This item is only available as the following downloads:
1 ROAD DEVELOPMENT, BUSMEAT EXTRACTION AND JAGUAR CONSERVATION IN YASUNI BIOSPHERE RESERVE ECUADOR By SANTIAGO RAFAEL ESPINOSA ANDRADE A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2012
2 2012 Santiago Rafael Espinosa Andrade
3 A mis dos flores, Nuria y Juliana, a mi madre, padre y hermanas
4 ACKNOWLEDGMENTS I would like to f irst acknowledge the communities of Guiyero, Timpoka, Ganketa, Dikaro, Oa, Apaika, Keweriono and Lorocachi. I truly appreciated their collaboration and participation that were essential to the fulfillment of this study. S pecial thanks go to Humbero Ahua, Daniel Alvarado, Matas Alvarado, Roque Alvarado, Boya Enkeri, Arturo Enomenga and Bolvar Enomenga. I consider all of them dear friends who provided me with wise advice in the field and were always ready to help me when I most needed it. I want to thank the institutions and programs that provided me with the financial support for my graduate studies and research: at the University of Florida, the Tropical Conservation and Development Program, the Gordon and Betty Moore Amazon Conservation Leadership Initi ative, and the Department of Wildlife Ecology and Conservation; World Wildlife Fund Russel E. Train Education for Nature Program; Wildlife Conservation Society Research Fellowship Program; and Panthera Jaguar Small Grants Program. I am extremely grateful to my academic advisor, Dr. Lyn Branch, for her support and guidance throughout my doctorate program. With her wisdom and experience, Lyn helped me to clarify my ideas and to achieve my goals. I would also like to thank my supervisory committee, Dr. Mike B inford, Dr. Brian Child, Dr. Marin Main and Dr. Melvin Sunquist. They provided me with thoughtful comments and suggestions that greatly improved the quality of this work. I feel honored to have had the guidance throughout my graduate program of such an out standing group of academics. I also would like to thank t he Pontificia Universidad Catlica del Ecuador, Universidad San Francisco de Quito, Fuerzas Armadas del Ecuador Batalln de Selva
5 48 Sangay, Wildlife Conservation Society Ecuador Program and Conserv ation in Action Tropic who provid ed me with important logistical support for my fieldwork. I am especially grateful to Esteban Surez, Andrew Noss and Rubn Cueva at WCS Ecuador Program for their willingness to collaborate with my research. I am also thank ful to Julia Salvador and Miguel Durango for their assistance in the field, which was crucial for the success of my study. I want to express my love and gratitude to Nuria and Juliana for their love, support and company, and for making my days happier. Add itionally, Nuria had to have a lot of patience and do extra work taking care of our little Juliana while I worked to finish this manuscript. Finally, I am grateful to my parents, Dolores and Alfonso, for their support and for providing me with the values and principles that have guided me throughout my life.
6 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 9 LIST OF FIGURES ................................ ................................ ................................ ........ 10 ABSTRACT ................................ ................................ ................................ ................... 12 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 14 2 ROAD DEVELOPMENT WITHIN HISTORICALLY INHABITED NEOTROPICAL PROTECTED AREAS: IMPLICATIONS FOR GAME MANAGEMENT AND CONSERVATION ................................ ................................ ................................ ... 21 General Overview ................................ ................................ ................................ ... 21 Methods ................................ ................................ ................................ .................. 23 Study Area ................................ ................................ ................................ ........ 23 Research Design ................................ ................................ .............................. 25 Data Collection ................................ ................................ ................................ 27 Analyses ................................ ................................ ................................ ........... 28 Roads change the spatial extent of hunting (H1) ................................ ....... 28 Rates of bushmeat extraction and trade at varying distance from market (H2) ................................ ................................ ................................ ......... 30 Differential use of game at varying distance from market (H3) .................. 31 Results ................................ ................................ ................................ .................... 31 Roads and the Spatial Extent of Wildlife Harvest (H1) ................................ ..... 32 Extraction and Trading R ates of Game (H2) ................................ .................... 33 Differential Use of Game Species (H3) ................................ ............................ 33 Discussion ................................ ................................ ................................ .............. 34 The Spatial Extent of Hunting ................................ ................................ ........... 35 Extraction and Trading Rates ................................ ................................ ........... 36 Differential Use of Game ................................ ................................ .................. 38 3 EFFECTS OF INCREASED ACCESIBILITY TO HUNTERS ON THE SPATIAL DISTRIBUTION OF GAME WITHIN A LARGE NEOTROPICAL PROTECTED AREA ................................ ................................ ................................ ...................... 49 General Overview ................................ ................................ ................................ ... 49 Methods ................................ ................................ ................................ .................. 52 Study Area ................................ ................................ ................................ ........ 52 Study Design ................................ ................................ ................................ .... 53 Analyzing the Effects of Access and Roads on Game Occurrence .................. 56
7 Results ................................ ................................ ................................ .................... 60 Landscape Access to Hunters and Game Occurrence ................................ ..... 60 Road Vicinity and Game Occurrence ................................ ............................... 62 Discussion ................................ ................................ ................................ .............. 63 Settlement Distribution, Landscape Access to Hunters and Game Occurrence ................................ ................................ ................................ .... 63 Road Placement and Game Occurrence ................................ .......................... 66 4 INCREAS ED ACCESS TO NEOTROPICAL PROTECTED AREAS, BUSHMEAT EXTRACTION AND ITS IMPLICATIONS FOR JAGUAR CONSERVATION ................................ ................................ ................................ ... 80 General Overview ................................ ................................ ................................ ... 80 Methods ................................ ................................ ................................ .................. 83 Study Area ................................ ................................ ................................ ........ 83 Study Design ................................ ................................ ................................ .... 84 Camera Trapping ................................ ................................ ............................. 87 Analyses of Prey Availability ................................ ................................ ............. 88 Estimating Jaguar Density ................................ ................................ ................ 90 Results ................................ ................................ ................................ .................... 94 Prey Abundance ................................ ................................ ............................... 94 Jaguar Abundance in Yasun ................................ ................................ ........... 95 Discussion ................................ ................................ ................................ .............. 97 Availability of Prey and Jaguar Abundance ................................ ...................... 97 Higher Accessibility and Jaguar Conservation in Yasun Biosphere Reserve 100 5 CONCLUSION ................................ ................................ ................................ ...... 112 General Conclusions ................................ ................................ ............................. 112 Conservation Recommendations ................................ ................................ .......... 114 APPENDIX A LIST OF SPECIES HUNTED ALONG THE MAXUS ROAD ................................ 117 B SPSS OUTPUT OF ROC CURVE ANALYSES ................................ .................... 119 C TOTAL DETECTIONS OF SPECIES IN YASUNI BIOSPHERE RESERVE ......... 120 D OCCUPANCY MODELS TO EXPLORE GAME OCCURRENCE AS A FUNTION OF LANDSCAPE ACCESS BY HUNTERS IN YASUNI BIOSPH ERE RESERVE ................................ ................................ ................................ ............. 121 E ANALYSES OF SPATIAL DEPENDENCE ................................ ........................... 128 F OCCUPANCY MODELS TO EXPLORE GAME OCCURRENCE AS A FUNTION OF DISTANCE FROM M AXUS ROAD ................................ ................ 132
8 G IDENTIFICATION OF JAGUAR INDIVIDUALS FROM THEIR ROSETTE PATTERNS ................................ ................................ ................................ ........... 136 H DISTRIBUTION OF DAILY BIOMASS OF PREY PER C AMERA TRAP STATION AT FOUR SITES OF YASUN BIOSPHERE RESERVE ...................... 137 LIST OF REFERENCES ................................ ................................ ............................. 138 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 153
9 LIST OF TABLES Table page 2 1 Most important species harvested as bushmeat during this study in Yasun Biosphere Reserve by 5 Waorani settlements along the Maxus road and traded in the market of Pompeya. ................................ ................................ ....... 40 2 2 Model selection of candidate models to describe the probability of hunting in an area as a function of accessibility.. ................................ ................................ 41 2 3 Untransformed parameter estimates of best fit model to estimate the e ffects of accessibility on landscape used by hunters.. ................................ .................. 41 3 1 Best occupancy models and untransformed model parameters to predict game probability of site occupancy ( ) as a function of accessibility across four study sites.. ................................ ................................ ................................ 71 3 2 I mportance of covariates in predicting for occupancy models that included four areas with var ying degree of accessibility an d for occupancy models along the Maxus Road .. ................................ ................................ ...................... 72 3 3 Average predicted probability of site occupancy of best fit models and nave occupancy estimates for individual species in four study sites. .......................... 73 3 4 Best occupancy models and untransformed model parameters to predict game occurrence by the Maxus Road. ................................ ............................... 74 4 1 Survey effor t with camera trap stations at four study sites in Yasun Biosphere Reserve. ................................ ................................ .......................... 103 4 3 Multiple comparison test follow ing a Kruskal Wallis of prey biomass at four sites in Yasun Biosphere Reserv e. ................................ ................................ .. 105 4 4 Capture history of 30 jaguars used in abundance estimation in the four areas in Yasun Biosphere Reserve ................................ ................................ ........... 106 4 5 Estima tes of jaguar density using non spatial models and spatially explicit models. ................................ ................................ ................................ ............. 107 A 1 Complete list of species hunted by 5 Waorani settlements between January 2008 and April 2009 along the Maxu s Road in Yasun Biosphere Reserve. .... 117 C 1 Total number of detections in 10 day survey occasion periods for species in four study areas in Yasun Biosphere Reserve. ................................ ................ 120 C 2 Total number of detections in 4 day survey occasion periods in cameras placed every 0.5 km in 13 5 km transects along the Maxus R oad. ................... 120
10 LIST OF FIGURES Figure page 1 1 Current known range of jaguar and actual extent of protected areas ................. 19 1 2 Yasun Biosphere Reserve and oil concession blocks in E region ................................ ................................ ................................ ................. 20 2 1 Study area and surveyed settlements along the Maxus R oad in Yasun Biosphere Reserve ................................ ................................ ............................. 42 2 2 Kill sit es of game between J anuary 2008 April 2009 by 5 Waorani settlements along the Maxus Road in Yasun Biosphere Reserve.. ................... 43 2 3 Hunting area based on extraction rate by 5 Waorani settlements alon g the Maxus R oad in Yasun Biosphere Reserve ................................ ........................ 44 2 4 Probability of hunting b ased on landscape accessibility for Waorani from 5 settlements along the Maxus Road in Yasun Biosphere Reserve.. ................... 45 2 5 Contribution of the 14 most hunted species to the total kills at settlements close and far from markets ................................ ................................ ................ 46 2 6 Waorani are ha rassed by middlemen who purchase bushmeat ....................... 47 3 1 Study sites and camera trap arrays to measure the effects of accessibility on game occurre nce in Yasun Biosphere Reserve ................................ ................. 75 3 2 Camera placement to analyze the effect of road on game distribution acro ss ................................ ................................ .......... 76 3 3 Accumulation curves of number of locations where a species is detected as survey effort progresses at four study si tes in Yasun Biosphere Reserve ......... 77 3 4 Probability of site occup ancy of white lipped peccary and collared pecc ary within Yasun Biosphere Reserve based on best site occupancy models ......... 79 4 1 Camera trap arrays to measure jaguar density in four si tes in Yasun Biosphere Reserve ................................ ................................ ........................... 108 4 2 Probability of site occupancy of ungulates and three medium sized mammals reported to be important jaguar prey at four si tes in Yasun Biosphere Reserve ................................ ................................ ................................ ............ 109 4 3 Density estimates with ML SECR using varying buffer distances of camera trap envelope. ................................ ................................ ................................ ... 110 4 4 nd two spatially explicit models ......... 110
11 4 5 Markov chains of total individual jaguars, Nsuper, estimated within the state space by Bayesian SECR models. ................................ ................................ ... 111 E 1 Correlograms for residuals of from models measuring the effect of landscape accessibility on the occurrence of 7 large bodied species in Yasun Biosphere Reserve ................................ ................................ ............... 129 E 2 Correlograms for residuals of from models measuring the effect of landscape accessibility on the occurrence of 5 medium sized bodied species in Yasun Biosphere Reserve ................................ ................................ ........... 130 E 3 Correlograms for residuals of from models measuring the effect of Maxus Road on the occurr ence of nine banded armadillo, agouti, paca, red brocket deer, tapir and collared peccary ................................ ................................ ....... 131 G 1 Identification of jaguar individuals from their rosette patterns.. ......................... 136 H 1 Distribution of daily biomass of prey per camera trap station at four sites of Yasun Bi osphere Reserve.. ................................ ................................ ............. 137
12 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy ROAD DEVELOPME NT, BUSMEAT EXTRACTION AND JAGUAR CONSERVATION IN YASUNI BIOSPHERE RESERVE ECUADOR By Santiago Rafael Espinosa Andrade August 2012 Chair: Lyn Branch Major: Wildlife Ecology and Conservation Large protected areas are essential for conservation of wild life populations, particularly species with large spatial requirements such as large carnivores. In the Neotropics, these areas face numerous threats including road building associated with extraction of natural resources. Additionally, many Neotropical pr otected areas are inhabited by indigenous people strongly dependent on bushmeat for subsistence. Roads provide greater access to landscapes for hunters and are associated with trade of wildlife in markets. I examine how roads can lead to cascading effects that initiate with changes in hunting practices by former subsistence hunters and lead to reduction of game and top predator populations. Understanding these cascading effects that involve social and ecological change is critical for effective conservation and management of wildlife populations within protected areas threatened with infrastructure development. My study was based in Yasun Biosphere Reserve ( ca. 18,000 km 2 ), the protec ted area with highest potential for jaguar conservation in Ecuador Yasun is inhabited by the Waorani a traditionally semi nomadic indigenous group strongly dependent on bushmeat. Two roads were constructed in Yasun to extract oil in the early 1980s and
13 early 1990s. Waorani who have settled along margins of these roads nowad ays use markets to trade bushmeat. I monitored hunting practices of Waorani along one road and measured prey occurrence and jaguar abundance in four areas with different degrees of accessibility across Yasun. Major findings included: 1) Roads doubled the hunting area used by Waorani; 2) Waorani located closer to markets extracted and traded a higher amount of bushmeat than those located farther from them; 3) Waorani living closer to markets hunted a significantly higher proportion of large bodied species (i.e., species with high market value), principally ungulates, than those farther from markets; 4) Higher landscape accessibility to hunters was linked with lower occurrence of game, particularly ungulates, the primary game for hunters and prey for jaguar; and 5) jaguar density was lower in the most accessible areas where bushmeat extraction was higher and game occurrence and biomass was lower. These results indicate that further road development in Yasun, and other protected areas with similar conditions, will have detrimental effects for game and jaguar populations.
14 CHAPTER 1 INTRODUCTION Historically, the jaguar ( Panthera onca ) has coexisted with indigenous people in the Neotropics, sharing the same space and resources (Figure 1 1) Persistence of th is large carnivore and prey species in lands occupied by indigenous people can be attributed primarily to low human population densities, scattered distribution of settlements in the landscape, primitive hunting technologies, and subsistence economies (Alv ard et al. 1997; McMichael et al. 2012; Robinson & Bennett 2004) However, these factors are changing rapidly as roads are built throughout the Amazon Basin and indigenous groups become more integrated into market based economies, increasing pressure on wi ldlife as it becomes a source of cash (Brandon 1996; Sanderson et al. 2002b; Schwartzman & Zimmerman 2005; Sierra et al. 1999; Wilkie et al. 2000) Many studies have shown detrimental effects of hunting by local people on wildlife populations across the Ne otropics (e.g., Bodmer et al. 1997; Hill et al. 2003; Hill & Padwe 2000; Mena Valenzuela et al. 2000; Peres 2000a) Species that constitute main prey for jaguar also are harvested intensively by hunters in the Neotropics (Jorgenson & Redford 1993) The neg ecology by reducing prey populations has been suggested and reported by several authors (e.g., Crawshaw & Quigley 2002; Novack et al. 2005; Polisar et al. 2003) The Amazon basin constitutes the main stronghold for jaguar conservation (Sanderson et al. 2002b) (Figure 1 1) The unique large protected areas or mega reserves (areas > 10,000 km 2 ) in Amazonia increase the probability of successful conservation of species with large spatial requirements, such as jaguar s (Laurance 2005) (Figure 1 1) However, p rotected areas and indigenous reserves in Amazonia are
15 threatened by the building of roads to access valuable resources such as soils for agriculture, timber, hydropower, oil, gas and minerals (Fearnside 2001; Fear nside 2006; Finer et al. 2008; Laurance et al. 2001) Roads are major drivers of disturbances in both ecological and social systems within and outside protected areas in the Amazon (Laurance et al. 2004) Some direct ecological impacts caused by road devel opment are habitat loss and fragmentation, increased fires, road killed animals and barriers affecting animal dispersal, among others (Forman & Alexander 1998; Forman et al. 2003) However, anthropogenic impacts originating from road development can be eve n more important in affecting the integrity of ecosystems, as these impacts are likely to increase over time after infrastructure development has occurred (Zeng et al. 2005) Road density is associated with market accessibility, economic growth, and increa ses in natural resource exploitation, for example, by providing access to areas that are otherwise inaccessible to hunters (Peres & Lake 2003; Wilkie et al. 2000) Because most protected areas in the Neotropics are inhabited by indigenous people (Dugelby & Libby 1998) and in the future these areas may become the only safe refuges for many game animals and large carnivores, key questions are: What are the consequences of changes in indigenous livelihood systems (e.g., shift from subsistence to commercial hu nting) for wildlife conservation? How do the cascading effects of road development affect social and ecological change in these inhabited parks? Our understanding of how humans and large carnivores interact on a landscape level through sharing of common s paces and food resources, and how these interactions are affected by changes in traditional livelihood systems, is limited. However, this understanding is critical for conservation of large carnivores, such as jaguars, as
16 protected areas may become the las t refuges where these animals can exist. Additionally, empirical studies are lacking that address these issues from an interdisciplinary perspective, through linking social and ecological components. This type of analysis is necessary for understanding ch ange in complex systems, such as in systems comprised of indigenous people, prey species and top predators (Forester & Machlis 1996; Gunderson & Holling 2002) Amazon region (Fi gure 1 2) The Biosphere is formed by Yasun National Park (10,000 km 2 ) and the Waorani Ethnic Reserve (8,000 km 2 ). Yasun is the largest protected area in Ecuador, one of the most biodiverse ecosystems in the world, and the most important refuge for jagua r conservation in the country (Bass et al. 2010; Espinosa et al. In press) Yasun has been historically inhabited by the Waorani indigenous group, a unique culture strongly associated with hunting (Yost & Kelley 1983) Despite its importance as a wildlife refuge, the integrity of Yasun is threatened by future road development associated with oil exploration (Finer et al. 2008) (Figure 1 2) Previous studies in Yasun suggest that market integration by Waorani could lead to higher rates of bushmeat extrac tion, and that the degree of integration with markets is a function of distance to roads (Sierra et al. 1999) Harvest of some game species by Waorani in Yasun may be above sustainable thresholds (Franzen 2005; Mena Valenzuela et al. 2000) This is the ca se for white lipped ( Tayassu pecari ) and collared peccaries ( Pecari tajacu (Aranda 2002; Oliveira 2002) If road density continues increasing within Yasun, I predict hunting activities likely will alter the spatial distribution of both prey and jaguars
17 in the system. My study addresses how roads may be modifying hunting practices of the Waorani and how increased access to the landscape by hunters may affect the distribution of game and jaguar w ithin Yasun. I use spatially explicit analyses to better understand and visualize the spatial distribution and consequences of hunting. In Chapter 2, I examine effects of increased accessibility to hunting areas and markets due to road development on hun ting patterns of the Waorani. Spatial aspects of hunting practices and responses of hunting patterns to infrastructure development within protected areas and indigenous territories are poorly understood. I evaluate the impact of road development on the spa tial extent of hunting by: 1) comparing spatial extent of a projected hunting area in the absence of roads with the observed spatial extent used by the Waorani in areas with roads, and 2) assessing probability of hunting as a function of sources of access. I examine how increased access to markets affects the amount of bushmeat extracted and traded by the Waorani. Finally, I examine how access to markets influences the intensity of harvest of different species used by the Waorani. In Chapter 3, I explore changes in game occurrence with different levels of landscape access to hunters. Empirical evidence shows reductions of game populations near human settlements (Hill et al. 1997; Mena Valenzuela et al. 2000) However, the interaction of landscape features that provide access to hunters (i.e., roads and navigable rivers) and settlement distribution likely alter the spatial distribution of game in inhabited protected areas, and this is less understood. I use camera traps and site occupancy models to evaluate the occurrence of game as a function of roads, rivers and settlement distribution (MacKenzie et al. 2006) My first analysis evaluates game
18 occurrence as a function of distance to nearest means of access (road or navigable river) and distance to nearest se ttlement in four areas with different degrees of accessibility by rivers and roads. As roads are major man made landscape features that provide access, my second analysis focuses in detail on occurrence of game along the margin of the road where harvest as sessments were made ( Chapter 2). In Chapter 4, I examine how increased landscape access to hunters relates to changes in prey occurrence and biomass and how prey availability influences jaguar abundance. The effects of bushmeat extraction on jaguar abundan ce need to be better understood, particularly within the context of large, inhabited protected areas. I use three different methodologies to estimate density of jaguars and compare the consistency of density estimates at four sites in Yasun. I finish my d issertation with C hapter 5 where I present general conclusions and recommendations for management and conservation of game and jaguar populations within Yasun and similar inhabited protected areas in the Neotropics.
19 Figure 1 1. Current known range o f jaguar and actual extent of protected areas. Jaguar range from Sanderson et al. (2002b) Map of protected areas from World Database on Protected Areas (IUCN & UNEP WCMC 2010)
20 Figure 1 2. Yasun Biosphere Reserve and oil concession blocks in Ecuador Amazon region. Polygons of oil concession blocks from Finer et al. 2008.
21 CHAPTER 2 ROAD DEVELOPMENT WITHIN HISTORICALLY INHABITED NEOTROPICAL PROTECTED AREAS: IMPLICATIONS FOR GAME MANAGEMENT AND CONSERVATION General Overview Large protected areas a nd, arguably, indigenous territories, are important refuges for future wildlife conservation in the Neotropics (Laurance 2005; Schwartzman & Zimmerman 2005) Although Neotropical rain forests are well represented within the s, the effectiveness of these protected areas is questionable (Brooks et al. 2004; Peres & Terborgh 1995) Protected areas are threatened by a variety of human activities, including road building to access valuable resources such as soils for agriculture, timber, hydropower, oil, gas and minerals (Fearnside 2001; Fearnside 2006; Finer et al. 2008; Laurance et al. 2001) Additionally, Neotropical regions encompass countries with emergent economies that rely on these resources to develop. As human population s, and thus resource demands continue to grow, road development is predicted to continue within natural lands throughout the Neotropics in the coming decades. For example plans have been developed for increasing roads within Ecuadorian protected areas and the Amazon basin in Per u to gain access to oil and gas reserves (Bass et al. 2010; Finer & Orta Martinez 2010) Increased accessibility to natural lands through road development is associated with negative impacts on wildlife communities (Fahrig & Rytwins ki 2009; Peres & Lake 2003) R oads initiate cascading effects by starting colonization processes that lead to habitat loss, fragmentation and degradation (Laurance et al. 2004) Roads cause direct impacts on wildlife distributions, for example, by introduc ing mortality through vehicle collisions, limiting animal dispersal, or disturbing animal behavior (Benitez Lopez et al.
22 2010; Forman & Alexander 1998; Forman et al. 2003; Trombulak & Frissell 2000) Roads provide access not only to previously unhunted are as but also to markets that increase extraction when game is used as a commodity (Sierra et al. 1999; Wilkie et al. 2000; Wilkie et al. 1992) Additionally, market participation fosters exploitation of species that provide high economic return such as larg e bodied species (Macdonald et al. 2011; Robinson & Bennett 2000a) A high proportion of the protected areas in the Neotropics are inhabited by indigenous peoples (hereafter traditional hunters) who are strongly dependent on wildlife as a source of meat and cash (Brandon et al. 1998; Robinson & Bennett 2004) Inhabited protected areas and indigenous lands often function as source sink systems, where areas accessible to hunters function as sinks (i.e., mortality rates of game species exceed birth rates) an d inaccessible areas function as sources or refuges (i.e., births outnumber deaths) (Novaro et al. 2005) To avoid the collapse of game populations within a source sink system, sufficient land inaccessible to hunters must be maintained to sustain populatio ns large enough to compensate for high mortality in hunted areas (Joshi & Gadgil 1991; McCullough 1996; Novaro et al. 2000) In face of further road development within historically inhabited protected areas, I ask: How do roads affect the spatial extent o f harvested areas by traditional hunters? What are the consequences of increased access to markets on hunting practices by traditional hunters? For example, are hunters living closer to markets harvesting more and concentrating on more profitable species t han those who live farther from them? To address these questions, I examined hunting practices of the Waorani indigenous group in Yasun Biosphere Reserve, located in the Amazon region of Ecuador. Yasun is one
23 of the most biodiverse areas on earth and thr eatened with further road development to exploit oil reserves (Bass et al. 2010; Finer et al. 2009) Methods Study Area Yasun Biosphere Reserve (hereafter Yasun) covers an area of 18,000 km 2 and is formed by Yasun National Park and the adjacent Waorani Ethnic Reserve (hereafter Waorani territory) (Figure 2 1). Seasons in Yasun are not clearly marked A nnual rainfall is close to 3,000 mm and no calen dar month has precipitation under 100 mm M ean monthly temperatures are within 22 34C (Valencia 2004). V e getation cover is dominated by tall evergree n terra firme tropical forest with canopy height between 25 40 m Flood plains and swamps occur along the margins of the main rivers (Valencia et al. 2004) Historically, Yasun has been occu pied by the Waorani indigenous group, which had a semi nomadic lifestyle and an economic system based on gathering and sharing forest products and horticulture (Rival 1996) Bushmeat is the most important source of protein and hunting is an essential part of Waorani culture, especially for men, who consider themselves warriors and hunters. Traditional hunting tools used by the Waorani were spears to kill large prey, such as peccaries and tapir, and blowguns for arboreal species, such as monkeys or birds (Yost & Kelley 1983) Contact between the Waorani and western world can be traced back to the early 1900s, although these first encounters were characterized by aggression from the Waorani in response to intrusion by outsiders (Cabodevilla 1999) In the late 1960s, evangelical missionaries achieved peaceful contact with the Waorani. Subsequently, missionaries and the Ecuadorian government promoted the relocation of Waorani to permanent settlements within a
24 limited area (1,600 km 2 vs. 20,000 km 2 dominated by Waorani prior to cont act), the (Yost & Kelley 1983) In the early 1980s, the first road (Auca Road) was opened into the Waorani territory by Texaco Oil Company, and in the early 1990s, a 120 km road was created by the Maxus Oil Company within Yasun (Finer et al 2009) Since Waorani have been in contact with the western Ecuadorian society, they have mostly replaced blowguns and spears with the more effective shotguns (Yost & Kelley 1983) Most Waorani now live in permanent settlements along rivers or the two roa ds created within Yasun. My study is located in the area of the Maxus Road, which is inhabited by Kichwa and Waorani indigenous groups (Figure 2 1). Settlements of the Kichwa ethnicity are located in the first 30 kilometers of the road and Waorani settle ments are located from km 32 to the end of the road, which is another 90 km. My study area is limited to the hunting territory of the Waorani, which does not overlap with the Kichwa territory. Approximately 320 W aorani distributed in 11 settlements, live Access to the Maxus Road is strictly controlled by the oil company operating in the area and every person who does not belong to one of the indigenous groups residing in the area must obtain a permit to access this region. The Max us Road is separated from the road network by the Napo River (Figure 2 1), and the only means to access the road by vehicle is using the oil company ferry, which also requires a special permit. Free transportation is provided by the oil company for the Wao rani who live along the Maxus Road and travel on a daily basis along the road. The Waorani use this transportation service to access hunting areas along the road. This service includes a bus ride that takes Waorani every Saturday to the closest market, whi ch is in Pompeya (Figure 2 1),
25 a small town outside Yasun, where they sell bushmeat obtained for that purpose during weekdays. I surveyed hunting activities in t hree settlements located closer to markets (Guiyero, 32 km from Pompeya; Ganketa, 38 km; and Timpoka, 51 km) and two settlements fa rt her from markets (Dikaro, 100 km; and Oa, 120 km) (Figure 2 1). I nhabitants of the three settlements close to markets need 1.5 to 2 hou rs to get to Pompeya (Figure 2 1). Guiyero Ganketa and Timpoka ha ve population s of 35 6 a nd 22 inhabitants organized in 6,1 and 3 households, respectively. Inhabitants from Dikaro and Oa need 3.5 to 4.5 hours to get to Pompeya. Dikaro is the largest settlement along the Maxus road, with a population of 180 organized in 35 families. Oa is a settlement composed of a single household of 6 people. Research Design To evaluate the effect of roads on the spatial extent of hunting and the effect of increased access to markets on hunting practices by the Waorani, I developed three working h ypothesis: 1. Roads change the spatial extent of wildlife exploitation by traditional subsistence hunters I expect ed to observe that roads would both significantly increase the proportion of landscape used and that the probability of hunting would be higher closer to roads than along other sources of access such as rivers. 2. Market access influences harvest and commercialization rates of bushmeat. I expect ed to observe that hunters closer to markets would harvest more and sell more meat than those fa rther from them 3. Market access causes differential harvest of available game species. I expect ed to observe that hunters who live closer to markets harvest a higher proportion of species preferred by non indigenous peoples in markets (e.g., large bodied species suc h as ungulates ) rather than species that are preferred by the Waorani (e.g. monkeys).
26 Hunting can be influenced by other factors than accessibility and vicinity to markets. Variation in hunting patterns can be induced by biological (e.g., species distribu tion and abundance ) and cultural (e.g hunting techniques, taboos, habitat modification, and hunting regulations) characteristics of the socio ecological system (Redford & Robinson 1987) Additionally, in the context of market use, alternative sources of cash for local people could also modify hunting patterns and the amount of wildlife extracted (Godoy et al. 1995; Godoy et al. 2009) These factors could confound my observations if they were distributed in space in the same way settlements are with respec t to markets. My research site permit s control of these sources of variation. First, all studied settlements and catchment areas are within the same ecosystem, with similar landscape composition and configuration T herefore, biological factors should contr ibute minimally to differences in hunting practices of the Waorani Second, only one ethnic group, the Waorani, occupies my study area, which minimizes variation in hunting practices due to cultural factors The Waorani in the five settlements share taboos and have equal hunting technologies H unting regulations do not exist. Third, variation in hunting patterns because of alternative sources of income is minimal. The only continuous source of jobs for the Waorani along the Maxus road is the oil company, w hich has agreements with the Waorani communities. Every Waorani who needs a job can get one at the oil company, which pays the same daily rate to all community members. Finally, another source of variability in hunting patterns could be availability of tra nsportation among households. However, th e oil company provides a free bus service that moves Waorani along the entire Maxus Road o n a daily basis, including a special trip on weekends to visit markets.
27 Data Collection Between January 2008 and April 2009 I obtained data on hunting activities of Waorani hunters through self reporting. Prior to data collection, I conducted a workshop in each community to discuss study objectives and methods, as well as benefits of the information for the Waorani I recruit ed 15 Waorani assistants who were in charge of reporting hunting activities on a daily basis within their households or households of related kin Each assistant was trained and co mpensated with US$ 30 per month. S ettlements were visited every 20 days to d ouble check completed questionnaires. As hunting is not an illegal activity for the Waorani, participants were open to report ing their hunting activities. Survey effort varied across settlements. The entire populations of Guiyero and Timpoka were monitored for 14 months. The single household of Ganketa was not present in the area for a great part of the study and was surveyed for 6 months. In Dikaro, I surveyed 22 out of 35 households for 12 months (144 people) The single household in Oa was surveyed for 11 months Data for each animal killed were collected on a separate data sheet that included species name, age class, weight, sex and reproductive stage Date and time of start and end of each hunting trip was reported to calculate harvest effort, along wi names and weapon used in each kill. The use of each carcass was also recorded F or example, a family commonly consumed the head and internal organs of peccaries and sold the ribs, front and hind limbs. Kill sites for each animal were marked on a map (scale 1:200,000) printed o n the back of each data sheet and habitat type was recorded. The map included landmarks for spatial reference such as kilometer mark ers on the Maxus Road settlements and rivers. By accompanying the Waorani in hunting trips or wildlife surveys, I observed they had an extremely good sense of spatial location. Their
28 accuracy on marking locations of kills in maps generally varied from 50 300 m when compared with locations obtained with a handheld GPS unit (Garmin GPSMAP 76Cx). Locations recorded on maps were digitized into a GIS database for analysis. Hunters reported parts that were sold from each animal hunted including front and hind limbs, ribs and heads. Animals sold in parts were mainly ungulates, whereas smaller game suc h as medium sized rodents, monkeys or birds commonly would be bushmeat traded, I assigned to each ungulate part a value representing its percentage weight To estimate these percentages, I used as guides the body shape of these species and carcass composition and dressing yields of domestic pig, cattle and venison ( Odoco i l e us virginianus ) presented in meat literature (Kauffman 2001; Warriss 2000) For all five species, the assigned proportions were 8% for head, 14% for both front limbs, 24% for both hind limbs and 10% for both rib flanks. I left 15% for internal organs and 19% for other parts including the axial skeleton, abdomen and stomach contents. When smaller mammals were sold in parts, I used these same proportions. Birds, except on two occasions, were always sold as a whole or half. Analyses Roads c hange the spatial e xtent of h unting (H1) To evaluate the effect of roads on the spatial extent of h unting by the Waorani, I used two approaches. First, I compared the spatial extent of projected harvest area (km 2 ) in the absence of the road with the spatial extent of the observed hunting area. Second, I evaluated the probability of hunting across the la ndscape as a function of sources of access (i.e., distance from roads, navigable rivers and settlements) and defined a harvest area based on hunting probability as a function of access by hunters.
29 For my first approach I projected the accessible area pri or to road building as the area within an 8 km radius from surveyed settlements. The maximum Euclidean distance walked from a point of access by Waorani hunters in the Maxus Road was 7 km, although only 37 out of 2,997 hunting records were at a distance gr eater than 5 km. The 8 km distance is similar to the farthest distance hunters have been recorded to walk from a point of access in other areas of Amazonia (e.g., Peres & Lake 2003) To estimate the current area harvested by Waorani, I used two methods, mi nimum convex polygon (MCP) and kernel density estimation (Silverman 1986) I used UTM locations of kills to estimate MCPs for the 5 settlements, that were subsequently merged to have an overall harvested area. To estimate the harvest area with kernel analy sis, I used locations of kills with their associated body mass to obtain an output of cells representing the amount of bushmeat extracted per unit of area (kg of bushmeat/km 2 /year). I performed kernel analysis using a spatial resolution of 250 m and a sear ch radius or smoothing factor of 8 km. MCP and Kernel analyses were conducted with software ESRI ArcGIS v10. For my second approach I used logistic regression to assess the probability of hunting as a function of accessibility For this model I used obser ved kill sites (n = 2,997) and random points (n = 3 000) represent ing hypothetical un hunted sites as the response variable Random points were placed within the Waorani hunting territory, which is located south of the Tiputini River (the area north of the Tiputini River is the hunting territory of Kichwa and was not part of this study) and within an accessible area limited by: a) an 8 km buffer from roads or navigable rivers used by inhabitants of the 5 settlements, and b) a maximum distance of 4 0 km from s ettlements which is the
30 maximum Euclidean distance Waorani traveled from their settlements to the farthest points of access along the Maxus Road and rivers, using either the bus provided by the oil company or dugout canoes. Each observed kill site and ran dom point were associated with three predictor variables representing accessibility: a) distance from road, b) distance from river, and c) distance from settlement where the hunter lived or nearest settlement for random points I developed seven models tha t included all possible combinations of predictor variables and used AIC to select the best fit model (Burnham & Anderson 2004) I used a threshold independent receiver operating characteristic (ROC) curve to assess the predictive performance of the best m odel (Fielding & Bell 1997) A ROC curve plots the fraction of true positives values (sensitivity) against the fraction false positive values (1 specificity). The area under the ROC curve (AUC) is used as a measure of accuracy that is not dependent on a single threshold. The AUC varies from 0.5 (no discrimination) to 1.0 (perfect discrimination) (Fielding & Bell 1997) Finally, I used the best model parameter estimates to plot the probability of hunting across the landscape using the raster calculator too l in ArcGIS v10. For this purpose, I created three raster layers where the centers of pixels (spatial resolution of 50 m) represent distance to the road, navigable rivers used by hunters and nearest settlements. I estimated a harvest area based on the prob ability of hunting by grouping all the pixels with a probability above 20%, which could be interpreted as sites hunted once every 5 days. I considered a 20% probability of hunting to be adequate to define an area as hunted for conservation purposes. Rates of bushmeat extraction and trade at varying distance from market (H2) I compare d per capita daily extraction (kg of bushmeat/person/day) and trading rates (kg of bushmeat sold/person/day) between settlements near and far from markets.
31 To obtain estimates o f daily per capita extraction and trade, I grouped settlements from Guiyero, Ganketa and Timpoka (32 51 km from market) as those close to markets, and settlements from Dikaro and Oa (100 120 km from market) as those far from markets. I estimated monthly p er capita daily extraction and trading rates for each surveyed month close and far from markets. I used a paired t test to compare means of daily per capita extraction and trading rates for 12 consecutive months between settlements close and far from marke ts. Differential use of game at varying distance from market (H3) I evaluated differences in the composition of taxa harvested by hunters as a function of distance from market. For this purpose, I group ed species by their market value. I considered a speci es had high market value when more than 20% of its total harvested biomass was traded and low market value otherwise. I calculated the proportion of individuals from species of high value that corresponded to the total animals killed by each household. I c ompared the average proportion of highly marketable animals harvested by households near (n = 10) and far (n = 23) from markets with a two sample t test. Results A total of 3101 animals (53,700 kg) representing 51 species were harvested between January 20 08 and April 2009 by the 33 Waorani households. This included 24 birds, 23 mammals and 4 reptiles (Appendix A). The five species of u ngulates present in the area white lipped peccary ( Tayassu pecari ) collared peccary ( Pecari tajacu ) red ( Mazama american a ) an d grey brocket deer ( M. gouazoubira ), and tapir ( Tapirus terrestris ), were the most important source of bushmeat, contributing 90% of the total harvested biomass (Table 2 1). W hite lipped and collared peccaries were the most
32 intensively hunted species T hey accounted for 65% of total biomass extracted and 75% of the total biomass traded (Table 2 1). I estimate d 35% of total biomass of harvested game was commercialized. Roads and the Spatial Extent of Wildlife Harvest (H1 ) The Maxus Road led to a signi ficant increase of the spatial extent used by hunters in Yasun. The hunting area estimated by combining the MCPs of the 5 surveyed settlements was 1,616 km 2 which is double the projected accessible area ( 790 km 2 ) in the absence of roads (Figure 2 2). Ker nel analysis provided estimates of biomass extraction that varied from 0 104 kg/km 2 during the 1 year period of this study (Figure 2 3). Based on a minimum extraction rate of 5 kg/km 2 /yr, the total harvested area was 1 560 km 2 (Figure 2 3). Hunters increas ed the harvested area by moving along rivers in motorized dugout canoes and the Maxus Road to reach otherwise inaccessible areas. Waorani moved along the road by using the transportation system provided by the oil company, hitchhiking, or using their own v ehicles (mostly motorcycles). The Waorani used the local market to purchase gas or obtained it from the oil company. The best model in predicting hunting incorporated the three predictor variables: st navigable rivers and Maxus Road. No other combination of predictors was a competitive model ( AIC > 600) (Table 2 2). The ROC plot analysis indicated that my best model had good predictive ability (AUC = 0.92) (Appendix C ). An AUC above 0.9 indicates very good discrimination because the rate of true positives is high relative to the rate of false positives (Pearce & Ferrier 2000) Based on my model of the landscape accessible to hunters, the probability of hunting was 20% or more in an area of 1,684 km 2 (Figure 2 4). The probability of hunting was negatively related with distance to rivers and roa ds, that is,
33 hunting was less likely to occur at farther distances from sources of access (Table 2 3 ) In contrast, the probability of hunting increased with distance from settlement (Table 2 3). Extraction and Trading Rates of Game (H2 ) The overall extra ction rate in the 5 studied settlements was 0.72 ( SE = 0.06) kg/person/day and average trading rate was 0.27 (SE = 0.03) kg/ person/day As I expected, the daily per capita extraction rate in settlements close to market was higher than in those fa rther away ( i.e., close = 0.81, SE = 0.09; far = 0.58, SE = 0.05; t = 2.67 df = 11 P < 0.05). A similar pattern was observed for the total amount of game traded (i.e., close = 0.33, SE = 0.04; far = 0.18, SE = 0.02; t = 2.61 df = 11, P < 0.05). By dedu cting the amount traded from the amount harvested, I found that bushmeat kept for self consumption did not differ between Waorani close to and far from markets (i.e., close = 0.48, SE = 0.06; far = 0.40, SE = 0.04; t = 1.28 df = 11, P > 0.2). D ifferential Use o f Game Species (H 3 ) The most commercialized game incl uded the five species of ungulates (white lipped and collared peccaries, tapir, red and grey brockets), two species of rodents (paca [ Cuniculus paca ] and agouti [ Dasyprocta fuliginosa ]) and yellow footed tortoise ( Chelonoidis denticulata ) Although less than 20% of harvested biomass of grey brocket was traded, I considered this species had high market value and attributed the lower proportion traded (11.4%) to small sample size (Table 2 1 ). Species with low commercial value but important for their biomass contribution to total harvest were large primates including woolly monkey ( Lagothrix poeppigii ) spider monkey ( Ateles belzebuth ) and howler monkey ( Alouatta seniculus ) and big terrestri al birds such as ( Mitu salvini ) ( Penelope jacquacu ) and blue throated
34 piping guan ( Aburria cumanensis ) (Table 2 1). Other 27 less commercialized or non commercialized species harvested in low numbers included 14 mammal 20 b ird and 3 reptile species (Appendix A ). Harvests of households close to markets were comprised of a higher proportion of individuals of species that have high commercial value than harvests of hunters far away from markets (i.e., close = 77%, SE = 3.85; far = 54%, SE = 3.82; t = 3.695, df = 31, P < 0.0 1) For 6 out of 8 species with high market value, individuals of these species comprised a higher proportion of the total harvest close to markets than farther from markets (Figure 2 5). Collared peccaries presented the largest difference comprising ~25% of the total individuals hunted by households close to markets, but only 10% of the individuals harvested far from markets (Figure 2 5). The two exceptions were red and grey brocket deer, which we re hunted more in areas farther from market. As expected, species with low commercial value were consistently less harvested close to to be equally hunted by both groups of hunters (Figure 2 5). Discussion A spatially explicit approach provides a means to visualize and to better understand the consequences of road creation within protected areas for wildlife conservation and game management. The major results were: 1. Roads s ubstantially increased the spatial extent of hunting of Waorani; 2) Increased access to markets was related with higher rates of bushmeat extraction and trade by Waorani households; and 3) Higher access to and use of markets was associated with increased e xploitation of species with greater value in markets.
35 The S patial E xtent of Hunting By increasing the spatial extent of hunted areas, roads can directly threaten the persistence of game populations within inhabited protected areas. When hunting quotas ar e not established the preservation of animal refugia, or sources, becomes vital for the conservation of exploited animal populations (Joshi & Gadgil 1991; McCullough 1996) In the context of this study, refugia are areas that continue to be inaccessible to hunters. Accordingly to Joshi and Gadgil (1991) if the proportion of a protected area that is inaccessible to hunters is large enough, game populations within the accessible area In this study 33 househol ds (i.e., 213 people) within 5 small settlements are harvesting an area of ~1,600 km 2 or approximately 8% of the total 18,000 km 2 area of Yasun. This is a significant area considering the surveyed population corresponds to roughly 7 8% of the total Waora ni population within Yasun, which is estimated as 2,500 3,000 (AMWAE ; CONAIE) Additionally the observed harvest area doubles the projected area in absence of roads. Despite this increase in the area used by hunters, the Waorani along the Maxus Road cont inue extracting large amounts of bushmeat, including vulnerable species such as large primates, tapir and white lipped peccaries (Bodmer et al. 1997) The Maxus Road is within a vast area of virtually intact forest functioning as a wildlife refuge, which m ay explain how such an intense bushmeat extraction is possible (Figure 2 1). However, this is not the case in other areas of Yasun, such as in the western portion of the Waorani Ethnic Reserve, also affected by a road and with a higher density of Waorani settlements (i.e., lower proportion of inaccessible areas or refuges) (Figure 2 1). For example, in the community of Keweriono some vulnerable species such as tapir presented low probability of site
36 occupancy and white lipped peccaries have not been report ed in the last 15 years (Chapter 3, Mena Valenzuela et al. 2000) The increase of the spatial extent of hunting is facilitated by synergistic effects between roads, rivers and markets. These synergistic effects are reflected in the higher probability of h unting along road and river margins. Once roads are open, and if markets are available, hunters can trade meat to obtain goods in markets, including gas for dugout canoes. In the absence of roads, traditional hunters in the Neotropics behave as central pla ce foragers, with most of their game obtained near settlements (Levi et al. 2011; Sirn et al. 2004; Smith 2008) However, when roads are built (i.e., low cost transportation networks), hunters will travel considerable distances by roads to reach areas whe re game is still abundant. As a consequence, the proportion of inaccessible areas, and therefore refuges, will be reduced. These results reveal the importance of prior planning and careful evaluation before developing new infrastructure within or nearby pr otected areas. In Amazonian landscapes, the impacts of roads on the spatial extent of hunting may increase significantly because of access provided directly by road placement, but also because of the synergism between roads and navigable rivers. Extraction and Trading R ates Bushmeat extraction and trade from Waorani hunters along the Maxus Road has increased significantly throughout the last decade. In 2002 the settlements of Guiyero, Timpoka and Dikaro extracted 6,360 kg of bushmeat in a 5 month period, a nd less than 4% of this harvest was traded at the market of Pompeya (Franzen 2006) By 2007 the annual bushmeat trade in Pompeya was estimated at 10,500 kg/year with the Waorani from these same settlements contributing nearly half of this biomass (Surez e t al. 2009) In my study, Waorani from Guiyero, Timpoka, Dikaro and two additional
37 households provided close to 18,500 kg of bushmeat for trade in the same market. Although the amount of bushmeat traded by Waorani clearly increased, results from these stud ies were obtained using different methodologies and therefore may not be completely comparable. The sample size from Franzen (2006) was limited in the number of hunters and time span of survey (5 months). The Suarez et al. (2009) estimate is based on obser vation of bushmeat transactions at the market of Pompeya, which can be challenging. The competition among middlemen is so high that they do not wait until the canoe with the Waorani is on shore in Pompeya to take the meat, making the observation of these t ransactions difficult (Figure 6). As a result, both Franzen (2006) and Suarez et al. (2009) may have underestimated the amount destined for trade by the Waorani. At the time of this study the price of bushmeat was US$3 6/kg, a good price considering that the value of poultry and beef was within a similar range at grocery stores in cities. Likely, if there is a good market for bushmeat, Waorani will continue using wildlife as a commodity. My data indicate that augmenting the road network within or near Yas un likely will have detrimental consequences on game populations by augmenting their extraction and reducing the proportion of refuges. Whether current extraction levels are sustainable or not needs further evaluation. Using Robinson and Redford (1991) pr oduction model, Franzen (2005) estimated the extraction rates of large monkeys, tapir and white lipped peccaries in Yasun were above sustainable levels when refugia were not considered. At the time of my study these species continued to be harvested at ev en higher rates. Understanding harvest at a large spatial scale is critical for appropriate game management efforts for traditional people. The study of Franzen
38 (2005) and my study point to the importance of refugia for maintaining high harvest levels that otherwise would lead to the extirpation of resource populations (McCullough 1996) Even though the Waorani along the Maxus Road are using a significant amount of bushmeat for trade, they maintain appropriate levels of meat consumption. Other studies est imate Neotropical hunters consume 65 that a healthy consumption of meat is about 0.25 kg/person/day (Hill & Hawkes 1983; Robinson & Bennett 2000a; Yost & Kelley 1983) Three decades ago the Waorani used 0.28 kg/person/day for self consumption (Yost & Kelley 1983) When correcting for the edible proportion of game in my study, the Waorani along the Maxus Road consume 0.30 0.32 kg/person/day, a similar amount to that previously reported by Yost and Kelley (1983). More import antly, the amount destined for self consumption did not differ between Waorani close to and far from markets. Increasing access to Yasun by new infrastructure development may threaten not only the persistence of game populations but also the future abili ty of Waorani to procure their source of protein. Differential Use of Game Access to markets leads to significant increases in the proportion of highly valued game, principally peccaries, harvested by the Waorani. Peccaries are among the most important te rrestrial game for Amazonian hunters (Bodmer et al. 1997; Peres 1996; Redford & Robinson 1987; Robinson & Bennett 2000a) By targeting large bodied organisms, hunters may lead to shifts in species composition that may affect the structure and function of N eotropical forest systems (Terborgh et al. 2008) For example, by creating and maintaining wallows where tadpoles can develop, peccaries are important ecosystem engineers for anurans (Beck et al. 2010) Additionally,
39 peccaries and tapirs are important for dispersal of palm seeds and other tropical plant species that produce large fruits (Beck 2006; Janzen & Martin 1982) Additionally, peccaries are generally the most important prey for jaguar ( Panthera onca ), the largest terrestrial predator in Neotropical terrestrial systems (Sanderson et al. 2002a) Decreases in peccary populations could reduce the abundance of this large predator by affecting top down ecosystem processes (Asquith et al. 1997; Terborgh 1988; Terborgh et al. 2001) Hence, development of com munity based programs is important to guarantee the long term harvest of wildlife by the Waorani without threatening the integrity of the ecosystem they inhabit. In addition to community based programs, an adequate proportion of Yasun needs to remain inac cessible to hunting. Given the historical lack of resources to manage protected areas in the Neotropics, preserving wildlife refugia may be the most secure option for maintaining wildlife if the areas are large enough to compensate for high mortality of la rge bodied organisms within accessible areas. Further road development within or near Yasun, and similar areas throughout the Neotropics, need to be very carefully evaluated and planned as they likely will threaten the integrity of these systems character ized by a high biological and cultural diversity.
40 Table 2 1 Most important species harvested as bushmeat during this study in Yasun Biosphere Reserve by 5 Waorani settlements along the Maxus road and traded in the market of Pompeya Hunting activiti es were surveyed between January 2008 and April 2009. O nly species that have a total harvest above 100 kg are listed Species Harvest Trade kg n %TH kg %ST %TT White lipped peccary ( Tayassu pecari ) 26,493 975 49.3 10,510 39.7 56.7 Collared peccary ( Pecari tajacu ) 8 498 448 15.8 3 316 39.0 17.9 Tapir ( Tapirus terrestris ) 8 200 58 15.3 2 166 26.4 11.7 Red brocket deer ( Mazama americana ) 4 517 155 8.4 1 524 33.7 8.2 Woolly monkey ( Lagothrix poeppigii ) 1 838 280 3.4 249 13.6 1.3 Lowland paca ( Cuniculus paca ) 1 032 117 1.9 335 32.4 1.8 ( Mitu salvini ) 693 173 1.3 60 8.6 0.3 Spider monkey ( Ateles belzebuth ) 556 73 1.0 61 10.9 0.3 Howler monkey ( Alouatta seniculus ) 340 51 0.6 33 9.8 0.2 Black agouti ( Dasyprocta fuligi nosa ) 238 49 0.4 54 22.6 0.3 Yellow footed tortoise ( Chelonoidis denticulata ) 169 31 0.3 110 65.4 0.6 ( Penelope jacquacu ) 165 155 0.3 3 1.6 0.0 Grey brocket deer ( Mazama gouazoubira ) 162 11 0.3 18 11.4 0.1 Blue throated Piping guan ( Aburria cumanensis ) 104 102 0.2 2 2.0 0.0 Other species (37) 688 423 1.3 91 13.2 0.5 Total 53,693 3101 100 .0 18,532 100 .0 %TH = percentage of total harvested biomass contributed by these species; %ST = percentage of harvested biomass that was trade d; %TT = percentage of the total biomass traded that was contributed by these species.
41 Table 2 2. Model selection of candidate models to describe the probability of hunting in an area as a function of accessibility. Predictor variables are distance (km ) to road, river and nearest settlement. Model AIC K road + river + settlement 4 196.5 4 0 .0 road + river 4 975.8 3 26. 5 road + settlement 5 979.1 3 1 029.8 road 5 968. 3 2 1 018.9 river + settlement 6 610. 7 3 1 661. 4 settlement 6 859.9 2 1 910.6 river 7 833.6 2 2 884. 3 Table 2 3 Untransformed p arameter estimates of best fit model to estimate the e ffects of accessibility on landscape use d by hunters. Best model: probability of hunting = road + river + settlement Likelihood Ratio Beta S E X 2 df p Intercept 2.6 9 0 .0 8 0 .0 8 1 0.00 R oad 0 .5 2 0 .01 0 .01 1 0.00 River 0 .65 0 .0 2 0 .0 2 1 0.00 Settlement 0 .32 0 .01 0 .01 1 0.00 Pseudo R 2 = 0.49
42 Figure 2 1. Study area (dashed rectangle) and surveyed settlements along the Maxus Road (Guiyero Ganketa, Timpoka, Dikaro and O a) in Yasun Biosphere Reserve.
43 Figure 2 2 Kill sites of game between January 2008 April 2009 (n=2 997) by 5 Waorani settlements along the Maxus Road in Yasun Biosphere Reserve. Polygons represent projected hunting ar ea in the absence of roads (8 km buffer from settlements) and combined hunting area based on minimum convex polygons of kills for the 5 settlements.
44 Figure 2 3. Hunting area based on extraction rate (kg of game/km 2 /year) by 5 Waorani settlements along the Maxus Road in Yasun Biosphere Reserve
45 Figure 2 4. Probability of hunting based on landscape accessibility (output using logistic regression model) for Waorani from 5 settlements along the Maxus Road in Yasun Biosphere Reserve. Hunting area is de fined as the area where the probability of hunting is above 20%.
46 Figure 2 5 Contribution of the 14 most hunted species to the total kills at settlements close (n of individuals harvested = 938) and far (n = 2,163) from markets. *Most commercialized ga me species.
47 A Figure 2 6. Waorani are harassed by middlemen who purchase bushmeat. Waorani is harassed by middlemen as soon as he crosses the Napo River with a bag of bushmeat (A). Waorani discusses the price of the meat middlemen had taken (B). Photo s courtesy of Santiago Espinosa
48 B Figure 2 6. Continued
49 CHAPTER 3 EFFECTS OF INCREASED ACCESIBILITY TO HUNTERS ON THE SPATIAL DISTRIBUTION OF GAME WITHIN A LARGE NEOTROPICAL PROTECTED AREA General Overview Large protected areas are critical in th e Neotropics for the conservation of species that require vast extents of natural lands to persist (Laurance 2005; Peres 2005) These areas also are important for maintain ing populations of species that have high value for local people and therefore are su bject to potential overharvest (Peres & Lake 2003) M ost protected areas in the Neotropics are inhabited by indigenous or traditional peoples who strongly depend on wildlife resources for their survival (Brandon et al. 1998; Robinson & Bennett 2004). L arge bodied mammals such as ungulates have large habitat requirements and are the species most valued by hunters because of their high economic return per unit of effort when sold in markets (Lindstedt et al. 1986; Macdonald et al. 2011) Not surprisingly, ung ulates are frequently the most sensitive species to local extirpation from overharvest in Neotropical ecosystems (Altrichter & Boaglio 2004; Bodmer et al. 1997; Peres 1996) Additionally, these species are among the most sensitive to habitat loss and fragm entation emphasizing the need for large protected areas to maintain viable populations ( Thornton et al. 2011 ). A decrease in abundance of game often occurs in natural areas when human access is high (Peres & Lake 2003) Rivers, which are well represented in Amazonian landscapes, have always been used by indigenous populations to access hunting areas. However, roads have increased access to many areas with historically poor accessibility and are highly detrimental to local game populations because they are associated with commercial exploitation, leading to intensive harvest of animal populations (Chapter 2; Laurance et al. 2006; Wilkie et al. 2000) Additionally, roads can
50 directly influence wildlife abundance in an area by increasing mortality rates from r oad kills, deterring species sensitive to traffic noise or attracting species fond of vegetation growing on road margins, and functioning as barriers for animal dispersal (Forman & Alexander 1998; Forman et al. 2003) Despite their importance for future w ildlife conservation, protected areas in the Neotropics are constantly threatened by new road construction associated with development and resource extraction activities (Finer et al. 2008; Laurance et al. 2006) In some cases, roads have not only changed access to hunting areas, but have also influenced the distribution of humans on the landscape by promoting permanent settlements in areas that previously had few humans (Maki et al. 2001) To better manage protected lands for wildlife conservation where pe ople have rights to hunt within these lands, we need to clearly understand the spatial relationship between means of access (roads and rivers) and settlement distribution and the occurrence of game across space. Although previous studies have documented th e relationship between game depletion and roads (Laurance et al. 2006; Wilkie et al. 2000) to my knowledge, the synergistic effects of principal landscape features that provide access (e.g., settlement distribution, roads and rivers) for hunters to tropic al forests and settlement distribution have not been examined. My research objectives were: (1) to evaluate the effects of landscape accessibility to hunters on game occurrence, and (2) to examine spatial patterns of game occurrence as a function of road v icinity. Hunting is more intense near human settlements and along roads and river margins, therefore, distance to settlements and means of access can be used as indicators of hunting intensity to predict game occurrence in the
51 landscape (Chapter 2; Levi et al. 2011; Smith 2008) Differences in species biology, such as home range, sociality, diet, and body size, will affect the response of organisms to threats such as hunting (Isaac & Cowlishaw 2004) For example, species with large home ranges may be more v ulnerable to hunting as they may pass through inhabited or hunted areas in search of resources and therefore be more susceptible to human induced mortality (Woodroffe & Ginsberg 1998) Social species, particularly those that live in large groups, are easie r to detect by hunters. Species with specific dietary needs, such as frugivorous, that depend on patchy resources can be easier to locate by hunters than species with homogeneously distributed resources (Hill & Padwe 2000) Large body size is related to lo wer rates of population increase and lower population densities; both factors make organisms more vulnerable to extirpation (Pimm et al. 1988) Additionally, large bodied species are important targets of hunters (Robinson & Bennett 2000a) I discuss how di fferences in species traits may interact with effects of increased access to the landscape and road placement to influence the occurrence of game. My study was conducted within Yasun Biosphere Reserve in eastern Ecuador, which is one of the most biodivers e areas in the world and an area threatened by future road development associated with extraction of large oil reserves (Bass et al. 2010; Finer et al. 2008) Road building is currently increasing landscape access for resource exploitation throughout Amaz onia (Perz et al. 2008; Pfaff et al. 2007) The problems I address are common across this region, as well as in other tropical forests worldwide. I examined effects of roads, rivers and settlement distribution on game occurrence using a spatially explicit approach and site occupancy models from camera trap data
52 (MacKenzie et al. 2006). My goal was to identify insights that can be used to improve management of this protected area and others facing similar developmental threats. Methods Study Area Yasun Bios phere Reserve (hereafter Yasun) is formed by Yasun National Park (10,000 km 2 ) the largest protected area in Ecuador, and the Waorani Ethnic Reserve (8,000 km 2 ) (Figure 3 1) Yasun possess important oil reserves underground and the first road (Auca road ) for extraction was created in the early 1980 s within the ancestral Waorani territor y. A second 120 km road (Maxus R oad) was built in the early 19 90s within Yasun National Park and Waorani Ethnic Reserve. Yasun is the ancestral land of the Waorani ethni c group who are distributed throughout all of Yasun. Traditionally, the Waorani were a semi nomadic group strongly dependent on forests resources such as plant fibers, fruits, and especially bushmeat. To procure game they used blowguns for arboreal specie s, such as woolly ( Lagothirx poeppigii ) and spider ( Ateles belzebuth ) monkeys, and spears for larger terrestrial game such as white lipped peccaries ( Tayassu pecari ) and tapir ( Tapirus terrestris ) (Yost & Kelley 1983) After road development, some Waorani settled along road margins, replaced blowguns and spears with shotguns, and abandoned subsistence hunting practices to harvest large quantities of bushmeat regularly traded in local markets (Chapter 2, Surez et al. 2009) In addition to the Waorani, s ever al Kichwa populations are located along the edge of Yasun principally by the Napo River in the north and Curaray River in the south E levation of Yasun ranges between 200 400 m T opography is characterized by numero us ridges (20 40 m high) and a dens e system of streams and rivers. V egetation
53 is dominated by tall evergreen terra firme forest with a canopy h e igh t of about 25 m punctuated with emergent trees of 40 50 m Flood plains and swamps occur at the margins of principal rivers (Valencia et al. 200 4). Additionally, two main habitat types, defined by topography, can be distinguished in terra firme forest across Yasun: 1) ridge vegetation that has highe r stem density and taller trees, and 2) valley vegetation that has lower stem density and lower can opy h e igh t (Valencia et al. 2004) Annual rainfall is 3,000 3,400 mm and minimum and maximum average temperatures of 22 36. Seasons are not clearly marked. May and June are the rainiest months wit h average precipitations of 330 390 mm and no month has a verage precipitation less than 130 mm ( Yasun Research Station meteorological, data 2000 2009). Study Design I focused my analysis on terrestrial mammals, which include the taxa most preferred by Neotropical subsistence hunters, such as ungulates (Peres 2 000b) My working hypotheses were (1) greater landscape access for hunters leads to lower game occurrence, and (2) roads lead to spatial patterns in game occurrence, where the probability of encountering game is lower near road margins and higher at distan ces farther away. To evaluate my first hypothesis, I conducted wildlife surveys in four areas with different degrees of accessibility by rivers and roads. I conducted a separate analysis to test my second hypothesis at one site accessible by road. In my st udy, the most accessible site to hunters was located along the 120 km Maxus R oad (Figure 3 1). The first 30 km of the road is inhabited by the Kichwa and the remaining 90 km is controlled by the Waorani Both indigenous groups settled in this area soon aft er road creation in 1993. My wildlife survey took place in the hunting area of four small Waorani settlements (Guiyero, Tiwe, Ganketa and Timpoka) with a
54 combined population of 70. The Waorani along this road use wildlife as a commodity, trading approximat ely 35% of the total harvest at local markets ( Chapter 2). Wildlife extraction was most intense at this site. My second site, Keweriono exhibited an intermediate level of access, was within the Waorani Ethnic Reserve, near the communities of Keweriono (6 0 inhabitants) and Apaika (10 inhabitants) located along the Shiripuno River. These two communities are accessible by river (3 6 hrs trip) or foot ( 6 8 hrs walk) from the Auca Road, which is located at 25 and 15 km from Keweriono and Apaika respectively ( Figure 3 1). All the surveyed area was accessible to hunters by foot from their own settlements Hunting in this area has been continuous since the establishment of Keweriono in 1989. H unting in these communities occurs principally for subsistence purposes However, s moked bushmeat is occasionally sold in nearby towns, the closest on the Auca Road, by hunters who have had exceptionally good kills (Lu 1999; Mena Valenzuela et al. 2000; Sierra et al. 1999) My third site, Tiputini was located by the Tiputin i River, in the northern boundary of Yasun National Park (Figure 3 1) Hunter access to this site was possible only by dugout canoe and hunting effort was lower than at other sites. Although not inhabited, this area is visited sporadically by Waorani and Kichwa hunters. Waorani hunters enter t h rough the Tiputini River from the west (M axus Road) and Kichwa hunters ca me from settlements close to the intersection of the Tiputini and Napo Rivers in the east Evidence from the field (e.g., huntin g camps, hunter trails and tracks, remains of animal parts) and direct observations of hunters showed that this area is used to extract
55 bushmeat for both subsistence and commercial purposes. T o access this site hunters must travel 3 5 hours from nearest settlements us ing dugo ut canoes My fourth site, Lorocachi was also a low access site adjacent to the Curaray River in the southern border of Yasun and near the Kichwa community of Lorocachi and the army base BS 48 Sangay (Figure 3 1) Lorocachi is a small community (~120 inhabitants) that was formed with the creation of BS 48 Sangay in 1953. This area is accessible to outsiders only by an army airplane that supplies the base. My survey area was accessible to hunters only by foot and was located from 3 to 21 km from t heir settlement. The maximum distance hunters walk to procure game in Amazonian landscapes is ab out 7 9 km and, therefore, approximately 50% of my study area was out of the reach of hunters (Chapter 2, Peres & Lake 2003) Hunting in Lorocachi is conducted exclusively by the Kichwa for subsistence purposes. Army personnel are not allowed to hunt, although wildlife is sporadically killed when found within the base facilities (SE personal observations). To test my first hypothesis and measure the effects of l andscape accessibility on game occurrence, I conducted a semi systematic sampling using remotely triggered camera traps at the four study areas. I n each site I created a transect system o f 44 67 km ( total of 215 km) to establish a grid of 26, 23, 25 and 26 camera trap stations in Maxus Road, Keweriono, Tiputini and Lorocachi respectively. Camera trap s tations consisted of two camera traps facing each other on opposite sides of a transect and located 2 3 km from each other to cover a polygon of 100 km 2 in ea ch site. I installed cameras in places along transects that would maximize the capture of mammals such
56 as at intersections with game trails or areas were the topography might funnel animals towards cameras. I used film camera traps equipped with 35 m m le nses and a passive heat motion sensor (Leaf River BU) I programmed camera traps to work continuously (24 hrs per day) with a minimum lag time of one minute between pictures I mounted cameras o n trees at 30 40 cm above ground and tested them to be sure that large mammals and mammals as small as agoutis ( Dasyprocta fuliginosa ) were photographed consistently I conducted camera trap surveys between November 2007 and December 2009. Camera trap grids worked for 90 consecutive days at each site for a cumulative effort of 8,389 camera trap nights across my four study areas. I visited camera trap stations every 10 15 days to replace film (ISO 200 or 400, 36 exposures) and to check for proper functioning To test my second hypothesis and measure the effe cts of roads on game distribution, I conducted systematic sampling at the Maxus Road area. I placed 13 5 km transects perpendicular to the road and installed one camera trap every 0.5 km from the beginning to the end of transects (i.e., a total of 10 camer a traps per transect ). Cameras were mounted on trees at 30 40 cm above ground facing transects. I cleared the herbaceous vegetation within 5 m of the front of the cameras to improve visibility All cameras were set to work for 16 continuous days. Surveys w ere conducted during two periods ( December 2007 March 2008 May June 2009 ) and 115 camera locations were used for analyses totaling 1,661 camera trap days (15 cameras failed). Analyzing the Effects of Access and Roads on Game Occurrence I used probability of site occupancy, as a measure of species occurrence and developed single season, single species occupancy models to estimate with program
57 PRESENCE 2 (Hines 2006; MacKenzie et al. 2006) A major assumption of site occupancy models is the state of site s species present or absent does not change during survey. When the objects of study are highly mobile species (e.g., large mammals) that could be temporarily absent from the surveyed sites, this assumption is likely violated. I viewed change in occupanc y state as a random process and relaxed convenience ( MacKenzie et al. 2006 ) To evaluate t he effect of landscape accessibility to hunters on the occurrence of game, I compared data. These species included t he five most hunted large mammals in my study area all of them ungulate s (Chapter 2): white lipped peccary ( Tayassu pecari ), collared peccary ( Pecari tajacu ), tapir ( Tapirus terrestris ), red brocket deer ( Mazama americana ) and grey brocket deer ( M. gouazoubira ). I compared ungulate occurrence with two other large mammals tha t live at low densities and are rarely hunted, giant armadillo ( Priodontes maximus ) and giant anteater ( Myrmecophaga tridactyla ). I contrasted the occurrence of large game with other mammalian species that are hunted and have small/medium body sizes and th erefore live at higher densities such as, paca ( Cuniculus paca ), agouti ( Dasyprocta fuliginosa ) and nine banded armadillo ( Dasypus novemcinctus ) (Robinson & Redford 1986a) Finally, I included two species of terrestrial urassow ( Mitu salvini ) and the less hunted grey winged trumpeter ( Psophia crepitans ) (Chapter 2). All species, except giant anteater, were detected on a minimum of 45 trapping occasions (Appendix C, Table C 1).
58 sample size of 100 surveyed points (i.e., camera trap stations). To input data in PRESENCE2, I broke up 90 each and developed a matrix of 100 rows (survey points) by 9 columns (survey occasions). Cells in matrices were filled with 1 if a species was detected within a 10 day survey occasion or 0 if that was not the case. To predict occupancy I used two measures of acces sibility, the Euclidean distance of camera trap stations to settlements and to the nearest source of access (road or navigable river). Hunting by traditional peoples is concentrated around settlements, therefore, distance to settlement is a good predictor of hunting intensity (Chapter 2; Levi et al. 2011) Also, hunting is more intense at distances close to the margins of rivers and roads (Chapter 2; Peres & Lake 2003) I added habitat type as a third predictor of which was represented by a dummy variabl e indicating two main terra firme habitat types in Yasun vegetation cover at ridge s and valle ys (Valencia et al. 2004) As paired cameras at each station were placed at varying distances, typically 7 9 m from each other, I included distance between pair ed cameras as a predictor of detection probability I developed 15 models for each species that included all combinations of explanatory variables for and detection probability. I used Akaike Information Criterion corrected for small sample ( AIC c ) to select the best fit model (Burnham & Anderson 2002) I assessed the relative importance of each predictor variable by adding the AIC c weights ( ) of model s where the predictor of interest was present ( + ; Burnham & Anderson 2002). correlograms to test for spatial independence of the residuals of occupancy models
59 (Fortin & Dale 2005) Correlograms were developed with software PASSaGE 2 (Rose nberg & Anderson 2011) As survey effort is an important factor in determining the probability of detection of species, I was concerned about the appropriateness of my survey effort for detecting differences between surveyed areas (Tobler et al. 2008) I developed accumulation curves to explore changes in the proportion of occupied sites (i.e., nave occupancy) by individual species as survey effort progressed. I created accumulation curves with VEGAN package in program R and used the exact method to estim ate m ean and standard deviation s of nave occupancy (Kindt et al. 2006; Oksanen et al. 2011) To evaluate occurrence of game at varying distances from the road, I used data from the 115 camera trap stations placed in 13 transects perpendicular to the Maxu s Road. I created detection histories for species by dividing 16 day survey periods into 4 survey occasions of 4 days each. Similar to my first analysis, a matrix of 115 rows (survey points) by 4 (survey occasions) was formed for each analyzed species wher e 1 represented detection and 0 non detection. To model occupancy I used three predictor variables: distance to road, distance to settlement and habitat type. Because of the limited number of detections per species, I did not model detection probability an d rather assume d it was constant across survey points I developed 8 models for each species and used AIC c to select the most representative model and to evaluate the importance of each covariate. I modeled for the following species: collared peccary, tapir, red brocket deer, paca, agouti, and nine banded armadillo. Because of low capture rates, I omitted two important game species, white lipped pecc aries and grey
60 brocket d eer from this analysis (Appendix C, Table C 2). As in my first analysis, I used Results Landscape Access to Hunters and Game Occurrence When analyzing wildlife occurrence as a function of distance to nearest road or river, distance to settlement, and habitat type, the best predictor of wildlife occurrence for all five species of ungulate was distance to settlement. Distance to settlement was included in all best models for un gulates and positively correlated with site occupancy, that is the farther from settlements the higher the probability of finding one of these species (Table 3 1). D istance to settlement summed 100% of AIC c weights in white lipped peccary, collared peccary and tapir, and 76 and 96% in red and grey brocket deer, respectively, indicating the relative importance of this predictor in ove rall models (Table 3 2, Appendix D). Distance to settlement also was important in determining the presence of medium sized gam agouti. In contrast, distance to settlement was not an important predictor of the occurrence of large but rarely hunted species such as giant anteater and giant armadillo, or of small and less intensive ly hunted species such as nine banded armadillo and grey winged trumpeter (Table 3 2). Distance to the nearest road or river was less important in predicting game occurrence than distance to settlement, only appearing in 3 of the best models for the 12 an alyzed species (Table 3 1). Distance to road accounted for 89% of AIC c weights for models to predict site occupancy of collared peccaries, but only for 34 and 33% of models weights for white lipped peccary and tapir respectively. Habitat also generally was a less important predictor, although in some cases more important than access, for
61 example, it accounted for 46% of the model weights for white lipped peccaries (Table 3 2). Distance between cameras was related positively with detection probability of ung ulates (Table 3 1), and models that accounted for variation of detection probability across sites were in general better than those t hat assumed constant p (Appendix D). Ungulates presented clear differences in nave occupancy between most accessible and least accessible areas as survey effort progressed (Figure 3 3 A E). Differences in nave site occupancy of ungulates between areas were most noticeable after 20 40 days of survey period and maintained throughout the total extent of surveys (Figure 3 3 A E ). In contrast, there were no differences in nave occupancy for the two rarely hunted large mammals, giant armadillo and giant anteater, and for the hunted agouti (i.e., these species presented almost identical occupancy accumulation curves) (Figure 3 3 F G, I). In the case of the hunted paca and nine banded armadillo, some differences in nave occupancy between areas were observed after 30 and 60 days, respectively (Figure 3 Waorani in Yasun, pr esented clear differences in nave occupancy between more accessible and less accessible areas from near the beginning of the surveys (ca. 10 days), and for grey winged trumpeter differences were not as apparent (Figure 3 3 K L). Predicted probability of s ite occupancy for hunted species was generally higher in less accessible areas than in most accessible areas (Table 3 3). Probability of site occupancy for white lipped peccaries, collared peccaries and tapir was remarkably higher in the remote areas, Loro cachi and Tiputini, than in the more accessible Keweriono and Maxus Road (Table 3 3). White lipped peccary appeared to be the most sensitive species to hunting and higher access. Also, was 10 times lower for this
62 species in the most accessible areas of M axus Road and Keweriono (it was never recorded in Keweriono) than in Lorocachi and Tiputini (Table 3 3). Although red brocket deer and grey brocket deer were more frequently encountered in less accessible areas differences were not as strong. In contrast, giant armadillo and giant anteater presented similar estimates of across all four areas (Table 3 3). Medium sized mammals (paca, and armadillo) presented less marked differences of and nave occupancy between least accessible and most accessible areas, and in the case of agouti no differences across areas were observ ed (Table 3 winged trumpeter presented higher estimates of areas, however, differences in the first species were much more strong. There was no evidence of spatial depen dence for species anal yzed at the four areas (Appendix E, Table E 1, Figures E 1, E 2) Road Vicinity and Game Occurrence Distance to road was clearly the most important predictor for of collared peccaries and red brocket deer, two of the three large bodi ed game analyzed, in areas within 5 km from the Maxus Road (Table 3 2). Distance to road was part of the best models of collared peccary, red brocket deer and paca and positively associated with their occurrence (i.e., the greater the distance the greater the probability of occurrence) (Table 3 4). Additionally, distance to road was among competitive models ( AIC c < 2) of tapir, and agouti ( Appendix F). The most important predictor for tapir was habitat type (Table 3 2), and tapirs were most likely to occur in forest valleys than on ridges (Table 3 4). Agouti was better predicted by and positively associated w ith distance form settlements (Tables 3 2, 3 4). In the case of armadillos, the null model had the highest ranked AIC with no other competitive models, suggesting that none of the three
63 covariates were important in determinin g armadillo occurrence (Appendi x F). Residuals for paca and agouti exhibited s ome spatial dependence (Appendix E, Table E 1, Figure E 3). Discussion Settlement Distribution, Landscape Access to Hunters and Game Occurrence My results demonstrate that the occurrence of game species across Neotropical landscapes is principally affected by settlement distribution when roads are scarcely represented. When all four areas were used to build single species occupancy models, vicinity to settlements was a better predictor of the presence of game s pecies than distance to nearest road or river (i.e., means of access). Traditional hunters are typically described as central place foragers because in the absence of infrastructure the intensity of hunting by traditional hunters is concentrated near settl ements (Chapter 2; Levi et al. 2011; Smith 2008) In my study, three of the four areas used to build occupancy models were located in areas without infrastructure. Therefore, in my first analyses, when using the nearest distance to river or road to create a predictor of accessibility, the effect of rivers was better represented than roads. This can be considered as a representative sample of the Yasun landscape, which still has a large proportion far from the influence of roads. However, if road density is higher, likely, the importance of settlement distribution and means of access in determining game occurrence will be similar. Large bodied species (i.e., low population growth rates) and those that are highly social appeared to be the most affected by in creased access to the landscape by hunters. The most hunted terrestrial game species by the Waorani in Yasun are white lipped and collared peccaries (Chapter 2, Franzen 2006; Mena Valenzuela et al. 2000;
64 Surez et al. 2009) Although both species appeared to be affected by increased access to hunters, the decrease in white lipped peccary occurrence in more accessible sites was remarkable. White lipped peccaries live in large herds, commonly of 10 300 individuals, have relatively large home ranges with reco rded maximum estimates of 109 120 km 2 and are known for their high reliance on wetlands and palm aggregations which concentrate s their distribution around these areas (Fragoso 1998; Fragoso 1999; Mayer & Wetzel 1987; Reyna Hurtado et al. 2009) In contra st, collared peccaries form smaller groups, typically of 6 8 animals, and are not dependent on the patchy palm aggregations (Robinson & Redford 1986b; Sowls 1997) Additionally, white lipped peccaries have lower rates of population increase ( r = 0.84 ) than collared peccaries ( r = 1.25 ), which make them more vulnerable to overharvest (Bodmer et al. 1997; Robinson & Redford 1986b) The higher vulnerability to hunting of white lipped peccaries compared to collared peccaries has been observed in other areas whe re these species coexist (Endo et al. 2010; Naranjo & Bodmer 2007; Peres 1996; Reyna Hurtado & Tanner 2007) From the three solitary ungulates tapir was the most sensitive to higher accessibility by hunters. This result was expected because tapirs are the largest terrestrial mammal in the Neotropics, have the lowest population growth rates ( r = 0.2), exhibit the longest generational time among Neotropical ungulates, and are among the preferred game for hunters. As a consequence, this species is susceptible to overexploitation (Bodmer et al. 1997; Robinson & Redford 1991) Additionally, tapirs are known to be linked with swampy or aquatic areas, particularly aggregations of the palm Mauritia flexuosa which makes them easier to locate by hunters, as observed with
65 white lipped peccaries (Bodmer 1990) Previous studies in other regions with both the lowland tapir Tapirus terrestris T. bairdii have found these species are more abundant in unhunted or moderately hunted areas than in intensively hunted areas (Hill et al. 1997; Naranjo & Bodmer 2007; Tobler 2002) In contrast to tapir, the solitary red and grey brocket deer have shorter generational times and higher reproductive rates than tapirs (r = 0.40 0.49) (Bodmer et al. 1997; Robinson & Red ford 1986b) Additionally, they are fast animals that flee shortly after being detected by hunters, which make them a difficult target (Fa et al. 2005) As expected from these traits, red brocket deer and grey brocket deer appeared to be less vulnerable to access to the landscape when compared with tapir and peccaries (Robinson & Redford 1986b) The occurrence of giant armadillos and giant anteaters was similar across sites for the former and did not present clear patterns for the latter, which was expected given that these species are rarely hunted in the studied areas (Chapter 2; Mena Valenzuela et al. 2000) The occurrence of medium sized game appeared to be less explained by accessibility than in large bodied species. Although pacas and armadil los appeared to be somewhat affected by hunting, and nave occupancy estimates were not as consistent as in the case of ungulates, suggesting their occurrence may not be well predicted by accessibility to hunters. Additionally, the occurrence of agouti a nd grey winged trumpeter was similar across the four different areas. Smaller species have higher reproductive rates and are predicted to live at higher densities than large bodied organisms (Robinson & Redford 1986a, b) My results suggest that hunting le vels of these smaller species are not strong enough to create spatial patterns of their
66 occurrence that can be described by distance to settlements or means of access. 4 kg bird, was well predicted by dista biological traits that make them vulnerable to hunting; for example, they are social (commonly found in couples) and have relatively low reproductive rates (r = 0.38). Additionally, they are highly sought after by Amazonian hunters and intensively hunted by the Waorani and Kichwa (Chapter 2, Begazo & Bodmer 1998; Thiollay 2005; Vickers 1991) Road Placement and Game Occurrence When roads are not available, hunters concentrate their a ctivities near settlements and as shown earlier, distance from settlement can be a more important predictor than means of access. However, once roads are developed they can provide low cost transportation for hunters to access new areas where game is still abundant (Chapter 2; Benitez Lopez et al. 2010; Wilkie et al. 2000) Studies on the abundance of mammalian species as a function of distance to road are biased toward temperate areas. For example, a recent meta analysis found that out of 49 available stud ies, 44 correspond to Europe and North America, whereas 3 are from Africa, 1 from Oceania and none from the Neotropics (Benitez Lopez et al. 2010) This meta analysis concludes that the abundance of mammalian species decreases with proximity to road margin s. However, the causes of the effects of roads on animal distribution are briefly discussed by Benitez Lopez et al. (2010) and mainly attributed to the ecological effects that roads have on animal distributions (Forman et al. 2003) My study expands this k nowledge by including an area in the Neotropics, where the effects that roads have on animal
67 abundance may be linked more to increased landscape access to subsistence hunters than to ecological effects. In my analyses of the occurrence of game as a functio n of increased distance from the Maxus Road, distance from settlements, and habitat type, the distance from road was either the best model or among the competitive models predicting the occurrence of five of the six species analyzed. Distance to road was p ositively associated with occurrence, that is, the farther from road, the greater the probability of species presence. Although multiple factors may explain this pattern (Forman et al. 2003) increased hunting pressure along roads may largely explain patte rns of occurrence of game by roads in Yasun. Roads lead to defaunation of Paleotropical landscapes by reducing transportation costs of bushmeat to local markets and by providing hunters with access to areas otherwise unexploited (Laurance et al. 2006; Wil kie et al. 2000; Wilkie et al. 1992) In Chapter 2, I show that roads have the same effect on bushmeat extraction by local inhabitants of Yasun. Three other factors that can cause decreases in wildlife near roads are behavioral responses of animals to the presence of roads, roads acting as barriers for animal dispersal, and mortality associated with vehicles (Forman & Deblinger 2000; Forman et al. 2003) Behavioral responses could play an important role in influencing the occurrence of game along roads in Yasun. Caribou ( Rangifer tarandus granti ) avoid areas within 0 4 km from roads and platforms used for oil extraction in Alaska. Even though areas closer to roads have better terrain for forage, caribou prefer to use areas 4 10 km from infrastructure where resources are scarcer (Joly et al. 2006; Nellemann & Cameron 1996; Nellemann et al. 2003) Given that the Maxus Road is surrounded by a continuous matrix of intact forest, dispersal limitations
68 are not likely to play an important role in determining game occurrence at this particular site. Nevertheless, this could be an important factor in other areas of Yasun. The effects of wildlife mortality caused by collisions with vehicles have not been measured in Yasun. Given the low traffic frequency along the M axus Road, which is restricted to trucks working for the oil company, and the low maximum speed (40 km/h) strictly enforced by the oil company, road kills also probably do not contribute significantly to game mortality when compared with mortality from hun ting. My study expands our understanding of how settlement distribution, roads and rivers interact to determine the occurrence of game in landscapes used by traditional hunters (Laurance et al. 2008; Laurance et al. 2006; Peres & Lake 2003) My results in dicate that the distribution of settlements is a more important predictor of game occurrence than the distribution of means of access (i.e., roads and navigable rivers) when roads do not dominate the landscape. This is a logical result because permanent se ttlements lead to intense hunting in the surrounding areas where hunters either walk or use rivers, a more costly means of access than roads, to reach hunting areas (Levi et al. 2011; Sirn et al. 2004; Smith 2008) However, when roads are available, hunte rs can increase their hunting area by moving along roads and incre asing the use of rivers (Chapter 2). In this case, roads can be equally or more important than settlements in determining the distribution of game Through spatial analyses I demonstrate ho w the distribution of settlements and roads, the main infrastructure created by humans to provide access, can have differential effects on the occurrence of game species within a human inhabited natural reserve. For example, results of site occupancy analy sis for white lipped and collared
69 peccaries (Table 3 1) can be used to extrapolate occurrence across the entire Yasun (Figure 3 4). The probability of site occupancy of white lipped peccaries is low within the western portion of Yasun, which has a higher settlement density compared to rest of the Biosphere (Figure 3 4). About 35 Waorani settlements are distributed within this area of ~3100 km 2 (i.e., 1 settlement per 90 km 2 ) with a population of ~1,000 Waorani, which is a lower human density than the the oretical carrying capacity in tropical forests of 1 person per km 2 (Robinson & Bennett 2000a) However, current settlement and population density in the western portion of Yasun may be great enough to affect the persistence of the most vulnerable white li pped peccary, but probably not that of collared peccaries (Figure 3 4), a species more resistant to hunting (Bodmer et al. 1997) These results suggest the importance of conducting species specific analyses for assessment of impacts of infrastructure expan sion, wildlife harvest plans, and generally to better inform future management decisions within Yasun. Inhabited protected areas with h unting that stems from s ettlements within park borders often exhibit source sink dynamics. In hunted areas near settleme nts, death rates are higher than birth rates, an d thus become sinks; whereas at farther distances (i.e., inaccessible areas), births outnumber deaths and these areas act as sources (Novaro et al. 2005). Sinks are replenished with game through animal disper sal from sources. The presence of sources permits hunters to continue harvesting vulnerable species, such as tapir, that otherwise would become extinct within hunting areas (Joshi & Gadgil 1991; McCullough 1996; Novaro et al. 2000) For these systems to ma intain wildlife populations and support wildlife harvest, two main conditions are needed: 1) the proportion of a protected area set aside as refuge, or inaccessible to hunters, has to be
70 large enough to maintain a stock able to compensate for high mortalit y in hunted areas; and 2) animal dispersal between sources and sinks needs to be maintained (Joshi & Gadgil 1991; McCullough 1996; Novaro et al. 2000) Roads increase the proportion of accessible areas to hunters, initiate colonization processes (i.e., hig her settlement density) and act as barriers for animal dispersal (Forman et al. 2003; Peres & Lake 2003) If further road development occurs within Yasun, the risk of collapse of this system functioning under a source sink dynamic will increase. A crash o f game populations in Yasun will not only affect its rich biodiversity, but also the Waorani, a unique culture strongly associated with hunting and one dependent on these resources to subsist. This unfortunate fate can be overcome by 1) avoiding further r oad building within Yasun to maintain an adequate proportion of its area inaccessible to hunters, and by 2) developing community based programs for the sustainable use of wildlife.
71 Table 3 1 Best occupancy models and untransformed model parameters to predict game probability of site occupancy ( ) as a function of accessibility across four study sites. Covariates for i nclude : Road/River = distance to nearest source of access (road or river); Settlement = distance to nearest settlement; Habitat = hill (1) and valley (0). Detection probability (p ) is modeled as a function of distance between paired cameras (Dist. Cam.). p p 1 Intercept Road/River Settlement Habitat Intercept Dist. Cam. Tayassu pecari (white lipped peccary) 5.73 0.77 2.78 0.19 0.2 7 Pecari tajacu (collared peccary) 2.26 1.63 0.43 1.49 0.10 0. 35 Tapirus terrestris (tapir) 0.88 0.20 3.13 0.21 0.20 Mazama americana (red brocket deer) 0.59 0.13 3.49 0.30 0.2 8 Mazama gouazoubira (grey brocket deer) 1.38 0.13 1.03 1.62 0.17 Myrmecophaga trida ctyla 2 (giant anteater) 1.09 1.48 1.02 1.10 0.21 0.0 9 Priodontes maximus 2 (giant armadillo) 0.96 0.27 2.69 0.06 Cuniculus paca (paca) 0.14 0.14 1.37 1.52 0.18 Dasyprocta fuliginosa (agouti) 1.91 0.82 0.52 0.37 Dasypus no vemcinctus 2 (nine banded armadillo) 0.01 0.07 0.60 0.09 0.21 Mitu salvini 1.82 0.40 1.84 3.18 0.22 0.2 4 Psophia crepitans (grey winged trumpeter) 0.88 0.13 1.09 0.13 0.52 1 Unmodeled detection probability. 2 First mo del that converges and where is modeled (Appendix B).
72 Table 3 2. I mportance of covariates in predicting for occupancy models that included four areas with varying degree of accessibility (A), and for occupancy models along the Maxus Road (B). For each species, importance was de termined by the sum of AIC c weights ( + ) of all models where each variable was contained. A. Species evaluated at four areas + road/river + settlement + habitat Tayassu pecari 0.34 1.00 0.46 Pecari tajacu 0.89 1.00 0.29 Tapirus terrestris 0.33 1.00 0.23 Mazama america na 0.35 0.76 0.49 Mazama gouazoubira 0.40 0.96 0.60 Myrmecophaga tridactyla 0.95 0.94 0.92 Priodontes maximus 0.43 0.44 0.54 Cuniculus paca 0.44 0.94 0.80 Dasyprocta fuliginosa 0.30 0.30 0.56 Dasypus novemcinctus 0.36 0.42 0.30 Mitu salvini 0.25 1.00 0.90 Psophia crepitans 0.53 0.69 0.38 B. Species evaluated at Maxus Road + road + settlement + habitat Pecari tajacu 0.5 3 0.28 0.26 Tapirus terrestris 0.39 0.33 0. 60 Mazama americana 0.86 0.59 0.25 Cuniculus paca 0.42 0.39 0.41 Dasyprocta fuliginosa 0.36 0.47 0.45 Dasypus novemcinctus 0.24 0.43 0.26
73 Table 3 3. Averag e predicted probability of site occupancy, (SE), of best fit models and nave occupancy 1 estimates for individual species in four study sites. Maxus Road Keweriono Tiputini Lorocachi N ave N ave N ave N ave Tayassu pecari 0.08 (0.05) 0.08 0.05 (0.04) 0.00 0.98 (0.04) 0.88 0.75 (0.10) 0.62 Pecari tajacu 0.66 (0.10) 0.69 0.73 (0.07) 0.61 0.99 (0.02) 0.96 0.97 (0.01) 0.96 Tapirus terrestris 0.46 (0.08) 0.54 0.43 (0.09) 0.22 0.89 (0.07) 0.84 0.77 (0.07) 0.62 Mazama america na 0.74 (0.08) 0.54 0.72 (0.08) 0.74 0.92 (0.05) 0.88 0.87 (0.05) 0.81 Mazama gouazoubira 0.24 (0.08) 0.27 0.17 (0.07) 0.00 0.54 (0.13) 0.44 0.37 (0.10) 0.31 Myrmecophaga tridactyla 0.65 (0.19) 0.35 0.26 (0.14) 0.00 1.00 (0.00) 0.28 0.62 (0.23) 0.23 Pr iodontes maximus 0.78 (0.21) 0.31 0.82 (0.21) 0.35 0.80 (0.21) 0.36 0.91 (0.16) 0.42 Cuniculus paca 0.55 (0.11) 0.42 0.42 (0.11) 0.39 0.83 (0.10) 0.76 0.65 (0.11) 0.42 Dasyprocta fuliginosa 0.83 (0.06) 0.73 0.79 (0.06) 0.83 0.82 (0.06) 0.76 0.78 (0.06) 0 .85 Dasypus novemcinctus 0.57 (0.08) 0.54 0.56 (0.08) 0.43 0.75 (0.09) 0.76 0.70 (0.08) 0.50 Mitu salvini 0.32 (0.10) 0.27 0.18 (0.08) 0.13 0.94 (0.07) 0.68 0.69 (0.11) 0.65 Psophia crepitans 0.79 (0.06) 0.92 0.78 (0.06) 0.70 0.94 (0.04) 0.88 0.90 (0.04 ) 0.88 1 Total camera trap stations where a species is detected divided by total number of camera trap stations placed surveyed.
74 Table 3 4 Best occupancy models and untransformed model parameters to predict game occurrence by the Maxus Road. Covariate s for include : Road = distance to nearest road; Settlement = distance to nearest settlement; and Habitat = hill (1) and valley (0). Detection probability (p) was kept constant. p 1 (SE) Nave 2 Intercept Road Settlement Habitat Pecari tajacu 0.56 0.3 7 0.24 0.55 (0.12) 0.35 Tapirus terrestris 1.69 2.06 0.11 0.6 9 (0.28) 0.25 Mazama americana 2.81 0.55 0.23 0.29 0.40 (0.09) 0.30 Cuniculus paca 0.61 1.65 0.12 0.84 (0.13) 0.31 Dasyprocta fuliginosa 0.71 0.26 0.21 0.60 (0.13) 0.36 Dasypus novemcinctus 0.65 0. 14 0.66 (0.20) 0.14 1 2 Total camera trap stations where a species is detected divided by total number of camera trap stations placed surveyed.
75 Figure 3 1. Study sites and camera trap arrays to measure the effects of a ccessibility on game occurrence in Yasun Biosphere Reserve Study sites A D are ordered from most accessible to least accessible.
76 Figure 3 2 Camera placement to analyze the effect of road on game distribution across
77 A B C D E F Figure 3 3. Accumulation curves of number of locations ( i.e., camera trap stations) where a species is detected as survey effort progresses at four study sites in Yasun Biosphere Reserve. A) white lipped peccary, B) collared peccary C) tapir, D) red brocket deer, E) grey brocket deer F) giant anteater G) gi ant armadillo H) paca I) agouti J) nine banded armadillo K) L) grey winged trumpeter. Leaf area indicates 1 standard deviation. White lipped peccaries, grey brocket deer and giant ant eater were not detected in Keweriono.
78 G H I J K L Figure 3 3. Continued.
79 Figure 3 4. Probability of site occup ancy of white lipped peccary (A) and collared peccary (B) within Yasun Biosphere Reserve based on best site occupancy models (Table 3 1).
80 CHAPTER 4 INCREASED ACCESS TO NEOTROPICAL PROTECTED AREAS, BUSHMEAT EXTRACTION AND ITS IMPLICATIONS FOR JAGUAR CON SERVATION General Overview Worldwide all populations of large felids, including lions ( Panthera leo ), tigers ( P tigris ), leopards ( P. pardus ) and jaguars ( P. onca ), are decreasing or at risk of extinction (IUCN 2011) Major causes for large carnivore pop ulation declines worldwide include: 1) habitat degradation, loss and fragmentation; 2) direct killings, either by poachers looking for animal parts or by ranchers attempting to eliminate problem animals; and 3) prey hoods are dependent on bushmeat (Inskip & Zimmermann 2009; Karanth & Chellam 2009; Treves & Karanth 2003) As a consequence of these threats, large protected areas or mega reserves i.e., those larger than 10,000 km 2 are becoming essential for maintenance of large carnivore populations and the ecosystems they inhabit (Peres 2005; Woodroffe & Ginsberg 1998) By sustaining large quantities of prey, these reserves have a higher probability of maintaining viable populations of large carnivores (Shaffer 1981) Additionally, by having a higher proportion of core, large protected areas can attenuate detrimental effects on carnivore populations caused by high mortality on reserve edges (Woodroffe & Ginsberg 1998) However, threats, particularly in the tropics (Brandon et al. 1998; Bruner et al. 2001) Some main threats to protected areas include removal of individual elements (e.g., extraction of particular species used as game), impoverishment of ecosystems within protecte d area (e.g., by pollution or overhunting), habitat loss and degradation (e.g., by forest clearing, road development and colonization processes) and isolation (e.g., by degradation of surrounding areas) (Chape et al. 2005)
81 The jaguar was originally foun d from southwestern United States to southern Argentina, occupying approximately 19.1 million km 2 Currently, their range has contracted by 46% to 8.75 million km 2 (Sanderson et al. 2002b; Seymour 1989) Although rarely found in areas above 1,200 m, jaguar s can live in a variety of ecosystems including xeric areas in Mexico, pine forests in Belize, the Venezuelan llanos, the Pantanal wetlands and Amazonian rain forests (Sunquist & Sunquist 2002) Although jaguars are opportunistic predators, feeding on more than 85 species, they prefer large prey (i.e., body size >15 kg) such as brocket deer ( Mazama spp.), capybara ( Hydrochoerus hydrochaeris ), caiman ( Cayman crocodilus ) and especially white lipped and collared peccaries ( Tayassu pecari and Pecari tajacu ) (Ar anda 2002; Azevedo 2008; Emmons 1987; Oliveira 2002; Polisar et al. 2003; Scognamillo et al. 2003; Seymour 1989) However, in some parts of their range jaguars live principally on smaller prey, such as armadillos ( Dasypus novemcinctus ), pacas ( Cuniculus pa ca ) and coatis ( Nasua nasua ) (Novack et al. 2005; Rabinowitz & Nottingham 1986) main stronghold for its conservation (Sanderson et al. 2002b) Also, Amazonia has been occupied f or the last 10,000 years by various indigenous groups and currently more than 370 indigenous groups remain scattered throughout the region (Dugelby & Libby 1998; RAISG 2009; Roosevelt et al. 1996) Although an important proportion of Amazonia is under nati onal systems of protected areas or indigenous reserves (e.g., approximately 2.3 million km 2 or 43% of Brazilian Amazon), this region faces numerous threats (Kirby et al. 2006; Walker et al. 2009) Amazonia is threatened by increased road development to acc ess resources such as soils for agriculture, oil reserves,
82 minerals and timber (Finer et al. 2008; Kirby et al. 2006) Roads are major drivers of deforestation and are associated with detrimental ecological effects (Forman & Alexander 1998; Forman et al. 2 003; Laurance et al. 2004) Roads also promote wildlife extraction by increasing the proportion of accessible landscape and facilitating market participation by rural inhabitants (Chapter 2, Laurance et al. 2006; Sierra et al. 1999; Wilkie et al. 2000) La rge game species, such as ungulates, are among the most exploited game by Amazonian hunters, who practice subsistence hunting, as well as by those who use wildlife as a commodity (Chapter 2, Bodmer et al. 1997; Peres 1996; Vickers 1991) Density of large c arnivore populations is closely related to the abundance of prey and particularly to the availability of large bodied species (Fuller & Sievert 2001; Karanth et al. 2004; Sunquist & Sunquist 1989) My work is based in Yasun Biosphere Reserve (hereafter Ya sun), a large protected area in Ecuador threatened with road development to extract oil (Finer et al. 2009) Yasun is inhabited by the Waorani, an indigenous group whose hunting practices have switched from subsistence to commercial hunting following the creation of roads in the region (Chapter 1). This condition a protected area or indigenous territory threatened with road development to access resources is common across the Neotropics (Finer et al. 2008) In face of the need to conserve carnivores and their prey within protected areas inhabited by humans in Amazonia, I ask: 1) Are decreases in local prey abundance, particularly the more exploited large bodied vertebrates, related to higher human access to the landscape?; 2) Does jaguar density change a cross protected habitat as a result of localized decreases in large bodied vertebrates and prey biomass associated with human access and hunting?; 3) What is the likely impact of
83 increased access to hunters and bushmeat extraction on the persistence of jag uar populations in the long run within a large protected area in the Amazon? Understanding these relationships can inform management strategies to increase the likelihood of success in conserving jaguar populations, as well as the prey base, within inhabit ed Neotropical protected areas. I addressed these questions by comparing jaguar and prey populations across four sites in Yasun with varying degree of access to hunters. Methods Study Area Yasun Biosphere Reserve formed by Yasun National Park (10,000 km 2 ) and the adjacent Waorani Ethnic Reserve ( 8,000 km 2 ) (Figure 4 1). V egetation cover in Yasun is continuous and largely homogeneous, dominated by tall evergreen tropical terra firme forest with canopy h eight betw een 25 40 m (Valencia et al. 2004) Flood plains (varzea) and swamp y areas, generally dominated by the palm Mauritia flexuosa occur along the margins of main rivers E levation in Yasun ranges between 200 300 m, although mountain ridges in the western por tion of the Waorani Ethnic Reserve can reach up to 600 m. Seasons i n Yasun are not clearly marked. A nnual rainfall is close to 3,000 mm N o calendar month has precipitation below 100 mm However, typically rainiest months are from April to June with avera ge precipitation of 300 400 mm, and August and January are the driest months with average precipitation of 100 150 mm (10 year data set from Yasun Research Station PUCE). M ean mon thly temperatures are within 22 34C (Valencia 2004). Yasun is one of the m ost biodiverse areas in the world and the protected area with highest potential for conserving jaguars in Ecuador (Bass et al. 2010; Espinosa et al. In press) Yasun posses s es important oil reserves and the first road to extract these
84 reserves (Auca Road ) was constructed in early 1980 s by Texaco within the Waorani territory. A second 120 km road was created within the park boundaries in early 1990 s by the oil company Maxus A third road was initiated to access reserves located in the core of Yasun Nation al Park by the oil company Petrobras in 2005. The construction of this road was stopped by the Ecuadorian government in the same year. However, oil concessions currently occur within Yasun boundaries, therefore, the threat of road building persists (Finer et al. 2010; Finer et al. 2009) Yasun historically has been inhabited by the Waorani who are scattered throughout the entire bioreserve. Traditionally, the Waorani were a semi nomadic group strongly dependent on forests resources such as plant fibers, fruits, and especially bushmeat. To procure game they used blowguns for arboreal species, such as woolly ( Lagothirx poeppigii ) and spider ( Ateles belzebuth ) monkeys, and spears for larger terrestrial game such as white lipped peccaries ( Tayassu pecari ) and tapirs ( Tapirus terrestris ) (Yost & Kelley 1983) After road development, some Waorani settled along the margins of the roads, replaced blowguns and spears with shotguns, and abandoned subsistence hunting practices to harvest large quantities of bushmeat regularly traded in local markets (Chapter 2, Surez et al. 2009) More than 53,000 kg of meat were extracted in one year during my study by approximately 220 Waorani living along the Maxus Road in Yasun. About 90% of that biomass came from ungu lates, mai nly peccaries (Chapter 2). Study Design I hypothesized that the occurrence and biomass of prey are inversely related to accessibility to the landscape and markets by hunters. I also hypothesized that reductions of large bodied prey cause local declines i n jaguar density. For this last
85 hypothesis, I assumed jaguar populations are regulated by large prey. I predicted that in more accessible areas occurrence and biomass of prey would be lower and, consequently, I would observe a lower local density of jaguar s. To evaluate these predictions, I surveyed prey and jaguar populations in four sites with varying degree of access. Ordered from the least to the most accessible, these sites were Lorocachi, Tiputini, Keweriono, and Maxus Road (Figure 1). I grouped the first two sites, Lorocachi and Tiputini, as low access and the last two sites, Keweriono and Maxus Road, as high access. The Lorocachi site was adjacent to the Curaray River o n the southern border of Yasun and adjacent to the Kichwa community of Lorocach i and the 300 man army base BS 48 Sangay (Figure 4 1) Lor ocachi is a small community (<12 0 inhabitants) that was formed with the creation of BS 48 Sangay in 1953. This area is accessible to outsiders only by an army airplane that supplies the base. To mee t their protein demands, the Kichwa in Lorocachi fish, raise poultry and practice subsistence hunting. Hunting has been regulated by creating a no take zone located beyond 10 km from the w ( Mitu salvini ) and tapir ( Tapirus terrestris ) is banned and the numbers of white lipped peccaries ( Tayassu pecari ) that can be taken by hunters is limited (Vallejo Real 2007) Army personnel are not allowed to hunt, although wildlife is sporadically kill ed when found within the base facilities (SE personal observations). My survey area extended from 3 to 21 km from the community of Lorocachi and was access ible to hunters only by foot The maximum distance hunters walk to procure game in Amazonian landscap es is about 7 9 km
86 (Chapter 2, Peres & Lake 2003) More than 50% of the study area was out of the reach of hunters and within the no take zone (Chapter 2, Peres & Lake 2003) The Tiputini site was located in the Northern margin of Yasun and only accessib le by the Tiputini River (Figure 4 1). Although not inhabited, this area is sporadically visited by Waorani and Kichwa hunters. Waorani hunters enter t h rough the Tiputini River from the west (Maxus Road) and Kichwa hunters come from settlements close to th e intersection of the Tiput ini and Napo Rivers in the east. T o access this study site hunters must travel 3 5 hours from the nearest settlements using dug out canoes anim al parts) and direct observations of hunters showed that this area is used to extract bushmeat for both subsistence and commercial purposes (S.E., personal observations) The Keweriono site was within t he Waorani Ethnic Reserve, in the western portion of Y asun (Figure 4 1), and included the Waorani communities of Keweriono (60 inhabitan ts) and Apaika (10 inhabitants) along the Shiripuno River. Keweriono and Apaika are 25 and 15 km from the Auca Road, respectively, and accessible by river (3 6 hrs) or by fo ot (~ 8 hrs walk). Hunting in this area has been continuous since the establishment of Keweriono in 1989 principally for subsistence purposes (Mena Valenzuela et al. 2000) However, s moked bushmeat is sold occasionally in the nearest towns which occur alo ng the Auca Road, by hunters who have had exceptionally good kills (Lu 1999) The Maxus Road site was located along this road in the northern border of Yasun (Figure 4 1). This site was within the hunting areas of four small Waorani settlements (Guiyero, Tiwe, Ganketa and Timpoka) with a combined population of 70 (Figure 4 1).
87 These settlements formed in the early 1990s when the Maxus Road was constructed for oil extraction. Waorani along this road harvest bushmeat for subsistence and for trade in local m arkets outside Yasun (Chapter 1). I assumed direct kills of jaguars had a minimal effect on jaguar abundance compared to reduction of prey by bushmeat extraction. The jaguar is an important symbol for the Waorani. They strongly believe that shamans (loca l healers), highly respected individuals within Waorani society, can alternate between human and jaguar forms. They also believe elder and respected Waorani warriors become jaguars after their death. As a result of these beliefs, jaguars are rarely killed by Waorani hunters. Kichwa in Lorocachi said they do not shoot jaguars unless they feel the cat is going to attack them. They also said jaguars commonly retreat in another direction when encountered. As livestock is not raised by the Waorani or the Kichwa in Lorocachi, conflicts due to predation did not exist in my study sites. Camera Trapping To measure jaguar and prey abundance, I conducted semi systematic sampling with remotely triggered film cameras BU) equipped with a passive he at and motion sensor At each of my four study sites, I created 44 67 km of transects ( total 215 km) and established 23 26 camera trap stations 2 3 km apart within a polygon of approximately 100 km 2 (Figure 4 1, Table 4 1). The minimum home range area repo rted for jaguars is 10 km 2 for a female (Rabinowitz & Nottingham 1986) Therefore, my camera spacing ensured that every jaguar within a study area would have some probability of capture (Rabinowitz & Nottingham 1986; Silver et al. 2004) A camera trap stat ion consisted of two cameras on opposite sides of a transect so that both flanks of an animal could be photographed, thus increasing the probability of
88 identifying jaguars from their rosette patterns (Appendix G, Figure G 1). I mounted cameras on trees at 30 40 cm above ground and tested them to ensure capture of prey species as small as an agouti ( Dasyprocta fuliginosa ) or armadillo (e.g., D. novemcinctus ). I placed trap stations in optimal sites (e.g., where I had previously observed jaguar tracks or wher e prey trails were present). Camera traps were only placed in terra firme forest to prevent equipment flooding in case of heavy rain. I programmed camera traps to work 24 hrs per day and with a lag time (i.e., minimum time allowed between pictures) of one minute. I visited cameras every 10 15 days to replace film and check for correct functioning. Camera trap grids functioned for 90 consecutive days with few camera failures in Maxus Road, Lorocachi and Keweriono (Table 4 1). I had a higher rate of camera fa ilure in Tiputini (Table 4 1). To have comparable trapping effort at Tiputini, I left the cameras to work for 10 additional days. I conducted surveys during the driest months because heavy rains increased logistical constraints and camera failure (Table 4 1). Analyses of Prey Availability I used two measures to estimate prey availability for jaguars: 1) prey occurrence measured as probability of site occupancy and 2) biomass per camera trap station (MacKenzie et al. 2006) Probability of site occupancy may reflect chances of jaguars encountering prey at a particular site and prey biomass is related to carnivore abundance (Fuller & Sievert 2001) To evaluate prey occurrence, I developed single season site occupancy models for the main terrestrial prey specie s using program PRESENCE2 (Hines 2006; MacKenzie et al. 2002) A major assumption of site occupancy models is the state of sites species present or absent does not change during survey. Because medium sized and large mammals are highly mobile, they could
89 enter and leave the survey area during the sampling period. Therefore, results should ( MacKenzie et al. 2006 ) I use the site occupancy analysis for species in four areas at Y asun Biosph ere Reserve presented in Chapter 3, where I developed separate occupancy models for white lipped peccary, collared peccary ( Pecari tajacu ), tapir, red brocket deer ( Mazama americana ), grey brocket deer ( M. gouazoubira ), paca, agouti and armadil lo. These species have been reported as main prey for jaguar in numerous studies (e.g., Aranda 2002; Azevedo 2008; Chinchilla 1997; Emmons 1987; Novack et al. 2005; Polisar et al. 2003; Rabinowitz & Nottingham 1986; Taber et al. 1997) Also, e xcluding arma dillos, these species are among the most intensively hunted species by nati ve Amazonians in Yasun (Chapter 2). To estimate biomass per camera trap station per night, I added the body weights of all animals recorded at a camera trap station in independen t photographs and divided the total sum by the number of nights the station was active. I considered photographs of the same species to be independent when a minimum of 1 hour occurred between detections. Average weight of each species was obtained from an imals hunted along the Maxus Road or from the database PanTHERIA (Chapter 2, Jones et al. 2009) For my analysis I included all recorded species of terrestrial mammals and birds with body weight 1 kg, excluding puma ( Puma concolor ), the other large cat in the area (Table 4 2). I tested for differences in biomass per day per camera trap station (kg/day/station) between the four areas with Kruskal Wallis followed by a multiple comparison test with (Giraudoux 2011; Siegel & Castellan 1998 )
90 Estimating Jaguar Density I used two approaches to estimate density of jaguars at each of my four sites. First, I estimated density as D = / ETA where is an estimate of population size derived from a capture recapture model of closed populations, and ETA is an estimate of effective trapping area. This method has two critical assumptions: 1) demographic closure (i.e., no migration, births or deaths) during the study period; and 2) all individuals have a probability greater than zero to be captured (Otis et al. 1978) I used program CAPTURE to estimate and test for population closure (Otis et al. 1978; Rexstad & Burnham 1991) To estimat e I used M h model with a jackknife estimator because this model permits individuals to have different capture probabilities which has more biological meaning than assuming capture homogeneity (Karanth & Nichols 1998) I developed one matrix per site wh ere rows correspond to i individuals identified at each site and columns to j trapping occasions. A trapping occasion was defined as three consecutive trapping nights. The entry in the X ij matrix was 1 if an individual was recorded, or 0 if no individual w as recorded during a trapping occasion. No jaguars were captured in the additional time allotted for camera failure in Tiputini. Therefore, to estimate jaguar density, I only included adult individuals captured within a 90 day period in all four sites. Thi s method for estimating density was first developed to study tiger populations and later widely applied to measure jaguar abundance in the Americas (Karanth & Nichols 1998; Maffei et al. 2011; Silver et al. 2004) Although the /ETA method has some major limitations, this method has been widely used to estimate jaguar density in other areas allowing for comparisons of results (see Maffei et al. 2011) /ETA limitations are due to high dependency of estimates on trap spacing and the number of
91 recaptures o f individuals (Krebs 1999) which can be problematic when studying large felids that have large home ranges and naturally low abundance. Additionally, although estimation of is straightforward, estimation of ETA is difficult when the sampling area has no physical boundaries. To address this problem, the common approach is to put a boundary strip, A W around the sampling area which typically equals either half or the full mean maximum distance moved (MMDM) by individuals trapped at different sites (Karan th & Nichols 1998; Soisalo & Cavalcanti 2006; Wilson & Anderson 1985) However, it is not clear which of these two alternatives (MMDM or MMDM) provides a better estimate, and probably this will vary among species and areas. Telemetry studies indicate the MMDM obtained from camera traps underestimates distances moved by jaguars and recommend the use of the full MMDM to estimate density (Soisalo & Cavalcanti 2006) To estimate ETA, I used the full MMDM calculated from the pooled maximum distances of all four study sites with only recaptures at different camera stations included. I obtained ETA by placing a buffer of the full MMDM around each camera station and calculated the standard error of density with method provided in Karanth and Nichols (1998) I used spatially explicit capture recapture models (SECR) as a second approach to estimate density. SECR models have two major advantages over the traditional / ETA method. First, they do not rely on estimating an ETA and, therefore, are particularly useful in the absence of geographic closure. Second, they implicitly take into consideration heterogeneity in capture probability of individuals. SECR analyses can be performed using maximum likelihood (ML SECR) or Bayesian (Bayesian SECR)
92 inference (Borchers & Effo rd 2008; Efford 2004; Efford et al. 2009; Royle et al. 2009; Royle & Young 2008) ML SECR models use maximum likelihood to estimate density by integrating two submodels. One submodel describes the distribution of animals (i.e., activity centers or home ra nge centers) within an area that includes the trapping array and is simulated by a homogeneous spatial Poisson process. A second submodel, a spatial detection function, describes the probability of catching an animal as a function of the distance between t (Efford et al. 2009) The second submodel requires a minimum of two parameters: 1) g 0 the detection probability trap equals zero; and 2) which represents the spatial scale at which detection probability decreases. Both parameters can be modeled with covariates and using a half normal, hazard or exponential distribution (Efford et al. 2004) I developed ML SECR models for each site using program DENSITY 4.4 (Efford et al. 2004) Due to limited captures and recaptures, I did not include covariates to model the detection function and assumed it was the same for all captured individuals [ g 0 (.), (.)]. DENSITY creates an area of integration th at is filled with uniformly distributed points that represent potential home range centers of animals captured in the trap array. For this purpose, a buffer is placed around the trap array envelope. After a given distance, further increases in this buffer will have no effect on density estimates. I developed five models for each site using buffers of 0.1, 1, 5, 10, and 15 km. Bayesian SECR models use similar principles to ML SECR. However, because Bayesian SECR models use non asymptotic inference, they ar e less sensitive to
93 situations where sample sizes are small, such as in large carnivore studies (Royle et al. 2009; but see Efford 2011) area of integration in ML SECR) is created with uniformly distributed points representing n observed individuals to obtain a total of M individu als. M is assumed to contain the true population number, N within the state space. In the context of Bayesian inference, M can be interpreted as an upper bound of an uninformative uniform prior (0, M ) for N (Gopalaswamy et al. 2011; Royle et al. 2009) To develop B SECR models I used R package SPACECAP (Gopalaswamy et al. 2011) An important limitation of this kind of analysis is the demand for computer power. For example, each model took between 16 22 hrs to run on a PC equipped with an Intel i7 1.60 GHz processor, and the ML SECR models generally took about 1 3 minutes to run. For this reason, and after testing the effect of different buffers on ML SECR models, for each site I ran only one null model that described the detection function with a half norma l distribution and capture encounters with a Bernoulli process. I created a state space area with a 15 km buffer around each trap array envelope. For data augmentation I used a density of 25 individuals per 100 km 2 which corresponds to about six times the highest density estimated using / ETA As SPACECAP runs only one chain at a time, I ran each model twice (60,000 iterations and burn in of 2,000) and assessed convergence by visual analysis of plotted chains.
94 Results Prey Abundance Occurrences of ungulat e species were significantly higher in the two least accessible sites, Lorocachi and Tiputini, than in the most accessible Keweriono and Maxus Road (Figure 4.2). The most extreme case was white lipped peccaries, scarcely detected in Maxus road and never de tected in Keweriono. The occurrence of armadillos and pacas was also higher in the two least accessible sites, whereas the occurrence of agoutis was similar in all four sites (Figure 4 2). In general, the occurrence of prey was best predicted by and positi vely related with distance to settlements, that is, the farther from settlements, the higher the probabili ty of encountering prey (Chapter 3, Table 3 1). Prey biomass was significantly higher at the two least accessible sites compared to the two most acce ssible sites (Tables 4 2 4 3 ; Appendix H Figure H 1). Biomass (kg/day/station) did not differ significantly between the two least accessible sites (Lorocachi and Tiputini) or between the two most accessible sites (Keweriono and Maxus Road) (Table 4 3). H owever, biomass differed significantly when compared between least and most accessible sites (Table 4 3). Average prey biomass in Lorocachi was 10.25 kg/day/station ( SE = 1.38) and in Tiputini, 16.58 kg/day/station ( SE = 1.43), whereas in Keweriono it was 3.65 kg/day/station (SE = 0.42) and in Maxus Road, 4.98 kg/day/station ( SE = 0.69). Ungulates were the most important contributors of biomass (kg/100 trap nights) among the 25 species analyzed at a site level. They accounted for 82 and 89% of biomass in t he low access sites, Lorocachi and Tiputini, and for 67 and 52% of the biomass in the high access sites, Maxus Road and Keweriono, respectively (Table 4 2). Ungulate biomass ranged from a low of 330 kg/100 trap nights in Keweriono to a high of
95 1,463 kg/100 trap nights in Tiputini (Table 4 2). White lipped peccaries accounted for 20 and 37% of biomass in low access sites. However, they accounted for only 3% of biomass in the high access Maxus Road and were not found in Keweriono. In contrast, collared peccar ies were the most important contributor to biomass in Keweriono (18% of total biomass) and they were also important in Maxus Road (18%) and Lorocachi (19%). They were less important in Tiputini (9%). Tapirs were also an important contributor to total bioma ss, which accounted for 31, 32, 38 and 17% of total kg/100 trap nights in Lorocachi, Tiputini, Maxus Road and Keweriono, respectively (Table 4 2). The amount of biomass provided by non ungulates was similar across the four sites, ranging from 166 to 184 kg /100 trap nights. Pacas, agoutis and armadillos, important medium sized prey, contributed 5 and 4% of available biomass in Lorocachi and Tiputini, and 9 and 10% in Maxus Road and Keweriono, respectively. Jaguar Abundance in Yasun A total of 30 adult j aguars (18 males, 7 females and 5 animals whose sex determination was not possible) were photographed during the 90 day survey periods at each site (Table 4 4). Non spatially explicit and spatially explicit models were, in general, consistent with respect to density estimates for the least and most accessible sites. The highest jaguar abundance occurred in the least accessible site, Lorocachi, and the lowest jaguar abundance in the most accessible site, Maxus Road (Table 4 6, Figure 4 4). The highest es timated with CAPTURE was in Lorocachi and the lowest at Maxus Road (Table 4 6). Accordingly to the test of closure, populations in the four sites were closed (Table 4 4). However, Otis et al. (1978) caution about the limitations of this test in rejecting c losure when sample sizes are small. To estimate the ETA necessary to
96 calculate density ( / ETA ), I used the maximum distance moved from 15 jaguars trapped in at least two different camera trap stations (Table 4 4). I obtained an overall MMDM of 6.08 SE 0.73 km that I applied as a buffer around each camera location. ETA varied from 458 486 k m 2 per site (Table 4 5). The highest density using / ETA occurred in Lorocachi, which was about six times higher than the estimate for Maxus Road (Table 4 5). However, jaguar density in the second least accessible site, Tiputini, was similar to jaguar den sity in the second most accessible site, Keweriono (Table 4 5). Density estimates using ML SECR were higher when using short buffer distances around trap array envelopes (i.e., 100 and 1000 m) and stabilized at Maxus Road, Keweriono and Lorocachi when buf fer distance was increased to 5 15 km (Figure 4 3). Density estimates in Tiputini decreased continuously as buffer distance was increased suggesting a problem with using ML SECR with this dataset (Figure 4 3). The highest density estimate with ML SECR was at the least accessible site, Lorocachi, which was about four times higher than estimate in the most accessible site, Maxus Road (Table 4 5). Jaguar density in the second most accessible site, Keweriono, was similar to Maxus road. Compared to the tradition al approach ( / ETA ), density estimates from ML SECR were about 60% higher in Maxus Road, 50% lower in Keweriono and about the same in Lorocachi (Table 4 5). However, based on standard errors, results from these two methods did not differ within sites (Figure 4 4). Ba yesian SECR provided similar results to the previous two methods. The highest jaguar density estimate was in Lorocachi and the lowest in Maxus Road (Table 4 5, Figure 4 4). Markov chains were well mixed, or converged, in models for Maxus Road, Keweriono an d Lorocachi, suggesting estimates reached a stable equilibrium (Figure 4
97 5). Markov chains at Tiputini did not converge suggesting an unreliable estimate for this site (Figure 4 5). Therefore, I did not estimate population size at this site with Bayesian S ECR models. When compared to the traditional approach, Bayesian SECR estimates were 135% higher in Maxus Road, 10% lower in Keweriono and 46% higher in Lorocachi (Table 4 5, Figure 4 4). However, based on standard errors, density estimates from Bayesian SE CR did not differ significantly from estimates using / ETA and ML SECR. Discussion Availability of Prey and Jaguar Abundance Both measures of prey availability, occurrence and biomass (kg/day/station), were higher at the two least accessible sites, Lorocachi and Tiputini, than at the two most accessible si tes, Maxus Road and Keweriono. The occurrence of white lipped peccaries, an important prey for jaguar and the most harvested species by Waorani, was dramatically reduced i n most accessible sites (Chapter 2). These results were predicted given that hunting is more intensive around human settlements and sources of access (Chapter 3, Peres & Lake 2003; Sirn et al. 2004; Smith 2008) Additionally, these observations concur with past studies showing decreases in the abundance of most hunted species in hunted ar eas across the Neotropics (Bodmer et al. 1997; Hill & Padwe 2000; Naranjo & Bodmer 2007; Peres 1996) Differences in prey biomass (kg/day/station) between the four sites in Yasun occurred for ungulates with biomass noticeably reduced in the most accessib le sites. Biomass of non ungulates was virtually the same in all four sites. These patterns are likely caused by the intensive extraction of ungulates, particularly peccaries, by the Waorani in Keweriono and Maxus Road (Chapter 2, Franzen 2006; Mena Valenz uela et
98 al. 2000; Surez et al. 2009) There is a close relationship between the abundances of large cats and large prey, particularly ungulates (Karanth et al. 2004; Sunquist & Sunquist 1989) Although jaguars are considered generalists, ungulates are imp ortant part of their diet (Aranda & Sanchez Cordero 1996; Azevedo 2008; Oliveira 2002) Jaguar densities based on / ETA were up to six times higher in the remote Lorocachi than in the most accessible site, Maxus Road. Using both, ML and Bayesian SECR methods, jaguar density was about four times higher in Lorocachi than in Keweriono and Maxus Road. These two last site s were the most accessible and presented 20 50% of the prey biomass (kg/100 trap nights) available in the more remote Lorocachi and Tiputini. However, an unexpected result was low capture rates of jaguars at Tiputini, where ungulate occurrence and prey bio mass were the highest. The Tiputini site was located in the northern border of Yasun, in an area in relatively close vicinity to settlements located along the Napo River and mainly formed by Kichwa people (Figure 4 1). Low density of jaguars in Tiputini m ay be explained by several factors. Kichwa people along the Napo River participate in commercial hunting. Therefore, jaguars may be killed for skins or other body parts at this edge of Yasun, and hunting of jaguars maybe the cause of reduced density here (Schlaepfer et al. 2002; Woodroffe & Ginsberg 1998) Landscape composition and configuration are important determinants of species distribution and abundance. Although I selected relatively similar areas, where habitat loss and fragmentation do not exist ( or is minimal on the Maxus Road), sites varied in their habitat composition. Wetlands, such as those dominated by the palm Mauritia flexuosa are strongly associated with white lipped peccaries and tapir across Amazonia (Bodmer 1990; Fragoso 1999) Palm ag gregations corresponded to
99 0.9, 15.7, 0 and 8.3% of the area covered by minimum convex polygon of camera trap arrays in Lorocachi, Tiputini, Keweriono and Maxus Road, respectively (MAE 2012) Variation in habitat composition could explain the high biomass of prey in Tiputini. However, jaguar density should have been higher as well, which was not the case. Heterogeneous detection probability across sites, which was not modeled for jaguar density estimation because of limited data, also may play a role in obs erved patterns (Royle et al. 2007) Finally, given the limited number of sites for comparing jaguar densities, variation from stochastic process could be important. For example, a fraction of Tiputini may have been controlled by a dominant male jaguar that deterred the presence of other individuals. Considering the above mentioned methodological constraints, my overall results provide some support for the hypothesis that large prey, represented by ungulates in the terra firme forest of Yasun, are importan t to maintain higher jaguar population numbers. In Yasun, roads increased access of Waorani hunters to the landscape and to markets, which augmented the spatial extent of hunting and led to higher extra ction rates of wildlife (Chapter 2). Also, higher ext raction rates were associated with increased exploitation of ungulates, particularly peccaries (Chapter 2). Finally, as presented here and analy zed in further detail in Chapter 3, increased access to Yasun was associated with reductions in prey occurrence and biomass, which was largely driven by ungulates. Higher access to hunters by new road development likely will seriously impact the abundance of jaguars by decreasing its main prey. Although jaguars are described as generalist predators, large bodied s pecies constitute main dietary items throughout most of their range, indicating jaguars
100 frequently depend on large vertebrates for their survival. In Central America, jaguars prey mainly on medium sized mammals, such as armadillos (Novack et al. 2005; Rabi nowitz & Nottingham 1986) However these previous studies in Central America were conducted in areas within the Maya region, which have sustained a large human population since pre Hispanic times (Gomez Pompa & Kaus 1999) A coexistence of humans and wild life for thousands of years may have led organisms to adapt to these highly anthropogenized systems. Environmental change can lead to rapid phenotypic change of organisms permitting their adaptation to new conditions (Ozgul et al. 2009) We can speculate t hat the smaller size of jaguars ( P. onca goldmani ) in this region of Central America could partially respond to adaptations of an environment where large prey was depleted by intensive hunting (Pocock 1939; Seymour 1989) In contrast, new evidence indicate s that the western Amazonia has been historically inhabited at low human population densities, with native Amazonians scattered in small and non permanent settlements (McMichael et al. 2012) This finding implies hunting of large game was never intense in this region. In light of this new evidence, a sudden decrease of important prey may negatively impact jaguar populations in western Amazonia. Worldwide, space is an important limiting factor for the conservation of large carnivores and as natural habitats continue to be reduced, protected areas become more important for their survival (Karanth & Chellam 2009; Treves & Karanth 2003) Therefore, managing protected areas so they maintain adequate prey populations is vital for maintaining viable carnivore popu lations. Higher Accessibility and Jaguar Conservation in Yasun Biosphere Reserve If the density of jaguars in Lorocachi is representative of the region under well conserved conditions, Yasun has the potential for long term jaguar conservation.
101 According ly to Eizirik et al. (2002) a jaguar population of 300 and 650 individuals has a 97 100% probability of persisting for the next 100 and 200 years, respectively. However, the probability of persistence of jaguar populations of 100 200 individuals decreases significantly. For example, in remnant populations of jaguars elsewhere in South America, an unmanaged jaguar population of 100 is projected to have only a 25% probability of persistence in a 100 year period (Eizirik et al. 2002) In an ideal conservation scenario (i.e., overall jaguar density similar to Lorocachi), Yasun National Park (10,000 km 2 ) could hold a population of 400 600 jaguars whereas the entire Yasun Biosphere Reserve (18,000 km 2 ) could maintain 700 1,000 individuals. These jaguar populati on numbers are based on / ETA and Bayesian SECR models, which provided the lowest and highest density estimates for Lorocachi, respectively. Conversely, if more roads are built within Yasun to extract resources such as oil, jaguar abundance certainly will decline. If jaguar den sities reach those of the Maxus Road and 150 in the park or 100 300 individuals in the entire biosphere reserve. Currently, the jaguar population within Yasun Biosphere Reserve seems to have a high chance of persisting for the next 200 years. However, reaching a jaguar population level of less than 200 individuals is a possibility considering a worse case scenario in the currently remote, but threatened Yasun (see Finer et al. 2010) Additionally, i mportant consideration must be given to the role of indigenous lands as conservation areas. The management of indigenous areas relies on the decisions of their original occupants who can align, or not, with conservation objectives (Alcorn 1993; Redford & S tearman 1993) In the population estimates above, I include the area of the Waorani
102 Ethnic Reserve (8,000 km 2 ) as available habitat for jaguars. However, habitat conditions could change in the future if the Waorani decide to manage their land differently. Under that scenario, maintaining the integrity of Yasun National Park becomes even more important for future jaguar conservation in Ecuador. A first start to achieve a better management of Yasun is to prevent or limit further road development in the regi on and controlling current exploitation of wildlife for commercial purposes (Chapter 2).
103 Table 4 1. Survey effort with camera trap stations at four study sites in Yasun Biosphere Reserve. Lorocachi Tiputini Keweriono Maxus Road Camera trap s tations 26 25 23 26 Starting survey date 07/23 /2008 12/01 /2008 08/15/2009 12/01 /2007 Ending survey date 10/20 /2008 03/10/ 2009 11 /1 2 /2009 02/28 /2008 Trap nights in 90 day period 2 275 1 901 (2 136 1 ) 1 972 2 241 Camera failure 2.78% 15.51% 4.73% 4.23% Minimum convex polygon of camera trap array (km 2 ) 110 110 106 104 1 Trap nights in 100 day period.
104 Table 4 2. Biomass (kg/100 trap days) of 25 potential prey species at four study sites in Yasun Biosphere Reserve. Species Lorocachi Tiputini Keweriono Maxus R oad Ungulates Tayassu pecari 198.99 600.95 0.00 13.01 Pecari tajacu 195.24 158.99 66.73 87.80 Tapirus terrestris 310.61 527.99 59.68 186.18 Mazama americana 117.37 142.68 60.14 33.72 Mazama gouazoubira 10.78 28.72 0.00 9.99 Mazama sp. 0.89 3.64 0.00 0.00 Subtotal ungulates 833.88 1 462.98 186.55 330.71 Non ungulates Aburria cumanensis 0.00 0.08 0.00 0.04 Atelocynus microtis 3.74 2.43 8.24 3.00 Cuniculus paca 9.69 16.48 9.52 11.27 Dasyprocta fuliginosa 31.96 29.78 24.57 22.64 Dasy pus novemcinctus 4.50 12.45 3.70 10.56 Didelphis marsupialis 0.83 4.33 1.38 1.31 Eira barbara 1.35 4.98 2.52 1.49 Leopardus pardalis 57.03 40.42 74.67 49.30 Leopardus wiedii 0.40 1.22 0.61 0.39 Mitu salvini 19.54 10.63 4.32 1.92 Myrmecophaga tridacty la 14.42 11.03 0.00 21.21 Nasua nasua 2.76 2.04 1.06 1.66 Nothocrax urumutum 0.22 0.51 1.02 0.22 Penelope jacquacu 0.04 0.97 0.50 0.51 Priodontes maximus 21.49 23.61 26.69 21.08 Procyon cancrivorus 0.00 0.58 0.00 1.11 Psophia crepitans 13.26 13.77 15 .13 15.92 Puma yagouaroundi 1.12 1.43 1.29 0.55 Speothos venaticus 0.51 0.00 0.00 0.25 Tamandua tetradactyla 1.37 1.20 0.68 1.53 Subtotal non ungulates 184.25 177.93 175.90 165.96 Grand Total 1 018.14 1 640.91 362.46 496.67
105 Table 4 3. Multiple co mparison test following a Kruskal Wallis of prey biomass (kg/day/station) at four sites in Yasun Biosphere Reserve. Output obtained R package. Tested pair Observed difference Critical difference Difference (p value= 0.5) Keweriono Lorocac hi 33.3 6 21.9 1 TRUE Keweriono Maxus Road 8.1 3 21.9 1 FALSE Keweriono Tiputini 53.9 8 22.11 TRUE Lorocachi Maxus Road 25.23 21.2 3 TRUE Lorocachi Tiputini 20.62 21.4 4 FALSE Maxus Road Tiputini 45.85 21.4 4 TRUE
106 Table 4 4. Capture history of 30 jaguars u sed in abundance estimation in the four areas in Yasun Biosphere Reserve LR = Lorocachi, TI = Tiputini, KW = Keweriono, MR = Maxus Road, MDM = maximum distance moved. Individual Sex Captures MDM LR 1 Male 4 8 990 LR 2 Male 1 7 531 LR 3 Female 1 0 LR 4 Male 2 2 529 LR 5 Male 4 4 526 LR 6 Male 5 4 464 LR 7 Undefined 1 0 LR 8 Male 2 0 LR 9 Male 4 4 078 LR 10 Male 1 0 LR 11 Female 1 0 LR 12 Male 1 0 LR 13 Undefined 1 0 TI 1 Female 2 9 458 TI 2 Male 2 8 838 TI 3 Female 1 0 TI 4 Male 2 2 685 TI 5 Undefined 1 0 TI 6 Undefined 1 0 KW 1 Male 3 8 100 KW 2 Female 3 2 847 KW 3 Male 2 0 KW 4 Male 3 11 715 KW 5 Undefined 1 0 KW 6 Male 1 0 KW 7 Female 1 0 KW 8 Male 1 0 MR 1 Female 2 6 011 MR 2 Male 3 5 055 MR 3 Male 2 4 393
107 Table 4 5. Estimates of jaguar density using non spatial models and spatially explicit models. Method for density estimate Site Lorocachi Tiputini Keweriono Maxus Road /ETA Capture probability 0.05 0.04 0.05 0.08 Closur e test P 0.92 0.07 0.31 0.75 SE 19 5.32 7 2.93 10 4.74 3 1.19 ETA (km 2 ) 486 46 7 458 463 D (jaguars/100 km 2 ) SE 3.91 1.11 1.50 0.63 2.18 1.04 0.65 0.26 CV (%) of D 28.4 42.0 47.8 40.0 ML SECR D (jaguars/100 km 2 ) SE 4.1 1.67 NA 1.08 0.61 1.04 0.89 CV (%) of D 40.7 5 6.5 85 6 Bayesian SECR D (jaguars/100 km 2 ) SE 5.70 2.08 NA 1.96 1.05 1.52 1.44 CV (%) of D 36.4 53.6 94.7 = population size estimated with program CAPTURE (Model M h ); ETA = effective trapping area; D = density; CV = coefficient of variation; ML SECR = spatially explicit model using maximum likelihood inference; Bayesian SECR = spatially explicit model using Bayesian inference.
108 Figure 4 1. Camera trap arrays to measure jaguar density in four sites in Yasun Biosphere Reserve. Sites are arranged from least accessible (Lorocachi) to most accessible (Maxus Road).
109 Figure 4 2. Probability of site occupancy of ungulates and three medium sized mammals reported to be impor tant jaguar prey at four sites in Yasun Biosphere Reserve. Sites are ordered from least to most accessible. Whiskers on top of bars are 1 SE. Figure d eveloped from results of Chapter 3, Table 3 3.
110 Figure 4 3. Density estimates with ML SECR using var ying buffer distances of camera trap envelope. Figure 4 4 Jaguar d ensity estimates (N/100 km 2 SE) using / ETA and two spatially explicit models (ML SECR with program DENSITY and Baysian SECR with R package SPACECAP). Sites are arranged from least accessible (Lorocachi) to most accessible (Maxus Road).
111 A. Lorocachi B. Tiputini C. Keweriono D. Maxus Road Figure 4 5. Markov chains of total individual jaguars, Nsuper, estimated within the state space by Bayesian SECR models. Horizontal axis indicates number of iterations. Only the last 30,000 iterations are plotted.
112 CHAPTER 5 CONCLUSION Gener al Conclusions My study demonstrates that increased access to inhabited protected areas within Amazonia by road development will result in a series of cascading effects with deleterious consequences for the conservation of large game species important to b oth traditional hunters and jaguar populations. Road development within Yasun was related to a two fold increase in the spatial extent of hunting by humans. This increase was linked to synergistic effects between means of access, roads and rivers, and use of markets by Waorani. Higher extraction and trading rates of bushmeat were associated with closer pr oximity to markets. My results concur with previous studies indicating roads increase extraction rates of wildlife by their association with markets, and that commercial sale of bushmeat by indigenous hunters is related with distance to markets (Sierra et al. 1999; Wilkie et al. 2000) Previous studies have concluded that bushmeat extraction rates along the Maxus Road were above sustainable levels (Franzen 2005) and the extraction rates I measured were even higher than previous rates. The high quantities of bushmeat extracted by Waorani along the Maxus Road are likely maintained by immigration of animals from inaccessible areas functioning as wildlife refu ges (Joshi & Gadgil 1991; McCullough 1996) My results point to the importance of avoiding new infrastructure development inside or near protected areas that can compromise the necessary proportion of protected areas needed as wildlife refuges to maintain sufficiently large game populations to prevent risk of extinction. Additionally, avoiding road development would not only aid in preservation of wildlife but also help
113 maintain a continued supply of bushmeat for the Waorani, an important nutritional resour ce that is highly associated with the Waorani identity. Increased landscape accessibility to hunters clearly influenced the distribution of game across space. When the first data set grouping four areas with different degree of access was used to model spe cies occurrence, the effect of landscape access to hunters (i.e., access defined by distance to settlement and distance to nearest road or river) on game occurrence was better predicted by distance to human settlements than by distance to nearest road or r iver. In contrast to this previous result, in models of species occurrences using the second data set gathered at the Maxus Road (i.e., access defined as distance to settlement and distance to road), distance to road appeared to be a better predictor for g ame occurrence than vicinity to settlements. In areas without roads, most hunting activities will be concentrated within a walking distance from settlements, and for hunters in the Neotropics distances above 7 9 km are prohibitive (Chapter 2; Levi et al. 2 011; Peres & Lake 2003) However, when roads are available, hunters can extend their harvest area along roads, and also increase the use of rivers for the same purpose. Hunters living along the Maxus Road moved up to 40 km from their settlements using road s and rivers to reach hunting areas. Additionally, roads are associated with markets that will lead to higher bushmeat extraction rates (Wilkie et al. 2000) Likely, as a result of these interactions between roads and markets, distance to roads became as i mportant as distance from settlements in predicting the occurrence of game when the site with road was analyzed separately. Species that are social and have low population growth rates (e.g., white lipped peccaries and tapir) were the most affected by high er landscape accessibility to
114 hunters, as corroborated by previous studies on the impact of hunting in Neotropical fauna (e.g., Bodmer et al. 1997; Peres 1996) Therefore, to attain conservation objectives, careful evaluation of possible impacts on wildlif e communities need to be performed prior to new infrastructure development within or near inhabited protected areas in the Neotropics. Jaguar density was low in areas where estimates of prey occurrence and prey biomass were low. Low biomass across sites w as linked with low occurrence of ungulates, which are among the most hunted species by Amazonian hunters (Redford & Robinson 1987; Robinson & Bennett 2000b) In contrast, the highest jaguar density estimate was in the most remote area with high occurrence and biomass of prey. However, in the site with the highest biomass of prey, Tiputini, jaguar density was as low as in the two sites with lower game biomass. Likely, the abundance of jaguar is affected by other factors, for example hunting of jaguars, which can decrease jaguar density even though prey is abundant (Woodroffe & Ginsberg 1998) The small sample size for comparing jaguar densities does not permit conclusive results. However, the considerable differences of jaguar density estimates between the re motest area (high occurrence of large prey) and the most accessible area (low occurrence of large prey) suggest that jaguars depend on large prey in Yasun. These results point to the importance of careful evaluation prior to new transportation infrastruct ure placement within or near inhabited protected areas similar to Yasun, where cascading effects initiated by road building could result in the extirpation of game and jaguar populations. Conservation Recommendations The threat of further infrastructure development within Yasun will continue as long as oil reserves exist within its borders. Construction of the most recent road to extract
115 oil within Yasun was initiated in 2005 by Petrobras to access oil in Block 31. The construction of this road was st opped by the Ecuadorian government in the same year. For the last 5 years, Ecuador has looked for support from the international community to http://yasuni itt.go b.ec ). In addition to protecting wildlife populations and maintaining prey resources for both humans and jaguars, not exploiting ITT will prevent the release of 400 million metric tons of CO 2 to the atmosphere. In exchange for not exploiting ITT, Ecuador asked the international community for 50% of the revenues the country would have if oil were extracted. So far, the Yasun ITT proposal has not received enough support. Also, a project exists to connect the Brazilian port of Belem on the Atlantic Ocean wit h the Ecuadorian port of Manta in the Pacific Ocean. This project will use the Napo River (northern margin of Yasun) as a means of access (Bank Information Center, www.bicusa.org ). This study emphasizes the importance of maintaining a roadless policy within Yasun to protect its rich biological and cultural diversity. The construction of roads should be avoided even if oil extraction increases, for example, to exploit Blocks 31 and ITT. Additionally, the impact of furt her road building near Yasun borders should be carefully evaluated in the context of wildlife exploitation, because roads may increase the use of rivers by local hunters, which can decrease the proportion of areas previously functioning as wildlife refuge s. Given that the Waorani in Yasun, and other indigenous groups in similar protected areas across Amazonia, strongly depend on bushmeat, development of strategies for sustainable management of game populations are critical. Traditional approaches designed to avoid game overexploitation, for example, restricting the access to firearms,
116 establishing hunting quotas, or using trained personnel to enforce the law have proven ineffective in the tropics (Gibson & Marks 1995) For example, even though hunting ins ide protected areas and the trade of bushmeat is forbidden in Ecuador, these actions o ccur on a daily basis in Yasun (Chapter 2, Surez et al. 2009) The ineffectiveness of these approaches in avoiding wildlife overexploitation partly results from a lack of resources to enforce them, but more importantly by the lack of involvement of local inhabitants to manage these resources, who perceive wildlife as an open access resource that is rapidly depleted (Hardin 1968) Community based management is another a pproach that can support conservation goals by devolving resources to the historical occupants of Amazonian protected areas. For example, the well know CAMPFIRE program has been successful in devolving a community based program for the management of wildli fe and other natural resources in Zimbabwe, where communities directly benefit from trophy hunting, while keeping game populations out of extinction risk (Child 1996; Taylor 2006) Game management in Yasun could be conducted under a community based framew ork using spatial controls (McCullough 1996) For example, Waorani could continue using their current harvest area and, by monitoring changes in the catch per unit of effort through time, they can determine if the extent of harvested area needs to be modif ied (McCullough 1996) By effectively managing game populations, Yasun and other large protected areas in Amazonia can increase their likelihood of maintaining viable jaguar populations in the long run.
117 APPENDIX A LIST OF SPECIES HUNTED ALONG THE MAXU S ROAD Table A 1. Complete list of species hunted by 5 Waorani settlements (Guiyero, Ganketa, Timpoka, Dikaro and Oa) between January 2008 and April 2009 along the Maxus Road in Yasun Biosphere Reserve. Taxa n Harvest (kg) Trade (kg) % Traded Columbifo rmes 4 0.6 0.0 0.0 Columba sp. 1 0.1 0.0 0.0 Leptotila rufaxilla 3 0.5 0.0 0.0 Galliformes 436 968.9 65.4 6.7 Mitu salvini 173 692.5 59.8 8.6 Nothocrax urumutum 4 5.4 0.9 16.7 Odontophorus gujanensis 2 1.8 0.4 19.5 Penelope jacquacu 155 165.3 2.3 1. 4 Pipile pipile 102 103.9 2.1 2.0 Gruiformes 35 29.3 3.6 12.4 Psophia crepitans 35 29.3 3.6 12.4 Passeriformes 9 1.5 0.0 0.0 Cyanocorax violaceus 1 0.2 0.0 0.0 Formicarius sp. 1 0.1 0.0 0.0 Myiarchus tuberculifer 1 0.1 0.0 0.0 Psarocolius angustifr ons 2 0.4 0.0 0.0 Psarocolius oseryi 4 0.7 0.0 0.0 Piciformes 154 66.8 2.8 4.2 Campehilus rubricollis 1 0.5 0.0 0.0 Pteroglossus pluricinctus 3 1.4 0.0 0.0 Ramphastos sp. 17 8.8 0.7 7.7 Ramphastos tucanus 133 56.1 2.1 3.7 Psittaciformes 72 59.7 1.1 1.9 Amazona farinosa 10 6.4 0.0 0.0 Ara ararauna 11 11.8 0.2 1.7 Ara macao 42 38.2 0.9 2.4 Ara sp. 2 1.8 0.0 0.0 Brotogeris cyanoptera 1 0.1 0.0 0.0 Pionopsitta barrabandi 1 0.2 0.0 0.0 Pionus menstruus 5 1.1 0.0 0.0 Tinamiformes 41 52.1 0.0 0.0 C rypturellus cinereus 18 22.4 0.0 0.0 Crypturellus sp. 5 5.2 0.0 0.0 Tinamus major 18 24.5 0.0 0.0
118 Table A 1. Continued Taxa n Harvest (kg) Trade (kg) % Traded Artiodactyla 1 589 39 670.0 15 369.0 38.7 Mazama americana 155 4 516.9 1 524.4 33.7 Maz ama gouazoubira 11 162.0 18.5 11.4 Pecari tajacu 448 8 497.8 3 315.6 39.0 Tayassu pecari 975 26 493.3 10 510.5 39.7 Carnivora 26 177.0 19.8 11.2 Eira Barbara 1 3.0 0.0 0.0 Nasua nasua 19 94.1 14.9 15.9 Panthera onca 2 72.5 0.9 1.2 Potos flavus 4 7.4 4.0 54.3 Cingulata 3 72.0 36.0 50.0 Priodontes maximus 3 72.0 36.0 50.0 Perissodactyla 58 8 200.3 2 166.2 26.4 Tapirus terrestris 58 8 200.3 2 166.2 26.4 Primates 431 2 791.1 347.8 12.5 Alouatta seniculus 51 340.1 33.4 9.8 Ateles belzebuth 73 555.7 60.7 10.9 Callicebus discolor 5 4.8 0.9 18.9 Callithrix pygmaea 1 0.9 0.0 0.0 Cebus albifrons 8 30.0 0.0 0.0 Cebus sp. 1 5.0 2.0 39.0 Lagothrix poeppigii 280 1 838.3 249.4 13.6 Pithecia monachus 6 9.9 1.5 15.1 Saimiri sciureus 6 6.5 0.0 0.0 Rodent ia 201 1 341.1 391.4 29.2 Cuniculus paca 117 1 032.5 334.6 32.4 Dasyprocta fuliginosa 49 237.6 53.7 22.6 Hydrochaerus hydrochaeris 1 36.0 0.0 0.0 Myoprocta pratti 12 16.1 1.4 8.5 Sciurus sp. 10 8.4 1.8 21.8 Sciurus spadiceus 12 10.5 0.0 0.0 Crocodyl ia 8 82.4 7.0 8.5 Caiman crocodylus 6 68.4 7.0 10.2 Caiman niger 1 9.0 0.0 0.0 Caiman sp. 1 5.0 0.0 0.0 Testudines 34 180.2 121.8 67.6 Chelonoidis denticulata 31 168.8 110.5 65.4 Podocnemis unifilis 3 11.4 11.4 100.0 Grand Total 3 101 53 693 .0 18 53 2 .0
119 APPENDIX B SPSS OUTPUT OF ROC CURVE ANALYSES Case Processing Summary kill Valid N (listwise) Positive a 2997 Negative 3009 Larger values of the test result variable(s) indicate stronger evidence for a positive actual state. a. The positive actu al state is 1. Area Under the Curve Test Result Variable(s):Predicted Value of Mean of Response Area Std. Error a Asymptotic Sig. b Asymptotic 95% Confidence Interval Lower Bound Upper Bound .920 .003 .000 .913 .926 a. Under the nonparametric ass umption b. Null hypothesis: true area = 0.5
120 APPENDIX C TOTAL DETECTIONS OF SPECIES IN YASUNI BIOSPHERE RESERVE Table C 1. Total number of detections in 10 day survey occasion periods for species in four study areas in Yasun Biosphere Reserve. Species Maxus Road Keweriono Tiputini Lorocachi Total Mammals Cuniculus paca 21 19 35 20 95 Dasyprocta fuliginosa 50 66 60 85 261 Dasypus novemcinctus 35 17 43 23 118 Mazama america na 18 34 67 71 190 Mazama gouazoubira 12 0 20 15 47 Myrmecophaga trida ctyla 15 0 8 7 30 Pecari tajacu 51 36 81 81 249 Priodontes maximus 10 12 11 12 45 Tapirus terrestris 21 6 48 38 113 Tayassu pecari 5 0 75 19 99 Birds Mitu salvini 7 7 36 50 100 Psophia crepitans 115 70 90 106 381 Table C 2. Total number of de tections in 4 day survey occasion periods in cameras placed every 0.5 km in 13 5 km transects along the Maxus R oad. Species Maxus Road Cuniculus paca 42 Dasyprocta fuliginosa 55 Dasypus novemcinctus 39 Mazama america na 50 Mazama gouazoubira 12 Pecari tajacu 56 Tapirus terrestris 34 Tayassu pecari 10
121 APPENDIX D OCCUPANCY MODELS TO EXPLORE GAME OCCURRENCE AS A FUNTION OF LANDSCAPE ACCESS BY HUNTERS IN YASUNI BIOSPHERE RESERVE Occupancy models developed to explore the effects of access to the landscape by hunters on game occurrence in Yasun Biosphere Reserve. Models are ranked based on AIC c (the lowest the AIC c the highest the model rank). K = number of parameters estimated in model; c + = support of predictor variable; site occupancy probability ; p = detection probability Covariates include: settle = distance from camera to settlement (km); access = distance from camera to road or na vigable river (km); habitat = ha bitat type given by topography (hill and valley); DC = distance (m) between paired cameras. Tayassu pecari Model AIC c c K 481.88 0.00 4 0.183 481.90 0.02 3 0.181 (settle+habitat),p(DC) 482.22 0.34 5 0.154 +settle+habitat),p(.) 482.60 0.72 5 0.128 482.42 0.54 4 0.140 482.88 1.00 4 0.111 483.80 1.92 5 0.070 485.28 3.40 6 0.033 537.47 55.59 4 0.000 538.67 56.79 5 0.000 539.48 57.60 3 0.000 540.59 58.71 4 0.000 544.77 62.89 4 0.000 546.45 64.57 2 0.000 548.47 66.59 3 0.000 Predi ctor + Access 0.34 Settlement 1.00 Habitat 0.46
122 Pecari tajacu Model AIC c c K (access+settle),p(DC) 975.30 0.00 5 0.33 (access+settle),p(.) 975.52 0.22 4 0.30 (access+settle+habitat),p(DC) 977.01 1.71 6 0.14 (access+settle+habit at),p(.) 977.21 1.91 5 0.13 (settle),p(DC) 979.34 4.04 4 0.04 (settle),p(.) 979.80 4.50 3 0.03 (settle+habitat),p(DC) 981.52 6.22 5 0.01 (settle+habitat),p(.)* 981.93 6.63 4 0.01 (access),p(DC) 989.66 14.36 4 0.00 (access),p(.) 990.19 14.89 3 0 .00 (access+habitat),p(DC) 990.58 15.28 5 0.00 (access+habitat),p(.) 991.19 15.89 4 0.00 (.),p(.) 1006.30 31.01 2 0.00 (habitat),p(DC) 1007.37 32.07 4 0.00 (habitat),p(.) 1008.43 33.13 3 0.00 Predictor + Access 0.89 Settlement 1.00 Habitat 0.29 Model does not converge Tapirus terrestris Model AIC c c K (settle),p(DC) 630.52 0.00 4 0.50 (access+settle),p(DC) 632.36 1.84 5 0.20 (settle+habitat),p(DC) 632.72 2.20 5 0.17 (access+settle+habitat),p(DC) 634.62 4.10 6 0.06 (access+settle),p(.) 634.65 4.13 4 0.06 (access),p(DC) 642.82 12.30 4 0.00 (access+habitat),p(DC) 644.82 14.30 5 0.00 (habitat),p(DC) 645.04 14.52 4 0.00 (access),p(.) 649.09 18.57 3 0.00 (access+settle+habitat),p(.) 650.64 20.12 5 0.00 (.),p(.) 650.49 19.97 2 0.00 (habitat),p(.) 652.62 22.10 3 0.00 (settle),p(.) 656.90 26.38 3 0.00 (settle+habitat),p(.) 659.07 28.55 4 0.00 (access+habitat),p(.) 660.05 29.53 4 0.00 Predictor + Access 0.33 Settlement 1.00 Habitat 0.23 Model does not converge
1 23 Mazama americana Model AIC c c K (settle),p(DC) 870.61 0.00 4 0.32 (settle+habitat),p(DC) 871.22 0.61 5 0.24 (access+settle),p(DC) 872.78 2.17 5 0.11 (access+settle+habitat),p(DC) 873.12 2.51 6 0.09 (acce ss+habitat),p(DC) 873.42 2.81 5 0.08 (habitat),p(DC) 873.28 2.67 4 0.08 (access),p(DC) 873.59 2.98 4 0.07 (settle),p(.) 883.50 12.89 3 0.00 (settle+habitat),p(.) 884.08 13.47 4 0.00 (access+settle+habitat),p(.) 885.07 14.46 5 0.00 (access+settl e),p(.) 885.45 14.84 4 0.00 (access+habitat),p(.) 887.32 16.71 4 0.00 (access),p(.) 888.19 17.58 3 0.00 (.),p(.) 888.73 18.12 2 0.00 (habitat),p(.) 889.52 18.91 3 0.00 Predictor + Access 0.35 Settlement 0.76 Habitat 0.49 Mazama gouazoubira Model AIC c c K (settle+habitat),p(.) 332.21 0.00 4 0.27 (access+settle+habitat),p(.) 333.38 1.17 5 0.15 (access+settle),p(.) 333.51 1.30 4 0.14 (settle),p(.) 333.41 1.20 3 0.15 (settle+habitat),p(DC) 334.21 2.00 5 0.10 (acces s+settle+habitat),p(DC) 335.44 3.23 6 0.05 (access+settle),p(DC) 335.62 3.41 5 0.05 (settle),p(DC) 335.42 3.21 4 0.05 (habitat),p(.) 338.48 6.27 3 0.01 (.),p(.) 338.94 6.73 2 0.01 (habitat),p(DC) 339.65 7.44 4 0.01 (access+habitat),p(.) 340.18 7.97 4 0.00 (access),p(.) 340.97 8.76 3 0.00 (access+habitat),p(DC) 341.52 9.31 5 0.00 (access),p(DC) 342.46 10.25 4 0.00 Predictor + Access 0.40 Settlement 0.96 Habitat 0.60
124 Myrmecophaga tridactyla Model AIC c c K (acc ess+settle+habitat),p(.) 250.08 0.00 5 0.61 (access+settle+habitat),p(DC) 251.67 1.59 6 0.28 (access+settle),p(DC) 256.70 6.62 5 0.02 (access+settle),p(.) 257.90 7.82 4 0.01 (.),p(.) 257.63 7.56 2 0.01 (access),p(.) 258.08 8.00 3 0.01 (habit at),p(.) 258.39 8.31 3 0.01 (habitat),p(DC) 259.34 9.26 4 0.01 (access),p(DC) 259.40 9.32 4 0.01 (access+habitat),p(.) 259.53 9.45 4 0.01 (settle),p(.) 259.48 9.40 3 0.01 (settle+habitat),p(DC) 260.08 10.00 5 0.00 (settle),p(DC) 259.95 9.87 4 0 .00 (settle+habitat),p(.) 259.96 9.88 4 0.00 (access+habitat),p(DC) 260.83 10.75 5 0.00 Predictor + Access 0.95 Settlement 0.94 Habitat 0.92 Model does not converge Priodontes maximus Model AIC c c K (.),p(.) 356.4637 0.0 0 2 0.15 (habitat),p(.) 356.64 0.18 3 0.14 (access+settle+habitat),p(DC) 356.9832 0.52 6 0.12 (access+habitat),p(.) 357.6011 1.14 4 0.09 (access),p(.) 357.9 1.44 3 0.08 (settle),p(DC) 358.0411 1.58 4 0.07 (settle),p(.) 358.08 1.62 3 0.07 (settle+habitat),p(.) 358.6911 2.23 4 0.05 (habitat),p(DC) 358.8111 2.35 4 0.05 (settle+habitat),p(DC)* 359.2283 2.76 5 0.04 (access+settle),p(DC) 359.2483 2.78 5 0.04 (access+habitat),p(DC) 359.7183 3.25 5 0.03 (access+settle+habitat),p(.) 3 59.7883 3.32 5 0.03 (access+settle),p(.) 360.0411 3.58 4 0.03 (access),p(DC) 360.0711 3.61 4 0.03 Predictor + Access 0.43 Settlement 0.44 Habitat 0.54 Model does not converge
125 Cuniculus paca Model AIC c c K (settle+habit at),p(.) 573.63 0.00 4 0.32 (access+settle+habitat),p(.) 574.21 0.58 5 0.24 (settle+habitat),p(DC) 575.66 2.03 5 0.11 (access+settle+habitat),p(DC) 576.30 2.67 6 0.08 (access+settle),p(.) 576.47 2.84 4 0.08 (settle),p(.) 576.96 3.33 3 0.06 (acc ess+settle),p(DC) 578.49 4.86 5 0.03 (habitat),p(.) 578.64 5.01 3 0.03 (settle),p(DC) 578.94 5.31 4 0.02 (.),p(.) 580.47 6.84 2 0.01 (access+habitat),p(.) 580.77 7.14 4 0.01 (habitat),p(DC) 580.81 7.18 4 0.01 (access),p(.) 582.54 8.91 3 0.00 (access+habitat),p(DC) 582.99 9.36 5 0.00 (access),p(DC) 584.71 11.08 4 0.00 Predictor + Access 0.44 Settlement 0.94 Habitat 0.80 Dasyprocta fuliginosa Model AIC c c K (habitat),p(.) 1013.89 0.00 3 0.20 (.),p(.) 1013.98 0.09 2 0.19 (habitat),p(DC) 1015.28 1.39 4 0.10 (access+habitat),p(.) 1015.85 1.96 4 0.07 (settle+habitat),p(.) 1015.93 2.04 4 0.07 (settle),p(.) 1015.92 2.03 3 0.07 (access),p(.) 1016.09 2.20 3 0.07 (settle+habitat),p(DC) 1017.32 3.43 5 0.04 (ac cess+habitat),p(DC) 1017.32 3.43 5 0.04 (settle),p(DC) 1017.21 3.32 4 0.04 (access+settle+habitat),p(.) 1017.54 3.65 5 0.03 (access),p(DC) 1017.46 3.57 4 0.03 (access+settle),p(.) 1017.92 4.03 4 0.03 (access+settle+habitat),p(DC) 1018.97 5.08 6 0 .02 (access+settle),p(DC) 1019.26 5.37 5 0.01 Predictor + Access 0.30 Settlement 0.30 Habitat 0.56
126 Dasypus novemcinctus Model AIC c c K (.),p(.) 664.82 0.00 2 0.26 (settle),p(DC) 665.57 0.75 4 0.18 (access+settle),p(DC) 666.05 1.22 5 0.14 (habitat),p(.) 666.76 1.94 3 0.10 (access),p(.) 666.87 2.05 3 0.09 (settle+habitat),p(DC) 667.64 2.81 5 0.06 (access+settle+habitat),p(DC) 668.29 3.47 6 0.05 (habitat),p(DC) 668.37 3.55 4 0.04 (access),p(DC) 668.49 3.67 4 0.0 4 (access+habitat),p(.) 668.89 4.07 4 0.03 (access+habitat),p(DC) 670.57 5.74 5 0.01 (settle+habitat),p(.) 687.47 22.65 4 0.00 (access+settle),p(.) 687.47 22.65 4 0.00 (settle),p(.) 688.02 23.20 3 0.00 (access+settle+habitat),p(.) 689.69 24 .86 5 0.00 Predictor + Access 0.36 Settlement 0.42 Habitat 0.30 Model does not converge Mitu salvini Model AIC c c K (settle+habitat),p(DC) 533.66 0.00 5 0.48 (settle+habitat),p(.) 535.40 1.74 4 0.20 (access+settle+habita t),p(DC) 535.92 2.26 6 0.15 (access+settle+habitat),p(.) 537.59 3.93 5 0.07 (settle),p(DC) 538.19 4.53 4 0.05 (access+settle),p(DC) 540.07 6.41 5 0.02 (settle),p(.) 540.09 6.43 3 0.02 (access+settle),p(.) 541.54 7.88 4 0.01 (access+habitat),p(D C) 551.79 18.13 5 0.00 (access),p(DC) 557.07 23.41 4 0.00 (access+habitat),p(.) 558.21 24.55 4 0.00 (habitat),p(DC) 559.30 25.64 4 0.00 (access),p(.) 563.46 29.80 3 0.00 (habitat),p(.) 567.18 33.52 3 0.00 (.),p(.) 568.72 35.07 2 0.00 Pre dictor + Access 0.25 Settlement 1.00 Habitat 0.90
127 Psophia crepitans Model AIC c c K (settle),p(DC) 1097.93 0.00 4 0.22 (access+settle),p(DC) 1099.09 1.16 5 0.12 (settle+habitat),p(DC) 1099.38 1.45 5 0.11 (access),p(DC) 1099.18 1.25 4 0.12 (access+settle+habitat),p(DC) 1100.05 2.12 6 0.07 (access+habitat),p(DC) 1099.91 1.98 5 0.08 (settle),p(.) 1100.16 2.23 3 0.07 (access+settle),p(.) 1101.19 3.26 4 0.04 (access),p(.) 1101.26 3.33 3 0.04 (settle+habitat),p(.) 1101.6 3 3.70 4 0.03 (access+settle+habitat),p(.) 1102.15 4.22 5 0.03 (access+habitat),p(.) 1101.97 4.04 4 0.03 (habitat),p(DC) 1102.74 4.81 4 0.02 (.),p(.) 1103.21 5.28 2 0.02 (habitat),p(.) 1104.88 6.95 3 0.01 Predictor + Access 0.53 Set tlement 0.69 Habitat 0.38 Model does not converge
128 APPENDIX E ANALYSES OF SPATIAL DEPENDENCE Table E measuring the effect of landscape accessibility on the occurrence of 12 species in Yasun Biosphere Reserve, and B) models measuring the effect of Maxus Road on the occurrence of 6 species. Species A. ( P ) ( P ) Tayassu pecari 0.162 Pecari tajacu 0.190 0.968 Tapirus terre stris 0.63 9 1.000 Mazama americana 0.378 0.59 5 Mazama gouazoubira 0.95 6 Priodontes m aximus 0.97 2 Myrmecophaga tridactyla 0.219 Cuniculus paca 0.47 1 0.007 Dasyprocta fuliginosa 0.908 0.010 Dasypus novemcinctus 0.34 7 1.000 Mitu salvini 0.102 Psophia crepitans 1.000
129 Figure E 1. Correlograms for residuals of from models measuring the effect of landscape accessibility on the occurrence of 7 large bodied species in Yasun Biosphere Reserve (collared peccary [ Pecari tajacu ], white lipped peccary [ Tayassu pecari ], tapir [ Tapirus terrestris ], red brocket deer [ Ma zama americana ], grey brocket deer [ M. gouazoubira ], giant armadillo [ Priodontes maximus ], and giant anteater [ Myrmecophaga tridactyla ]).
130 Figure E 2. Correlograms for residuals of from models measuring the effect of landscape accessibility on the occu rrence of 5 medium sized bodied species in Yasun Biosphere Reserve (paca [ Cuniculus paca ], agouti [ Dasyprocta fuliginosa ], nine banded armadillo [ Dasypus novemcinctus Mitu salvini ] and grey winged trumpeter [ Psophia crepitans ]).
131 F igure E 3. Correlograms for residuals of from models measuring the effect of Maxus Road on the occurr ence of nine banded armadillo ( Dasypus novemcinctus ), agouti ( Dasyprocta fuliginosa ), paca ( Cuniculus paca ), red brocket deer ( Mazama americana ) tapir ( Tapirus terrestris ) and collared peccary ( Peccary tajacu )
132 APPENDIX F OCCUPANCY MODELS TO EXPLORE GAME OCCURRENCE AS A FUNTION OF DISTANCE FROM MAXUS ROAD Occupancy models developed to explore the effects of vicinity to Maxus Road on game occurrence across space. Models are ranked based o n AIC c (the lowest the AIC c the highest the model rank). K = number of parameters estimated in model; c + = support of predictor variable; probability of detection is kept constant. Model covariates include: road = distance from camera to road (km); river = distance from camera to river (km); habi tat = habitat type given by topography (hill and valley). Pecari tajacu Model AIC c c K psi(road),p(.) 328.85 0.00 3 0.287 psi(.),p(.) 329.20 0.35 2 0.241 psi(road+habitat),p(.) 330.83 1.98 4 0.107 psi(settle),p(.) 330.75 1.90 3 0.111 psi(settle +road),p(.) 331.01 2.16 4 0.098 psi(habitat),p(.) 331.32 2.47 3 0.084 psi(settle+road+habitat),p(.) 333.05 4.20 5 0.035 psi(settle+habitat),p(.) 332.91 4.06 4 0.038 Predictor + habitat 0.26 road 0.53 settlement 0.28
133 Tapirus terres tris Model AIC c c K psi(habitat),p(.) 239.77 0.00 3 0.269 psi(road+habitat),p(.) 240.83 1.06 4 0.158 psi(road),p(.) 241.27 1.50 3 0.127 psi(settle+habitat),p(.) 241.46 1.69 4 0.116 psi(.),p(.) 241.41 1.64 2 0.118 psi(settle),p(.) 241.70 1.93 3 0.103 psi(settle+road+habitat),p(.) 242.95 3.18 5 0.055 psi(settle+road),p(.) 242.97 3.20 4 0.054 Predictor + habitat 0.60 road 0.39 settlement 0.33 *Model does not converge Mazama americana Model AIC c c K psi(settle+road),p( .) 284.97 0.00 4 0.346 psi(road),p(.) 285.24 0.27 3 0.302 psi(settle+road+habitat),p(.) 287.19 2.22 5 0.114 psi(road+habitat),p(.) 287.40 2.43 4 0.103 psi(settle),p(.) 287.51 2.54 3 0.097 psi(settle+habitat),p(.) 289.45 4.48 4 0.037 psi(.),p(.) 297.3 7 12.40 2 0.001 psi(habitat),p(.) 297.92 12.95 3 0.001 Predictor + habitat 0.25 road 0.86 settlement 0.59
134 Cuniculus paca Model AIC c c K psi(road),p(.) 278.87 0.00 3 0.191 psi(settle),p(.) 278.93 0.06 3 0.186 psi(habitat),p(.) 279.27 0.40 3 0.157 psi(.),p(.) 279.50 0.63 2 0.139 psi(road+habitat ),p(.) 279.82 0.95 4 0.119 psi(habitat+settle),p(.) 280.20 1.33 4 0.098 psi(road+settle),p(.) 280.85 1.98 4 0.071 psi(road+settle+habitat),p(.) 282.03 3.16 5 0.039 Predictor + habitat 0.41 road 0.42 settlement 0.39 *Model does not c onverge Dasyprocta fuliginosa Model AIC c c K psi(settle),p(.) 327.82 0.00 3 0.211 psi(habitat),p(.) 328.21 0.39 3 0.173 psi(.),p(.) 328.30 0.48 2 0.165 psi(road),p(.) 328.55 0.73 3 0.146 psi(habitat+settle),p(.) 328.73 0.91 4 0.134 psi(road+h abitat),p(.) 329.45 1.63 4 0.093 psi(road+settle),p(.) 329.83 2.01 4 0.077 psi(road+settle+habitat),p(.) 330.92 3.10 5 0.045 Predictor + habitat 0.41 road 0.28 settlement 0.46 *Model does not converge
135 Dasypus novemcinctus Model AIC c c K psi(.),p(.) 264.56 0.00 2 0.358 psi(settle),p(.) 266.62 2.06 3 0.128 psi(habitat),p(.) 266.63 2.07 3 0.127 psi(road),p(.) 266.69 2.13 3 0.124 psi(habitat+settle),p(.) 266.96 2.40 3 0.108 psi(road+settle),p(.) 267.18 2.62 3 0.097 psi(road +habitat),p(.) 268.79 4.23 4 0.043 psi(road+settle+habitat),p(.) 270.88 6.31 5 0.015 Predictor + habitat 0.26 road 0.24 settlement 0.43
136 APPENDIX G IDENTIFICATION OF JAGUAR INDIVIDUALS FROM THEIR ROSETTE PATTERNS A B Figu re G 1. Identification of jaguar individuals from their rosette patterns. Areas within red squares can be used for easy comparison. Photos courtesy of Santiago Espinosa.
137 APPENDIX H DISTRIBUTION OF DAILY BIOMASS OF PREY PER CAMERA TRAP STATION AT FOUR SI TES OF YASUN BIOSPHERE RESERVE Figure H 1. Distribution of daily biomass of prey per camera trap station at four sites of Yasun Biosphere Reserve. Sites are ordered from less accessible to most accessible (1 = Lorocachi, 2 = Tiputini, 3 = Keweriono, 4 = Maxus Road). Boxplot description: Bar across boxes shows median; bottom and top of boxes show 25 th and 75 th percentiles, respectively; whiskers show 1.5 times the interquartile range; circles are values greater than 1.5 times the interquartile range and defined as outliers.
138 LIST OF REFERENCES Alcorn, J. B. 1993. Indigenous peoples and conservation. Conservation Biology 7 :424 426. Altrichter, M., and G. I. Boaglio. 2004. Distribution and relative abundance of peccaries in the Argentine Chaco: associatio ns with human factors. Biological Conservation 116 :217 225. Alvard, M. S., J. G. Robinson, K. H. Redford, and H. Kaplan. 1997. The sustainability of subsistence hunting in the neotropics. Conservation Biology 11 :977 982. AMWAE. Asociacin de Mujeres Waoran i de la Amazona Ecuatoriana. Available from http://www.waoraniwomen.org/ (accessed June 11 2012) Aranda, M. 2002. Importancia de los pecares para la conservacin del jaguar en Mxico. Pages 101 105 in R. A. Medellin, C. Equihua, C. L. B. Chetkiewicz, P. G. Crawshaw Jr, A. Rabinowitz, K. H. Redford, J. G. Robinson, E. W. Sanderson, and A. B. Taber, editors. El jaguar en el nuevo milenio. Fondo de Cultura Econmica, Universidad Nacional Autnoma de Mxico, Wildlife Conservation Society, Mxico, D.F. Aranda M., and V. Sanchez Cordero. 1996. Prey spectra of jaguar ( Panthera onca ) and puma ( Puma concolor ) in tropical forests of Mexico. Studies on Neotropical Fauna and Environment 31 :65 67. Asquith, N. M., S. J. Wright, and M. J. Clauss. 1997. Does mammal comm unity composition control recruitment in Neotropical forests? Evidence from Panama. Ecology 78 :941 946. Azevedo, F. C. C. 2008. Food habits and livestock depredation of sympatric jaguars and pumas in the Iguacu National Park area, south Brazil. Biotropica 40 :494 500. Bass, M. S., et al. 2010. Global conservation significance of Ecuador's Yasuni National Park. PLoS ONE 5 :Doi 10.1371/Journal.Pone.0008767. Beck, H. 2006. A review of peccary palm interactions and their ecological ramifications across the Neotro pics. Journal of Mammalogy 87 :519 530. Beck, H., P. Thebpanya, and M. Filiaggi. 2010. Do Neotropical peccary species (Tayassuidae) function as ecosystem engineers for anurans? Journal of Tropical Ecology 26 :407 414. Begazo, A. J., and R. E. Bodmer. 1998. U se and conservation of Cracidae (Aves: Galliformes) in the Peruvian Amazon. Oryx 32 :301 309.
139 Benitez Lopez, A., R. Alkemade, and P. A. Verweij. 2010. The impacts of roads and other infrastructure on mammal and bird populations: a meta analysis. Biological Conservation 143 :1307 1316. Bodmer, R. E. 1990. Fruit patch size and frugivory in the lowland tapir ( Tapirus terrestris ). Journal of Zoology 222 :121 128. Bodmer, R. E., J. F. Eisenberg, and K. H. Redford. 1997. Hunting and the likelihood of extinction of A mazonian mammals. Conservation Biology 11 :460 466. Borchers, D. L., and M. G. Efford. 2008. Spatially explicit maximum likelihood methods for capture recapture studies. Biometrics 64 :377 385. Brandon, K. 1996. Traditional peoples, nontraditional times: Soc ial change and the implications for biodiversity conservation. Pages 219 336 in K. Redford, and J. A. Mansour, editors. Traditional peoples and biodiversity conservation in large tropical landscapes. The Nature Conservancy, Arlington. Brandon, K., K. H. Re dford, and S. E. Sanderson, editors. 1998. Parks in peril: people, politics, and protected areas. Island Press, Washington, D.C. Brooks, T. M., et al. 2004. Coverage provided by the global protected area system: is it enough? BioScience 54 :1081 1091. Bruner, A. G., R. E. Gullison, R. E. Rice, and G. A. B. da Fonseca. 2001. Effectiveness of parks in protecting tropical biodiversity. Science 291 :125 128. Burnham, K. P., and D. R. Anderson 2002. Model selection and multimodel inference: a practical inform ation theoretic approach. Springer, New York. Burnham, K. P., and D. R. Anderson. 2004. Multimodel inference: understanding AIC and BIC in model selection. Sociological Methods & Research 33 :261 304. Cabodevilla, M. A. 1999. Los Waorani en la historia de l os pueblos del oriente. CICAME, Quito. Chape, S., J. Harrison, M. Spalding, and I. Lysenko. 2005. Measuring the extent and effectiveness of protected areas as an indicator for meeting global biodiversity targets. Philosophical Transactions of The Royal Soc iety B: Biological Sciences 360 :443 455. Child, B. 1996. The practice and principles of community based wildlife management in Zimbabwe: the CAMPFIRE programme. Biodiversity and Conservation 5 :369 398. Chinchilla, F. A. 1997. La dieta del jaguar ( Panthera onca ), el puma ( Felis concolor ) y el manigordo ( Felis pardalis ) (Cartivora: Felidae) en el Parque Nacional Corcovado, Costa Rica. Revista de Biologia Tropical 45 :1223 1229.
140 CONAIE. Confederacin de Nacionalidades Indgenas del Ecuador. Available from http ://www.conaie.org/ (accessed June 11 2012) Crawshaw, P. G., and H. Quigley. 2002. Hbitos alimentarios del jaguar y el puma en el Pantanal, Brasil, con implicaciones para su manejo y conservacin. Pages 223 235 in R. A. Medellin, C. Equihua, C. L. B. Chetk iewicz, P. G. Crawshaw, A. Rabinowitz, K. H. Redford, J. G. Robinson, E. W. Sanderson, and A. B. Taber, editors. El jaguar en el nuevo milenio. Fondo de Cultura Econmica, Universidad Nacional Autnoma de Mxico, Wildlife Conservation Society, Mxico, D.F. Dugelby, B. L., and M. Libby. 1998. Analyzing the social context at PiP sites. Pages 63 75 in K. Brandon, K. Redford, and E. Sanderson, editors. Parks in peril: people, politics, and protected areas. Island Press, Washington D.C. Efford, M. 2004. Density estimation in live trapping studies. Oikos 106 :598 610. Efford, M. G. 2011. Estimation of population density by spatially explicit capture recapture analysis of data from area searches. Ecology 92 :2202 2207. Efford, M. G., D. L. Borchers, and A. E. Byrom. 2009. Density estimation by spatially explicit capture recapture: likelihood based methods. Pages 255 269 in D. L. Thomson, E. G. Cooch, and M. J. Conroy, editors. Modeling demographic processes in marked populations. Springer, New York. Efford, M. G., D. K. Dawson, and C. S. Robbins. 2004. DENSITY: software for analysing capture recapture data from passive detector arrays. Animal Biodiversity and Conservation 1 :217 228. Eizirik, E., C. B. Idrusiak, and W. E. Johnson. 2002. Anlisis de la viabilidad de la s poblaciones de jaguar: evaluacin de parmetros y estudios de caso en tres poblaciones remanentes del sur de Sudamrica. Pages 501 516 in R. A. Medellin, C. Equihua, C. L. B. Chetkiewicz, P. G. Crawshaw Jr, A. Rabinowitz, K. H. Redford, J. G. Robinson, E W. Sanderson, and A. B. Taber, editors. El jaguar en el nuevo milenio. Fondo de Cultura Econmica, Universidad Nacional Autnoma de Mxico, Wildlife Conservation Society, Mxico, D.F. Emmons, L. H. 1987. Comparative feeding ecology of felids in a neotrop ical rain forest. Behavioral Ecology and Sociobiology 20 :271 283. Endo, W., C. a. Peres, E. Salas, S. Mori, J. L. Sanchez Vega, G. H. Shepard, V. Pacheco, and D. W. Yu. 2010. Game vertebrate densities in hunted and nonhunted forest sites in Manu National P ark, Peru. Biotropica 42 :251 261. Espinosa, S., L. Albuja, D. G. Tirira, G. Zapata Ros, E. Araguillin, V. Utreras, and A. Noss. In press. Anlisis del estado de conservacin del jaguar en el Ecuador in R. A. Medellin, C. Chvez, J. A. de la Torre, H. Zarz a, and G. Ceballos, editors. El jaguar en el siglo XXI: la perspectiva continental. Instituto de Ecologa Universidad Nacional Autnoma de Mxico, Mxico, D.F.
141 Fa, J. E., S. F. Ryan, and D. J. Bell. 2005. Hunting vulnerability, ecological characteristics and harvest rates of bushmeat species in afrotropical forests. Biological Conservation 121 :167 176. Fahrig, L., and T. Rytwinski. 2009. Effects of roads on animal abundance: an empirical review and synthesis. Ecology and Society 14 Fearnside, P. M. 2001. Soybean cultivation as a threat to the environment in Brazil. America 28 :23 38. Fearnside, P. M. 2006. Dams in the Amazon: Belo Monte and Brazil's hydroelectric development of the Xingu River basin. Environmental Management 38 :16 27. Fielding, A. H., and J. F. Bell. 1997. A review of methods for the assessment of prediction errors in conservation presence/absence models. Environmental Conservation 24 :38 49. Finer, M., C. N. Jenkins, S. L. Pimm, B. Keane, and C. Ross. 2008. Oil and gas projects in the weste rn Amazon: threats to wilderness, biodiversity, and indigenous peoples. PLoS ONE 3 :Doi 10.1371/Journal.Pone.0002932. Finer, M., R. Moncel, and C. N. Jenkins. 2010. Leaving the oil under the Amazon: Ecuador's Yasuni ITT initiative. Biotropica 42 :63 66. Fine r, M., and M. Orta Martinez. 2010. A second hydrocarbon boom threatens the Peruvian Amazon: trends, projections, and policy implications. Environmental Research Letters 5 :Doi 10.1088/1748 9326/1085/1081/014012. Finer, M., V. Vijay, F. Ponce, C. N. Jenkins, and T. R. Kahn. 2009. Ecuador's Yasuni Biosphere Reserve: a brief modern history and conservation challenges. Environmental Research Letters 4 :Doi 10.1088/1748 9326/1084/1083/034005. Forester, D. J., and G. E. Machlis. 1996. Modeling human factors that af fect the loss of biodiversity. Conservation Biology 10 :1253 1263. Forman, R. T. T., and L. E. Alexander. 1998. Roads and their major ecological effects. Annual Review of Ecology and Systematics 29 :207 226. Forman, R. T. T., and R. D. Deblinger. 2000. The e cological road effect zone of a Massachusetts (USA) suburban highway. Conservation Biology 14 :36 46. Forman, R. T. T., et al. 2003. Road ecology: science and solutions. Island Press, Washington, DC. Fortin, M. J., and M. R. T. Dale 2005. Spatial analysis: a guide for ecologists. Cambridge University Press, Cambridge.
142 Fragoso, J. M. V. 1998. Home range and movement patterns of white lipped peccary ( Tayassu pecari ) herds in the northern Brazilian Amazon. Biotropica 30 :458 469. Fragoso, J. M. V. 1999. Percepti on of scale and resource partitioning by peccaries: behavioral causes and ecological implications. Journal of Mammalogy 80 :993 1003. Franzen, M. 2005. Huaorani resource use in the Ecuadorian Amazon: hunting, food sharing, and market participation. PhD diss ertation, University of California, Davis. Franzen, M. 2006. Evaluating the sustainability of hunting: a comparison of harvest profiles across three Huaorani communities. Environmental Conservation 33 :36 45. Fuller, T. K., and P. R. Sievert. 2001. Carnivor e demography and the consequences of changes in prey availability. Pages 163 178 in J. L. Gittleman, S. M. Funk, D. Macdonald, and R. K. Wayne, editors. Carnivore conservation. Cambridge University Press, Cambridge. Gibson, C. C., and S. A. Marks. 1995. Tr ansforming rural hunters into conservationists: An assessment of community based wildlife management programs in Africa. World Development 23 :941 957. Giraudoux, P. 2011. R Package 'pgirmess', URL http://perso.orange.fr/giraudoux Godoy, R., N. Brokaw, and D. Wilkie. 1995. The effect of income on the extraction of non timber tropical forest products: model, hypotheses, and preliminary findings from the Sumu Indians of Nicaragua. Human Ecology 23 :29 52. Godoy, R., E. a. Undurraga, D. Wilkie, V. Reyes Garca, T. Huanca, W. R. Leonard, T. McDade, S. Tanner, and V. Vadez. 2009. The effect of wealth and real income on wildlife consumption among native Amazonians in Bolivia: estimates of annual trends with longitudinal household data (2002 2006). Animal Conservati on 13 :265 274. Gomez Pompa, A., and A. Kaus. 1999. From pre Hispanic to future conservation alternatives: lessons from Mexico. Proceedings of the National Academy of Sciences of the United States of America 96 :5982 5986. Gopalaswamy, A. M., J. A. Royle, J. Hines, P. Singh, D. Jathanna, N. S. Kumar, and K. U. Karanth. 2011. SPACECAP: a program to estimate animal abundace and density using Bayesian spatially explicit capture recapture models. Wildlife Conservation Society India Program, Centre for Wildlife Studies, Bangalure, India. Gunderson, L. H., and C. S. Holling 2002. Panarchy: understanding transformations in human and natural systems. Island Press, Washington, DC.
143 Hardin, G. 1968. The tragedy of the commons. Science 162 :1243 1248. Hill, K., and K. Ha wkes. 1983. Neotropical hunting among the Ach of eastern Paraguay. Pages 139 188 in R. Hames, and W. T. Vickers, editors. Adaptive responses of native Amazonians. Academic Press, New York. Hill, K., G. McMillan, and R. Faria. 2003. Hunting related change s in game encounter rates from 1994 to 2001 in the Mbaracayu Reserve, Paraguay. Conservation Biology 17 :1312 1323. Hill, K., and J. Padwe. 2000. Sustainability of Ache hunting in the Mbaracayu Reserve, Paraguay. Pages 79 105 in J. G. Robinson, and E. L. Be nnet, editors. Hunting for sustainability in tropical forests. Columbia University Press, New York. Hill, K., J. Padwe, C. Bejyvagi, A. Bepurangi, F. Jakugi, R. Tykuarangi, and T. Tykuarangi. 1997. Impact of hunting on large vertebrates in the Mbaracayu re serve, Paraguay. Conservation Biology 11 :1339 1353. Hines, J. E. 2006. PRESENCE2 Software to estimate patch occupancy and related parameters. USGS PWRC. Inskip, C., and A. Zimmermann. 2009. Human felid conflict: a review of patterns and priorities worldw ide. Oryx 43 :18 34. Isaac, N. J. B., and G. Cowlishaw. 2004. How species respond to multiple extinction threats. Proceedings of The Royal Society B: Biological Sciences 271 :1135 1141. IUCN. 2011. IUCN Red List of Threatened Species. Version 2011.2. Availab le from www.iucnredlist.org (accessed 13 March 2012) IUCN, and UNEP WCMC. 2010. The world database on protected areas (WDPA). UNEP WCMC, Cambridge. Available from www.protectedplanet.net (accessed July 24 2012) Janzen, D. H., and P. S. Martin. 1982. Neot ropical anachronisms: the fruits the gomphotheres ate. Science 215 :19 27. Joly, K., C. Nellemann, and I. Vistnes. 2006. A reevaluation of caribou distribution near an oilfield road on Alaska's north slope. Wildlife Society Bulletin 34 :866 869. Jones, K. E. et al. 2009. PanTHERIA: a species level database of life history, ecology, and geography of extant and recently extinct mammals. Ecology 90 :2648. Jorgenson, J. P., and K. H. Redford. 1993. Humans and big cats as predators in the Neotropics. Symposia of t he Zoological Society of London 65 :367 390.
144 Joshi, N. V., and M. Gadgil. 1991. On the role of refugia in promoting prudent use of biological resources. Theoretical Population Biology 40 :211 229. Karanth, K. U., and R. Chellam. 2009. Carnivore conservation at the crossroads. Oryx 43 :1 1. Karanth, K. U., and J. D. Nichols. 1998. Estimation of tiger densities in India using photographic captures and recaptures. Ecology 79 :2852 2862. Karanth, K. U., J. D. Nichols, N. S. Kumar, W. a. Link, and J. E. Hines. 2004. Tigers and their prey: predicting carnivore densities from prey abundance. Proceedings of the National Academy of Sciences of the United States of America 101 :4854 4858. Kauffman, R. G. 2001. Meat composition in Y. H. Hui, W. Nip, R. W. Rogers, and O. A. Young, editors. Meat science and applications. Marcel Dekker Inc., New York. Kindt, R., P. Van Damme, and A. J. Simons. 2006. Patterns of species richness at varying scales in western Kenya: planning for agroecosystem diversification. Biodiversity and Cons ervation 15 :3235 3249. Kirby, K. R., W. F. Laurance, A. K. Albernaz, G. Schroth, P. M. Fearnside, S. Bergen, E. M. Venticinque, and C. da Costa. 2006. The future of deforestation in the Brazilian Amazon. Futures 38 :432 453. Krebs, C. J. 1999. Ecological me thodology. Addison Welsey Educational Publishers Inc., Menlo Park. Laurance, W. F. 2005. When bigger is better: the need for Amazonian mega reserves. Trends in Ecology & Evolution 20 :645 648. Laurance, W. F., A. K. M. Albernaz, P. M. Fearnside, H. L. Vasco ncelos, and L. V. Ferreira. 2004. Deforestation in Amazonia. Science 304 :1109 1109. Laurance, W. F., M. A. Cochrane, S. Bergen, P. M. Fearnside, P. Delamonica, C. Barber, S. D'Angelo, and T. Fernandes. 2001. The future of the Brazilian Amazon. Science 291 : 438 439. Laurance, W. F., B. M. Croes, N. Guissouegou, R. Buij, M. Dethier, and A. Alonso. 2008. Impacts of roads, hunting, and habitat alteration on nocturnal mammals in African rainforests. Conservation Biology 22 :721 732. Laurance, W. F., B. M. Croes, L Tchignoumba, S. A. Lahm, A. Alonso, M. E. Lee, P. Campbell, and C. Ondzeano. 2006. Impacts of roads and hunting on central African rainforest mammals. Conservation Biology 20 :1251 1261.
145 Levi, T., F. Lu, D. W. Yu, and M. Mangel. 2011. The behaviour and di et breadth of central place foragers: an application to human hunters and Neotropical game management. Evolutionary Ecology Research 13 :171 185. Lindstedt, S. L., B. J. Miller, and S. W. Buskirk. 1986. Home range, time, and body size in mammals. Ecology 67 :413 418. Lu, F. 1999. Changes in subsistence patterns and resource use of the Huaorani indians in the Ecuadorian Amazon. PhD dissertation, University of North Carolina Chapel Hill. Macdonald, D. W., P. J. Johnson, L. Albrechtsen, A. Dutton, S. Seymour, J. Dupain, A. Hall, and J. E. Fa. 2011. Association of body mass with price of bushmeat in Nigeria and Cameroon. Conservation Biology 25 :1220 1228. MacKenzie, D. I., J. D. Nichols, G. B. Lachman, S. Droege, J. A. Royle, and C. A. Langtimm. 2002. Estimating s ite occupancy rates when detection probabilities are less than one. Ecology 83 :2248 2255. MacKenzie, D. I., J. D. Nichols, J. A. Royle, K. H. Pollock, L. L. Bailey, and J. E. Hines 2006. Occupancy estimation and modeling: inferring patterns and dynamics of species occurrence. Elsevier, Boston. MAE. 2012. Sistema de clasificacin de los ecosistemas del Ecuador continental. Escala 1:100.000. Ministerio del Ambiente del Ecuador, Subsecretara de Patrimonio Natural, Quito. Maffei, L., A. J. Noss, S. C. Silver, and M. J. Kelly. 2011. Abundance/density case study: jaguars in the Americas. Pages 119 144 in A. F. O'Connell, J. D. Nichols, and K. U. Karanth, editors. Camera traps in animal ecology, methods and analyses. Springer, New York. Maki, S., R. Kalliola, and K. Vuorinen. 2001. Road construction in the Peruvian Amazon: process, causes and consequences. Environmental Conservation 28 :199 214. Mayer, J. J., and R. M. Wetzel. 1987. Tayassu pecari Mammalian Species 293 :1 7. McCullough, D. R. 1996. Spatially structu red populations and harvest theory. Journal of Wildlife Management 60 :1 9. McMichael, C. H., D. R. Piperno, M. B. Bush, M. R. Silman, A. R. Zimmerman, M. F. Raczka, and L. C. Lobato. 2012. Sparse pre columbian human habitation in western Amazonia. Science 336 :1429 1431. Mena Valenzuela, P., J. R. Stallings, J. Regalado, and R. Cueva. 2000. The sustainability of current hunting practices by the Huaorani. Pages 57 78 in J. G. Robinson, and E. L. Bennet, editors. Hunting for sustainability in tropical forests. Columbia University Press, New York.
146 Naranjo, E., and R. Bodmer. 2007. Source sink systems and conservation of hunted ungulates in the Lacandon Forest, Mexico. Biological Conservation 138 :412 420. Nellemann, C., and R. D. Cameron. 1996. Effects of petrole um development on terrain preferences of calving caribou. Arctic 49 :23 28. Nellemann, C., I. Vistnes, P. Jordhoy, O. Strand, and A. Newton. 2003. Progressive impact of piecemeal infrastructure development on wild reindeer. Biological Conservation 113 :307 3 17. Novack, A. J., M. B. Main, M. E. Sunquist, and R. F. Labisky. 2005. Foraging ecology of jaguar ( Panthera onca ) and puma ( Puma concolor ) in hunted and non hunted sites within the Maya Biosphere Reserve, Guatemala. Journal of Zoology 267 :167 178. Novaro, A. J., M. C. Funes, and R. S. Walker. 2005. An empirical test of source sink dynamics induced by hunting. Journal of Applied Ecology 42 :910 920. Novaro, A. J., K. H. Redford, and R. E. Bodmer. 2000. Effect of hunting in source sink systems in the neotropi cs. Conservation Biology 14 :713 721. Oksanen, J., et al. 2011. Vegan: Community Ecology Package version 2.0 2. Oliveira, T. G. 2002. Ecologa comparativa de la alimentacin del jaguar y puma en el neotrpico. Pages 101 105 in R. A. Medellin, C. Equihua, C. L. B. Chetkiewicz, P. G. Crawshaw Jr, A. Rabinowitz, K. H. Redford, J. G. Robinson, E. W. Sanderson, and A. B. Taber, editors. El jaguar en el nuevo milenio. Fondo de Cultura Econmica, Universidad Nacional Autnoma de Mxico, Wildlife Conservation Societ y, Mxico, D.F. Otis, D. L., K. P. Burnham, G. C. White, and D. R. Anderson. 1978. Statistical inference from capture data on closed animal populations. Wildlife Monographs:1 135. Ozgul, A., S. Tuljapurkar, T. G. Benton, J. M. Pemberton, T. H. Clutton Broc k, and T. Coulson. 2009. The dynamics of phenotypic change and the shrinking sheep of St. Kilda. Science 325 :464 467. Pearce, J., and S. Ferrier. 2000. Evaluating the predictive performance of habitat models developed using logistic regression. Ecological Modelling 133 :225 245. Peres, C. A. 1996. Population status of the white lipped Tayassu pecari and collared peccaries T. tajacu in hunted and unhunted Amazonian forests. Biological Conservation 3207 :115 123. Peres, C. A. 2000a. Effects of subsistence hunti ng on vertebrate community structure in Amazonian forests. Conservation Biology 14 :240 253.
147 Peres, C. A. 2000b. Evaluating the impact and sustainability of subsistence hunting at multiple Amazonian forest sites in J. G. Robinson, and E. L. Bennett, editors Hunting for sustainability in tropical forests. Columbia University Press, New York. Peres, C. A. 2005. Why we need megareserves in Amazonia. Conservation Biology 19 :728 733. Peres, C. A., and I. R. Lake. 2003. Extent of nontimber resource extraction in tropical forests: accessibility to game vertebrates by hunters in the Amazon basin. Conservation Biology 17 :521 535. Peres, C. A., and J. W. Terborgh. 1995. Amazonian nature reserves: an analysis of the defensibility status of existing conservation units a nd design criteria for the future. Conservation Biology 9 :34 46. Perz, S., et al. 2008. Road building, land use and climate change: prospects for environmental governance in the Amazon. Philosophical Transactions of The Royal Society B: Biological Sciences 363 :1889 1895. Pfaff, A., et al. 2007. Roads and deforestation in the Brazilian Amazon. Journal of Regional Science 47 :109 123. Pimm, S. L., H. L. Jones, and J. Diamond. 1988. On the risk of extinction. American Naturalist 132 :757 785. Pocock, R. I. 1939. The races of the jaguar, Panthera onca Novitates Zoologicae 41 :406 422. Polisar, J., I. Maxit, D. Scognamillo, L. Farrell, M. E. Sunquist, and J. F. Eisenberg. 2003. Jaguars, pumas, their prey base, and cattle ranching: ecological interpretations of a ma nagement problem. Biological Conservation 109 :297 310. Rabinowitz, A. R., and B. G. Nottingham. 1986. Ecology and behaviour of the jaguar ( Panthera onca ) in Belize, Central America. Journal of Zoology 210 :149 159. RAISG. 2009. Red Amaznica de Informacin Socioambiental Georreferenciada. Available from http://www.raisg.socioambiental.org (accessed March 2012) Redford, K. H., and J. G. Robinson. 1987. The game of choice: patterns of indian and colonist hunting in the neotropics. American Anthropologist 89 :6 50 667. Redford, K. H., and A. M. Stearman. 1993. Forest dwelling native Amazonians and the conservation of biodiversity: interests in common or in collision? Conservation Biology 7 :248 255.
148 Rexstad, E., and K. P. Burnham 1991. Users's guide for interactiv e program CAPTURE. Colorado Cooperative Fish & Wildlife Research Unit, Colorado State University, Fort Collins, USA. Reyna Hurtado, R., E. Rojas Flores, and G. W. Tanner. 2009. Home range and habitat preferences of white lipped peccaries ( Tayassu pecari ) i n Calakmul, Campeche, Mexico. Journal of Mammalogy 90 :1199 1209. Reyna Hurtado, R., and G. W. Tanner. 2007. Ungulate relative abundance in hunted and non hunted sites in Calakmul Forest (Southern Mexico). Biodiversity and Conservation 16 :743 756. Rival, L. M. 1996. Hijos del sol, padres del jaguar: los huaorani de ayer y hoy. Ediciones Abya Yala, Quito, Ecuador. Robinson, J. G., and E. L. Bennett. 2000a. Carrying capacity limits to sustainable hunting in tropical forests. Pages 13 30 in J. G. Robinson, and E. L. Bennett, editors. Hunting for sustainability in tropical forests. Columbia University Press, New York. Robinson, J. G., and E. L. Bennett, editors. 2000b. Hunting for sustainability in tropical forests. Columbia University Press, New York. Robinson, J. G., and E. L. Bennett. 2004. Having your wildlife and eating it too: an analysis of hunting sustainability across tropical ecosystems. Animal Conservation 7 :397 408. Robinson, J. G., and K. H. Redford. 1986a. Body size, diet, and population density of N eotropical forest mammals. American Naturalist 128 :665 680. Robinson, J. G., and K. H. Redford. 1986b. Intrinsic rate of natural increase in Neotropical forest mammals: relationship to phylogeny and diet. Oecologia 68 :516 520. Robinson, J. G., and K. H. Re dford. 1991. Sustainable harvest of neotropical forest mammals. Pages 415 429 in J. G. Robinson, and K. H. Redford, editors. Neotropical wildlife use and conservation. The University of Chicago Press, Chicago. Roosevelt, A. C., et al. 1996. Paleoindiancave dwellers in the Amazon: The people in the Americas. Science 272 :373 384. Rosenberg, M. S., and C. D. Anderson. 2011. PASSaGE: Pattern Analysis, Spatial Statistics and Geographic Exegesis. Version 2. Methods in Ecology and Evolution 2 :229 232.
149 Royle, J. A. K. U. Karanth, A. M. Gopalaswamy, and N. S. Kumar. 2009. Bayesian inference in camera trapping studies for a class of spatial capture recapture models. Ecology 90 :3233 3244. Royle, J. A., M. Kry, R. Gautier, and H. Schmid. 2007. Hierarchical spatial mod els of abundance and occurrence from imperfect survey data. Ecological Monographs 77 :465 481. Royle, J. A., and K. V. Young. 2008. A hierarchical model for spatial capture recapture data. Ecology 89 :2281 2289. Sanderson, E. W., C. L. B. Chetkiewicz, R. A. Medelln, A. Rabinowitz, K. H. Redford, J. G. Robinson, and A. B. Taber. 2002a. Un anlisis geogrfico del estado de conservacin y distribucin de los jaguares a travs de su rea de distribucin. Pages 551 583 in R. A. Medelln, C. Equihua, C. L. B. Chet kiewicz, P. G. Crawshaw Jr, A. Rabinowitz, K. H. Redford, C. A. J. Robinson, E. Sanderson, and A. B. Taber, editors. El jaguar en el nuevo milenio. Fondo de Cultura Econmica, Universidad Nacional Autnoma de Mxico, Wildlife Conservation Society, Mxico, D.F. Sanderson, E. W., K. H. Redford, C. L. B. Chetkiewicz, R. A. Medellin, A. R. Rabinowitz, J. G. Robinson, and A. B. Taber. 2002b. Planning to save a species: the jaguar as a model. Conservation Biology 16 :58 72. Schlaepfer, M. A., M. C. Runge, and P. W Sherman. 2002. Ecological and evolutionary traps. Trends in Ecology & Evolution 17 :474 480. Schwartzman, S., and B. Zimmerman. 2005. Conservation alliances with indigenous peoples of the Amazon. Conservation Biology 19 :721 727. Scognamillo, D., I. E. Max it, M. Sunquist, and J. Polisar. 2003. Coexistence of jaguar ( Panthera onca ) and puma ( Puma concolor ) in a mosaic landscape in the Venezuelan llanos. Journal of Zoology 259 :269 279. Seymour, K. L. 1989. Panthera onca Mammalian Species 340 :1 9. Shaffer, M. L. 1981. Minimum population sizes for species conservation. BioScience 31 :131 134. Siegel, S., and N. J. Castellan 1998. Non parametric statistics for the behavioural sciences. MacGraw Hill New York. Sierra, R., F. Rodriguez, and E. Loses. 1999. Forest re source use change during early market integration in tropical rain forests: the Huaorani of upper Amazonia. Ecological Economics 30 :107 119.
150 Silver, S. C., L. E. T. Ostro, L. K. Marsh, L. Maffei, A. J. Noss, M. J. Kelly, R. B. Wallace, H. Gomez, and G. Aya la. 2004. The use of camera traps for estimating jaguar Panthera onca abundance and density using capture/recapture analysis. Oryx 38 :148 154. Silverman, B. W. 1986. Density estimation for statistics and data analysis. Chapman and Hall, New York. Sirn, A. H., P. Hambck, and J. Machoa. 2004. Including spatial heterogeneity and animal dispersal when evaluating hunting: a model analysis and an empirical assessment in an Amazonian community. Conservation Biology 18 :1315 1329. Smith, D. A. 2008. The spatial pa tterns of indigenous wildlife use in western Panama: implications for conservation management. Biological Conservation 141 :925 937. Soisalo, M., and S. Cavalcanti. 2006. Estimating the density of a jaguar population in the Brazilian Pantanal using camera t raps and capture recapture sampling in combination with GPS radio telemetry. Biological Conservation 129 :487 496. Sowls, L. K. 1997. Javelinas and other peccaries: their biology, management, and use. Texas A&M University Press, College Station, Texas. Sur ez, E., M. Morales, R. Cueva, V. Utreras Bucheli, G. Zapata Ros, E. Toral, J. Torres, W. Prado, and J. Vargas Olalla. 2009. Oil industry, wild meat trade and roads: indirect effects of oil extraction activities in a protected area in north eastern Ecuador Animal Conservation 12 :364 373. Sunquist, M. E., and F. Sunquist 2002. Wildcats of the world. The University of Chicago Press, Chicago. Sunquist, M. E., and F. C. Sunquist. 1989. Ecological constraints on predation by large felids. Pages 283 301 in J. L. Gittleman, editor. Carnivore behavior, ecology, and evolution. Chapman & Hall, London & Cornell University Press, New York. Taber, A. B., A. J. Novaro, N. Neris, and F. H. Colman. 1997. The food habits of sympatric jaguar and puma in the Paraguayan Chaco. Biotropica 29 :204 210. Taylor, R. D. 2006. Case studies on successful southern African NRM initiatives and their impacts on poverty and governance. Case study: CAMPFIRE (Communal Areas Management Programme for Indigenous Resources), Zimbabwe. USAID FRAME, IUCN TPARI, IRG. Terborgh, J. 1988. The big things that run the world a sequel to E. O. Wilson. Conservation Biology 2 :402 403. Terborgh, J., et al. 2001. Ecological meltdown in predator free forest fragments. Science 294 :1923 1926.
151 Terborgh, J., G. Nun ez Iturri, N. C. A. Pitman, F. H. C. Valverde, P. Alvarez, V. Swamy, E. G. Pringle, and C. E. T. Paine. 2008. Tree recruitment in an empty forest. Ecology 89 :1757 1768. Thiollay, J. M. 2005. Effects of hunting on guianan forest game birds. Biodiversity and Conservation 14 :1121 1135. Thornton, D., L. Branch, and M. Sunquist. 2011. Passive sampling effects and landscape location alter associations between species traits and response to fragmentation. Ecological Applications 21 :817 829. Tobler, M. W. 2002. Hab itat use and diet of Baird's tapirs ( Tapirus bairdii ) in a montane cloud forest of the Cordillera de Talamanca, Costa Rica. Biotropica 34 :468 474. Tobler, M. W., S. E. Carrillo Percastegui, R. Leite Pitman, R. Mares, and G. Powell. 2008. An evaluation of c amera traps for inventorying large and medium sized terrestrial rainforest mammals. Animal Conservation 11 :169 178. Treves, A., and K. U. Karanth. 2003. Human carnivore conflict and perspectives on carnivore management worldwide. Conservation Biology 17 :14 91 1499. Trombulak, S. C., and C. A. Frissell. 2000. Review of ecological effects of roads on terrestrial and aquatic communities. Conservation Biology 14 :18 30. Valencia, R., et al. 2004. Tree species distributions and local habitat variation in the Amazo n: large forest plot in eastern Ecuador. Journal of Ecology 92 :214 229. Vallejo Real, I. 2007. Experiencias etnoambientales, etnozoneamiento y derechos territoriales en la amazona centro. Pages 507 541. II Congreso ecuatoriano de antropologa y arqueolog a. Tomo 2. Balance de la ltima dcada: aportes, retos y nuevos temas. Abya Yala, Banco Mundial Ecuador, Quito. Vickers, W. T. 1991. Hunting yields and game composition over ten years in an Amazon indian territory. Pages 53 81 in J. G. Robinson, and K. H. Redford, editors. Neotropical wildlife use and conservation. The University of Chicago Press, Chicago. Walker, R., N. J. Moore, E. Arima, S. Perz, C. Simmons, M. Caldas, D. Vergara, and C. Bohrer. 2009. Protecting the Amazon with protected areas. Proceedin gs of the National Academy of Sciences of the United States of America 106 :10582 10586. Warriss, P. D. 2000. Meat science: an introductory text. CABI Publishing, New York. Wilkie, D., E. Shaw, F. Rotberg, G. Morelli, and P. Auzel. 2000. Roads, development, and conservation in the Congo basin. Conservation Biology 14 :1614 1622.
152 Wilkie, D. S., J. G. Sidle, and G. C. Boundzanga. 1992. Mechanized logging, market hunting, and a bank loan in Congo. Conservation Biology 6 :570 580. Wilson, K. R., and D. R. Anderson 1985. Evaluation of two density estimators of small mammal population size. Journal of Mammalogy 66 :13 21. Woodroffe, R., and J. R. Ginsberg. 1998. Edge effects and the extinction of populations inside protected areas. Science 280 :2126 2128. Yost, J. A., and P. Kelley. 1983. Shotguns, blowguns and spears: the analysis of technological efficiency. Pages 198 224 in R. B. Hames, and W. T. Vickers, editors. Adaptive responses of native Amazonians. Academic Press, New York. Zeng, H., D. Z. Sui, and X. B. Wu. 2 005. Human disturbances on landscapes in protected areas: a case study of the Wolong Nature Reserve. Ecological Research 20 :487 496.
153 BIOGRAPHICAL SKETCH Santiago Espinosa rec iology at the Pontificia Universidad Catlica del Ecuador in 2000. In 2002 he moved to Gainesville to start his graduate studies at the University of Florida. In August 2004 Santiago received his program. He received hi s Ph.D. in wildlife ecology and conservation in August 2012