amazônia - A última fronteira. Belém do Pará: Sociedade Brasileira de Primatologia, p. 21.

Silva, S. & Ferrari, S., 2007. Notes on the reproduction, behavior and diet of Saguinus niger (Primates: Callitrichidae) in a forest remnant at the National Primate Centre, Ananindeua, Pará. Biologia Geral Experimental, 7(1), pp. 19-28.

Simmen, B., 1992. Competitive utilization of bagasse fruits by sympatric howler and spider monkeys. Folia Primatologica, Volume 58, pp. 155-161.

Simmen, B., Julliot, C., Bayart, F. & Pages, E., 1998. Primate densities in a rain forest of French Guiana: Test of a transect method. Comptes Rendus de L'Academie des Science (Ser III), 321(8), pp. 699-704.

Simmen, B., Julliot, C., Bayart, F. & Pages-Feuillade, E., 2001. Diet and population densities of the primate community in relation to fruit supplies. In: F. Bongers, P. Charles-Dominique, P. Forget & M. Thery, eds. Noragues: Dynamics and plant- animal interactions in a Neotropical rainforest. Dordrecht: Kluwer Academic, pp. 89-101.

231

Simmen, B. & Sabatier, D., 1996. Diets of some French Guiana primates: Food composition and food choices. International Journal of Primatology, 17(5), pp. 661-693.

Skorupa, J., 1986. Responses of rainforest primates to selective logging in Kibale Forest, Uganda: A summary report. In: K. Benirschke, ed. Primates: The road to self- sustaining populations. New York: Springer Verlag, pp. 57-70.

Skorupa, J., 1988. Unpublished data cited in Whitesides et al. 1988, Unpublished.

Snowdon, C. & Soini, P., 1988. The tamarins, genus Saguinus. In: R. Mittermeier, A. Rylands, A. Coimbra-Filho & A. da Fonseca, eds. Ecology and behavior of Neotropical primates. Volume 2. Washington, D.C.: World Wildlife Fund, pp. 223- 298.

Somerville, A., 1980. Wind stability: Forest layout and silviculture. New Zealand Journal of Forestry Science, Volume 10, pp. 476-501.

Sorensen, T., 1999. Tropical dry forest regeneration and its influence on three species of Costa Rican monkeys. Ph.D. thesis. Alberta: University of Alberta.

Sorensen, T. & Fedigan, L., 2000. Distribution of three monkey species along a gradient of regenerating tropical dry forest. Biological Conservation, Volume 92, pp. 227- 240.

Sork, V., 1983. Distribution of pignut hickory (Carya glabra) along a forest to edge transect, and factors affecting seedling recruitment. Bulletin of the Torrey Botanical Club, 110(4), pp. 494-506.

Spironello, W., 1987. Range size of a group of Cebus apella in central Amazonia. American Journal of Primatology, Volume 8, p. 522.

232

Spironello, W., 1991. Importância dos frutos de palmeiras (Palmae) na dieta de um grupo de Cebus apella (Cebidae: Primates) na Amazônia central. Primatologia no Brasil, Volume 3, pp. 285-296.

Spironello, W., 2001. The brown capuchin monkey (Cebus apella): Ecology and home range requirements in central Amazonia. In: J. Bierregaard, C. Gascon, T. Lovejoy & R. Mesquita, eds. Lessons from Amazonia: The ecology and conservation of a fragmented forest. New Haven: Yale University Press, pp. 271-283.

Sprugel, D., 1991. Disturbance, equilibrium, and environmental variability: What is 'natural' vegetation in a changing environment? Biological Conservation, Volume 58, pp. 1-18.

Steffens, T., K, M. & Lehman, S., 2009. Avoiding the edge: Group distribution of Propithecus coquereli in Ankarafantsika National Park [Abstract]. American Journal of Primatology, 71(Suppl 1), p. 70.

Steininger, M., 2000. Secondary forest structure and biomass following short and extended land-use in central and southern Amazonia. Journal of Tropical Ecology, Volume 16, pp. 689-708.

Stouffer, P. & Bierregaard Jr., R., 1993. Spatial and temporal abundance patterns of Ruddy Quail-doves (Geotrygon montana) near Manaus, Brazil. Condor, Volume 95, pp. 896-903.

Strusaker, T., 1981. Census methods for estimating densities. In: Techniques for the study of primate population ecology. Washington D.C.: National Academy Press, pp. 36-80.

Symington, M., 1988. Black spider monkeys (Ateles paniscus chamek) in the Manu National Park, Peru. American Journal of Primatology, Volume 1, pp. 45-67.

Terborgh, J., 1983. Five New World primates: A study in comparative ecology. Princeton: Princeton University Press.

233

Terborgh, J., 1986. Conserving New World primates: Present problems and future solutions. In: J. Else & C. Lee, eds. Primate ecology and conservation. Cambridge: Cambridge University Press, pp. 355-366.

Terborgh, J., 1990. Mixed flocks and polyspecific associations: Costs and benefits of mixed groups to birds and monkeys. American Journal of Primatology, Volume 21, pp. 87-100.

The Plant List, 2010. The Plant List. Version 1. [Online] Available at: www.theplantlist.org [Accessed 10 11 2011].

Thorington, R. J., 1968. Observations of the tamarin Saguinus midas. Folia Primatologica, Volume 9, pp. 95-98.

Tollefson, J., 2012. President prunes forest reforms. Nature, 7 June, 486(7401), p. 13.

Tucker, N. & Murphy, T., 1997. The effects of ecological rehabilitation on vegetation recruitment: Some observations from the wet tropics of north Queensland. Forest Ecology and Management, 99(1-2), pp. 133-152.

Tweheyo, M., Lye, K. & Weladijc, R., 2004. Chimpanzee diet and habitat selection in the Budongo Forest Reserve, Uganda. Forest Ecology and Management, Issue 188, pp. 267-278.

UCLA Academic Technology Services Statistical Consulting Group, 2012. What is complete or quasi-complete separation in logistic/probit regression and how do we deal with them? [Online] Available at: http://www.ats.ucla.edu/stat/mult_pkg/faq/general/complete_separation_logit _models.htm [Accessed 10 July 2012].

234

Uhl, C., Buschbacher, R. & Serrano, E., 1988. Abandoned pastures in eastern Amazonia. I. Patterns of plant succession. Journal of Ecology, Volume 76, pp. 663-681.

Uhl, C., Clark, K., Clark, H. & Murphy, P., 1981. Early plant succession after cutting and burning in the upper Rio Negro region of the Amazon Basin. Journal of Ecology, Volume 69, pp. 631-649. van Roosmalen, M., 1980. Habitat preferences, diet, feeding strategy and social organization of the black spider monkey (Ateles paniscus paniscus Linnaeus 1758) in Surinam (Ph.D. dissertation). Wageningen: Agricultural University of Wageningen. van Roosmalen, M., 1985. Habitat preferences, diet, feeding strategy and social organization of the black spider monkey (Ateles paniscus paniscus Linnaeus 1758) in Surinam. Acta Amazonica, 15(3/4 suppl), pp. 1-238. van Roosmalen, M. & Klein, L., 1988. The spider monkeys, genus Ateles. In: R. Mittermeier, A. Rylands, A. Coimbra-Filho & A. da Fonseca, eds. Ecology and Behavior of Neotropical primates. Volume 2. Washington D.C.: World Wildlife Fund, pp. 455-537. van Roosmalen, M., Mittermeier, R. & Milton, K., 1981. The bearded sakis, genus Chiropotes. In: A. Coimbra-Filho & R. Mittermeier, eds. Ecology and behavior of Neotropical primates. Rio de Janeiro: Academia Brasileira de Ciências, pp. 419- 441.

Veiga, L. & Marsh, L., 2008. Pithecia pithecia. [Online] Available at: http://www.iucnredlist.org/details/41569/0 [Accessed 15 10 2012].

235

Veiga, L., Silva Jr., J., Mittermeier, R. & Boubli, J.-P., 2008. Chiropotes chiropotes. [Online] Available at: http://www.iucnredlist.org/details/43891/0 [Accessed 15 10 2012].

Veracini, C., 2000. Dados preliminares sobre a ecologia de Saguinus niger na Estação Científica Ferreira Penna, Caxiunana, Pará, Brasil. Neotropical Primates, 8(3), pp. 108-113.

Vié, J.-C., Richard-Hansen, C. & Fournier-Chambrillon, C., 2001. Abundance, use of space, and activity patterns of white-faced sakis (Pithecia pithecia) in French Guiana. American Journal of Primatology, Volume 55, pp. 203-221.

Vogel, S., 1988. Life's devices: The physical world of animals and plants. Princeton: Princeton University Press.

Vulinec, K., Lambert, J. & Mellow, D., 2006. Primate and dung beetle communities in secondary growth rain forest: Implications for conservation of seed dispersal systems. International Journal of Primatology, 27(3), pp. 855-879.

Wachter, B., Schabel, M. & Nöe, R., 1997. Diet overlap and poly-specific associations of red colobus and diana monkeys in the Tai National Park, Ivory Coast. Ethology, Volume 103, pp. 514-526.

Walker, S., 1993. Habitat use by Pithecia pithecia and Chiropotes satanas. American Journal of Physical Anthropology, Volume 16(suppl.), pp. 202-203.

Walker, S., 1996. The evolution of positional behavior in the saki-uakaris (Pithecia, Chiropotes, and Cacajao). In: M. Norconk, A. Rosenberger & P. Garber, eds. Adaptive radiations of Neotropical primates. New York & London: Plenum Press, pp. 335-367.

236

Wallace, R., Painter, L. & Taber, A., 1998. Primate diversity, habitat preferences, and population density estimates in Noel Kempff Mercado National Park, Santa Cruz Department, Bolivia. American Journal of Primatology, Volume 46, pp. 197-211.

Wandelli, E., 1991. Reposta ecofisiológica de palmeira de sub-bosque Astrocaryuym sociale Barb. Rodr. às mudanças ambientais resultantes do efeito de borda de floresta (Master's thesis). Manaus: INPA & Fundação Universidade do Amazonas.

Warner, M., 2002. Assessing habitat utilization by Neotropical primates: A new approach. Primates, 43(1), pp. 59-71.

Warton, D., 2005. Many zeros does not mean zero inflation: Comparing the goodness- of-fit of parametric models to multivariate abundance data. Environmetrics, Volume 16, pp. 275-289.

White, L., 1994. The effects of commercial mechanised selective logging on a transect in lowland rainforest in the Lopé Reserve, Gabon. Journal of Tropical Ecology, Volume 10, pp. 313-322.

Whitesides, G., Oates, J., Green, S. & Kluberdanz, R., 1988. Estimating primate densities from transects in a west African rain forest: A comparison of techniques. Journal of Animal Ecology, Volume 57, pp. 345-367.

Wiederholt, R., Fernandez-Duque, E., Diefenbach, D. & Rudran, R., 2010. Modeling the impacts of hunting on the population dynamics of red howler monkeys (Alouatta seniculus). Ecological Modeling, 221(20), pp. 2482-2490.

Wijdeven, S. & Kuzee, M., 2000. Seed availability as a limiting factor in forest recovery processes in Costa Rica. Restoration Ecology, Volume 8, pp. 414-424.

Williams-Linera, G., 1990a. Origin and early development of forest edge vegetation in Panama. Biotropica, 22(3), pp. 235-241.

237

Williams-Linera, G., 1990b. Vegetation structure and environmental conditions of forest edges in Panama. Journal of Ecology, Volume 78, pp. 356-373.

Williams-Linera, G., 1993. Bordes de bosque nublado en el Parque Ecológico Clavijero, Xalapa, Veracruz, México. Revista de Biología Tropical, Volume 41, pp. 107-117.

Wink, M., 1993a. Production and application of phytochemicals from an agricultural perspective. In: T. van Beek & H. Breteler, eds. Phytochemistry and agriculture. Proceedings of the Phytochemical Society of Europe Volume 34. Oxford: Claredon Press, pp. 171-223.

Wink, M., 1993b. Quinolizidine alkaloids. In: P. Waterman, ed. Methods in plant biochemistry. London: Academic Press, pp. 197-239.

Wissmann, M., Toutenburg, H. & Shalabh, 2007. Role of cetegorical variables in multicollinearity in the linear regression model, Munich: University of Munich.

Wolfheim, J., 1983. Primates of the world: Distribution, abundance and conservation. Seattle: University of Washington Press.

Wrangham, R., Gittleman, J. & Chapman, C., 1993. Constraints on group size in primates and carnivores: Population density and day range as assays of exploitation competition. Behavioral Ecology and Sociobiology, Volume 32, pp. 199-209.

Wright, B., 2004. Ecological distinctions in diet, food toughness, and masticatory anatomy in a community of six Neotropical primates in Guyana, South America (Ph.D. dissertation). Urbana-Champaign: University of Illinois.

Wunderle Jr, J., 1997. The role of animal seed dispersal in accelerating native forest regeneration on degraded tropical lands. Forestry Ecology and Management, 99(1-2), pp. 223-235.

Zhang, S., 1995a. Sleeping habits of brown capuchin monkeys (Cebus apella) in French Guiana. American Journal of Primatology, Volume 36, pp. 327-335.

238

Zhang, S., 1995b. Activity and ranging patterns in relation to fruit utilization by brown capuchins (Cebus apella) in French Guiana. International Journal of Primatology, Volume 16, pp. 489-507.

Zuur, A. et al., 2009. Mixed effects models and extensions in ecology with R. New York: Springer.

239

Biography

Bryan Bernard Lenz was born and raised in Milwaukee, Wisconsin. In 2001 he completed a B.A. in Biological Aspects of Conservation and a certificate in Environmental

Studies at the University of Wisconsin-Madison. These degrees were followed by an

M.Sc. in Primate Conservation from Oxford Brookes University in 2004, an M.A. in

Physical Anthropology from Tulane University in 2008, and this dissertation is the basis for a Ph.D. from Tulane University (2013). Bryan has conducted fieldwork focused on

Neotropical primates at the Lamanai Archaeological Reserve in Belize, the Cocha Cashu

Biological Station in the Peruvian Amazon, Santa Rosa National Park in Costa Rica, and the fieldwork component of this dissertation was carried out at the Biological Dynamics of Forest Fragments Project, which is located 80 km north of the city of Manaus in the central Brazilian Amazon.