University of Florida Thesis Or Dissertation Formatting

ECOLOGY AND CONSERVATION OF MAMMALIAN COMMUNITIES AND THE IMPACT OF ILLEGAL HUMAN ACTIVITIES IN NYUNGWE NATIONAL PARK, RWANDA

JENNIFER F. MOORE

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

2018

To my family, friends, and colleagues

ACKNOWLEDGMENTS

First, and foremost, I would like to thank my collaborators, Wildlife Conservation

Society – Rwanda Program for their support, both logistically and financially, and their friendships over the last four years. Specifically, I thank M. Masozera, A. Kayitare, T.

Tear, C. Cipolletta, M. Bana, F. Mulindahabi, P. Niyigaba, R. Mugabo, F. Tugendahayo,

C. Tuyishime, M. Nyiratuza, A. Inshuti, and G. Gatorano. In addition, I thank the WCS field researchers for collecting data and for accompanying me in the field. Lastly, I am immensely thankful to T. O’Brien for connecting me with WCS for my dissertation work.

Next, I would like to thank the Rwanda Development Board for supporting my work in Nyungwe National Park and for providing necessary permits, data, and equipment. Specifically, I thank P. Ntihemuka, T. Mudakikwa, K. Ildephonse, E.

Turikunkiko, I. Ndikubwimana, E. Musabyimana, R. Hategekimana, and N. Karegire.

I would like to thank my dissertation committee for their unwavering support of me as a scientist while I worked towards my doctoral degree, and their thought- provoking conversations and guidance. I have learned so much from each and every one of them: B. Pine, J. Nichols, J.M. Ponciano, M. Ernest, and M. Masozera. In addition to my official committee, I thank the following people for their support and guidance through my time at the University of Florida: J. Hines, J. Martin, K. Sieving, and E. Hellgren.

Finally, I thank those who have funded any portion of my dissertation work:

University of Florida Department of Wildlife Ecology and Conservation, University of

Florida Tropical Conservation and Development, University of Florida Center for African

Studies, Sigma Xi Honor Society, Lewis and Clark Fund for Exploration and Field

Research from the American Philosophical Society, American Society of Mammalogists,

University of Florida International Center’s Research Abroad for Doctoral Students

Award, Wildlife Conservation Society – Rwanda Program, and the Rwanda

Development Board.

TABLE OF CONTENTS

page

ACKNOWLEDGMENTS ...... 4

LIST OF TABLES ...... 8

LIST OF FIGURES ...... 9

ABSTRACT ...... 11

CHAPTER

1 INTRODUCTION ...... 13

2 ARE RANGER PATROLS EFFECTIVE IN REDUCING POACHING-RELATED THREATS WITHIN A PROTECTED AREA? ...... 17

Materials and Methods...... 20 Study Area ...... 20 Field Methodology ...... 21 Results ...... 26 Discussion ...... 28

3 FACTORS AFFECTING SPECIES RICHNESS AND DISTRIBUTION SPATIALLY AND TEMPORALLY WITHIN A PROTECTED AREA USING MULTI-SEASON OCCUPANCY MODELS ...... 42

Materials and Methods...... 45 Study Area ...... 45 Field Methodology ...... 45 Species List ...... 46 Data Analysis ...... 47 Results ...... 50 Species Richness ...... 50 Species Distribution ...... 51 Discussion ...... 53

4 SHIFTING THROUGH THE FOREST: HOME RANGE, MOVEMENT PATTERNS, AND DIET OF THE EASTERN CHIMPANZEE (PAN TROGLODYTES SCHWEINFURTHII) IN NYUNGWE NATIONAL PARK, RWANDA ...... 67

Methods ...... 69 Study site ...... 69 Data collection ...... 70 Data analysis ...... 71

Home Range Analysis ...... 71 Movement ...... 72 Diet ...... 73 Results ...... 74 Home Range ...... 74 Movement ...... 74 Diet ...... 75 Discussion ...... 77

5 CONCLUSIONS AND MANAGEMENT IMPLICATIONS ...... 93

APPENDIX

A MODEL SELECTION TABLE FOR POACHING ANALYSIS ...... 96

B MAMMAL SPECIES LIST FOR NYUNGWE NATIONAL PARK, RWANDA ...... 99

C MODEL SELECTION TABLES FOR SPECIES RICHNESS AND SPECIES DISTRIBUTION ANALYSES ...... 101

D FULL EASTERN CHIMPANZEE DIET LIST FOR NYUNGWE NATIONAL PARK, RWANDA ...... 109

LIST OF REFERENCES ...... 118

BIOGRAPHICAL SKETCH ...... 133

LIST OF TABLES

Table page

2-1 Model comparison statistics for multi-season occupancy models testing for covariate effects on initial occupancy (휓2006), probability of extinction (휖푖), probability of colonization (훾푖), and detection probability (푝푖, 푗) ...... 35

2-2 The effect of covariates on the probability of extinction (ϵi) and probability of colonization (γi) ...... 36

3-1 Model comparison table for multi-species multi-season occupancy models for species richness testing for the effect of covariates on initial occupancy (휓푡), probability of colonization (훾), probability of extinction (휖) and detection probability (푝) ...... 59

3-2 Detection probability for species richness analysis ...... 60

3-3 Model comparison table for multi-season occupancy models for species distribution testing for the effect of covariates on initial occupancy (휓2009), probability of colonization (훾), probability of extinction (휖), and detection probability (푝푖) ...... 61

4-1 Home range sizes for the Mayebe and Cyamudongo chimpanzee communities ...... 84

4-2 Hourly step length (m) overall, by season, and by month (A) and daily distance moved (m) overall and by season (B) for the Mayebe and Cyamudongo chimpanzee communities ...... 85

4-3 Proportion of days (± SE) within each month averaged over the years of data collection when the chimpanzees were observed consuming each tree species ...... 86

A-1 Model comparison statistics for multi-season occupancy models for poaching activity ...... 96

B-1 Full mammal species list for Nyungwe National Park, Rwanda ...... 99

C-1 Model comparison statistics for multi-season occupancy models for the species richness analysis ...... 101

C-2 Model comparison statistics for multi-season occupancy models for the species distribution analysis ...... 105

D-1 Full diet list for the Mayebe community of chimpanzees ...... 109

D-2 Full diet list for the Cyamudongo community of chimpanzees ...... 114

LIST OF FIGURES Figure page

1-1 The number of threats of 7 different categories detected per ranger patrol between 2006 and 2015 ...... 16

2-1 Nyungwe National Park, Rwanda showing the 1 km2 grid cells, which served as study sites, the elevation categories, roads, trails, ranger posts, park boundary, and the locations of detected poaching-related threats ...... 37

2-2 The total number of ranger patrols and the total number of detected poaching-related threats (ignoring imperfect detection) encountered each year of the study (2006-2015) in Nyungwe National Park, Rwanda ...... 38

2-3 Parameter estimates and the influence of covariates based on the top-ranked model from the candidate model set ...... 39

2-4 The projected probability of occurrence of poaching-related threats for 2016 .... 40

2-5 The average annual probability of occupancy of poaching-related threats at a randomly selected site in NNP...... 41

3-1 Overview map of Nyungwe National Park, Rwanda showing 41 park-wide transects, roads, and tourist trails ...... 62

3-2 Map of species richness in (A) 2009, (B) 2014, and (C) change between 2009 and 2014 ...... 63

3-3 Effect of the probability of poaching activity (A), the distance to the nearest tourist trail (B), and the maximum elevation (C) on the probability of colonization...... 64

3-4 Probability of occupancy for the park as a whole for each species in 2009 and 2014 ...... 65

3-5 Probability of occupancy at each transect location for the 7 species of interest ...... 66

4-1 Overview of Nyungwe National Park and its location within Rwanda including tourist trails, roads, and the overall home range of each of the two habituated chimpanzee communities ...... 88

4-2 Seasonal home range for the (A) Mayebe and (B) Cyamudongo chimpanzee communities ...... 89

4-3 Monthly home range for the Mayebe community (A) and Cyamudongo community (B) of chimpanzees ...... 90

4-4 Average hourly step length (± SE) by wet and dry season for (A) Mayebe and (B) Cyamudongo chimpanzee communities ...... 91

4-5 The cumulative number of plant species consumed over months of observation for each chimpanzee community ...... 92

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

ECOLOGY AND CONSERVATION OF MAMMALIAN COMMUNITIES AND THE IMPACT OF ILLEGAL HUMAN ACTIVITIES IN NYUNGWE NATIONAL PARK, RWANDA

Jennifer F. Moore

August 2018

Chair: Bill Pine Major: Wildlife Ecology and Conservation

Monitoring wildlife populations in protected areas and quantifying illegal human threats to wildlife is crucial for achieving long-term conservation goals. Using long-term data from Nyungwe National Park (NNP), Rwanda, we explored spatial and temporal trends in illegal poaching activity, investigated changes in species richness and distribution of individual species over time, and studied ranging, movement, and diet of the endangered eastern chimpanzee (Pan troglodytes schweinfurthii). We found that the probability of occurrence of poaching activity was highest at lower elevations, near roads and tourist trails, and away from the park boundary and ranger posts. In addition, the probability of extinction of poaching-related threats was affected by the number of ranger patrols, with the probability of extinction increasing from only 7% with zero patrols, up to 20% with 20 patrols, and 57% with 50 patrols to each site each year.

Mammalian species richness and the distributional range of five of the seven species increased between 2009 and 2014 in NNP. The probability of colonization of a species into a new area in 2014 where it was not present in 2009 was highest in sites with a lower probability of poaching activity, close to tourist trails, and at lower elevations.

Duiker species had the largest increase in distribution during the study, while there was a decrease in distribution of the eastern chimpanzee and the blue monkey. Lastly, chimpanzee home range was 21 km2 for the Mayebe community, which inhabits the main forest block, and only 4 km2 for the Cyamudongo community, which is restricted to a 4 km2 forest fragment. Both communities fed primarily on Ficus spp., with other important dietary items including Symphonia globulifera, Syzygium guineense, and

Chrysophyllum gorungosanum for the Mayebe community and Trilepisium madagascariense for the Cyamudongo community. Future ranger patrol protocols for

NNP should focus on areas of the park with a high probability of occurrence of poaching activity, low species richness, and within the core home range of the chimpanzee communities. Methodologies used in this study are broadly applicable to other protected areas and can be used to optimize management plans and protect species and national parks worldwide.

CHAPTER 1 INTRODUCTION

Illegal human activities are a threat to species and protected areas worldwide causing the decline of species, the shifting or reduction in species distributional range, and the alteration or loss of ecosystems as a whole through cascading effects across trophic levels (Dudley & Stolton, 2008; Newmark, 2008; Estes et al., 2011; Effiom,

Nuñez-Iturri, Smith, Ottosson, & Olsson, 2013). Mitigation of these threats is crucial for biodiversity conservation. Thus, we need a thorough understanding of the patterns in illegal activity, how they affect different wildlife species, and the potential management actions for alleviation of these threats.

The Albertine Rift of central Africa is home to the highest number of vertebrate species on the continent, including many endemic and globally threatened species

(Plumptre et al., 2007). The Rift covers 6 countries (the Democratic Republic of Congo,

Uganda, Rwanda, Burundi, Tanzania, and Zambia) and covers a variety of habitats ranging from glaciers at the top of the Rwenzori mountains (5100 m) down to montane forests, bamboo forests, bogs, and savanna grasslands (60 – 2500 m; Plumptre et al.,

2007). Rwanda, located at the center of the Rift, is a small country (~26,000 km2), but harbors a growing human population of almost 12 million people (CIA, 2018). This equates to a population density of over 450 people/km2, among the highest in Africa

(The World Bank, 2017). About one-third of Rwanda used to be covered in natural montane forest, but today the natural habitats are restricted to three national parks:

Volcanoes National Park, Akagera National Park, and the focus of this dissertation,

Nyungwe National Park (NNP).

Nyungwe National Park is a 1009 km2 montane forest located in southwestern

Rwanda, bordered by Kibira National Park, Burundi to the south. It ranges in elevation from 1450 – 2950 m and covers a variety of habitat types including dense forest, swamp, bamboo forest, and savannah. NNP is a biodiversity hotspot home to populations of the endangered eastern chimpanzee (Pan troglodytes schweinfurthii), the near-endemic L’hoest monkey (Cercopithecus lhoesti), and super-groups of the

Angolan black and white colobus (Colobus angolensis ruwenzorii). It is suspected that biodiversity in Rwanda and NNP, specifically, is being lost due to high human population density coupled with an extreme level of poverty; ~ 39% of the population lives in poverty and 16% in extreme poverty (UNDP, 2014). The local people depend on the forest for their livelihoods, including their need for food, firewood, timber, and land.

Animals in the park are poached for meat, and fires are set to smoke bees out of their hives to get honey. Trees cut in the forest are used for firewood, construction materials, beans poles in the garden, and timber; and bamboo and medicinal plants are also collected. The land within the park is converted for agriculture and livestock grazing, and recently mining operations for gold and coltan have increased as well. These activities cause deforestation, habitat loss, increased erosion, and the loss of biodiversity (WCS, 2009). Of all of these threats, poaching activity is the most abundant in NNP, and is thought to have the largest effect on wildlife populations (Figure 1-1).

Therefore, in collaboration with Wildlife Conservation Society – Rwanda (WCS), the overall goal of my dissertation research is to contribute to the management of

Nyungwe National Park, Rwanda by exploring the trends in poaching activity, species richness and distribution, and chimpanzee movement and ranging over time. Specific

objectives are to (1) quantify the temporal and spatial patterns of poaching activity, factors influencing these patterns, and the effect of ranger patrols in mitigating this threat; (2) provide estimates of mammalian species richness and distribution of key species throughout the park, and discern factors influencing the aforementioned population and community-level parameters; and (3) examine home range, movement parameters and diet of the eastern chimpanzee, the flagship species of NNP.

Figure 1-1. The number of threats of 7 different categories detected per ranger patrol between 2006 and 2015.

CHAPTER 2 ARE RANGER PATROLS EFFECTIVE IN REDUCING POACHING-RELATED THREATS WITHIN A PROTECTED AREA?

Illegal human activities (e.g. poaching, natural resource extraction) often represent the single most important challenge to wildlife conservation and protected area management (Martin & Caro, 2013; Effiom et al., 2013). These anthropogenic threats are often correlated with human population growth and poverty levels, with protected areas situated in regions characterized by rapidly growing, poverty-stricken human populations generally facing the highest levels of threats (Butchart et al., 2010;

Craigie et al., 2010; Challender & MacMillan, 2014). Effective management of protected areas and law enforcement to mitigate these threats are vitally important for global biodiversity conservation (Brandon, Redford, & Sanderson, 1998; Dudley & Stolton,

2008). However, managing anthropogenic threats to wildlife often depends on cultural, social, and economic contexts, and there is no single solution that is likely to be appropriate for all regions (McNeely, Harrison, & Dingwall, 1994).

In many protected areas, poaching is the largest threat to wildlife, and has been shown to be the cause of population declines as well as shifts or reductions in distributional range of many species (Stoner et al., 2007; Newmark, 2008). Poaching can also have cascading effects across trophic levels, can alter structure and functions of ecological communities and ecosystems, and potentially can affect ecosystem services offered by many protected areas (Lawlor, 1979; Paine, 1980; Brodie, Helmy,

Brockelan, & Maron, 2009; Estes et al. 2011). Poaching occurs for three primary

Reprinted with permission from: Moore, J.F., Mulindahabi, F., Masozera, M.K., Nichols, J.D., Hines, J.E., Turikunkiko, E., & Oli, M.K. (2018). Are ranger patrols effective in reducing poaching-related threats within protected areas? Journal of Applied Ecology, 55(1), 99-107.

reasons. First, animals are poached in retribution for livestock predation (e.g., Oli,

Taylor, & Rogers, 1994; Kissui, 2008) or crop raiding (Desai & Riddle, 2015; Goswami,

Medhi, Nichols, & Oli, 2015). Second, hunters will poach wildlife species within protected areas for bushmeat, as a source of protein in poor communities that do not have access to alternative and affordable protein sources. Finally, poachers kill wildlife to supply demands for illegal wildlife products and trade, mostly to wealthy countries

(Lindsey et al., 2013). The illegal wildlife trade alone (excluding small-scale subsistence bushmeat hunting) is estimated to be worth US$20 billion globally (Challender &

McMillan, 2014); thus, fully eradicating poaching would be challenging, if not impossible, requiring cooperation among governments and conservation organizations across the globe. However, effective anti-poaching management programs in each protected area would be a necessary first step in combatting poaching-related threats to global biodiversity conservation.

Attempts have been made to quantify poaching through surveys of bushmeat sold in markets and villages (Edderai & Dame, 2006; Fa et al., 2006), household or hunter questionnaire surveys (Lindsey et al., 2011; Moro et al., 2012), and arrest records from protected areas (Knapp, Rentsch, Schmitt, Lewis, & Polasky, 2010;

Ijeomah, Ogogo, & Ogbara, 2013; Risdianto et al., 2016). These methods can lead to biased inference, as market and village surveys only measure supply, and not necessarily demand or source of bushmeat. Household and hunter questionnaires can underestimate poaching activity due to fear of being reported for illegal activity, and arrest records can be confounded by the effort put forth by law enforcement (St John et al., 2012; Moro et al., 2012). An alternative to these indirect methods would be to

directly measure poaching-related threats within national parks or protected areas

(Watson, Becker, McRobb, & Kanyembo, 2013; Plumptre et al., 2014; Critchlow et al.,

2015). Direct quantification of poaching-related threats has the potential to yield unbiased estimates if a probabilistic (e.g., random or stratified random) sampling protocol is followed and probability of detection is taken into account. It is highly unlikely that researchers or rangers will detect all poaching-related activities that are present within a surveyed area (Keane, Jones, & Milner-Gulland, 2011; Critchlow et al., 2015;

Nguyen et al., 2016). Failure to account for imperfect detection can potentially lead to incorrect or biased inference (Williams, Nichols, & Conroy, 2002; MacKenzie et al.,

2006; Goswami et al., 2015).

Using 10-years of data on poaching-related threats collected in Nyungwe

National Park (NNP), Rwanda by park rangers while on patrol, our objectives were to:

(1) quantify spatio-temporal patterns of poaching-related threats in NNP, (2) discern spatial factors that affect detection and dynamics of these threats, and (3) develop a threat map showing the probability of poaching-related threats based on spatial characteristics. In particular, as a part of (2), we aimed to directly investigate the relationships between various management actions (e.g., number of ranger patrols, locations of ranger stations) and threat dynamics; such relationships permit predictions needed to inform management decisions. Data collected by rangers while on patrol have previously been used to assess illegal activities within protected areas in Uganda,

Democratic Republic of Congo, Rwanda, Sumatra, and India among others (Gray &

Kalpers, 2005; Stokes, 2010; Mackenzie, Chapman, & Sengupta, 2011; Critchlow et al.,

2015; Linkie et al., 2015). However, whether and to what extent ranger patrols reduce

poaching-related threats generally remains untested. Thus, we also sought to (4) evaluate the efficacy of ranger patrols in reducing poaching-related threats in NNP. We used dynamic occupancy models to analyze the data (MacKenzie et al., 2002;

MacKenzie, Nichols, Hines, Knutson, & Franklin, 2003) because they provide a rigorous statistical framework for estimating relevant parameters while also accounting for imperfect detection of threats by rangers (MacKenzie et al., 2006; Sharma, Wright,

Joseph, & Desai, 2014; Goswami et al., 2015; Linkie et al., 2015).

We hypothesized that: (1) the probability of occurrence of poaching-related threats will be the highest in easily accessible areas of NNP, such as along the park boundary, near roads and tourist trails, and in areas of low elevation. These areas tend to be close to human habitation, and thus are the closest and easiest places to access the park for poaching and other illegal activities (Watson et al., 2013; Plumptre et al.,

2014); (2) poaching threats would be lower in areas closer to the ranger posts because presence of rangers in these posts generally deters illegal activities; and, for the same reason, (3) probability of extinction of poaching-related threats in an area would be positively influenced by the number of ranger patrols.

Materials and Methods

Study Area

This study was conducted in Nyungwe National Park, a tropical, montane forest located in southwestern Rwanda (Figure 2-1). Together with Kibira National Park,

Burundi, the Nyungwe-Kibira landscape is the largest remaining montane forest in

Africa (Plumptre et al., 2002). The main forest block of Nyungwe is 970 km2 in size, ranges in elevation from 1451 – 2950 m, and covers a range of habitat types including rainforest, bamboo forest, savannah, and swamp (Plumptre et al., 2002). Nyungwe is a

biodiversity hotspot because it harbors several endemic and globally threatened species such as the endangered eastern chimpanzee (Pan troglodytes schweinfurthii; Plumptre et al., 2007). It is presumed to be home to 85 mammal, 275 bird, 32 amphibian, 38 reptile, and 1058 plant species (Plumptre et al., 2002).

Field Methodology

Ranger-based monitoring programs are implemented in many protected areas because they are more cost-effective and financially feasible than long-term and independent monitoring and law-enforcement programs (Gary & Kalpers, 2005; Keane et al., 2011). A ranger-based monitoring (RBM) program was initiated in Nyungwe

National Park in 2006 and continues to date. Rangers had a minimum of three years of experience before the initiation of this program; rangers hired since 2006 were trained by the original rangers. The objectives of ranger patrols were to (1) detain poachers (or those involved in other illegal activities) within the park, (2) destroy or remove snares or other means of poaching, and (3) collect data on all illegal activities. They recorded the type and magnitude of illegal activities they encountered, including evidence of poaching, forest product extraction, livestock, agriculture, honey extraction, fire, and mining. For this study, we used data collected on poaching-related illegal activities, which included wire, metal, and rope snares, poachers’ camps, and arrest and detention of poachers.

Park rangers were sent on patrol in NNP during the day and at night, with an average of 6-7 patrols per day. Each patrol consisted of 4-6 rangers hired by the government; patrols in later years were also accompanied by a community informant (or an ex-poacher), about once a month. The target location for each patrol was determined by the head ranger for each management sector of the park, and rangers could follow

any route to and from that location. Data collected by rangers were stored and managed using the Management Information System (MIST) database (Stokes, 2010).

We applied a dynamic multi-season occupancy modeling framework (MacKenzie et al., 2003; MacKenzie et al., 2006) to these data to estimate the probability of occupancy (probability that poaching-related threats are present in a site), colonization

(probability that poaching-related threat are present in an area in year 푖 + 1, given that they were not present in year 푖), and extinction (probability that poaching-related threats were not present in year 푖 + 1, given that they were present in year 푖) of poaching threats within Nyungwe National Park. This framework requires a set of sites, at least some of which are visited multiple times, as well as detection/non-detection data on threats associated with those visits. We divided NNP into 1169 grid cells of up to 1-km2 in size (hereafter, “sites”); some sites along the park boundary were <1-km2 in size due to the irregular shape of the park (Figure 2-1). Each time a ranger patrol visited a site, it was given a value of 1 if at least one poaching-related threat was detected, or 0 if no poaching-related threat was detected. Each year of study (2006-2015) was considered a primary period (i) and each ranger patrol within a year was considered a secondary occasion (j). The number of secondary occasions varied each year, corresponding to the maximum number of times a ranger patrol entered any one site during that year; sites receiving fewer than the maximum number of visits for a given year were assigned missing values. Thus, our detection histories consisted of 1’s or 0’s (detection/non- detection), or missing data (-) for sites that were visited fewer than the maximum number of times that year.

Multi-season occupancy models usually assume “closure” across the secondary periods (i.e., ranger visits) that comprise each primary period (i.e., year; MacKenzie et al., 2003; MacKenzie et al., 2006). In this study, we viewed occupancy of a site as

“use”, in the sense that we did not assume that poaching was occurring throughout the entire primary period (year) at an occupied site. Instead, we viewed an occupied site as threatened by poaching throughout the year, with actual poaching activity occurring at random times within each year. A related issue is the absence of a clear time interval with no sampling to which the extinction and colonization parameters pertain. We explored the data trying to identify periods of the year that contained a large fraction of sampling occasions, with the idea that we would define a portion of each year as the annual primary sampling period. However, ranger patrols were distributed throughout the year, so identifying a specific period of the year would have resulted in substantial loss of data. We thus followed the approach sometimes used in band recovery (Smith and Anderson, 1987) and open capture-recapture modeling of treating the entire year as a primary sampling period and viewing local extinction and colonization parameters as corresponding to the period from the mid-point of one year (e.g., July 1) to the mid- point of the next. This approach is not ideal but was necessitated by our use of ranger patrol survey data rather than data from a designed sampling program.

If rangers detected illegal activities or sighted wildlife while on patrol, they obtained GPS locations at those sites. When illegal activities were not detected or wildlife were not observed, rangers recorded GPS locations every 30 minutes. Based on these GPS locations, we created the most likely trajectory followed by each ranger

patrol. These trajectories were used to determine the number of patrols received by a site for each year of the study.

For each site, we calculated the following covariates: (1) minimum and maximum elevation category (1461-1800 m, 1801-2200 m, 2201-2600 m, and 2601-3000 m; continuous elevation data were not available), (2) distance to the nearest ranger post

(mean = 5203 (SE 77); range: 0-12025 m), nearest road (mean = 6107 (SE 219); range:

0-31067 m), nearest tourist trail (mean = 6953 (SE 218); range: 0-32221 m), and nearest park boundary (mean = 1991 (SE 69); range: 0-9595 m); and, (3) the number of ranger visits (0-87 visits per year; Figure 2-1). All distances were calculated using

ArcGIS version 10.4 (ESRI, 2011). Distance to anthropogenic features and elevation categories were used as site-specific covariates. Each ranger visit to a site was viewed as a sampling occasion (i,j denoting occasion j of year i) and was used to draw inferences about detection. For the modeling of threat occupancy, extinction and colonization, the number of ranger visits was a time-varying, site-specific covariate, and was calculated by mapping the daily patrols in ArcGIS (ESRI, 2011) and summing the number of times a patrol visited each site. The continuous covariates were all standardized to a mean of zero and standard deviation of one by subtracting the mean and dividing by the standard deviation. We did not consider additive or interactive effects of multiple covariates on model parameters due to multicollinearity between covariates and insufficient data to support more complex model structures.

We used the initial parameterization of the multi-season occupancy model

(MacKenzie et al., 2006), which allowed us to estimate the following parameters: (1)

휓2006, initial occupancy, or the probability that a poaching-related threat existed in a site

at the beginning of the study period (year 2006); (2) 휖푖, extinction probability, or the probability that poaching-related threats were not present in year 푖 + 1, given that they were present in year 푖; (3) 훾푖, colonization probability, or the probability that poaching- related threats were present in year 푖 + 1, given that they were not present in year 푖; and (4) 푝푖,푗, detection probability for occasion j in year i, probability that poaching-related threats are detected at a site containing threats in year i. More specifically, given that threats occur at a site during year i, 푝푖,푗 represents the product of (1) the probability that threats are present during sampling occasion j and (2) the probability that at least one threat is detected given presence at occasion j. We modeled the threat-related parameters (휓2006, 휖푖, and 훾푖) as possible functions of minimum elevation category or maximum elevation category, as well as the distance to boundary, road, trail, and ranger posts. Additionally, we allowed 휖푖 and 훾푖 to be affected by the number of ranger visits in year 푖 . We allowed 푝푖,푗 to be affected by the area of the site, the year (푖) of the study, as well as the additive effect of area and year.

We estimated unconditional occupancy probabilities (휓푖) for each year 푖 (the probability that a threat is present in a site in year 푖) as a derived parameter using the following recursive equation (MacKenzie et al., 2006):

휓푖 = 휓푖−1(1 − 휖푖−1) + (1 − 휓푖−1)훾푖−1 2-1

Using the most parsimonious model (Table 2-1), we estimated the probability of poaching-related threats for each site for 2016. We then mapped these values using

ArcGIS (ESRI, 2011) to develop a “risk map”, which displays the probability of occurrence of poaching-related threats for each site in 2016.

We used an information-theoretic approach using the Akaike Information

Criterion corrected for small sample size (AICc) for statistical inference and to select the most parsimonious model in the candidate model set (Burnham and Anderson, 2002).

The influence of a covariate on model parameters was evaluated by comparing models with and without that covariate, and by checking to see if the 95% confidence interval

(95% CI) for the regression coefficient (β parameter) included zero. Models were run in program MARK (White & Burnham, 1999) using the ‘RMark’ package version 2.2.0

(Laake, 2013) in the R computing environment version 3.3.1 (R Development Core

Team, 2015).

Results

Over our study period (2006-2015), 17785 ranger patrols were deployed, and

39463 poaching-related threats were detected (mean number of threats per site per year = 4.24 (SE 0.14); range: 0 – 344; Fig. 2). The spatial locations of all poaching- related threats detected during the study period are displayed in Figure 2-1. The maximum number of ranger patrols to a site in a single year was 87 (mean = 10.9 (SE

0.04)). The total number of ranger patrols increased over time with more patrols in the later years (Figure 2-2).

Ignoring the effect of covariates (휓2006(. )휖(. )훾(. )푝(. , . )), the overall estimated probability that poaching-related threats occurred at any site at the beginning of the study period (initial occupancy; 휓2006) was 0.69 (SE 0.03). The probability that poaching-related threats were not present in a site in one year, given that they were present the previous year (probability of extinction; 휖) was estimated to be 0.15 (SE

0.01), and the probability that poaching-related threats were present in a site in one

year, given they were not present in the previous year (probability of colonization; 훾) was estimated at 0.40 (SE 0.02). Finally, the probability of detecting poaching-related threats during each ranger patrol when they were present (detection probability; 푝) was estimated to be 0.10 (SE 0.002).

In terms of covariates, the most parsimonious model (i.e., the model with the lowest AICc value) included the effect of: minimum elevation category on ψ2006, number of ranger visits on ϵ푖; distance to boundary on 훾푖; and an additive effect of year and grid cell size on 푝푖 (Table 2-1; weight = 0.81; top 100 models included in supplementary material). Based on the most parsimonious model (i) 휓2006 ranged from 0.25 (SE 0.23) at the highest elevation zone (2601-3000 m) to 0.83 (SE 0.06) at the mid-elevation

(1801-2200 m) zone (Figure 2-3A); (ii) 휖푖 was positively influenced by the number of ranger visits to a site in year 푖 (Table 2-2; Fig. 2-3B); (iii) 훾푖 was positively influenced by the distance to the park boundary (Table 2-2; Fig 2-3C); and (iv) 푝푖,푗 varied over time ranging from 0.06 (SE 0.005) in 2006 to 0.13 (SE 0.004) in 2015 (Fig. 2-3D). The sites that were 1 km2 in size had a marginally higher detection probability than the sites along the boundary of < 1km2 in size.

Because covariates chosen for our analyses had management implications, we also considered singular effects of covariates even if they were not included in the most parsimonious model. We used the top model (based on AICc scores) containing each covariate for these results. We found that 휖푖 was positively influenced by the distance to roads and trails, but negatively influenced by the distance to the park boundary and ranger post (Table 2-2). The effect of these covariates on 훾푖 was the opposite, with 훾푖 being negatively influenced by the distance to roads and trails, but positively influenced

by the distance to the park boundary and ranger post (Table 2-2). Additionally, 휖푖 was the highest in the lowest elevation zone (1461-1800 m), and 훾푖 was the highest in the mid-elevation zone (1801-2200 m). There was no discernible effect of number of ranger patrols on 훾푖.

The unconditional probabilities of poaching-related threats predicted for 2016 based on parameters estimated from the most parsimonious model (weight = 0.81,

Table 2-1) ranged from 0.10 to 0.97 (Figure 2-4). The sites with the highest probabilities of poaching-related threats were in the southern portion of the park near Burundi, while the sites with the lowest probabilities of poaching-related threats were near the park boundary, especially around ranger posts.

Discussion

Poaching is a global concern and is thought to be one of the primary causes of population declines of many wildlife species throughout the world (Stoner et al., 2007;

Newmark, 2008); yet, spatial and temporal patterns of poaching remain largely unknown. Using ranger-based monitoring data and a dynamic occupancy modeling framework, our goal was to quantify spatial and temporal patterns of poaching-related threats and to discern factors influencing such threats within Nyungwe National Park,

Rwanda. The occupancy modeling framework allowed us to directly quantify poaching- related threats, taking into account uneven sampling (both temporally and spatially) as well as imperfect detection while evaluating factors influencing the threats (MacKenzie et al., 2006).

During our study, >1900 poaching-related threats were detected in Nyungwe

National Park each year, with the number of threats detected increasing over time to

more than 9000 in 2015 (Figure 2-2). These threats were distributed throughout the national park but occurred in high numbers in the west and northwest portion of the park

(Figure 2-1). Fewer threats were detected in the southern and central east portions of the park. This low encounter with threats was because of inaccessibility of these areas and potential violence due to unrest in neighboring Burundi. However, we cannot rule out the possibility that few poaching-related activities occurred in these areas because little natural vegetation remains in this region due to arson fires, thus potentially eliminating the habitat for some species. Overall, the probability that poaching-related threats occurred in a randomly selected cell within NNP at the beginning of the study period was 69%; this probability decreased to 58% in 2016 (Figure 2-5). However, the raw count of poaching-related activities and the naïve threat occupancy increased over the study period (2759 threats or ~20% naïve occupancy in 2006 to 9473 threats or

~51% naïve occupancy in 2015). Even though the raw count and naïve occupancy of threats were higher in the later years, the overall probability of occurrence of poaching- related threats was lower because detection probability increased over time (Figure 2-

3D). Without accounting for imperfect detection, we would have erroneously concluded that poaching-related threats increased in NNP during our study (Figure 2-2).

Estimated probability of detection during a ranger patrol ranged from 6% in 2007 to 13% in 2015. When coupled with a low number of secondary sampling occasions, a low detection probability can lead to an overestimated occupancy probability

(MacKenzie et al., 2002; MacKenzie & Royle, 2005). Although the probability of detection in our study was < 15%, we had a fairly large number of secondary occasions

(≥ 31), which positively affects the overall detection probability for each year. To

illustrate, the overall probability of detection (probability that threats are detected on at least one occasion at a site, given that threats occur at the site) for a primary sampling

퐾 period can be computed as 푝푖* = 1 − (1 − 푝푖,푗) , where K is the number of secondary occasions in a given year. Thus, for 푝푖,푗 = 0.10 and 31 occasions, 푝푖* ≈ 0.96, so our estimates of dynamic occupancy parameters should be approximately unbiased, even with low detection probability during each ranger visit. The increase in detection probability over the study period suggests that the rangers’ ability to find poaching- related activities improved as they gained more experience, or that the density of threats within the site increased over time (or a combination thereof). An increase in the average number of rangers per patrol could also potentially lead to higher detection probability; however, the average number of rangers per patrol did not change during our study.

The probability of poaching-related threats being present in 2006 as well as the probability of colonization of threats was higher near roads and trails. Because these features make the park more accessible, poachers are likely to use these paths to enter the forest (Watson et al., 2013; Plumptre et al., 2014). Thus, we expected a higher occurrence of threats near the park boundary (i.e. closer to human habitations).

Contrary to our expectation, based on the probability of colonization and extinction of poaching-related threats, there was a lower probability of threats closer to the park boundary. Though villages are located around the park, the majority of ranger posts are also located near the park boundary; thus, poachers would be more likely to be apprehended if engaged in illegal activities near the edge of the park. This explanation is also consistent with our finding that the probability of occurrence of poaching-related

threats is negatively influenced by distance to ranger posts where ranger patrols originate; thus, areas around ranger posts are by default heavily patrolled.

Efficient law enforcement and basic management actions can directly enhance conservation success in the long-term (Laurance et al., 2012; Lindsey et al., 2013). Our study shows that increasing ranger patrols reduced the probability of poaching-related threats given that they were present in the previous year (휖푖) within the park. Estimated threat extinction probability for sites that are not visited by ranger patrols is 7%; this probability would increase to 12%, 20% and 57% with 10, 20 and 50 ranger visits per year, respectively. Thus, increasing the average number of ranger patrols per site to about 20 per year would almost double the probability of threat extinction (Figure 2-3B).

Increasing efficiency by prioritizing sites experiencing high threats (Figure 2-4) has the potential to reduce threats in a manner that is more cost-effective and logistically feasible (also see Hofer, Campbell, East, & Huish, 2000; Plumptre et al.,

2014). To identify areas with high poaching-related threats, we developed a threat map using dynamic occupancy parameters estimated from the top model (Figure 2-4). Our threat map revealed that the southern region of the park and areas along the border with Burundi experience the highest probability of poaching-related threats. Based on these results, and our findings that the occurrence of poaching-related threats is lower in close proximity to ranger posts and with increased ranger visits, we suggest adding ≥

1 post in the southern portion of the park and increasing ranger visits to this region. A previous study from the Serengeti showed that mapping the probability of occurrence of poaching activity was an effective way to identify target locations for law enforcement patrols (Hofer et al., 2000), and a study from the Greater Virunga Landscape used

threat detections to select sites for prioritizing law enforcement for more efficient patrols

(Plumptre et al. 2014). We concur with these findings but highlight the need to account for imperfect detection and uneven sampling while developing threat maps to identify areas experiencing high poaching-related threats.

More efficient ranger patrols will undoubtedly deter poaching-related threats and may lead to apprehension of poachers. However, efficient prosecution of poachers when apprehended is also crucial for increasing the morale of rangers, encouraging them to continue performing their duties at the highest levels, and to discourage potential poachers from engaging in illegal activities within the park. Additionally, integrated conservation and development projects (Brandon et al., 1998; McShane &

Wells, 2004), community-based conservation (Gibson & Marks, 1995; Child, 1996;

Steinmetz, Srirattanaporn, Mor-Tip, & Seuaturien, 2014), payment for ecosystem services (Wunder, 2007; Engel, Pagiola, & Wunder, 2008), or alternative livelihood projects (Wright et al., 2016) are also crucial, as they seek to involve the local community in the conservation of protected areas through various means. These approaches could be instrumental in areas with high levels of poverty such as Rwanda, because they provide local villagers with an alternative to illegal activities (Pagiola,

Arcenas, & Platais, 2005; Milder, Scherr, & Bracer, 2010). For example, preliminary results of experimental employment of ex-poachers to patrol with park rangers has resulted in an increased detection of poaching-related activities in Nyungwe National

Park, while simultaneously providing an alternative income source to incentivize reduction of poaching from villagers outside of the park (M. K. Masozera, pers. comm.).

This type of management scheme has been suggested as a way to reduce poaching

threats elsewhere (Cooney et al., 2017; Wilkie, Painter, & Jacob, 2016), but its effectiveness remains untested.

By applying dynamic occupancy models to ranger-based monitoring data, we were able to estimate the occurrence of poaching-related threats within NNP, as well as identify factors influencing the spatial and temporal patterns of these threats, while accounting for imperfect detection and uneven sampling over the study period. This is one of the first studies to objectively evaluate poaching threats and to show that they can be reduced or eradicated by spatially targeting areas with high occurrence of threats and carefully planning and deploying ranger patrols to those sites. Using models of threat dynamics as a function of management actions such as this is rare, but such models are the basis for serious management efforts that are based on decision theory

(e.g. Williams et al., 2002). These types of models enable park managers to objectively select management actions to meet conservation objectives. Our findings support the hypothesis that ranger patrols deter poaching-related activities within protected areas, which is a management action that, when combined with spatial and temporal data on poaching-related threats, could be incorporated into law enforcement protocols in the future. In addition, we show the importance of ranger posts in deterring illegal poaching activity; the threat map can be used as a guide to identify locations where additional posts can be established to reduce poaching-related threats. Finally, our results highlight the need for adequately modeling detection probability while quantifying anthropogenic threats, especially in dense forests and other areas, where illegal human activities could be easily missed by researchers and rangers. Without accounting for imperfect detection, we would have incorrectly concluded that poaching threats

increased over our study period, when in fact the overall distribution of poaching threats had declined.

Table 2-1. Model comparison statistics for multi-season occupancy models testing for covariate effects on initial occupancy (휓2006), probability of extinction (휖푖), probability of colonization (훾푖), and detection probability (푝푖,푗). Model structure, number of parameters, AICc (Akaike Information Criterion corrected for small sample size) statistic, ΔAICc (difference in AICc statistic between top model and a selected model), and model weight are also given. Models with ΔAICc <10 are presented. See Appendix A for top 100 models.

Model* npar 퐀퐈퐂퐜 ∆퐀퐈퐂퐜 Weight 훙ퟐퟎퟎퟔ(퐦퐢퐧퐄)훜(퐫퐯퐬퐭)후(퐛퐨퐮퐧퐝퐚퐫퐲)퐩(퐲퐞퐚퐫 + 퐚퐫퐞퐚) 19 38559.53 0.00 0.81

훙ퟐퟎퟎퟔ(퐛퐨퐮퐧퐝퐚퐫퐲)훜(퐫퐯퐬퐭)후(퐛퐨퐮퐧퐝퐚퐫퐲)퐩(퐲퐞퐚퐫 + 퐚퐫퐞퐚) 17 38565.29 5.76 0.04

훙ퟐퟎퟎퟔ(퐫퐚퐧퐠퐩퐨퐬퐭)훜(퐫퐯퐬퐭)후(퐛퐨퐮퐧퐝퐚퐫퐲)퐩(퐲퐞퐚퐫 + 퐚퐫퐞퐚) 17 38565.31 5.78 0.05

훙ퟐퟎퟎퟔ(ퟏ)훜(퐫퐯퐬퐭)후(퐛퐨퐮퐧퐝퐚퐫퐲)퐩(퐲퐞퐚퐫 + 퐚퐫퐞퐚) 16 38565.45 5.92 0.04

훙ퟐퟎퟎퟔ(퐫퐨퐚퐝)훜(퐫퐯퐬퐭)후(퐛퐨퐮퐧퐝퐚퐫퐲)퐩(퐲퐞퐚퐫 + 퐚퐫퐞퐚) 17 38566.42 6.90 0.03

훙ퟐퟎퟎퟔ(퐭퐫퐚퐢퐥)훜(퐫퐯퐬퐭)후(퐛퐨퐮퐧퐝퐚퐫퐲)퐩(퐲퐞퐚퐫 + 퐚퐫퐞퐚) 17 38566.60 7.08 0.02

훙ퟐퟎퟎퟔ(퐦퐚퐱퐄)훜(퐫퐯퐬퐭)후(퐛퐨퐮퐧퐝퐚퐫퐲)퐩(퐲퐞퐚퐫 + 퐚퐫퐞퐚) 19 38568.90 9.38 0.01

* Covariates are: minE (minimum elevation category), rvst (number of ranger visits during year 푖), boundary (distance to park boundary (m)), year (year of study), area (size of study site(m2)), ranger post (distance to nearest ranger post (m)), road (distance to nearest road (m)), trail (distance to nearest trail (m)), and maxE (maximum elevation category).

Table 2-2. The effect of covariates on the probability of extinction (ϵi) and probability of colonization (γi). The regression coefficients (β) with confidence intervals (in parentheses) based on the top-ranked model that included a particular covariate are presented for each parameter. See Table 2-1 for a description of covariates.

Model Distance to Distance to Distance to Distance to Number of Parameter Road Boundary Trail Ranger Post Ranger Visits 훜퐢 0.39 -0.98 0.40 -0.39 0.52 (0.26, 0.51) (-1.22, -0.73) (0.27, -0.53) (-0.55, -0.23) (0.43, 0.62) 후퐢 -0.32 0.82 -0.33 0.51 -0.001 (-0.44, -0.19) (0.50, 1.13) (-0.47, -0.19) (0.30, -0.72) (-0.01, 0.10) *Values in italics denote the covariates included in the most parsimonious model (Table 2-1; weight = 0.81)

Figure 2-1. Nyungwe National Park, Rwanda showing the 1 km2 grid cells, which served as study sites, the elevation categories, roads, trails, ranger posts, park boundary, and the locations of detected poaching-related threats.

Figure 2-2. The total number of ranger patrols and the total number of detected poaching-related threats (ignoring imperfect detection) encountered each year of the study (2006-2015) in Nyungwe National Park, Rwanda.

Figure 2-3. Parameter estimates and the influence of covariates based on the top- ranked model from the candidate model set (see Table 2-1). (A) the probability of occurrence of a poaching-related threat at the beginning of the study period (휓2006) based on elevation category; (B) the probability of poaching-related threats not being present in a site during year 푖 + 1 given threats were present in year 푖 (휖푖) based on the number of ranger patrols in year 푖; (C) the probability of poaching-related threats being present in a site in season 푖 + 1 given threats were not present in season 푖 (훾푖) based on the distance of the site to the park boundary; (D) the temporal trend in the probability of detecting a poaching-related threat during a ranger visit given that the threat is present (푝푖,푗), for each year of the study for a site of average size.

Figure 2-4. The projected probability of occurrence of poaching-related threats for 2016 derived from estimates of initial occupancy, probability of colonization and extinction, and the associated covariates based on the highest ranked model from Table 2-1.

Figure 2-5. The average annual probability of occupancy of poaching-related threats at a randomly selected site in NNP for 2007-2016 derived from the highest ranked model from Table 2-1. Annual occupancy (휓푖 = 휓푖−1 ∗ (1 − 휖푖−1) + (1 − 휓푖−1) ∗ 훾푖−1) was estimated as a derived parameter for each site using site-specific covariate values (see Table 2-1 for parameter definition).

CHAPTER 3 FACTORS AFFECTING SPECIES RICHNESS AND DISTRIBUTION SPATIALLY AND TEMPORALLY WITHIN A PROTECTED AREA USING MULTI-SEASON OCCUPANCY MODELS

Understanding the dynamics and shifts in community metrics, as well as the distribution of individual species over time through long-term monitoring, is crucial for identifying conservation priorities and protecting species from extinction (Noon 2003;

Lyons et al. 2008; Stein et al. 2013; Sutter et al. 2015). Additionally, evaluating the factors that affect these trends, such as environmental covariates or levels of human disturbance, is necessary to prioritize management actions (Ricketts & Imhoff 2003;

Lepczyk et al. 2008; Zipkin et al. 2009). Species richness is simply the number of species at a focal location. It the most frequently used measure of biodiversity and is often the metric used for monitoring programs and management of protected areas

(Nichols et al. 1998; Purvis & Hector 2000; Yoccoz et al. 2001). Spatial and temporal trends in species richness and the effect of different environmental factors in explaining these trends have been assessed over a wide range of taxa such as mammals (e.g.,

Kinnaird & O’Brien 2012; Rovero et al. 2014; Wearn et al. 2016), birds (e.g., Zipkin et al.

2009; Bergman et al. 2014; Dyer et al. 2017), amphibians (e.g., Behangana et al. 2009;

Mouchet et al. 2015), butterflies (e.g., Stefanescu et al. 2004; Eskildsen et al. 2015;

Gallou et al. 2017), and fish(e.g., Gratwicke & Speight, 2005; Vasconcelos et al. 2015).

Traditionally, species richness has been estimated by counting the number of species detected in a region (Krebs 1999). However, this common method fails to take into

Submitted for publication: Moore, J.F., Hines, J.E., Mulindahabi, F., & Masozera, M.K. In Review. Factors affecting species richness and distribution spatially and temporally within a protected area using multi- season occupancy models. Biological Conservation.

account imperfect detection, or the idea that a species might be present in an area but not detected at the time of the survey (Burnham and Overton 1979; Williams et al.

2002). Failing to account for non-detection can result in biased estimates of species richness that do not include rare or inconspicuous species that are infrequently detected in an area (Kery & Royle, 2008). Additionally, detection can vary by species; cumulative measures of species richness ignore individual species identities and thus detection differences among species are ignored (Fischer, Lindenmayer, & Cowling, 2004). Many studies on community metrics fail to take into account detection and species identities

(but see e.g., Nichols et al., 1998; Zipkin et al., 2009), which could lead to erroneous conclusions about the true species richness of a particular area.

Exploring species richness of an area provides an overall assessment of the biodiversity of a community over time. However, little to no variation in species richness over time does not necessarily correspond to no change in species composition over time (Ernest & Brown, 2001; Parody, Cuthbert, & Decker, 2001). In addition to species richness, assessing shifts in species composition at each site or the distribution of each species across an area over time is also important in understanding the effect of environmental factors or illegal human activity on a site (Buij et al., 2007; Martin et al.,

2009; Kery, Guillera-Arroita, & Lahoz-Monfort, 2013; MacKenzie et al., 2018). Estimates of species distribution must also account for imperfect detection to determine true species distribution (MacKenzie, 2006). Covariates such as elevation or vegetation type are commonly used to explain species distribution patterns (MacArthur & MacArthur,

1961; MacArthur, 1972; Morrison, Marcot, & Mannan, 2006; Lenoir, Gégout, Marquet, de Ruffray, & Brisse, 2008); yet when multi-species assessments are conducted,

covariates are less useful as they can vary among individual species. In some cases covariates that can strongly influence the presence/absences of a species, such as areas where poaching is known to occur, could have large effects on targeted species, but little or no effect on non-target species (Fitzgibbon, Mogaka, & Fanshawe, 1995).

Despite the frequency of studies exploring trends in species richness and distribution in a wide array of species, many studies still report raw numbers of species in an area and naïve distributions of species without taking into account imperfect detection or species identities. Multi-season occupancy models with multi-species data allow for integrated modeling of community metrics (i.e., species richness) or distribution and species-specific detection rates (Boulinier, Nichols, Sauer, Hines, &

Pollock, 1998; Dorazio, Royle, Soderstrom, & Glimskar, 2006; Zipkin, Royle, Dawson, &

Bates, 2010; Tobler, Hartley, Carrillo-Percastegui, & Powell, 2015; MacKenzie et al.,

2018). In Nyungwe National Park (NNP), Rwanda, multiple park-wide line transect surveys were conducted in 2009 and 2014 to survey for mammalian species. Using multi-season occupancy models and multi-species data, we explored trends in species richness and distribution spatially and temporally and factors influencing these trends.

We hypothesize that (1) species richness was highest at lower elevations, within the interior of the park away from access points, closer to tourist trails (because of the presence of park guides, tourists, and rangers, which protect these areas of illegal activities), and at sites with less human disturbance; (2) species richness was higher in

2014 than 2009 because of an increased number of ranger patrols combating illegal human activity; (3) species distribution decreased for species that are heavily poached such as duikers (Cephalophus spp.) and wildpigs (Potamochoerus larvatus,

Hylochoerus meinertzhageni), while species distribution increased for species that are not targeted for poaching such as primates; (4) species distribution expanded in 2014 to areas with an increased number of ranger patrols; and, (5) detection of species was higher for indirect sightings of dung or nests versus direct sightings of animals, larger species, and group-living species versus solitary individuals.

Materials and Methods

Study Area

This study was conducted in NNP, a montane forest located in southwestern

Rwanda (2˚15’ - 2˚55’ S, 29˚00’ - 29˚30’ E; Figure 3-1). Nyungwe is a top priority site for biodiversity conservation because it harbors populations of several endemic and globally threatened species, including the endangered eastern chimpanzee (Pan troglodytes schweinfurthii; Plumptre et al. 2007). At a size of only 970 km2 for the main forest block, NNP covers a variety of habitats including rainforest, bamboo forest, savannah, and swamp (Plumptre et al. 2002). The park ranges in elevation from 1,451 to 2,950 m. Contiguous with Kibira National Park in Burundi on the southern border,

NNP is surrounded by dense human populations on all other sides, with a population density of almost 500 people/km2 in some areas (The World Bank 2017). The park is important to local communities in Rwanda providing clean water, erosion control, flood protection, and climate regulation.

Field Methodology

Line transect surveys were conducted in 2009 and 2014 in NNP (Buckland et al.

1993). Forty-one transects of three km in length were established in a systematic pattern covering the entire park (Figure 3-1). Transects were walked four times in 2009 between June and September, and five times in 2014 between May and September,

which corresponds to the long dry season in Rwanda. Direct (sighting) and indirect

(dung, nest) observations of mammal species were recorded along the transects for each sampling occasion. For each transect, the following covariates were calculated: minimum elevation (mean = 2235 [SE 38]; range: 1730-2420 m), maximum elevation

(mean = 2387 [SE 35]; range: 1879-2762 m), minimum distance between transect and access point (park boundary or road; mean = 1025 [SE 210]; range: 0-4840 m), and minimum distance between transect and tourist trail (mean = 5577 [SE 986]; range: 0-

25510 m). Because NNP is mostly dense forest with small areas of other habitat types, we chose to use elevational range as an indicator of the type of habitat found along each transect. Additionally, we calculated the number of ranger patrols per square kilometer (2008: mean = 6.5 [SE 1.0]; range: 0-24.5 patrols/km2; 2013: mean = 8.8 [SE

1.1]; range: 1.3-28.3 patrols/km2) and the estimated average occupancy of poaching activity (snares, poaching camps, or poachers; 2008: mean = 0.68 [SE 0.03]; range:

0.25-0.94; 2013: mean = 0.52 [SE 0.03]; range: 0.16-0.89) in the area surrounding each transect for the year prior to the surveys (2008 or 2013) as temporal covariates. The number of ranger patrols was collected using a ranger-based monitoring program and entered into the Management Information System (MIST) database and poaching occupancy was estimated using a multi-season occupancy model (Stokes 2010; see

Moore et al. 2017 for full field and statistical methodology).

Species List

In total, 29 mammal species are thought to inhabit the park based on prior surveys and expert opinion from local researchers (Wildlife Conservation Society, personal communication). Species that are known to be extinct such as the African elephant (Loxodonta sp.) were excluded from the mammal list. In addition, bats and

small mammals, such as mice and rats, were excluded since they are not recorded on line transect surveys, as well as all species that are strictly nocturnal, since our line transect surveys were conducted during the day. Squirrels were only recorded in the

2014 survey. A detection referred to a direct observation or sighting of all species except for duiker species (Cephalophus nigrifrons, C. silvicultor, and C. weynsi lestradei) or pig species (Potamochoerus larvatus and Hylochoerus meinertzhageni), in which a detection was a sighting of dung, and the eastern chimpanzee (Pan troglodytes schweinfurthii), in which a detection was a sighting of chimpanzee nest. Dung and nests were marked so that the same observation was not recorded in multiple surveys. We used indirect measures for these particular species because the number of detections was much higher than for direct sightings. Additionally, because dung for the three duiker species and two pig species are difficult to differentiate in the field, we combined these species into two categories. We calculated the following covariates for each species: observation category and group-living vs. solitary. Observations were grouped into five mutually exclusive categories: (1) small animal, direct sighting (average adult body mass: <3000 g), (2) medium animal, direct sighting (3001 – 10000 g), (3) large animal, direct sighting (>10001 g), (4) indirect observation of dung, and (5) indirect observation of nest. For the species distribution analysis, we used observation method instead of observation category as a covariate, corresponding to either direct observation, indirect observation of dung, or indirect observation of nest. Full species list and covariate values are included in Appendix B.

Data Analysis

We used multi-season occupancy models with multi-species data to calculate species richness, species distribution, and colonization (or turnover) between years

(MacKenzie et al. 2003; MacKenzie et al. 2018). This framework requires sites which are visited multiples times, detection/non-detection data for each species at each site during each visit, as well as a species list for the area. For the multi-species adaptation of occupancy models the “sites” correspond to a specific species (29) and transect location (41), which gives a total of 1189 “sites”. Detection (given a value of 1) and non- detection (given a value of 0) are recorded for each species at each location for all sampling occasions. If a species from the mammal list was not detected at all during the surveys, it was given a value of 0 for all sampling occasions. We considered year of study as our primary sampling occasions (2009, 2014), and survey repetition as our secondary sampling occasions (4 for 2009, 5 for 2014).

For the species richness analysis, we used the multi-season modelling framework to estimate the following parameters: (1) 휓푡, initial occupancy, the proportion of species present at each transect t in year 2009; (2) 훾푖, colonization probability, or the probability that a species is present in year i + 1, given that it was not present in year i;

(3) 휖푖, extinction probability, or the probability that a species is not present in year i + 1, given that it was present in year i; and (4) 푝푡푖, detection probability, the probability that a species is detected on a particular transect t in year i. We modelled (1) initial occupancy

(휓), probability of colonization (훾), and probability of extinction (휖) as a possible of the following covariates: transect, minimum elevation, maximum elevation, distance to nearest access point, and distance to nearest tourist trail, occupancy of poaching activity, and number of ranger patrols, and (2) detection probability (푝푡푖) as a possible function of the detection covariates (observation category, solitary/group-living). We

estimated the probability of occupancy for 2014 (휓2014) as a derived parameter using the following recursive equation (MacKenzie et al. 2018):

휓2014 = 휓2009(1 − 휖) + (1 − 휓2009)훾 3-1

Species richness for 2009 and 2014 on each transect was estimated using the following equation (MacKenzie et al., 2018):

푀

푆푅푡푖 = ∑ 휓푡푖푚 3-2 푚=1 with M representing the total number of possible species (M=29), and 휓푡푖푚 representing the unconditional occupancy for each year i for each transect t for each species m.

For species distribution range, we narrowed our species list down to species that were detected at least 10 times during the study period. We eliminated squirrels from our mammal list because they were not included in the 2009 surveys. This resulted in a list of 7 species, including the endangered eastern chimpanzee (Pan troglodytes schweinfurthii), the near-endemic L’hoest monkey (Cercopithecus lhoesti), as well as the duiker and wild pig species, which are heavily targeted for poaching (see Appendix

B for full list). For the species distribution analysis, we estimated (1) initial occupancy

(휓푡) as a function of the additive effect of species and the following covariates: minimum elevation, maximum elevation, distance to nearest access point, and distance to nearest tourist trail, occupancy of poaching activity, and number of ranger patrols, as well as the singular effect of each covariate, (2) probability of colonization (훾) and probability of extinction (휖) as the singular effect of each of the same covariates, and (3) detection probability (푝푡푖) as possible function of the observation method (direct sighting, dung, nest). In this situation, occupancy corresponds to the species distribution or the total

number of sites at which each species was present. We then used the same set of equations to derive occupancy for 2014 (휓2014).

We used an information-theoretic approach using the Akaike information criterion corrected for small sample size (AICc) for statistical inference and for selecting the most parsimonious model from the candidate model set (Burnham & Anderson 2002). We tested for the influence of each covariate by comparing models with and without the covariate, and by checking to see if the 95% confidence interval for the regression coefficient (훽 parameter) included zero. Models were run in program MARK (White &

Burnham 1999) using the RMark package version 2.2.4 (Laake 2013) in the R computing environment version 3.3.1 (R Development Core Team 2015).

Results

Species Richness

The most parsimonious model included the effect of distance to the nearest tourist trail on seasonal occupancy, the minimum elevation on the probability of extinction, the probability of poaching activity on the probability of colonization and the additive effect of observation category and whether the species was group-living or solitary on detection probability (휓(푡푟푎푖푙)휖(푚푖푛퐸푙푒푣)훾(푝표푎푐ℎ)푝(표푏푠 + 푔푟표푢푝); Table 3-

1, top 100 models in Appendix C-1). For 2009, the proportion of species present at a single transect (휓푡2009) varied from 0.199 to 0.322 (mean = 0.293 (SE 0.005)), which corresponds to a species richness of 8 to 13 species (mean = 12 species (SE 0.202);

Figure 3-2). The naïve species richness by transect for 2009 varied from 0 to 5 species

(mean = 2.7 (SE 0.219)). For 2014, the proportion of species present at a single transect (휓푡2014) varied from 0.287 to 0.529 (mean = 0.392 (SE 0.010)), which corresponds to a species richness of 12 to 22 species (mean = 16 (SE=0.406); Figure

3-2). The naïve species richness by transect for 2014 varied from 0 to 9 species (mean

= 5.5 (SE 0.313)). All of the transects had an increase in species richness between

2009 and 2014 (mean = 4 (SE 0.353); range 1-10; Figure 3-2). The probability of colonization was 0.172 (SE 0.030) and was negatively influenced by the probability of poaching activity (훽 = −2.276 (푆퐸 0.846)). The probability of extinction was 0.191 (SE

0.062) and was positively influenced by the minimum elevation (훽 = 0.758 (푆퐸 0.433)).

Finally, the probability of detection varied from 0.002 (SE 0.001) for large, solitary animals up to 0.447 (SE 0.040) for dung of group-living animals (Table 3-2).

Because our main interest is in the factors that affect species richness both temporally and spatially, we were also interested in the singular effect of covariates on each model parameter, even if the covariate is not included in the most parsimonious model. Based on the top model (i.e., model with the lowest AIC score) that included each covariate, the probability of colonization (훾), which represents a species moving into a new area, was negatively influenced by the distance to the nearest tourist trail

(훽 = −0.473 (푆퐸 0.232)), the maximum elevation (훽 = −0.310 (푆퐸 0.152)), and the occupancy of poaching activity (훽 = −2.276 (푆퐸 0.846); Figure 3-3).

Species Distribution

The most parsimonious model included species on initial occupancy, minimum elevation on the probability of extinction, distance to the nearest tourist trail on the probability of colonization, and the observation method (direct/indirect) on detection probability (휓(푠푝푒푐푖푒푠)휖(푚푖푛퐸푙푒푣)훾(푡푟푎푖푙)푝(표푏푠푀); Table 3-3; top 100 models in

Appendix C-2). For 2009, the proportion of sites that a species was present at (휓푠2009) varied from 0.231 to 1.000 (mean = 0.479 (SE 0.105)), which corresponds to a

distribution of 9 to 41 sites out of 41 total sites (mean = 20 sites (SE 4.306); Figure 3-4).

The naïve species distribution by species for 2009 varied from 8 to 28 transects (mean

= 14.4 (SE 2.927)). For 2014, the proportion of sites that a species was present at

(휓푠2014) varied from 0.431 to 0.799 (mean = 0.550 (SE 0.050); Figure 3-4), which corresponds to a distribution of 18 to 33 sites out of 41 total sites (mean = 23 sites (SE

2.061); see Figure 3-5 for distribution maps). The naïve species distribution by species for 2014 varied from 3 to 31 transects (mean = 18.1 (SE 3.826)). All of the species except for the eastern chimpanzee and the blue monkey (Cercopithecus mitis) increased in distribution between 2009 and 2014 (mean = 3 (SE 2.245); range: -8 - 8).

The probability of extinction (i.e., the probability that a species that was present at a transect in 2009 was no longer present at that transect in 2014) was 0.201 (SE 0.064) and was positively influenced by the minimum elevation (훽 = 0.607 (푆퐸 0.451)), while the probability of colonization (i.e., the probability that a species that was not present at a transect in 2009 was present at that transect in 2014) was 0.321 (SE 0.060) and was negatively influenced by the distance to the nearest tourist trail (훽 =

−0.714 (푆퐸 0.347)). Finally, the probability of detection varied by observation method.

The probability of detection for (1) direct observations was 0.243 (SE 0.020), (2) dung was 0.364 (SE 0.033), and (3) nests was 0.409 (SE 0.037). Based on the top model

(i.e., model with the lowest AIC score) that included each covariate, the only covariate that significantly influenced one of the parameters was the distance to the nearest trail on the probability of colonization.

Discussion

Our study showed an overall increase in species richness throughout the park between 2009 and 2014. In addition, species distribution increase for 5 of 7 species. We related these increases in richness and distribution to the decrease in poaching activity, with areas of high richness related to spatial features of the park such as the distance to the nearest tourist trails and the elevation.

During our study, species richness increased from between 1 and 10 species at each of the 41 transects around Nyungwe National Park. In 2009, species richness was highest on the north side of the main road, decreasing to the southern point of the park.

In 2014, the sites with the highest species richness were along the main road, as well as at sites in the north and west of the park. Sites south of the road towards the border with Burundi still had the lowest species richness within the park. Because Nyungwe

National Park is contiguous with Kibira National Park in Burundi along this southern border, weak law enforcement and lack of large mammals in Burundi could explain increased poaching and thus low species richness in this area. In terms of change between the years, sites along the main road had the largest increase in species richness between 2009 and 2014 as well as a few sites on the north side of the park and along the eastern border. There was little increase in species richness in the sites just south of the main road and just north of the main road (Figure 3-2).

The probability of colonization, which is the probability that a species will move into a new area in 2014 where they were not present in 2009, was influenced by the probability of poaching activity, the maximum elevation, and the distance to the nearest tourist trail. After a ranger-based monitoring program was implemented in NNP in 2006, the amount of poaching activity declined due to increased ranger patrols (Moore et al.

2017). In this study the probability of colonization with no poaching is around 50%, but this drops to only about 10% with 100% chance of poaching (Figure 3-3). Previous studies have also linked poaching activity to species declines within an area (Newmark,

2008; Stoner et al. 2007).

We hypothesized that species richness would be higher close to tourist trails, away from access points, and at lower elevations within the park. Our results corroborated our hypothesis that species richness would be higher closer to tourist trails and at lower elevations. Previous studies have shown lower species richness near tourist trails because of disturbance (e.g. Cunha 2010; Zhou et al. 2013); however, the presence of tourists can also deter illegal human activity (e.g., Kӧndgen et al. 2008;

Jachmann et al. 2011). In NNP, the number of tourists visiting many of the trails within the park is very low. Therefore, the presence of a tourist trail deters illegal activity, but does not cause enough of a disturbance to decrease species richness. On tourist trails the probability of colonization was about 30%, but this declined to about 10% at 25 km from the nearest tourist trail (Figure 3). We observed a decline in species richness with elevation (see review by Rahbek (1995)); and estimated the probability of colonization near 20% at 1450 m but a decline to just above 10% at 3000 m (Figure 3). The distance to the nearest park access point (road or park boundary) was not a significant predictor for the probability of colonization.

We hypothesized that the probability of detection would be highest for group- living species, larger body-size species, and for indirect signs versus direct sighting of species. In our study, for each observation type, group-living species had a higher probability of detection than solitary species. With line transect surveys, larger groups

are often seen at further distances from the transects than small groups (Buckland et al., 1993; Plumptre, 2000). In our case we predicted higher detection for group-living species since they are seen over a wider area from each transect than for solitary species. Contrary to our hypothesis, the detection probability was highest for the smallest body-size species and lowest for the largest species. In NNP, the largest species corresponds to the three felid species (leopard (Panthera pardus), serval

(Leptailurus serval), and golden cat (Caracal aurata)) as well as the baboon (Papio anubis), while the smallest species corresponds primarily to squirrels. The felid species are uncommon in NNP and are rarely detected by researchers or using camera traps; however, squirrels are abundant (Wildlife Conservation Society, unpublished data). This variation in abundance for the different groups of species affects detection probability, with higher abundance leading to higher detection (Royle & Nichols, 2003). Lastly, for indirect vs. direct sightings, detection probability was highest for indirect sightings as hypothesized but with the exception of small animals (<3000 g average adult body mass), which had a higher probability of detection than dung of solitary animals. Within tropical forests, such as NNP, indirect estimation techniques are often used over direct observation because the visibility of species is low (Plumptre, 2000), such as using nest counts for great apes (Hall, Saltonstall, Inogwabini, & Omari, 1998; Marchesi, Marchesi,

Fruth, & Boesch, 1995; Plumptre & Reynolds, 1996). Therefore, indirect sightings would be higher than direct; small animals are an exception because of their high abundance as noted previously.

During our study, seven species were detected at least 10 times. In 2009, blue monkeys had the largest distribution, found at all transects throughout the park, followed

by chimpanzees, L’hoest monkeys, wild pig species, Angolan colobus (Colobus angolensis), Grey-cheeked mangabey (Lophocebus albigena), and duiker species with the smallest distribution. Duikers were only found at 9 of the 41 transects. In 2014, the order of species from largest to smallest distribution stayed the same, but the number of transects each species was present at increased except for blue monkeys and chimpanzees. In 2014, duikers were now found at 18 of the 41 transects. Between the two years, the duiker species saw the largest increase in distribution, followed by Grey- cheeked mangabey, Angolan colobus, wild pigs, and L’hoest monkeys. We saw a small decrease in distribution for chimpanzees (28 transects in 2009 down to 26 transects in

2014) and a larger decrease for blue monkeys (41 transect in 2009 down to 33 transects in 2014).

Our models showed an increase in the probability of colonization closer to tourist trails, as was also seen in our species richness analysis. Tourist trails can be linked with lower human disturbance (Kӧndgen et al. 2008; Jachmann et al. 2011) due to the mere presence of tourists, which could explain the movement of species into sites near these trails. Although not significant, there was a negative relationship between presence of poaching activity and the probability of colonization implying a decrease in poaching would lead to a higher probability of a species colonizing a new area that it was not present in during 2009. Duiker species and wild pig species are heavily targeted for poaching in NNP, while monkeys (excluding chimpanzees) are also targeted when convenient (Wildlife Conservation Society, personal communication). The large increase in distribution for the duiker species could be at least in part due to this decrease in poaching. In addition, this could explain why the distribution of chimpanzees did not

change much between the two years, since they are not targeted for poaching, but could be limited by other factors such as food availability or changes in habitat instead.

The decrease in blue monkey distribution could be due to the increase in other monkey species, or a focus of poaching on blue monkeys since they were more widely distributed than other monkey species. Lastly, like with the species richness analysis, detection probability was highest for indirect vs. direct observations.

The results of this study can be used to improve conservation planning within

NNP. Areas of the park with low species richness, as well as sites with low probability of occurrence for species of conservation concern, such as the endangered eastern chimpanzee, should be more heavily patrolled to reduce poaching activity. Poaching activity was seen to have a significant effect of the probability of colonization, or the probability of species moving into new areas where they were not present during the first year of the study. This includes an implementation of transboundary patrols with rangers from Rwanda and Burundi to reduce poaching along the southern border of the park. Transboundary cooperation is necessary otherwise poachers can cross the border to avoid being apprehended by rangers. However, it is also important to note that areas where a species is not present could also be due to unsuitable habitat or other factors that were not directly examined in this study.

For conservation planning it is important to examine not just community metrics, such as species richness, but also individual species distributions. Based on our species richness results, we see an increase of species at each site throughout the park. However, based on the distribution analysis, though increasing for many species, it is actually decreasing for the eastern chimpanzee, the only endangered mammal

species within the park. Because of the importance of chimpanzee for tourism revenue in Rwanda (Rwanda Development Board, personal communication), as well as their ecological role as a seed disperser and one of the largest remaining mammals in NNP

(Gross-Camp, Masozera, & Kaplin, 2009), it is important to further monitor this decrease in their distribution. This is a conservation concern for NNP and demonstrates the utility of our approach for informing monitoring efforts in other protected areas.

Lastly, our study highlights the use of multi-season occupancy models with multi- species data in exploring trends in richness and distribution, while taking into account imperfect detection and species-specific identities. This approach allows for inferences to be made about species that are rarely or never detected within a protected area based on shared characteristics with species that are easier to detect. These models can be used broadly to influence species monitoring programs and protected area management plans throughout the world.

Table 3-1. Model comparison table for multi-species multi-season occupancy models for species richness testing for the effect of covariates on initial occupancy (휓푡), probability of colonization (훾), probability of extinction (휖) and detection probability (푝). Model structure, number of parameters (npar), AICc (Akaike information criterion corrected for small sample size), ΔAICc (change in AICc between top model and selected model) and model weights are all presented. Sample size is 1189 (41 transect * 29 species). Models with ΔAICc < 2 are included with top 100 models included in Appendix C-1.

Model* npar AICc ΔAICc Weight 1 Psi(~Trail)Epsilon(~minElev)Gamma(~poach)p(~obs + group) 12 3474.326 0 0.0601 2 Psi(~maxElev)Epsilon(~minElev)Gamma(~poach)p(~obs + group) 12 3474.987 0.661 0.0432 3 Psi(~poach)Epsilon(~minElev)Gamma(~Trail)p(~obs + group) 12 3475.851 1.524 0.0281 4 Psi(~maxElev)Epsilon(.)Gamma(~poach)p(~obs + group) 11 3476.109 1.783 0.0246 5 Psi(~minElev)Epsilon(~minElev)Gamma(~poach)p(~obs + group) 12 3476.151 1.824 0.0241 6 Psi(~poach)Epsilon(~minElev)Gamma(~poach)p(~obs + group) 12 3476.236 1.910 0.0231 7 Psi(~Access)Epsilon(~minElev)Gamma(~poach)p(~obs + group) 12 3476.312 1.986 0.0223 * Covariates are: distance to nearest access point (Access), distance to nearest trail (Trail), minimum elevation (minElev), maximum elevation (maxElev), probability of occupancy of poaching activity (poach), number of ranger patrols (rp), observation category (obs), group-living/solitary (group).

Table 3-2. Detection probability for species richness analysis based on (1) observation type (direct sighting of small animal (<3000 g), direct sighting of medium animal (3001-10000 g), direct sighting of large animal (> 10000 g), indirect sighting of dung, indirect sighting of nest, and (2) whether the animal is solitary or lives in a group.

Observation Group/Solitary Detection probability 95% confidence interval Small Solitary 0.316 (0.269, 0.366) Medium Solitary 0.080 (0.051, 0.122) Medium Group 0.184 (0.152, 0.221) Large Solitary 0.002 (0.0005, 0.008) Large Group 0.005 (0.001, 0.021) Dung Solitary 0.237 (0.174, 0.314) Dung Group 0.447 (0.370, 0.526) Nest Group 0.433 (0.368, 0.500)

Table 3-3. Model comparison table for multi-season occupancy models for species distribution testing for the effect of covariates on initial occupancy (휓2009), probability of colonization (훾), probability of extinction (휖), and detection probability (푝푖). Model structure, number of parameters (npar), AICc (Akaike information criterion corrected for small sample size), ΔAICc (change in AICc between top model and selected model) and model weight are all presented. Sample size is 287 (41 transect * 7 species). Models with ΔAICc < 2 are included with top 100 models included in Appendix C-2.

Model* npar AICc ΔAICc Weight 1 Psi(~species)Epsilon(~minElev)Gamma(~Trail)p(~obsM) 14 2025.120 0.000 0.0368 2 Psi(~species + Trail)Epsilon(~minElev)Gamma(~Trail)p(~obsM) 15 2025.392 0.272 0.0321 3 Psi(~species)Epsilon(~Access)Gamma(~Trail)p(~obsM) 14 2025.589 0.469 0.0291 4 Psi(~species)Epsilon(.)Gamma(~Trail)p(~obsM) 13 2025.615 0.495 0.0287 5 Psi(~species + Trail)Epsilon(.)Gamma(~Trail)p(~obsM) 14 2025.761 0.641 0.0267 6 Psi(~species + Trail)Epsilon(~Access)Gamma(~Trail)p(~obsM) 15 2025.778 0.658 0.0265 7 Psi(~species + poach)Epsilon(~minElev)Gamma(~Trail)p(~obsM) 15 2026.318 1.198 0.0202 8 Psi(~species)Epsilon(~Trail)Gamma(~Trail)p(~obsM) 14 2026.484 1.364 0.0186 * Covariates are: distance to nearest access point (Access), distance to nearest trail (Trail), minimum elevation (minElev), maximum elevation (maxElev), probability of occupancy of poaching activity (poach), number of ranger patrols (rp), direct/indirect observation method (obsM)

Figure 3-1. Overview map of Nyungwe National Park, Rwanda showing 41 park-wide transects, roads, and tourist trails.

Figure 3-2. Map of species richness in (A) 2009, (B) 2014, and (C) change between 2009 and 2014.

Figure 3-3. Effect of the probability of poaching activity (A), the distance to the nearest tourist trail (B), and the maximum elevation (C) on the probability of colonization based on the top model (i.e., model with the lowest AIC score) containing each covariate for the species richness analyses

Figure 3-4. Probability of occupancy for the park as a whole for each species in 2009 and 2014

Figure 3-5. Probability of occupancy at each transect location for the 7 species of interest

CHAPTER 4 SHIFTING THROUGH THE FOREST: HOME RANGE, MOVEMENT PATTERNS, AND DIET OF THE EASTERN CHIMPANZEE (PAN TROGLODYTES SCHWEINFURTHII) IN NYUNGWE NATIONAL PARK, RWANDA

Understanding how animals use space or select resources is essential for devising and implementing management plans for conservation (Kertson & Marzluff,

2010; Manly, McDonald, Thomas, McDonald, & Erickson, 2002). Animals move around their environment to obtain the resources necessary for survival and reproduction, and thus, their movements are dictated by the location of available food resources, the need to avoid competitors and predators, and the whereabouts of possible mates (Houston,

Higginson, & McNamara, 2011; McIntyre & Wiens, 1999; Powell, 2012). Because of differences in external factors (i.e., available resources) and internal factors (i.e., nutritional needs), home range sizes may vary among species and among different populations of the same species (Börger, Danziel, & Fryxell, 2008).

Great apes generally have large home ranges; therefore, by protecting the range and critical food resources for a great ape, smaller, sympatric species such as other non-human primates are also protected (Berger, 1997; Brown, 1995; Butynski, 1997;

Struhsaker, 1997; Wilcox, 1984). Additionally, great apes often serve important ecological functions for the health of the ecosystems where they occur, such as contributing to the regrowth of the forest through seed dispersal and shaping the vegetative structure of their habitat by trampling and breaking vegetation as they travel or forage (Gross-Camp, Masozera, & Kaplin, 2009; Lambert & Garber, 1998; Plumptre,

Reprinted with permission from: Moore, J.F., Mulindahabi, F., Gatorano, G., Niyigaba, P., Ndikubwimana, I., Cipolletta, C., & Masozera, M.K. 2018. Shifting through the forest: home range, movement patterns, and diet of the eastern chimpanzee (Pan troglodytes schweinfurthii) in Nyungwe National Park, Rwanda. American Journal of Primatology. DOI: 10.1002/ajp.22897

1994). Most great ape species occur in biologically diverse tropical forests in Africa and

Asia; consequently, by conserving these species we also conserve habitats and other endemic species (Myers, 1988; Myers, Mittermeier, Mittermeier, da Fonseca, & Kent,

2000).

Eastern chimpanzees (Pan troglodytes schweinfurthii), one of four recognized sub-species of chimpanzee, occur in Central and East Africa, but are declining in numbers throughout their range (Plumptre et al. 2016). Previous studies of the eastern chimpanzee have shown that home ranges vary in size from around 7 km2 in moist, forest environments to over 500 km2 in drier environments, and that home range size varies seasonally (Basabose, 2005; Chapman & Wrangham, 1993; Furuichi,

Hashimoto, & Tashiro, 2001; Goodall, 1986; Kano, 1972; Newton-Fisher, 2003; Pusey,

Pintea, Wilson, Kamenya, & Goodall, 2007). However, little is known about the population of this subspecies in Rwanda, one of their range countries. Nyungwe

National Park (NNP), a montane rainforest in southwest Rwanda, harbors the largest population of the eastern chimpanzee in the country. Despite years of data collection by habituation teams, home range and movement behavior of this particular population, and how it compares to the other nearby populations of chimpanzees, is unknown.

Moreover, only one previous study has explored the important food resources of this population (Gross-Camp et al., 2009). Here, using 10-15 years of tracking data for two eastern chimpanzee communities in NNP, we describe the annual, seasonal, and monthly ranging and movement behaviors and identify important food resources for these communities. These communities differ by their location within the national park, with one inhabiting the forest center, and the other living in a 4km2 forest fragment. We

hypothesized that (1) the chimpanzee community living in the forest center would have a larger home range and longer hourly and daily movement ranges than the community living in a forest fragment because their movement is not constrained in space; (2) home range would shift throughout the year depending on the spatial distribution and phenology of their primary food resources; (3) both communities would primarily feed on

Ficus spp. throughout the year as do other eastern chimpanzee communities in the region (e.g., Kahuzi-Biega National Park, Democratic Republic of Congo and Bwindi

Impenetrable Forest, Uganda; Basabose, 2002; Stanford & Nkurunungi, 2003), and (4) the richness of food species consumed by the community inhabiting the main forest would be greater than the community living in the forest fragment.

Methods

Study site

We conducted this study in Nyungwe National Park (NNP), located in southwestern Rwanda (215’ - 255’ S, 2900’ - 2930’ E), which encompasses 1019 km2 including the forest fragments of Gisakura and Cyamudongo (Figure 4-1).

Nyungwe, combined with the adjacent Kibira National Park in Burundi to the south, is one of the largest remaining montane forests in Africa (Plumptre et al., 2002). The park

(1600 – 2950 m elevation) comprises multiple habitat types including primarily tropical rainforest with small areas of bamboo forest, savannah, and swamp. The park receives more than 2000 mm of rain annually, primarily during the two wet seasons (March to

May and September to December). Temperature varies little throughout the year, with average minimum and maximum temperatures of 10.9°C and 19.6°C, respectively. Fruit production peaks between March and May, leaf flush during July and August, and flower production peaks during December and January (Sun et al., 1996).

Nyungwe is a top priority site for biodiversity conservation in the Albertine Rift because it harbors a large number of endemic and globally threatened species, including the eastern chimpanzee (Plumptre et al., 2007). Nyungwe is home to more than 85 mammal species, and is considered a ‘primate hotspot’, with at least 13 species of primates living within the forest (Plumptre et al., 2002). Nyungwe is estimated to be home to ~400 eastern chimpanzees (Wildlife Conservation Society, unpublished data).

Data collection

Our study was based on the daily tracking of two communities of habituated chimpanzees that reside in NNP. The Mayebe community consisted of about 50-60 individuals and ranged near the center of the main forest block of Nyungwe (Figure 4-1).

Tracking of the Mayebe community began in the 1990s but was suspended from 1994 to 2000 due to civil unrest; data from 2000 to 2015 was used in this study. The

Cyamudongo community had about 35-40 individuals and resided within the

Cyamudongo forest fragment (4 km2), located about 10 km from the main forest block

(Figure 4-1). This community was restricted to this forest fragment because of dense surrounding human population, which separated this area from the main forest block; however, the chimpanzees did sometimes foray into neighboring agricultural fields to raid crops. Data collection for the Cyamudongo community began in 2005, with data from 2005-2015 used in this study.

During the study, trackers from the Wildlife Conservation Society and the

Rwanda Development Board tracked members of the two study communities from nest to nest on a daily basis (approximately 6:00 AM to 6:00 PM), taking a global position system (GPS) location point every 30 minutes. If a given party broke apart, trackers continued following the largest subgroup. Trackers attempted to follow members of both

communities 365 days per year over the entire study period. Sometimes they were unable to locate members of a given community, so there were days when no follows occurred (Mayebe: 2419 days total, average of 161 ± SD 110 days per year;

Cyamudongo: 2888 days total, average of 263 ± SD 84 days per year).

Trackers also collected dietary data through direct observation for Mayebe (2000

– 2011) and Cyamudongo (2005 – 2011) communities. At 30-minute intervals, they recorded all food items (tree species and part of tree, primate species, honey, or insects) consumed by any feeding chimpanzee.

All aspects of this study complied with protocols approved by the Rwanda

Development Board and Wildlife Conservation Society, and applicable research permits were acquired prior to the initiation of the study. All research adhered to the American

Society of Primatologists Principles for the Ethical Treatment of Primates.

Data analysis

Home Range Analysis

We estimated home range size based on GPS locations collected by trackers using two methods: minimum convex polygons (MCP; Hayne, 1949) and fixed kernel density (KDE; Seaman & Powell, 1996; Worton, 1989). For the KDE analyses, we used an ‘ad hoc’ method to estimate the smoothing factor, which assumes a bivariate normal kernel, because the least square cross validation (LSCV) method did not converge. We estimated 50%, 95%, and 100% MCP and 50% and 95% KDE home ranges for each community of chimpanzees over the entire study period, by season (wet season (March

– May; September – December)/dry season (January – February; June – August), and by month. The 100% MCP estimates were included to allow direct comparison to earlier studies (e.g., Budongo Forest and Kibale National Park, Uganda; Chapman &

Wrangham, 1993; Newton-Fisher, 2003). Additionally, we designated the 50% KDE estimates for each month as the ‘core’ home range for the chimpanzee communities.

We chose this area as the ‘core’ because it closely corresponded to the overlap between the 95% KDE estimate for each month of the year, and because there was no clear inflexion point when comparing the %KDE to the home range size. Locations taken at 30-minute intervals are temporally autocorrelated, which can lead to biased estimates of home range size (Swihart & Slade, 1985). To minimize the effect of autocorrelation in location data on home range estimates, we used locations at 6-hour intervals to estimate home ranges. This corresponded to at most three locations per day. Chimpanzees can potentially travel across their home ranges in 2-4 hours

(Newton-Fisher, 2003); therefore, by using locations at 6-hour intervals we minimized or eliminated temporal autocorrelation. We conducted home range analyses using the

‘adehabitatHR’ package (Calenge, 2006) in the R computing environment version 3.3.1

(R Development Core Team, 2015).

Movement

We used hourly locations to estimate hourly step length over the course of the study, by season, and by month for Mayebe and Cyamudongo chimpanzee communities. Because data collection points occurred at variable clock times each day, we rounded all location points to the nearest hour of the day. Step length is the straight- line distance between two successive hourly locations. In addition, we calculated the daily distance moved by summing the average hourly step length for each hour of the day over the course of the day. We also calculated daily distance moved over the course of the study and by season for each chimpanzee community. We tested for differences in hourly step length between hours of the day and seasons using

generalized linear models with a Gaussian distribution. Our dependent variable was the hourly step length, and we used hours of the day and season as two categorical independent variables (Gotelli & Ellison, 2004). We used the ‘adehabitatLT’ package

(Calenge, 2006) to conduct step length analyses.

Diet

Using direct observations of chimpanzee feeding behavior collected by chimpanzee trackers, we calculated the proportion of days per month when members of each chimpanzee community were recorded eating a particular food item (tree species and part of tree, primate species, honey, insects). Data were collected for 1895 days

(average of 13.1 – 18.3 days per month) between 2000-2011 for the Mayebe community, with data available on average for 10 months ± SD 3.73 of the year, and for

1925 days between 2005-2011 for the Cyamudongo community, with data collected on average for 12 months ± SD 0.49 of the year. For tree species, we also calculated the proportion of dietary records (i.e., days) comprising each part of the tree (fruit, leaves, flowers, marrow, bark, or trunk). For the most frequently consumed tree species (> 10 observation days for any one month), we calculated the average proportion of days when the species was consumed across the years of the study. In addition, we used a contingency table 휒2 test to test for seasonal variation in the top tree species (16 for

Mayebe, 4 for Cyamudongo) consumed by each chimpanzee community. Seasons were defined as follows: Dry 1 (January – February), Wet 1 (March – May), Dry 2 (June

– August), and Wet 2 (September – December). Lastly, we estimated the richness of food species for each chimpanzee community by plotting the cumulative number of unique species consumed based on the number of months of data collection. This shows the sampling effort required to determine the full repertory of species consumed.

Results

Home Range

Based on 6271 6-hourly locations, the 95% KDE home range for the Mayebe chimpanzee community was 21.23 km2 (Figure 4-1), while the 100% MCP was 60.98 km2 (Table 4-1). For the KDE analyses, the smoothing parameter used was 281.67 m.

Home range was larger in the wet season than in the dry season (Figure 4-2) and varied monthly from 12.67 km2 in April to 23.40 km2 in January (95% KDE; Figure 4-3;

Table 4-1). Home range of the Mayebe community covered altitudes from 1600 – 2600 m throughout the study period. The core home range for the Mayebe community (50%

KDE) was 5.17 km2, covering altitudes from 1800 – 2425 m (Figure 4-1).

Based on 3445 6-hourly locations, the 95% KDE home range for the

Cyamudongo community was 3.72 km2 (Figure 4-1), while the 100% MCP was 8.18 km2

(Table 4-1). For the KDE analyses, the smoothing parameter used was 128.90 m. The home range did not differ substantially between the wet and dry season (Figure 4-2) but varied monthly from 2.55 km2 in December to 5.36 km2 in March (95% KDE; Figure 4-3;

Table 4-1). Home range of the Cyamudongo community covered altitudes from 1530 –

2103 m throughout the study period. The core home range for the Cyamudongo community (50% KDE) was 0.56 km2, covering altitudes from 1725 – 2000 m (Figure 4-

1).

Movement

We obtained 8812 hourly step lengths (N = 4860 wet season, N = 3262 dry season, N = 465-904 per month) for the Mayebe community over the 16 years of tracking. Average hourly step length was 75 ± SE 5 m with a significant difference between average hourly step length at different hours of the day (Linear model: F =

1.873, df = 13 and 8108, P = 0.028), but no significant difference in average hourly step length between the wet and dry season (Linear model: 훽 = 9.8 ± SE 10.6, 푃 = 0.352;

Table 4-2; Figure 4-4). Average step length exhibited monthly variation ranging from 22

± SE 5 m in February to 133 ± SE 30 m in August (Table 4-2). Daily distance moved was 987 ± SE 71 m. This distance was shorter during the dry season (915 m ± SE 108) than the wet season (1030 m ± SE 94).

For the Cyamudongo community, we obtained 9172 hourly step lengths (N =

5286 wet season, N = 3886 dry season, N = 658-861 per month) over the 10 years of tracking. The average hourly step length was 52 ± SE 3 m and varied between hours of the day and season of the year (Linear model: F = 3.46, df = 14 and 9157, P < 0.001;

Figure 4-4). Average hourly step length for the dry season (63 ± SE 5 m) was longer than for the wet season (44 ± SE 3 m). Additionally, average hourly step length varied monthly from 29 ± SE 4 m in November up to 76 ± SE 14 m in January for Cyamudongo

(Table 4-2). Daily distance moved was 651 ± SE 71 m. This distance was shorter during the wet season (559 m ± SE 48) than the dry season (768 m ± SE 68).

Diet

Chimpanzees in the Mayebe community consumed 62 plant species, as well as blue monkey (Cercopithecus mitis doggetti), bees and honey, and other insects (see

Appendix D-1 for full diet list). Species of the Ficus genus were combined into one category because field assistants could not distinguish them in the field. The Mayebe chimpanzee community diet included 12 tree species that were consumed on at least

10 days during the study period (Table 4-3). The most frequently consumed taxon throughout 8 months of the year (January – May; July – September) was Ficus spp., with Chrysophyllum gorungosanum consumed the most in June, and Syzygium

guineense consumed the most from October to December. The most frequently consumed tree taxon varied across seasons (Table 4-3; 휒2 = 1563, 푑푓 = 33, 푃 < 0.001).

Ficus spp. were the dominant food source in the first dry season (Jan-Feb), first wet season (Mar-May), and second dry season (June-Aug) whereas S. guineense was the most frequently consumed tree species during the second wet season (Sep-Dec). The number of unique species consumed leveled out at 65 species after 29 months of data collection (Figure 4-5). Overall, the chimpanzees were mostly observed eating fruits

(81% of 1942 total observations), followed by leaves (5%), new fruit buds (5%), seeds

(3%), and flowers (3%).

The Cyamudongo community consumed 49 different plant species, as well as unspecified monkey species during the study period (see Appendix D-2 for full diet list).

Four tree species were consumed on at least 10 days during the study period (Table 4-

3). Ficus spp. were the most consumed species from January to May and Trilepisium madagascariense dominated the diet from June to December. The most frequently consumed tree species varied across seasons (Table 4-3; 휒2 = 400.23, 푑푓 = 9, 푃 <

0.001). Ficus spp. were the dominant food source in the first dry season (Jan-Feb) and the first wet season (Mar-May), and T. madagascariense was the dominant food source in the second dry season (Jun-Aug) and second wet season (Sep-Dec). The number of unique species consumed leveled out at 50 species after 23 months (Figure 4-5).

Overall, the chimpanzees were most often observed eating fruits (96% of the 1925 observations), followed by leaves (2%).

Discussion

We explored ranging behavior, movement, and diet of two habituated communities of chimpanzees in Nyungwe National Park, Rwanda. The Mayebe community inhabited the main forest block of Nyungwe, while the Cyamudongo community was restricted to a forest fragment located 10 km from the main forest block.

As expected, the home range of the Mayebe chimpanzees (21 km2; 95% KDE) was five times the home range size of the Cyamudongo community (4 km2; 95% KDE).

Considering that the Cyamudongo community inhabited a 4 km2 forest fragment surrounded by a dense human population, their home range size was certainly constrained in space. However, community size could also contribute to the difference in home range size between the two communities. The Mayebe community, which was larger (50-60 individuals) than the Cyamudongo community (35-40 individuals) would likely have greater demands for food, and thus travel further to meet these demands

(Clutton-Brock & Harvey, 1979; Herbinger, Boesch, & Rothe, 2001; Olupot, Chapman,

Brown, & Waser, 1994).

Methods for estimating home range size for primates have changed over time, with most early studies using the ‘grid-square’ analysis (e.g., Fossey, 1974; Chapman &

Wrangham, 1993; Hashimoto et al., 1998). This method can lead to overestimates of home range size; consequently, these estimates are not directly comparable to those obtained using the MCP or KDE methods that we used (Harris et al., 1990; Newton-

Fisher, 2003). However, two studies conducted in Uganda reported 100% MCP estimates, and thus are directly comparable. In the Sonso chimpanzee community of

Budongo Forest, Uganda, chimpanzee annual home range was estimated at 7 km2

(Newton-Fisher, 2003), and in the Kanyawara community in Kibale National Park,

Uganda home range was estimated at 15 km2 (Chapman & Wrangham, 1993). By comparison, our 100% MCP estimate for the Mayebe community was 61 km2. These differences in home range size could relate to two factors. First, as noted above, community size can affect home range size (Clutton-Brock & Harvey, 1979; Herbinger et al. 2001; Olupot et al., 1994). The Mayebe community at 50-60 individuals is larger than both the Sonso community (46 individuals) and the Kanyawara community (41 individuals). Second, differences in home range size can also arise because of differences in habitat productivity/quality and food availability (Clutton-Brock & Harvey,

1977; South, 1999; Van Orsdol, Hanby, & Bygott, 1985). Annual rainfall has been used as a proxy of primary productivity or fruit availability, with higher rainfall leading to higher productivity across forests (van Schaik, Madden, & Ganzhorn, 2005). However, productivity also declines with elevation; thus, high elevation montane forests are less productive than lowland forests (Tanner, Vitousek, & Cuevas, 1998; Veneklaas, 1991).

In Nyungwe, annual rainfall (on average > 2000 mm per year; Sun et al. 1996) is high compared to the annual rainfall in Budongo Forest (1780-1900 mm per year; Reynolds,

Plumptre, Greenham, & Harborne, 1998) and Kibale National Park (average of 1570 mm per year; Chapman & Wrangham, 1993); however, elevation is also higher (up to

2950 m) in Nyungwe compared to Budongo (700 – 1270 m) and Kibale (1100 – 1600 m). A decrease in primary productivity at the higher elevations, where the Mayebe community lives (1500 – 2600 m), could explain the larger home range size obtained for this population of the eastern chimpanzee.

We observed monthly and seasonal shifts in the location of the home range for the Mayebe community. Some food species the Mayebe community frequently

consumed (e.g., S. guineense) occurred only at high elevations, while C. gorungosanum was more abundant at lower elevations (Wildlife Conservation Society, unpublished data). This uneven distribution of food resources and the timing of fruit availability could explain these shifts. In Tanzania, Turner (2000) found that the distribution of food resources could be the main factor explaining how chimpanzees use their home range, and in Kahuzi-Biega National Park in the Democratic Republic of

Congo, Basabose (2005) found that chimpanzees adapted their ranging patterns to the seasonality of food resources. In Nyungwe, fruit production peaks in the wet season,

March-May, which leads to abundant ripe fruits during the dry season (Wildlife

Conservation Society, unpublished data). This timing of fruit availability explains the seasonal shifts and smaller home range during the dry season. The location of food resources based on elevational range explains the monthly shifts. In Kahuzi-Biega

National Park, there were also monthly and seasonal shifts in home range. These shifts were attributed to the seasonality of fruit and periods of fruit scarcity, which require longer movements to meet food demands (Basabose, 2005). However, the monthly home range size did not increase in times of food scarcity, but instead the location of the home range shifted. In the Mayebe group, we witnessed the same shifts, but also an increase in the home range during particular months. The shifts were not as apparent in the Cyamudongo community which used the majority of their available home range at all times.

The Cyamudongo community was an anomaly compared to other eastern chimpanzee communities because its home range was restricted in space. The area outside of the forest fragment had a high human population density, and thus the

chimpanzees could not range far from the protected area except for brief forays into neighboring agricultural fields. They occasionally consumed non-wild plants, but this is not thought to be a large portion of their diet because of the high human population density and the potential for human-wildlife conflict. However, diet data included only wild species in our study. Because of the space limitation, there was little variation in home range size or location across seasons or months of the year. However, we emphasize that home range size for the Cyamudongo community is the smallest recorded for any eastern chimpanzee population, and also the only community we know of constrained to a small forest fragment. This community may be able to persist because (1) the community size is fairly small (35-40 individuals), and (2) the high annual rainfall combined with the slightly lower elevation of Cyamudongo (1530 – 2130 m) corresponds with relatively high fruit availability.

Few studies of chimpanzees or great apes in general report movement metrics

(but see Bates & Byrne, 2009; Herbinger et al. 2001; Vedder, 1984). Our study showed that the hourly step lengths for the Mayebe and Cyamudongo communities were 75 m and 52 m, respectively. We were unable to find estimates of hourly step length from other chimpanzee populations. The daily distance travelled was 987 m for the Mayebe community and 651 m for the Cyamudongo community. These daily distance estimates were based on the average hourly step length for each hour of the day because there were no days when locations were available for the chimpanzee communities for the entire 12 hours. The only study of chimpanzee communities that reported comparable metrics was that of Herbinger et al. (2001) from Taï National Park, Ivory Coast. They reported an average daily distance travelled between 2 and 4 km based on at least 5

hours of consecutive data, thus likely to be an underestimation of the real distance.

Regardless, this distance is still much greater than our estimate for both the Mayebe and Cyamudongo communities. We expected hourly step length and daily distance travelled to be short for the Cyamudongo community because of their space constraint.

However, given the large home range size of the Mayebe community and the lower fruit availability in the higher elevations, we would expect much longer step lengths in this group compared to the communities in Taï National Park, which is lower in elevation (80

– 396 m) and comparable in average annual rainfall (1800 mm; Boesch & Boesch,

1989). This result could reflect (1) low population density in Nyungwe (~0.4 chimpanzees/sq km; Wildlife Conservation Society, unpublished data) compared to Taï

National Park (1.84 chimpanzees/sq km; Herbinger et al. 2001), which means less time is needed to defend and patrol home range boundaries, or (2) the high percentage of fruit in their diet. In some other primates, travel distances are shorter when fruit is a larger fraction of the diet (O’Brien & Kinnaird, 1997).

As expected based on the species-time-area relationships (Krebs, 1999), we found that we had enough months of data for both chimpanzee communities to cover the full repertoire of species consumed. In addition, we found that the cumulative number of unique species consumed was higher for the Maybe community, which has a larger home range size, than the Cyamudongo community, with a smaller home range.

Both the Mayebe and Cyamudongo chimpanzee communities ate primarily fruits throughout the year, with fruits of Ficus spp. being the most frequently consumed food item for 7-8 months of the year. In the other months, the Mayebe community fed primarily on the fruit of two other trees, Syzygium guineense and Chrysophyllum

gorungosanum. In Cyamudongo, the only other primary food source was the fruits of

Trilepisium madagascariense. The tree species consumed by the Mayebe community were also recorded in the diet of chimpanzees in Bwindi Impenetrable Forest, Uganda

(Stanford & Nkurunungi, 2003) and Budongo Forest Reserve, Uganda (Tweheyo, Lye,

& Weladji, 2004); the most commonly consumed food in both populations was also

Ficus spp. In Kahuzi-Biega National Park, Democratic Republic of Congo, Ficus spp. were also the most commonly consumed taxon, but there were fewer records of consumption of S. guineense, and no records of chimpanzees eating C. gorungosanum, though it was present in the area (Basabose, 2002). The only other eastern chimpanzee communities reported to eat T. madagascariense were in Kibale National Park and

Budongo Forest Reserve, Uganda (Krief et al., 2006; Tweheyo & Babweteera, 2007). In addition, the Nigerian-Cameroon subspecies of chimpanzee was recorded feeding on fruits of T. madagascariense especially during the wet season (Dutton & Chapman,

2014). In Cyamudongo, it is likely that T. madagascariense has the longest fruiting period (April – December) of any tree, which may explain why it was eaten throughout most of the year (Wildlife Conservation Society, unpublished data). However, it is important to note that our diet data consisted of the number of days when each plant species was consumed by the chimpanzees and not the amount of time per day spent feeding on each species.

Eastern chimpanzees are threatened across their range by poaching, habitat loss, mining, climate change, and disease (Plumptre et al., 2016). Knowledge of the home range of chimpanzees and how it shifts during the year, the distance moved by each community daily, as well as the dominant food source for each month, can help

park rangers monitor and protect chimpanzees against these threats. The core home range (calculated as 50% KDE; Figure 4-1) represents critical habitat for the chimpanzees and where they spend half of their time on an annual basis. Increasing ranger patrols in the core home range area for each community can minimize threats from poaching, as well as habitat destruction and mining. Because of the ecological significance of chimpanzees to the ecosystem as a whole, the importance of conserving chimpanzees in Nyungwe cannot be overemphasized. The designation of critical habitat and information on the diet and spatial patterns of the chimpanzees reported in this study also protects sympatric species, and thus is crucial to improve NNP management as well as to develop and implement effective conservation plans for the national park.

Table 4-1. Home range sizes for the Mayebe and Cyamudongo chimpanzee communities in Nyungwe National Park, Rwanda. Overall, seasonal, and monthly home range sizes (HR; km2) calculated using 50%, 95%, and 100% minimum convex polygon (MCP) and 50% and 95% fixed kernel density estimates (KDE). Monthly home range estimates on the bolded lines refer to the wet season months, with months on the unbolded lines referring to the dry season months.

Mayebe Cyamudongo 50% 95% 100% 50% 95% 50% 95% 100% 50% 95% MCP MCP MCP KDE KDE MCP MCP MCP KDE KDE Overall HR 5.63 25.31 60.98 5.17 21.23 0.62 5.11 8.18 0.56 3.72

Seasonal Wet Season 5.95 22.95 55.30 5.72 21.78 0.50 4.58 8.00 0.54 3.88 HR Dry Season 3.55 17.61 45.12 3.42 19.33 0.71 4.78 6.46 0.63 3.81

Monthly January 2.89 15.35 23.62 4.91 23.40 0.29 1.72 3.24 0.50 2.72 HR February 3.38 13.35 26.08 4.64 20.49 0.75 3.49 4.73 1.18 4.91 March 2.72 13.62 34.54 3.16 17.48 0.83 3.69 4.50 1.36 5.36 April 1.93 9.87 26.58 2.28 12.67 0.51 2.86 4.14 0.80 4.10 May 1.34 10.82 24.19 1.43 13.62 0.42 3.65 4.31 0.73 4.43 June 1.20 9.88 17.04 2.21 13.32 0.50 4.20 5.61 0.72 4,48 July 3.56 17.10 26.66 4.04 19.59 0.41 2.82 3.48 0.53 3.27 August 2.63 15.20 26.13 2.66 20.00 1.21 4.42 4.93 0.98 4.67 September 10.23 15.17 20.92 4.54 22.28 0.62 3.94 5.03 0.75 4.50 October 5.59 19.12 35.47 5.42 23.05 0.30 3.52 4.33 0.48 3.62 November 4.46 15.94 32.48 3.99 18.58 0.23 2.13 5.17 0.38 3.06 December 0.82 12.65 19.75 1.51 13.44 0.27 2.33 4.61 0.37 2.55

Table 4-2. Hourly step length (m) overall, by season, and by month (A) and daily distance moved (m) overall and by season (B) for the Mayebe and Cyamudongo chimpanzee communities in Nyungwe National Park, Rwanda. Seasonal and monthly estimates are averaged for all years of the study period. Daily distance moved is based on the average hourly step length at each hour of the day for 2000-2015 for the Mayebe community and 2005- 2015 for the Cyamudongo community. Monthly average hourly step lengths within the wet season are in bolded rows, while the unbolded rows correspond to the dry season. A. Mayebe Cyamudongo Number of Hourly Step Number of Hourly Step Hourly Step Length (m) ± Hourly Length (m) ± SE Lengths SE Step Lengths Overall 8122 75 ± 5 9172 52 ± 3 Seasonal Wet 4860 79 ± 7 5286 44 ± 3 Season Dry 3262 69 ± 7 3886 63 ± 5 Season Monthly January 669 42 ± 6 658 76 ± 14 February 769 22 ± 5 679 59 ± 13 March 904 93 ± 19 694 41 ± 7 April 670 39 ± 6 744 33 ± 5 May 764 103 ± 36 822 49 ± 6 June 742 77 ± 16 852 68 ± 12 July 617 102 ± 16 861 62 ± 11 August 465 133 ± 30 836 54 ± 7 September 564 75 ± 20 690 58 ± 8 October 643 102 ± 12 751 67 ± 14 November 734 87 ± 9 753 29 ± 4 December 581 41 ± 6 832 32 ± 5

B. Mayebe Cyamudongo Daily Distance Moved (m) ± SE Daily Distance Moved (m) ± SE Overall 987 ± 71 651 ± 71 Seasonal Wet Season 1030 ± 94 559 ± 48 Dry Season 915 ± 108 768 ± 68

Table 4-3. Proportion of days (± SE) within each month averaged over the years of data collection when the chimpanzees were observed consuming each tree species for the Mayebe community over 11.5 years from mid-2000 - 2011 (A) and the Cyamudongo community over 7 years from 2005 – 2011 (B). Table includes tree species that were consumed on at least 10 days during the study period. Bolded values represent the species with the highest average proportion of days consumed for each month. A. Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Number of 160 161 183 156 158 169 164 131 153 166 162 132 Observation Days Species Chrysophyllum 0.061 ± 0.068 ± 0.045 ± 0.134 ± 0.190 ± 0.273 ± 0.143 ± 0.099 ± 0 0 0 0 gorungosanum 0.056 0.051 0.028 0.078 0.103 0.094 0.098 0.088 Dombeya 0.015 ± 0.026 ± 0.115 ± 0.083 ± 0.069 ± 0.018 ± 0.016 ± 0.033 ± 0.008 ± 0.017 ± 0.012 ± 0 goetzenii 0.011 0.014 0.039 0.073 0.064 0.013 0.013 0.033 0.005 0.008 0.008 Ekebergia 0.005 ± 0.005 ± 0.086 ± 0.123 ± 0.076 ± 0 0 0 0 0 0 0 capensis 0.005 0.005 0.048 0.059 0.040 Ficus spp 0.439 ± 0.588 ± 0.562 ± 0.464 ± 0.265 ± 0.164 ± 0.254 ± 0.295 ± 0.329 ± 0.194 ± 0.034 ± 0.028 ± 0.112 0.098 0.098 0.112 0.082 0.055 0.055 0.106 0.094 0.111 0.028 0.016 Macaranga 0.006 ± 0.017 ± 0.078 ± 0.010 ± 0.007 ± 0.004 ± 0 0 0 0 0 0 kilimandscharica 0.006 0.017 0.052 0.010 0.007 0.004 Myrianthus holstii 0.055 ± 0.013 ± 0.035 ± 0.083 ± 0.134 ± 0.017 ± 0 0 0 0 0 0 0.035 0.013 0.019 0.031 0.074 0.017 Olea capensis 0.070 ± 0.109 ± 0.082 ± 0.117 ± 0.105 ± 0.093 ± 0 0 0 0 0 0 0.070 0.089 0.068 0.086 0.090 0.089 Pancovia 0.005 ± 0.050 ± 0.094 ± 0.008 ± 0 0 0 0 0 0 0 0 golungensis 0.005 0.050 0.074 0.008 Prunus africana 0.008 ± 0.033 ± 0.100 ± 0.014 ± 0.010 ± 0.042 ± 0.014 ± 0.042 ± 0.006 ± 0 0 0 0.005 0.033 0.067 0.010 0.007 0.033 0.007 0.028 0.006 Symphonia 0.042 ± 0.108 ± 0.224 ± 0.128 ± 0.040 ± 0 0 0 0 0 0 0 globulifera 0.022 0.072 0.118 0.064 0.033 Syzygium 0.266 ± 0.055 ± 0.073 ± 0.028 ± 0.006 ± 0.028 ± 0.199 ± 0.567 ± 0.780 ± 0.921 ± 0 0 guineense 0.130 0.050 0.073 0.028 0.006 0.028 0.100 0.133 0.107 0.036 Trema orientalis 0.010 ± 0.050 ± 0.006 ± 0.006 ± 0 0 0 0 0 0 0 0 0.010 0.046 0.006 0.006

Table 4-3. Continued B. Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Number of 155 141 136 149 181 176 182 173 154 151 158 169 Observation Days Species Ficus spp 0.519 ± 0.587 ± 0.757 ± 0.508 ± 0.519 ± 0.348 ± 0.349 ± 0.265 ± 0.279 ± 0.120 ± 0.107 ± 0.145 ± 0.145 0.091 0.035 0.101 0.139 0.113 0.101 0.111 0.057 0.037 0.107 0.133 Musanga leo- 0.102 ± 0.065 ± 0.048 ± 0.128 ± 0.059 ± 0.058 ± 0.028 ± 0.032 ± 0.052 ± 0.043 ± 0.010 ± 0 errerae 0.082 0.024 0.024 0.044 0.018 0.018 0.011 0.026 0.034 0.021 0.007 Trema orientalis 0.015 ± 0.038 ± 0.009 ± 0 0.011 ± 0.011 ± 0.016 ± 0.122 ± 0.040 ± 0.008 ± 0 0 0.015 0.038 0.009 0.011 0.007 0.016 0.106 0.027 0.008 Trilepisium 0.308 ± 0.160 ± 0.009 ± 0.187 ± 0.360 ± 0.470 ± 0.523 ± 0.459 ± 0.450 ± 0.650 ± 0.866 ± 0.829 ± madagascariense 0.135 0.089 0.009 0.064 0.128 0.113 0.099 0.122 0.128 0.117 0.104 0.128

Figure 4-1. Overview of Nyungwe National Park and its location within Rwanda including tourist trails, roads, and the overall home range of each of the two habituated chimpanzee communities. Outer line represents the 95% fixed kernel density estimates (KDE), and inner area represents the 50% KDE.

Figure 4-2. Seasonal home range for the (A) Mayebe and (B) Cyamudongo chimpanzee communities. Blue line denotes the wet season 95% fixed kernel density estimates (March-May, September-December), and brown area represents the dry season 95% fixed kernel density estimates (January – February, June – August).

Figure 4-3. Monthly home range for the Mayebe community (A) and Cyamudongo community (B) of chimpanzees. Black line denotes the 50% fixed kernel density estimate (KDE) for the stated month.

Figure 4-4. Average hourly step length (± SE) by wet and dry season for (A) Mayebe and (B) Cyamudongo chimpanzee communities. Data were not available for the Cyamudongo group for 5 am

Figure 4-5. The cumulative number of plant species consumed over months of observation for each chimpanzee community.

CHAPTER 5 CONCLUSIONS AND MANAGEMENT IMPLICATIONS

Species conservation programs and protected area management plans should be guided by the best available science (Mills, Quigley, & Everest, 2001). Using long- term data and innovative uses for occupancy models, we provided insight into spatial and temporal trends in illegal poaching activity and mammalian species richness and distribution. In addition, we explored the ranging and diet behaviors of the endangered eastern chimpanzee (Pan troglodytes schweinfurthii). These results are crucial for informing science-based management decisions for the long-term survival of both species and habitats within Nyungwe National Park, Rwanda.

Using 10-year of ranger-based monitoring data and multi-season occupancy models, we found that the probability of occurrence of poaching activity is highest at lower elevations (1801-2200 m), near roads, and near tourist trails. The probability of occurrence of poaching was lowest at high elevation (2601-3000 m), near the park boundary, and near ranger posts. Additionally, we found that the number of ranger patrols affects the probability of extinction of poaching-related threats in areas where they were present in the previous year. With zero ranger patrols, the probability of extinction of threat is only 7%, but this percentage increases to 20% and 57% with 20 and 50 ranger visits to each site per year. Therefore, by increasing the number of patrols to areas with a high probability of poaching activity, we can effectively reduce poaching within Nyungwe National Park. This analysis is a novel application of multi- season occupancy models that explicitly considers imperfect detection of poaching activity.

Mammalian species richness and the distributional range of five of seven species increased between 2009 and 2014 in NNP. Using data from 41 line transects covering the entire national park, we used multi-season occupancy models with multi-species data to estimate species richness and distribution for species of interest between 2009 and 2014. We found that the probability of poaching activity, the distance to the nearest tourist trail, and the maximum elevation influenced the probability of colonization of a species into a new area in 2014, where it was not present in 2009, and thus part of the increase in species richness. The probability of colonization with no poaching was about

50%, but this dropped to about 10% with a 100% chance of poaching activity. For individual species distribution, duikers had the largest increase in distributional range between 2009 and 2014, while there was a decline in range for the eastern chimpanzee and blue monkey. This analysis provided estimates of species richness and species distribution using multi-season occupancy models that explicitly accounted for imperfect detection as well as specific-species identities. In addition, we highlighted areas of the park with low species richness as well as species with a limited distributional range.

Lastly, using long-term data from 2000 – 2015, we studied home range, movement, and diet of two communities of eastern chimpanzees. The Mayebe community, which inhabits the main forest block, had a home range size of 21 km2, which the Cyamudongo community, which is restricted to a 4 km2 forest fragment, had a home range size of only 4 km2. Home range sizes were smaller during the dry season and varied monthly throughout the year. The Mayebe community had a daily movement range of 987 m, with an average hourly step length of 75 m, while the daily movement range for the Cyamudongo community was 651 m with an average hourly step length of

52 m. Both communities fed primarily on Ficus spp. throughout the year, with other important dietary items including the fruits of Symphonia globulifera, Syzygium guineense, and Chrysophyllum gorungosanum for the Mayebe community and

Trilepisium madagascariense for the Cyamudongo community. This study provided the first estimates of home range size and movement parameters for chimpanzees in

Rwanda, and also highlighted the core home range for each chimpanzee community, which should be protected to ensure the long-term survival of this species.

Future management plans for Nyungwe National Park should incorporate the findings from this research. Ranger patrol protocols should be managed adaptively to focus on areas of the park with a high probability of occurrence of poaching activity, low species richness, and within the core home range of the chimpanzee communities. As further data is collected in the park, these same models should be implemented to investigate continuing trends in poaching activity, species richness, and species distribution. Additionally, with the new insight on eastern chimpanzees, we can now better protect the habituated groups throughout the year based on our knowledge of their monthly home range and important diet resources. More broadly, the methodologies used in this research can be implemented elsewhere around the world.

Using similar long-term data and these models, we can optimize monitoring and ranger patrol protocols, and work towards better protecting our national parks, wildlife, and their habitats.

APPENDIX A MODEL SELECTION TABLE FOR POACHING ANALYSIS

Table A-1. Model comparison statistics for multi-season occupancy models testing for covariate effects on initial occupancy (휓1), probability of extinction (휖), probability of colonization (훾), and detection probability (푝). Model structure, number of parameters, AICc (Akaike Information Criterion corrected for small sample size) score, ΔAICc (difference in AICc score between top model and a selected model), and model weight are also given. Model comparison results are presented for top 100 models.

Model npar AICc ΔAICc Weight

1 휓1(minE)휖(rvst)훾(boundary)p(year + area) 19 38559.52 0 0.81

2 휓1(boundary)휖(rvst)훾(boundary)p(year + area) 17 38565.29 5.76 0.05

3 휓1(rangpost)휖(rvst)훾(boundary)p(year + area) 17 38565.30 5.78 0.05

4 휓1(1) 휖(rvst)훾(boundary)p(year + area) 16 38565.44 5.92 0.04

5 휓1(road)휖(rvst)훾(boundary)p(year + area) 17 38566.42 6.90 0.03

6 휓1(trail)휖(rvst)훾(boundary)p(year + area) 17 38566.60 7.08 0.02

7 휓1(maxE)휖(rvst)훾(boundary)p(year + area) 19 38568.90 9.38 0.01

8 휓1(minE)휖(rvst)훾(rangpost)p(year + area) 19 38576.60 17.08 1.58E-04

9 휓1(boundary)휖(rvst)훾(rangpost)p(year + area) 17 38580.26 20.73 2.55E-05

10 휓1(minE)휖(rvst)훾(road)p(year + area) 19 38581.52 22.00 1.35E-05

11 휓1(minE)휖(rvst)훾(trail)p(year + area) 19 38582.23 22.71 9.49E-06

12 휓1(1)휖(rvst)훾(rangpost)p(year + area) 16 38582.29 22.77 9.22E-06

13 휓1(rangpost)휖(rvst)훾(rangpost)p(year + area) 17 38582.97 23.44 6.58E-06

14 휓1(road)휖(rvst)훾(rangpost)p(year + area) 17 38583.01 23.49 6.44E-06

15 휓1(trail)휖(rvst)훾(rangpost)p(year + area) 17 38583.25 23.73 5.70E-06

16 휓1(boundary)휖(rvst)훾(road)p(year + area) 17 38583.55 24.02 4.92E-06

17 휓1(boundary)휖(rvst)훾(trail)p(year + area) 17 38584.01 24.49 3.90E-06

18 휓1(rangpost)휖(rvst)훾(road)p(year + area) 17 38584.96 25.43 2.43E-06

19 휓1(rangpost)휖(rvst)훾(trail)p(year + area) 17 38585.42 25.89 1.93E-06

20 휓1(maxE)휖(rvst)훾(rangpost)p(year + area) 19 38585.72 26.20 1.66E-06

21 휓1(1)휖(rvst)훾(road)p(year + area) 16 38586.15 26.62 1.34E-06

22 휓1(1)휖(rvst)훾(trail)p(year + area) 16 38586.95 27.43 8.97E-07

23 휓1(road)휖(rvst)훾(road)p(year + area) 17 38587.86 28.34 5.68E-07

24 휓1(trail)휖(rvst)훾(road)p(year + area) 17 38588.07 28.54 5.14E-07

25 휓1(road)휖(rvst)훾(trail)p(year + area) 17 38588.69 29.17 3.76E-07

26 휓1(trail)휖(rvst)훾(trail)p(year + area) 17 38588.91 29.39 3.36E-07

27 휓1(maxE)휖(rvst)훾(road)p(year + area) 19 38590.06 30.54 1.89E-07

28 휓1(maxE)휖(rvst)훾(trail)p(year + area) 19 38590.92 31.40 1.23E-07

29 휓1(minE)휖(rvst)훾(maxE)p(year + area) 21 38598.29 38.77 3.09E-09

30 휓1(boundary)휖(rvst)훾(maxE)p(year + area) 19 38599.17 39.65 1.99E-09

31 휓1(boundary)휖(rvst)훾(minE)p(year + area) 19 38600.38 40.86 1.09E-09

32 휓1(minE)휖(rvst)훾(minE)p(year + area) 21 38600.59 41.06 9.82E-10

Table A-1. Continued Model npar AICc ΔAICc Weight

33 휓1(rangpost)휖(rvst)훾(maxE)p(year + area) 19 38601.31 41.80 6.82E-10

34 휓1(rangpost)휖(rvst)훾(minE)p(year + area) 19 38602.51 42.99 3.76E-10

35 휓1(1)휖(rvst)훾(maxE)p(year + area) 18 38602.81 43.28 3.24E-10

36 휓1(minE)휖(rvst)훾(1)p(year + area) 18 38603.23 43.70 2.62E-10

37 휓1(road)휖(rvst)훾(maxE)p(year + area) 19 38603.59 44.07 2.19E-10

38 휓1(trail)휖(rvst)훾(maxE)p(year + area) 19 38603.96 44.43 1.82E-10

39 휓1(1)휖(rvst)훾(minE)p(year + area) 18 38604.18 44.66 1.62E-10

40 휓1(boundary)휖(rvst)훾(1)p(year + area) 16 38604.46 44.94 1.42E-10

41 휓1(road)휖(rvst)훾(minE)p(year + area) 19 38604.89 45.37 1.14E-10

42 휓1(minE)휖(rvst)훾(rvst)p(year + area) 19 38605.23 45.71 9.61E-11

43 휓1(trail)휖(rvst)훾(minE)p(year + area) 19 38605.24 45.72 9.57E-11

44 휓1(boundary)휖(rvst)훾(rvst)p(year + area) 17 38606.46 46.94 5.19E-11

45 휓1(rangpost)휖(rvst)훾(1)p(year + area) 16 38606.77 47.24 4.46E-11

46 휓1(maxE)휖(rvst)훾(maxE)p(year + area) 21 38606.97 47.44 4.04E-11

47 휓1(maxE)휖(rvst)훾(minE)p(year + area) 21 38607.57 48.05 2.98E-11

48 휓1(1)휖(rvst)훾(1)p(year + area) 15 38607.91 48.39 2.52E-11

49 휓1(road)휖(rvst)훾(1)p(year + area) 16 38608.61 49.08 1.78E-11

50 휓1(rangpost)휖(rvst)훾(rvst)p(year + area) 17 38608.77 49.25 1.64E-11

51 휓1(trail)휖(rvst)훾(1)p(year + area) 16 38609.05 49.52 1.43E-11

52 휓1(minE)휖(boundary)훾(rangpost)p(year + area) 19 38609.16 49.64 1.35E-11

53 휓1(1)휖(rvst)훾(rvst)p(year + area) 16 38609.90 50.38 9.30E-12

54 휓1(road)휖(rvst)훾(rvst)p(year + area) 17 38610.60 51.08 6.57E-12

55 휓1(trail)휖(rvst)훾(rvst)p(year + area) 17 38611.05 51.52 5.26E-12

56 휓1(maxE)휖(rvst)훾(1)p(year + area) 18 38611.58 52.06 4.02E-12

57 휓1(boundary)휖(boundary)훾(rangpost)p(year + area) 17 38612.21 52.69 2.94E-12

58 휓1(maxE)휖(rvst)훾(rvst)p(year + area) 19 38613.58 54.06 1.48E-12

59 휓1(1)휖(boundary)훾(rangpost)p(year + area) 16 38613.71 54.18 1.39E-12

60 휓1(rangpost)휖(boundary)훾(rangpost)p(year + area) 17 38614.73 55.21 8.32E-13

61 휓1(road)휖(boundary)훾(rangpost)p(year + area) 17 38615.05 55.53 7.08E-13

62 휓1(trail)휖(boundary)훾(rangpost)p(year + area) 17 38615.44 55.92 5.84E-13

63 휓1(minE)휖(boundary)훾(trail)p(year + area) 19 38617.49 57.97 2.09E-13

64 휓1(maxE)휖(boundary)훾(rangpost)p(year + area) 19 38617.61 58.08 1.97E-13

65 휓1(boundary)휖(boundary)훾(trail)p(year + area) 17 38618.57 59.04 1.22E-13

66 휓1(minE)휖(boundary)훾(road)p(year + area) 19 38619.33 59.81 8.34E-14

67 휓1(boundary)휖(boundary)훾(road)p(year + area) 17 38620.61 61.09 4.40E-14

68 휓1(rangpost)휖(boundary)훾(trail)p(year + area) 17 38620.69 61.17 4.23E-14

69 휓1(1)휖(boundary)훾(trail)p(year + area) 16 38620.84 61.32 3.92E-14

70 휓1(minE)휖(boundary)훾(rvst)p(year + area) 19 38621.45 61.93 2.90E-14

71 휓1(boundary)휖(boundary)훾(minE)p(year + area) 19 38621.93 62.41 2.27E-14

72 휓1(minE)휖(boundary)훾(minE)p(year + area) 21 38622.11 62.59 2.08E-14

73 휓1(boundary)휖(boundary)훾(rvst)p(year + area) 17 38622.58 63.06 1.64E-14

Table A-1. Continued Model npar AICc ΔAICc Weight

74 휓1(road)휖(boundary)훾(trail)p(year + area) 17 38622.59 63.07 1.63E-14

75 휓1(minE)휖(boundary)훾(boundary)p(year + area) 19 38622.60 63.08 1.63E-14

76 휓1(1)휖(boundary)훾(road)p(year + area) 16 38622.72 63.19 1.54E-14

77 휓1(rangpost)휖(boundary)훾(road)p(year + area) 17 38622.72 63.19 1.54E-14

78 휓1(trail)휖(boundary)훾(trail)p(year + area) 17 38622.84 63.32 1.44E-14

79 휓1(minE)휖(boundary)훾(maxE)p(year + area) 21 38623.99 64.47 8.12E-15

80 휓1(rangpost)휖(boundary)훾(minE)p(year + area) 19 38624.15 64.62 7.50E-15

81 휓1(road)휖(boundary)훾(road)p(year + area) 17 38624.43 64.91 6.51E-15

82 휓1(1)휖(boundary)훾(rvst)p(year + area) 16 38624.68 65.16 5.76E-15

83 휓1(trail)휖(boundary)훾(road)p(year + area) 17 38624.70 65.17 5.71E-15

84 휓1(1)휖(boundary)훾(minE)p(year + area) 18 38624.83 65.30 5.35E-15

85 휓1(boundary)휖(boundary)훾(maxE)p(year + area) 19 38624.84 65.32 5.30E-15

86 휓1(rangpost)휖(boundary)훾(rvst)p(year + area) 17 38625.11 65.59 4.64E-15

87 휓1(maxE)휖(boundary)훾(trail)p(year + area) 19 38625.30 65.77 4.23E-15

88 휓1(road)휖(boundary)훾(rvst)p(year + area) 17 38625.97 66.44 3.03E-15

89 휓1(boundary)휖(boundary)훾(boundary)p(year + area) 17 38626.19 66.67 2.70E-15

90 휓1(road)휖(boundary)훾(minE)p(year + area) 19 38626.20 66.68 2.69E-15

91 휓1(trail)휖(boundary)훾(rvst)p(year + area) 17 38626.55 67.03 2.26E-15

92 휓1(trail)휖(boundary)훾(minE)p(year + area) 19 38626.60 67.08 2.20E-15

93 휓1(1)휖(boundary)훾(boundary)p(year + area) 16 38626.78 67.26 2.01E-15

94 휓1(rangpost)휖(boundary)훾(maxE)p(year + area) 19 38626.90 67.38 1.90E-15

95 휓1(maxE)휖(boundary)훾(road)p(year + area) 19 38627.09 67.57 1.73E-15

96 휓1(rangpost)휖(boundary)훾(boundary)p(year + area) 17 38627.35 67.83 1.52E-15

97 휓1(1)휖(boundary)훾(maxE)p(year + area) 18 38627.45 67.92 1.44E-15

98 휓1(road)휖(boundary)훾(boundary)p(year + area) 17 38628.13 68.61 1.03E-15

99 휓1(trail)휖(boundary)훾(boundary)p(year + area) 17 38628.53 69.01 8.39E-16

100 휓1(maxE)휖(boundary)훾(minE)p(year + area) 21 38628.82 69.30 7.26E-16

APPENDIX B MAMMAL SPECIES LIST FOR NYUNGWE NATIONAL PARK, RWANDA

Table B-1. Full mammal species list for Nyungwe National Park, Rwanda including the associated covariate values used for species richness and distribution analyses.

Scientific* English Order IUCN Statusa Groupb Mass Groupc Obsd Aonyx congicus Congo Clawless Otter Carnivora NT 0 Lg Direct Atilax paludinosus Marsh Mongoose Carnivora LC 0 Med Direct Canis adustus Side-striped Jackal Carnivora LC 0 Lg Direct Cephalophus spp * Duiker Species Artiodactyla LC 0 Dung Dung Cercopithecus aethiops Vervet Monkey Primate LC 1 Med Direct Cercopithecus ascanius Redtail Monkey Primate LC 1 Med Direct Cercopithecus hamlyni Owl-faced Monkey Primate VU 1 Med Direct Cercopithecus lhoesti * L'hoest's Monkey Primate VU 1 Med Direct Cercopithecus mitis * Blue Monkey Primate LC 1 Med Direct Cercopithecus mona Mona Monkey Primate LC 1 Med Direct Colobus angolensis * Angolan Colobus Primate LC 1 Med Direct Felis aurata Golden Cat Carnivora NT 0 Lg Direct Felis serval Serval Carnivora LC 0 Lg Direct Felis silvestris Wild Cat Carnivora LC 0 Sm Direct Funisciurus carruthersi Carruther's Mountain Tree Squirrel Rodentia LC 0 Sm Direct Funisciurus pyrropus Cuvier's Fire-footed Squirrel Rodentia LC 0 Sm Direct Heliosciurus ruwenzorii Montane Sun Squirrel Rodentia LC 0 Sm Direct Herpestes ichneumon Ichneumon Mongoose Carnivora LC 0 Med Direct Herpestes sanguineus Slender Mongoose Carnivora LC 0 Sm Direct

Table B-1. Continued Scientific* English Order IUCN Statusa Groupb Mass Groupc Obsd Lophocebus albigena * Grey-cheeked Mangabey Primate LC 1 Med Direct Mellivora capensis Honey Badger Carnivora LC 0 Med Direct Panthera pardus Leopard Carnivora NT 0 Lg Direct Pan troglodytes * Chimpanzee Primate EN 1 Nest Nest Papio Anubis Olive Baboon Primate LC 1 Lg Direct Paraxerus alexandri Alexander's Squirrel Rodentia LC 0 Sm Direct Paraxerus boehmi Boehm's Squirrel Rodentia LC 0 Sm Direct Potamochoerus larvatus * Pig Species Artiodactyla LC 1 Dung Dung Hylochoerus meintzhageni Protoxerus stangeri African Giant Squirrel Rodentia LC 0 Sm Direct Tragelaphus scriptus Bushbuck Cetartiodactyla LC 0 Lg Direct * denotes species included in the species distribution analysis a IUCN categories include: LC (least concern), VU (vulnerable), NT (near threatened), EN (endangered) b Group covariate for species richness analysis: 0 (solitary), 1 (group-living) c Mass group covariate for species richness analysis: Sm (small animal < 3000 g average adult body mass), Med (medium animal >3000 g and < 10000 g), Lg (large animal > 10000 g), Dung (indirect sighting of dung), Nest (indirect sighting of nest) d Observation type covariate for species distribution analysis: Direct (direct sighting of animal), Dung (indirect sighting of dung), Nest (indirect sighting of nest)

100

APPENDIX C MODEL SELECTION TABLES FOR SPECIES RICHNESS AND SPECIES DISTRIBUTION ANALYSES

Table C-1. Model comparison statistics for multi-season occupancy models for the species richness analysis testing for covariate effects on seasonal occupancy (휓푡푖), probability of colonization (훾), and detection probability (푝). Model structure, number of parameters, AICc (Akaike Information Criterion corrected for small sample size) score, ΔAICc (difference in AICc score between top model and a selected model), and model weight are also given. Model comparison results are presented for top 100 models.

Model npar AICc ΔAICc Weight 1 Psi(~Trail)Epsilon(~minElev)Gamma(~poach)p(~obs + group) 12 3474.326 0.000 0.0601 2 Psi(~maxElev)Epsilon(~minElev)Gamma(~poach)p(~obs + group) 12 3474.987 0.661 0.0432 3 Psi(~poach)Epsilon(~minElev)Gamma(~Trail)p(~obs + group) 12 3475.851 1.524 0.0281 4 Psi(~maxElev)Epsilon(.)Gamma(~poach)p(~obs + group) 11 3476.109 1.783 0.0246 5 Psi(~minElev)Epsilon(~minElev)Gamma(~poach)p(~obs + group) 12 3476.151 1.824 0.0241 6 Psi(~poach)Epsilon(~minElev)Gamma(~poach)p(~obs + group) 12 3476.236 1.910 0.0231 7 Psi(~Access)Epsilon(~minElev)Gamma(~poach)p(~obs + group) 12 3476.312 1.986 0.0223 8 Psi(~Trail)Epsilon(~Trail)Gamma(~poach)p(~obs + group) 12 3476.370 2.044 0.0216 9 Psi(~Trail)Epsilon(~maxElev)Gamma(~poach)p(~obs + group) 12 3476.407 2.081 0.0212 10 Psi(~rp)Epsilon(~minElev)Gamma(~poach)p(~obs + group) 12 3476.541 2.215 0.0199 11 Psi(~maxElev)Epsilon(~Access)Gamma(~poach)p(~obs + group) 12 3476.641 2.315 0.0189 12 Psi(~maxElev)Epsilon(~minElev)Gamma(~Trail)p(~obs + group) 12 3476.734 2.408 0.0180 13 Psi(~maxElev)Epsilon(~maxElev)Gamma(~poach)p(~obs + group) 12 3476.745 2.418 0.0179 14 Psi(~Trail)Epsilon(~Access)Gamma(~poach)p(~obs + group) 12 3476.891 2.565 0.0167 15 Psi(~Trail)Epsilon(.)Gamma(~poach)p(~obs + group) 11 3476.905 2.578 0.0166 16 Psi(~Trail)Epsilon(~minElev)Gamma(~Trail)p(~obs + group) 12 3476.940 2.614 0.0163 17 Psi(~maxElev)Epsilon(~Trail)Gamma(~poach)p(~obs + group) 12 3477.006 2.680 0.0157 18 Psi(~Access)Epsilon(~minElev)Gamma(~Trail)p(~obs + group) 12 3477.172 2.846 0.0145 19 Psi(~poach)Epsilon(~Access)Gamma(~Trail)p(~obs + group) 12 3477.263 2.937 0.0138 20 Psi(~maxElev)Epsilon(~Access)Gamma(~Trail)p(~obs + group) 12 3477.273 2.947 0.0138 21 Psi(~poach)Epsilon(~Trail)Gamma(~Trail)p(~obs + group) 12 3477.543 3.216 0.0120 22 Psi(~poach)Epsilon(~maxElev)Gamma(~Trail)p(~obs + group) 12 3477.649 3.323 0.0114

101

Table C-1. Continued Model npar AICc ΔAICc Weight 23 Psi(~maxElev)Epsilon(~rp)Gamma(~poach)p(~obs + group) 12 3477.698 3.372 0.0111 24 Psi(~maxElev)Epsilon(~Trail)Gamma(~Trail)p(~obs + group) 12 3477.764 3.438 0.0108 25 Psi(~minElev)Epsilon(.)Gamma(~poach)p(~obs + group) 11 3477.816 3.490 0.0105 26 Psi(~minElev)Epsilon(~minElev)Gamma(~Trail)p(~obs + group) 12 3477.902 3.575 0.0101 27 Psi(~maxElev)Epsilon(~poach)Gamma(~poach)p(~obs + group) 12 3477.919 3.592 0.0100 28 Psi(~minElev)Epsilon(~maxElev)Gamma(~poach)p(~obs + group) 12 3477.993 3.667 0.0096 29 Psi(~maxElev)Epsilon(.)Gamma(~Trail)p(~obs + group) 11 3478.035 3.709 0.0094 30 Psi(~rp)Epsilon(~minElev)Gamma(~Trail)p(~obs + group) 12 3478.107 3.781 0.0091 31 Psi(~poach)Epsilon(.)Gamma(~Trail)p(~obs + group) 11 3478.130 3.804 0.0090 32 Psi(~poach)Epsilon(~maxElev)Gamma(~poach)p(~obs + group) 12 3478.219 3.893 0.0086 33 Psi(~minElev)Epsilon(~Access)Gamma(~poach)p(~obs + group) 12 3478.221 3.895 0.0086 34 Psi(~Access)Epsilon(~maxElev)Gamma(~poach)p(~obs + group) 12 3478.231 3.905 0.0085 35 Psi(~Trail)Epsilon(~Access)Gamma(~Trail)p(~obs + group) 12 3478.387 4.060 0.0079 36 Psi(~maxElev)Epsilon(~maxElev)Gamma(~Trail)p(~obs + group) 12 3478.419 4.093 0.0078 37 Psi(~Trail)Epsilon(~rp)Gamma(~poach)p(~obs + group) 12 3478.446 4.120 0.0077 38 Psi(~Access)Epsilon(.)Gamma(~poach)p(~obs + group) 11 3478.514 4.188 0.0074 39 Psi(~poach)Epsilon(.)Gamma(~poach)p(~obs + group) 11 3478.582 4.255 0.0072 40 Psi(~rp)Epsilon(~maxElev)Gamma(~poach)p(~obs + group) 12 3478.641 4.315 0.0070 41 Psi(~poach)Epsilon(~Access)Gamma(~poach)p(~obs + group) 12 3478.666 4.340 0.0069 42 Psi(~Trail)Epsilon(~Trail)Gamma(~Trail)p(~obs + group) 12 3478.718 4.391 0.0067 43 Psi(~minElev)Epsilon(~Trail)Gamma(~poach)p(~obs + group) 12 3478.754 4.428 0.0066 44 Psi(~Trail)Epsilon(~poach)Gamma(~poach)p(~obs + group) 12 3478.755 4.429 0.0066 45 Psi(~Access)Epsilon(~maxElev)Gamma(~Trail)p(~obs + group) 12 3478.876 4.550 0.0062 46 Psi(~Access)Epsilon(~Access)Gamma(~Trail)p(~obs + group) 12 3478.888 4.562 0.0061 47 Psi(~Trail)Epsilon(~maxElev)Gamma(~Trail)p(~obs + group) 12 3478.932 4.606 0.0060 48 Psi(~Access)Epsilon(~Access)Gamma(~poach)p(~obs + group) 12 3478.946 4.620 0.0060 49 Psi(~minElev)Epsilon(~Access)Gamma(~Trail)p(~obs + group) 12 3478.955 4.629 0.0059 50 Psi(~Access)Epsilon(~Trail)Gamma(~Trail)p(~obs + group) 12 3479.032 4.705 0.0057 51 Psi(~rp)Epsilon(~Access)Gamma(~poach)p(~obs + group) 12 3479.185 4.859 0.0053 52 Psi(~rp)Epsilon(.)Gamma(~poach)p(~obs + group) 11 3479.205 4.879 0.0052

102

Table C-1. Continued Model npar AICc ΔAICc Weight 53 Psi(~maxElev)Epsilon(~rp)Gamma(~Trail)p(~obs + group) 12 3479.252 4.925 0.0051 54 Psi(~minElev)Epsilon(~rp)Gamma(~poach)p(~obs + group) 12 3479.309 4.983 0.0050 55 Psi(~Access)Epsilon(.)Gamma(~Trail)p(~obs + group) 11 3479.355 5.029 0.0049 56 Psi(~poach)Epsilon(~rp)Gamma(~Trail)p(~obs + group) 12 3479.389 5.063 0.0048 57 Psi(~poach)Epsilon(~Trail)Gamma(~poach)p(~obs + group) 12 3479.451 5.125 0.0046 58 Psi(~Access)Epsilon(~Trail)Gamma(~poach)p(~obs + group) 12 3479.498 5.172 0.0045 59 Psi(~rp)Epsilon(~Access)Gamma(~Trail)p(~obs + group) 12 3479.572 5.245 0.0044 60 Psi(~poach)Epsilon(.)Gamma(~maxElev)p(~obs + group) 11 3479.641 5.315 0.0042 61 Psi(~minElev)Epsilon(~poach)Gamma(~poach)p(~obs + group) 12 3479.654 5.327 0.0042 62 Psi(~poach)Epsilon(~minElev)Gamma(.)p(~obs + group) 11 3479.676 5.350 0.0041 63 Psi(~minElev)Epsilon(~maxElev)Gamma(~Trail)p(~obs + group) 12 3479.697 5.371 0.0041 64 Psi(~Trail)Epsilon(~minElev)Gamma(.)p(~obs + group) 11 3479.699 5.373 0.0041 65 Psi(~minElev)Epsilon(~Trail)Gamma(~Trail)p(~obs + group) 12 3479.706 5.380 0.0041 66 Psi(~poach)Epsilon(~minElev)Gamma(~maxElev)p(~obs + group) 12 3479.720 5.394 0.0041 67 Psi(~Trail)Epsilon(~minElev)Gamma(~Access)p(~obs + group) 12 3479.758 5.432 0.0040 68 Psi(~poach)Epsilon(~Access)Gamma(~maxElev)p(~obs + group) 12 3479.762 5.436 0.0040 69 Psi(~Trail)Epsilon(.)Gamma(~Trail)p(~obs + group) 11 3479.777 5.450 0.0039 70 Psi(~maxElev)Epsilon(~poach)Gamma(~Trail)p(~obs + group) 12 3479.847 5.521 0.0038 71 Psi(~Access)Epsilon(~rp)Gamma(~poach)p(~obs + group) 12 3479.916 5.590 0.0037 72 Psi(~minElev)Epsilon(.)Gamma(~Trail)p(~obs + group) 11 3479.949 5.623 0.0036 73 Psi(~poach)Epsilon(~rp)Gamma(~poach)p(~obs + group) 12 3479.954 5.628 0.0036 74 Psi(~Trail)Epsilon(~minElev)Gamma(~maxElev)p(~obs + group) 12 3480.002 5.675 0.0035 75 Psi(~Trail)Epsilon(~Access)Gamma(~maxElev)p(~obs + group) 12 3480.020 5.694 0.0035 76 Psi(~rp)Epsilon(~maxElev)Gamma(~Trail)p(~obs + group) 12 3480.052 5.726 0.0034 77 Psi(~poach)Epsilon(~poach)Gamma(~Trail)p(~obs + group) 12 3480.091 5.765 0.0034 78 Psi(~rp)Epsilon(~Trail)Gamma(~poach)p(~obs + group) 12 3480.206 5.879 0.0032 79 Psi(~Trail)Epsilon(.)Gamma(~maxElev)p(~obs + group) 11 3480.374 6.047 0.0029 80 Psi(~rp)Epsilon(~rp)Gamma(~poach)p(~obs + group) 12 3480.392 6.066 0.0029 81 Psi(~Access)Epsilon(~poach)Gamma(~poach)p(~obs + group) 12 3480.403 6.076 0.0029 82 Psi(~poach)Epsilon(~poach)Gamma(~poach)p(~obs + group) 12 3480.458 6.132 0.0028

103

Table C-1. Continued Model npar AICc ΔAICc Weight 83 Psi(~poach)Epsilon(~minElev)Gamma(~Access)p(~obs + group) 12 3480.491 6.165 0.0028 84 Psi(~Access)Epsilon(~rp)Gamma(~Trail)p(~obs + group) 12 3480.504 6.178 0.0027 85 Psi(~poach)Epsilon(~minElev)Gamma(~minElev)p(~obs + group) 12 3480.506 6.180 0.0027 86 Psi(~rp)Epsilon(~Trail)Gamma(~Trail)p(~obs + group) 12 3480.530 6.204 0.0027 87 Psi(~poach)Epsilon(.)Gamma(~minElev)p(~obs + group) 11 3480.566 6.240 0.0027 88 Psi(~Trail)Epsilon(.)Gamma(~Access)p(~obs + group) 11 3480.696 6.369 0.0025 89 Psi(~Trail)Epsilon(~minElev)Gamma(~rp)p(~obs + group) 12 3480.701 6.374 0.0025 90 Psi(~poach)Epsilon(~Access)Gamma(~minElev)p(~obs + group) 12 3480.703 6.377 0.0025 91 Psi(~poach)Epsilon(~Access)Gamma(.)p(~obs + group) 11 3480.744 6.418 0.0024 92 Psi(~Trail)Epsilon(~Access)Gamma(.)p(~obs + group) 11 3480.830 6.504 0.0023 93 Psi(~Trail)Epsilon(~rp)Gamma(~Trail)p(~obs + group) 12 3480.839 6.513 0.0023 94 Psi(~rp)Epsilon(.)Gamma(~Trail)p(~obs + group) 11 3480.897 6.571 0.0022 95 Psi(~Trail)Epsilon(~minElev)Gamma(~minElev)p(~obs + group) 12 3480.921 6.595 0.0022 96 Psi(~poach)Epsilon(~maxElev)Gamma(~maxElev)p(~obs + group) 12 3480.947 6.621 0.0022 97 Psi(~poach)Epsilon(~minElev)Gamma(~rp)p(~obs + group) 12 3480.949 6.622 0.0022 98 Psi(~minElev)Epsilon(~rp)Gamma(~Trail)p(~obs + group) 12 3480.997 6.670 0.0021 99 Psi(~Trail)Epsilon(~Trail)Gamma(~Access)p(~obs + group) 12 3481.045 6.719 0.0021 100 Psi(~poach)Epsilon(~rp)Gamma(~maxElev)p(~obs + group) 12 3481.107 6.781 0.0020

104

Table C-2. Model comparison statistics for multi-season occupancy models for the species distribution analysis testing for covariate effects on seasonal occupancy (휓푡푖), probability of colonization (훾), and detection probability (푝). Model structure, number of parameters, AICc (Akaike Information Criterion corrected for small sample size) score, ΔAICc (difference in AICc score between top model and a selected model), and model weight are also given. Model comparison results are presented for the top 100 models.

Model npar AICc ΔAICc Weight 1 Psi(~species)Epsilon(~minElev)Gamma(~Trail)p(~obsM) 14 2025.120 0.000 0.0368 2 Psi(~species + Trail)Epsilon(~minElev)Gamma(~Trail)p(~obsM) 15 2025.392 0.272 0.0321 3 Psi(~species)Epsilon(~Access)Gamma(~Trail)p(~obsM) 14 2025.589 0.469 0.0291 4 Psi(~species)Epsilon(.)Gamma(~Trail)p(~obsM) 13 2025.615 0.495 0.0287 5 Psi(~species + Trail)Epsilon(.)Gamma(~Trail)p(~obsM) 14 2025.761 0.641 0.0267 6 Psi(~species + Trail)Epsilon(~Access)Gamma(~Trail)p(~obsM) 15 2025.778 0.658 0.0265 7 Psi(~species + poach)Epsilon(~minElev)Gamma(~Trail)p(~obsM) 15 2026.318 1.198 0.0202 8 Psi(~species)Epsilon(~Trail)Gamma(~Trail)p(~obsM) 14 2026.484 1.364 0.0186 9 Psi(~species)Epsilon(~maxElev)Gamma(~Trail)p(~obsM) 14 2026.541 1.421 0.0181 10 Psi(~species + Trail)Epsilon(~Trail)Gamma(~Trail)p(~obsM) 15 2026.546 1.425 0.0180 11 Psi(~species + minElev)Epsilon(~minElev)Gamma(~Trail)p(~obsM) 15 2026.659 1.539 0.0170 12 Psi(~species + poach)Epsilon(.)Gamma(~Trail)p(~obsM) 14 2026.681 1.561 0.0168 13 Psi(~species + poach)Epsilon(~Access)Gamma(~Trail)p(~obsM) 15 2026.742 1.622 0.0163 14 Psi(~species + Trail)Epsilon(~maxElev)Gamma(~Trail)p(~obsM) 15 2026.800 1.679 0.0159 15 Psi(~species + rp)Epsilon(~minElev)Gamma(~Trail)p(~obsM) 15 2026.803 1.683 0.0158 16 Psi(~species)Epsilon(~rp)Gamma(~Trail)p(~obsM) 14 2027.011 1.891 0.0143 17 Psi(~species + minElev)Epsilon(.)Gamma(~Trail)p(~obsM) 14 2027.115 1.994 0.0136 18 Psi(~species + minElev)Epsilon(~Access)Gamma(~Trail)p(~obsM) 15 2027.119 1.998 0.0135 19 Psi(~species + maxElev)Epsilon(~minElev)Gamma(~Trail)p(~obsM) 15 2027.170 2.050 0.0132 20 Psi(~species + Access)Epsilon(~minElev)Gamma(~Trail)p(~obsM) 15 2027.221 2.101 0.0129 21 Psi(~species + Trail)Epsilon(~rp)Gamma(~Trail)p(~obsM) 15 2027.278 2.158 0.0125 22 Psi(~species + rp)Epsilon(~Access)Gamma(~Trail)p(~obsM) 15 2027.293 2.172 0.0124 23 Psi(~species + rp)Epsilon(.)Gamma(~Trail)p(~obsM) 14 2027.336 2.216 0.0121 24 Psi(~species + poach)Epsilon(~Trail)Gamma(~Trail)p(~obsM) 15 2027.435 2.315 0.0116 25 Psi(~species + Trail)Epsilon(~minElev)Gamma(~minElev)p(~obsM) 15 2027.473 2.353 0.0113

105

Table C-2. Continued Model npar AICc ΔAICc Weight 26 Psi(~species + maxElev)Epsilon(~Access)Gamma(~Trail)p(~obsM) 15 2027.640 2.520 0.0104 27 Psi(~species + maxElev)Epsilon(.)Gamma(~Trail)p(~obsM) 14 2027.654 2.534 0.0104 28 Psi(~species + Access)Epsilon(~Access)Gamma(~Trail)p(~obsM) 15 2027.690 2.570 0.0102 29 Psi(~species + poach)Epsilon(~maxElev)Gamma(~Trail)p(~obsM) 15 2027.708 2.587 0.0101 30 Psi(~species + Access)Epsilon(.)Gamma(~Trail)p(~obsM) 14 2027.710 2.590 0.0101 31 Psi(~species)Epsilon(~poach)Gamma(~Trail)p(~obsM) 14 2027.716 2.596 0.0100 32 Psi(~species + Trail)Epsilon(~poach)Gamma(~Trail)p(~obsM) 15 2027.870 2.750 0.0093 33 Psi(~species + minElev)Epsilon(~Trail)Gamma(~Trail)p(~obsM) 15 2028.038 2.918 0.0085 34 Psi(~species + minElev)Epsilon(~maxElev)Gamma(~Trail)p(~obsM) 15 2028.061 2.940 0.0084 35 Psi(~species)Epsilon(~minElev)Gamma(~minElev)p(~obsM) 14 2028.065 2.945 0.0084 36 Psi(~species + poach)Epsilon(~rp)Gamma(~Trail)p(~obsM) 15 2028.201 3.081 0.0079 37 Psi(~species + rp)Epsilon(~maxElev)Gamma(~Trail)p(~obsM) 15 2028.237 3.116 0.0077 38 Psi(~species + rp)Epsilon(~Trail)Gamma(~Trail)p(~obsM) 15 2028.295 3.174 0.0075 39 Psi(~species + minElev)Epsilon(~rp)Gamma(~Trail)p(~obsM) 15 2028.502 3.382 0.0068 40 Psi(~species + Trail)Epsilon(~Access)Gamma(~minElev)p(~obsM) 15 2028.555 3.435 0.0066 41 Psi(~species + maxElev)Epsilon(~Trail)Gamma(~Trail)p(~obsM) 15 2028.557 3.437 0.0066 42 Psi(~species + Access)Epsilon(~Trail)Gamma(~Trail)p(~obsM) 15 2028.580 3.460 0.0065 43 Psi(~species + maxElev)Epsilon(~maxElev)Gamma(~Trail)p(~obsM) 15 2028.585 3.465 0.0065 44 Psi(~species + Access)Epsilon(~maxElev)Gamma(~Trail)p(~obsM) 15 2028.643 3.523 0.0063 45 Psi(~species + rp)Epsilon(~rp)Gamma(~Trail)p(~obsM) 15 2028.690 3.570 0.0062 46 Psi(~species + Trail)Epsilon(~minElev)Gamma(~poach)p(~obsM) 15 2028.704 3.583 0.0061 47 Psi(~species + poach)Epsilon(~poach)Gamma(~Trail)p(~obsM) 15 2028.788 3.667 0.0059 48 Psi(~species + Trail)Epsilon(.)Gamma(~poach)p(~obsM) 14 2029.003 3.883 0.0053 49 Psi(~species + maxElev)Epsilon(~rp)Gamma(~Trail)p(~obsM) 15 2029.049 3.929 0.0052 50 Psi(~species + Trail)Epsilon(~minElev)Gamma(.)p(~obsM) 14 2029.075 3.954 0.0051 51 Psi(~species + Access)Epsilon(~rp)Gamma(~Trail)p(~obsM) 15 2029.115 3.995 0.0050 52 Psi(~species)Epsilon(~minElev)Gamma(~poach)p(~obsM) 14 2029.214 4.094 0.0047 53 Psi(~species + minElev)Epsilon(~poach)Gamma(~Trail)p(~obsM) 15 2029.222 4.102 0.0047 54 Psi(~species + Trail)Epsilon(.)Gamma(~minElev)p(~obsM) 14 2029.235 4.115 0.0047 55 Psi(~species + Trail)Epsilon(~Access)Gamma(~poach)p(~obsM) 15 2029.253 4.132 0.0047

106

Table C-2. Continued Model npar AICc ΔAICc Weight 56 Psi(~species + Trail)Epsilon(.)Gamma(.)p(~obsM) 13 2029.329 4.209 0.0045 57 Psi(~species)Epsilon(~Access)Gamma(~minElev)p(~obsM) 14 2029.336 4.215 0.0045 58 Psi(~species + Trail)Epsilon(~Access)Gamma(.)p(~obsM) 14 2029.357 4.237 0.0044 59 Psi(~species + poach)Epsilon(~minElev)Gamma(~minElev)p(~obsM) 15 2029.392 4.272 0.0043 60 Psi(~species + rp)Epsilon(~poach)Gamma(~Trail)p(~obsM) 15 2029.444 4.324 0.0042 61 Psi(~species + Trail)Epsilon(~maxElev)Gamma(~minElev)p(~obsM) 15 2029.567 4.447 0.0040 62 Psi(~species)Epsilon(.)Gamma(~poach)p(~obsM) 13 2029.661 4.540 0.0038 63 Psi(~species + maxElev)Epsilon(~poach)Gamma(~Trail)p(~obsM) 15 2029.763 4.642 0.0036 64 Psi(~species + Access)Epsilon(~poach)Gamma(~Trail)p(~obsM) 15 2029.819 4.699 0.0035 65 Psi(~species + Access)Epsilon(~minElev)Gamma(~minElev)p(~obsM) 15 2029.842 4.721 0.0035 66 Psi(~species)Epsilon(~minElev)Gamma(.)p(~obsM) 13 2029.858 4.737 0.0034 67 Psi(~species)Epsilon(~Access)Gamma(~poach)p(~obsM) 14 2029.884 4.764 0.0034 68 Psi(~species + rp)Epsilon(~minElev)Gamma(~minElev)p(~obsM) 15 2029.891 4.770 0.0034 69 Psi(~species + maxElev)Epsilon(~minElev)Gamma(~minElev)p(~obsM) 15 2029.950 4.830 0.0033 70 Psi(~species)Epsilon(.)Gamma(~minElev)p(~obsM) 13 2030.047 4.927 0.0031 71 Psi(~species + Trail)Epsilon(~minElev)Gamma(~maxElev)p(~obsM) 15 2030.052 4.931 0.0031 72 Psi(~species + Trail)Epsilon(~maxElev)Gamma(~poach)p(~obsM) 15 2030.134 5.013 0.0030 73 Psi(~species + Trail)Epsilon(~Trail)Gamma(~poach)p(~obsM) 15 2030.154 5.034 0.0030 74 Psi(~species + minElev)Epsilon(~minElev)Gamma(~minElev)p(~obsM) 15 2030.156 5.036 0.0030 75 Psi(~species)Epsilon(~maxElev)Gamma(~minElev)p(~obsM) 14 2030.195 5.075 0.0029 76 Psi(~species)Epsilon(~Access)Gamma(.)p(~obsM) 13 2030.254 5.133 0.0028 77 Psi(~species)Epsilon(.)Gamma(.)p(~obsM) 12 2030.267 5.147 0.0028 78 Psi(~species + Trail)Epsilon(~rp)Gamma(~minElev)p(~obsM) 15 2030.388 5.267 0.0026 79 Psi(~species + Trail)Epsilon(~minElev)Gamma(~rp)p(~obsM) 15 2030.446 5.326 0.0026 80 Psi(~species + Trail)Epsilon(~minElev)Gamma(~Access)p(~obsM) 15 2030.468 5.348 0.0025 81 Psi(~species + Trail)Epsilon(~maxElev)Gamma(.)p(~obsM) 14 2030.480 5.360 0.0025 82 Psi(~species + Trail)Epsilon(~Access)Gamma(~Access)p(~obsM) 15 2030.530 5.409 0.0025 83 Psi(~species + Trail)Epsilon(~rp)Gamma(~poach)p(~obsM) 15 2030.565 5.445 0.0024 84 Psi(~species + poach)Epsilon(~Access)Gamma(~minElev)p(~obsM) 15 2030.607 5.487 0.0024 85 Psi(~species + Trail)Epsilon(~Trail)Gamma(.)p(~obsM) 14 2030.628 5.508 0.0023

107

Table C-2. Continued Model npar AICc ΔAICc Weight 86 Psi(~species + Trail)Epsilon(~Trail)Gamma(~minElev)p(~obsM) 15 2030.637 5.517 0.0023 87 Psi(~species)Epsilon(~minElev)Gamma(~maxElev)p(~obsM) 14 2030.641 5.520 0.0023 88 Psi(~species + Trail)Epsilon(~Access)Gamma(~maxElev)p(~obsM) 15 2030.653 5.533 0.0023 89 Psi(~species)Epsilon(~maxElev)Gamma(~poach)p(~obsM) 14 2030.664 5.544 0.0023 90 Psi(~species + rp)Epsilon(~minElev)Gamma(~poach)p(~obsM) 15 2030.710 5.590 0.0022 91 Psi(~species + Trail)Epsilon(~Access)Gamma(~rp)p(~obsM) 15 2030.750 5.630 0.0022 92 Psi(~species + minElev)Epsilon(~minElev)Gamma(~poach)p(~obsM) 15 2030.772 5.652 0.0022 93 Psi(~species + Trail)Epsilon(~rp)Gamma(.)p(~obsM) 14 2030.778 5.658 0.0022 94 Psi(~species + poach)Epsilon(~minElev)Gamma(~poach)p(~obsM) 15 2030.867 5.747 0.0021 95 Psi(~species + Trail)Epsilon(.)Gamma(~rp)p(~obsM) 14 2030.867 5.747 0.0021 96 Psi(~species + Trail)Epsilon(.)Gamma(~maxElev)p(~obsM) 14 2030.870 5.750 0.0021 97 Psi(~species + poach)Epsilon(~minElev)Gamma(.)p(~obsM) 14 2030.909 5.789 0.0020 98 Psi(~species + Trail)Epsilon(.)Gamma(~Access)p(~obsM) 14 2030.972 5.851 0.0020 99 Psi(~species)Epsilon(~rp)Gamma(~minElev)p(~obsM) 14 2030.996 5.876 0.0019 100 Psi(~species)Epsilon(~Trail)Gamma(~poach)p(~obsM) 14 2031.018 5.898 0.0019

108

APPENDIX D FULL EASTERN CHIMPANZEE DIET LIST FOR NYUNGWE NATIONAL PARK, RWANDA

Table D-1. Full diet list for the Mayebe community of chimpanzees in Nyungwe National Park, Rwanda. Values represent the proportion of total days for each month when any individual chimpanzee was observed consuming the specific food item.

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec TOTAL Number of Observation Days 160 161 183 156 158 169 164 131 153 166 162 132 1895 Species Part of Tree Species Tree Aframomum mala Fruit 0 0 0 0 0.006 0 0 0 0 0 0 0 0.001 Agauria salicifolia Fruit 0 0 0.011 0 0 0 0 0 0 0 0 0 0.001 Allophylus kiwuensis Fruit 0 0 0 0 0.006 0.053 0 0 0 0 0 0 0.005 Begonia meyeri-johannis Leaves 0 0.006 0 0 0 0 0 0 0 0 0 0 0.001 Beilschmiedia michelsonii Fruit 0 0 0 0 0.006 0.006 0 0.008 0 0 0 0 0.002 Bridelia bridelifolia Fruit 0 0 0 0 0 0 0 0.008 0.013 0 0 0 0.002 Cammelia sp Leaves 0 0 0.005 0 0 0 0 0 0 0 0 0 0.001 Cappa Sascic Leaves 0 0 0 0 0 0 0.006 0 0 0 0 0 0.001 Carapa grandiflora Fruit 0 0 0 0 0 0 0 0 0.007 0 0 0 0.001 Casearia runssorica Fruit 0.031 0.006 0 0 0 0 0 0 0 0 0 0 0.003 Cassipourea gummiflua Fruit 0.006 0 0 0 0 0 0 0.031 0.013 0.012 0.006 0.008 0.006 Cassipourea ruwensorensis Fruit 0.013 0.006 0 0 0 0 0 0 0 0 0.006 0 0.002 Chrysophyllum Fruit 0.119 0.081 0.038 0.147 0.184 0.154 0.049 0.023 0 0 0 0 0.068 gorungosanum Chrysophyllum Seeds 0 0 0 0 0 0.095 0.134 0.115 0 0 0 0 0.028 gorungosanum Cleistanthus polystachyus Fruit 0 0 0 0 0 0 0 0.008 0 0 0 0 0.001

109

Table D-1. Continued Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec TOTAL Part of Tree Species Tree Cyathea manniana Marrow 0 0.006 0.005 0 0 0 0.012 0 0 0 0 0 0.002 Dombeya goetzenii Bark 0 0.012 0.038 0.006 0 0 0 0 0 0 0 0 0.005 Dombeya goetzenii Leaves 0.025 0.019 0.104 0.026 0.013 0.012 0 0.008 0.013 0.012 0.006 0.008 0.022 Dombeya goetzenii Flowers 0 0 0.005 0.071 0.108 0 0 0 0 0 0 0 0.015 Dombeya goetzenii Fruit 0 0 0 0 0 0 0 0 0 0 0.006 0.008 0.001 Dombeya goetzenii Marrow 0 0 0 0 0 0 0 0.015 0 0 0.006 0 0.002 Drypetes occidentalis Fruit 0 0 0 0 0.025 0.006 0 0 0 0 0 0 0.003 Ekebergia capensis Fruit 0.006 0.006 0 0 0 0.083 0.116 0.107 0 0 0 0 0.026 Ensete ventricosum Marrow 0 0 0 0.006 0 0 0 0 0 0 0 0 0.001 Ficus spp Fruit 0.556 0.609 0.552 0.442 0.209 0.160 0.244 0.229 0.327 0.060 0.037 0.038 0.294 Girardinia bullosa Leaves 0 0.006 0 0 0 0 0 0 0 0 0 0 0.001 Gouania longispicata Leaves 0 0 0 0 0 0 0 0 0 0 0 0.008 0.001 Grewia mildbraedii Fruit 0 0 0 0 0 0.012 0 0 0.007 0 0 0 0.002 Gynura scandens Leaves 0 0 0.005 0 0 0 0 0 0 0 0 0 0.001 Impatiens niamniamensis Flowers 0 0 0 0 0.006 0 0 0 0 0 0 0 0.001 Laportea alatipes Leaves 0 0 0.005 0 0 0 0 0 0 0 0 0 0.001 Macaranga Flowers 0 0.006 0 0 0 0 0 0 0 0 0 0 0.001 kilimandscharica Macaranga Fruit 0 0 0 0 0 0 0.006 0.076 0.007 0.012 0 0.008 0.008 kilimandscharica Maesa lanceolata Fruit 0 0 0 0 0 0.012 0.043 0.046 0 0 0 0 0.008 Memecylon walikalense Fruit 0 0 0 0 0 0 0.006 0.008 0 0 0 0 0.001 Mimulopsis arborescens Marrow 0 0 0.005 0 0 0 0 0 0 0 0 0 0.001 Mimulopsis solmsii Leaves 0 0 0.005 0 0 0 0 0 0 0 0 0 0.001

110

Table D-1. Continued Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec TOTAL Part of Tree Species Tree Mimulopsis solmsii Marrow 0 0 0 0 0 0 0.006 0.015 0 0 0 0 0.002 Mimulopsis solmsii Trunk 0 0 0 0 0 0 0 0 0 0 0.006 0 0.001 Musa acuminata Marrow 0 0.006 0 0 0 0.024 0.012 0.008 0 0 0 0 0.004 Musanga leo-errerae Marrow 0 0 0 0 0.006 0 0 0 0 0 0 0 0.001 Myrianthus holstii Fruit 0.056 0 0 0.013 0.051 0.095 0.116 0.008 0 0 0 0 0.029 Myrianthus holstii Seeds 0 0 0 0 0.006 0 0.018 0 0 0 0 0 0.002 Myrica sp Leaves 0 0 0.005 0 0 0 0 0 0 0 0 0 0.001 Myrica sp Fruit 0 0 0.016 0.019 0 0 0 0 0 0 0 0 0.003 Olea capensis Fruit 0 0 0 0 0.044 0.071 0.085 0.107 0.190 0.157 0 0 0.054 Olinia rochetiana Fruit 0.006 0.037 0.027 0.006 0 0 0 0.015 0 0 0 0 0.008 Pancovia golungensis Fruit 0 0 0 0 0 0 0 0.008 0 0.054 0.105 0.008 0.015 Panicum pusillum Leaves 0 0 0 0 0.006 0 0 0 0 0 0 0 0.001 Parinari excelsa Bark 0 0 0 0 0 0 0 0 0 0.006 0 0 0.001 Parinari excelsa Fruit 0 0 0.005 0.013 0.006 0.012 0 0 0.007 0 0 0 0.004 Pennisetum purpureum Trunk 0 0 0 0 0 0 0 0 0 0 0 0.008 0.001 Piper capense Trunk 0 0 0.005 0 0 0 0 0 0 0 0 0 0.001 Podocarpus falcatus Fruit 0 0 0 0 0 0 0 0.008 0.007 0 0 0 0.001 Podocarpus latifolius Fruit 0 0 0 0 0 0 0 0.008 0.059 0 0 0 0.005 Prunus africana Fruit 0 0.012 0.044 0.071 0.013 0.012 0.030 0.023 0.033 0.006 0 0 0.021 Pterdium aquilinum Leaves 0 0.006 0 0.013 0.013 0 0 0 0 0 0 0.008 0.003 Pterdium aquilinum Marrow 0 0.006 0 0 0.006 0 0.018 0.031 0.020 0 0 0 0.006 Pterdium aquilinum Trunk 0.013 0.031 0.044 0 0.006 0.012 0 0 0.013 0 0 0 0.011

111

Table D-1. Continued Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec TOTAL Part of Tree Species Tree Senecio mannii Marrow 0 0 0 0 0.006 0 0 0 0 0 0 0 0.001 Senecio sp Trunk 0.006 0 0 0 0 0 0 0 0 0 0 0 0.001 Solanum sp Fruit 0 0 0.005 0 0 0 0 0 0 0 0 0 0.001 Strombosia scheffleri Fruit 0 0 0 0.006 0 0 0 0 0 0 0 0 0.001 Symphonia globulifera Buds 0 0 0.027 0.115 0.203 0.154 0.055 0 0 0 0 0 0.047 Symphonia globulifera Flowers 0 0 0.005 0.006 0.076 0.006 0 0 0 0 0 0 0.008 Symphonia globulifera Fruit 0 0 0 0.006 0 0.012 0 0 0 0 0 0 0.002 Syzygium guineense Leaves 0 0 0 0 0 0 0 0 0 0.006 0 0 0.001 Syzygium guineense Fruit 0.138 0.012 0.044 0.032 0 0.006 0 0.038 0.255 0.614 0.759 0.902 0.225 Trema orientalis Fruit 0.019 0.075 0 0 0 0 0 0 0.007 0.006 0 0 0.009 Triumfetta cordifolia Leaves 0.019 0.025 0.016 0.013 0.025 0.024 0.006 0.008 0.013 0.018 0.012 0.008 0.016 Triumfetta cordifolia Fruit 0 0.006 0 0 0 0 0 0 0 0 0 0 0.001 Urera cameroonensis Leaves 0 0 0.005 0 0 0 0 0 0 0 0 0 0.001 Urera cameroonensis Flowers 0 0 0 0 0 0 0 0 0.007 0 0 0 0.001 Urera cameroonensis Marrow 0 0 0 0 0 0 0 0.008 0 0 0 0 0.001 Urera hypelodendron Leaves 0 0 0.005 0 0 0 0 0 0 0 0.006 0 0.001 Urera hypelodendron Flowers 0 0 0 0 0 0 0 0 0.007 0.012 0.006 0 0.002 Urera hypelodendron Fruit 0 0.006 0.005 0.006 0 0 0.006 0.008 0.007 0.024 0.031 0.008 0.008 Vepris sp Fruit 0 0.006 0 0 0.006 0.030 0.055 0.046 0.007 0 0 0 0.012 Vernonia auriculifera Marrow 0.006 0 0 0 0 0 0 0 0 0 0 0 0.001 Vernonia lasiopus Marrow 0.006 0 0 0 0 0 0 0 0 0 0 0 0.001 Vernonia sp Trunk 0 0 0.005 0 0 0 0 0 0 0 0 0 0.001

112

Table D-1. Continued Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec TOTAL Part of Tree Species Tree Xymalos monospora Leaves 0 0 0 0 0 0 0 0 0 0.006 0 0 0.001 Non-Tree Species Bees Honey 0 0.012 0 0 0.013 0 0 0 0 0 0 0 0.002 Cercpithecus mitis 0.006 0 0 0 0 0 0 0 0 0 0 0 0.001 Insects 0 0 0 0.006 0 0 0 0 0 0 0.006 0 0.001

113

Table D-2. Full diet list for the Cyamudongo community of chimpanzees in Nyungwe National Park, Rwanda. Values represent the proportion of total days for each month when any individual chimpanzee was observed consuming the specific food item.

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec TOTAL Number of Observation Days 155 141 136 149 181 176 182 173 154 151 158 169 1925 Species Part of Tree Species Tree Aframomum angustifolium Trunk 0 0.028 0 0 0 0 0 0 0 0 0 0 0.002 Aframomum mildbraedii Trunk 0 0 0 0 0 0 0 0 0.006 0 0 0 0.001 Aframomum montanum Fruit 0 0 0 0 0 0.006 0 0.006 0 0 0 0 0.001 Aframomum montanum Trunk 0.006 0 0 0 0 0 0 0 0 0 0 0 0.001 Aframomum zambeziacum Fruit 0 0 0 0 0 0.006 0 0 0 0 0 0 0.001 Alangium chinense Fruit 0 0 0.015 0.020 0 0 0 0 0 0 0.006 0 0.003 Albizia gummifera Flowers 0 0 0.007 0 0 0 0 0 0 0 0 0 0.001 Allophylus sp Leaves 0 0 0 0 0 0 0.005 0 0 0 0 0 0.001 Brillantaisia cicatricosa Marrow 0 0 0.007 0 0 0 0 0 0 0 0 0 0.001 Casearia runssorica Fruit 0 0 0.007 0.007 0 0 0 0 0 0 0 0 0.001 Cassipourea gummiflua Fruit 0.006 0 0 0 0 0 0 0 0 0 0 0.006 0.001 Cassipourea Fruit 0.006 0 0 0 0 0 0 0 0 0 0 0 0.001 ruwensorensis Celtis africana Leaves 0 0 0.007 0.020 0 0 0 0 0 0 0.006 0 0.003 Celtis africana Flowers 0 0 0 0 0 0 0 0 0.019 0 0 0 0.002 Celtis africana Fruit 0 0 0.015 0.040 0.011 0 0 0 0.019 0 0 0 0.007 Celtis gomphophylla Leaves 0 0 0 0 0 0 0 0 0.006 0 0 0 0.001 Celtis gomphophylla Fruit 0 0.007 0 0.013 0.017 0 0 0 0 0 0 0 0.003

114

Table D-2. Continued Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec TOTAL Species Part Chrysophyllum Fruit 0 0 0 0.007 0 0.045 0.022 0.029 0.006 0 0 0 0.010 gorungosanum Chrysophyllum rwandense Fruit 0 0 0 0 0 0 0.011 0 0 0 0 0 0.001 Commelina sp Leaves 0 0 0.007 0 0 0 0 0 0 0 0 0 0.001 Dombeya goetzenii Bark 0 0 0 0.007 0 0 0 0 0 0 0 0 0.001 Dombeya goetzenii Leaves 0 0.014 0 0 0.006 0.011 0 0.006 0.006 0.020 0 0 0.005 Dombeya goetzenii Flowers 0 0 0 0 0.006 0 0 0 0 0 0 0 0.001 Dombeya goetzenii Fruit 0 0 0 0 0 0 0 0 0.006 0 0 0 0.001 Dombeya goetzenii Trunk 0 0 0 0.007 0 0 0 0 0 0.007 0 0 0.001 Drypetes occidentalis Fruit 0 0 0 0 0 0.006 0 0 0 0 0 0 0.001 Ekebergia capensis Fruit 0 0.043 0.059 0 0 0 0.011 0.023 0.039 0.013 0 0 0.015 Ensete ventricosum Leaves 0.006 0 0.022 0 0 0 0 0 0 0.007 0 0 0.003 Eucalyptus saligna Bark 0.006 0.007 0 0 0 0 0 0 0 0 0 0 0.001 Eucalyptus sp Bark 0 0 0 0 0.006 0.006 0 0 0 0 0 0 0.001 Ficus spp Fruit 0.497 0.610 0.757 0.503 0.530 0.369 0.374 0.306 0.279 0.113 0.076 0.136 0.373 Gouania longispicata Leaves 0 0 0 0 0 0 0 0 0 0.007 0 0 0.001 Gouania longispicata Fruit 0 0 0 0.007 0.006 0.006 0 0 0 0 0 0 0.002 Grevillea robusta Fruit 0 0 0 0 0 0.011 0 0 0 0 0 0 0.001 Grewia mildbraedii Fruit 0 0 0 0 0 0 0.011 0 0 0 0 0 0.001 Harungana montana Fruit 0 0 0 0 0 0 0 0 0.032 0.020 0 0 0.004 Macaranga Fruit 0 0 0 0.007 0 0 0 0.006 0 0 0 0 0.001 kilimandscharica Musanga leo-errerae Flowers 0.006 0 0 0 0 0 0 0 0 0 0 0 0.001 Musanga leo-errerae Fruit 0.084 0.057 0.044 0.141 0.055 0.057 0.027 0.023 0.039 0.046 0.013 0 0.048

115

Table D-2. Continued Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec TOTAL Species Part Musanga leo-errerae Trunk 0 0 0.007 0 0 0 0 0 0 0 0 0 0.001 Myrianthus holstii Fruit 0 0 0 0 0 0.006 0.016 0.012 0.006 0 0 0 0.004 Pancovia golungensis Fruit 0 0 0 0 0 0 0 0 0 0 0 0.012 0.001 Parinari excelsa Fruit 0 0 0 0 0 0 0 0.006 0 0 0 0 0.001 Piper capense Marrow 0 0 0.007 0 0 0 0 0 0 0 0 0 0.001 Polyscias fulva Fruit 0 0 0 0 0.006 0 0 0.006 0 0 0 0 0.001 Salacia erecta Fruit 0.006 0.007 0 0.007 0 0 0 0 0 0 0 0.006 0.002 Symphonia globulifera Flowers 0 0 0.007 0 0 0 0 0 0 0 0 0 0.001 Syzygium guineense Fruit 0 0.007 0 0 0 0 0 0 0 0.033 0.006 0 0.004 Tabernaemontana Fruit 0 0 0 0.007 0 0 0 0 0 0.007 0 0 0.001 stapfiana Teclea nobilis Fruit 0 0 0 0 0 0 0.011 0 0 0 0 0 0.001 Trema orientalis Fruit 0.013 0.028 0.007 0 0.011 0.011 0.011 0.081 0.032 0.007 0 0 0.017 Trilepisium Fruit 0.361 0.149 0.007 0.195 0.348 0.449 0.500 0.480 0.468 0.702 0.892 0.840 0.459 madagascariense Triumfetta cordifolia Leaves 0 0.014 0.022 0.007 0 0.011 0.005 0.006 0 0.020 0 0 0.007 Triumfetta cordifolia Fruit 0 0.007 0 0 0 0 0 0.006 0 0 0 0 0.001 Triumfetta cordifolia Trunk 0 0 0.015 0 0 0 0 0 0 0 0 0 0.001 Urera cameroonensis Trunk 0.006 0 0 0 0 0 0 0 0 0 0 0 0.001 Urera hypselodendron Flowers 0 0 0 0 0 0 0 0.006 0 0 0 0 0.001 Urera hypselodendron Fruit 0 0 0 0 0 0 0 0 0 0.007 0 0 0.001 Vaccinium stanleyi Fruit 0 0 0 0 0 0 0 0 0 0.007 0 0 0.001 Vepris stolzii Fruit 0 0 0 0 0.011 0.006 0 0 0 0 0 0 0.002

116

Table D-2. Continued Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec TOTAL Species Part Vernonia sp Trunk 0 0.007 0 0 0 0 0 0 0 0 0 0 0.001 Vine Fruit 0 0 0 0.020 0 0 0 0 0 0 0 0 0.002 Vine Marrow 0 0.028 0 0 0 0 0 0 0 0 0 0 0.002 Zanha golugensis Fruit 0 0 0 0 0 0 0 0 0.032 0.007 0 0 0.003 Non-Tree Species Monkey species 0.006 0 0 0 0 0 0 0 0 0 0 0 0.001 (unknown)

117

LIST OF REFERENCES

Basabose, A. K. (2002). Diet composition of chimpanzees inhabiting the montane forest of Kahuzi, Democratic Republic of Congo. American Journal of Primatology, 58(1), 1-21.

Basabose, A. K. (2005). Ranging patterns of chimpanzees in a montane forest of Kahuzi, Democratic Republic of Congo. International Journal of Primatology, 26(1), 33-54.

Bates, L. A. & Byrne, R. W. (2009). Sex differences in the movement patterns of free- ranging chimpanzees (Pan troglodytes schweinfurthii): foraging and border checking. Behavioral Ecology and Sociobiology, 64(2), 247-255.

Behangana, M., Kasoma, P.M.B., and Luiselli, L. (2009). Ecological correlates of species richness and population abundance patterns in the amphibian communities from the Albertine Rift, East Africa. Biodiversity and Conservation, 18(11), 2855-2873.

Berger, J. (1997). Population constraints associated with the use of black rhinos as an umbrella species for desert herbivores. Conservation Biology, 11(1), 69-78.

Bergman, T.P., Sekercioglu, C.H., and Tobias, J.A. (2014). Global patterns and predictors of bird species responses to habitat fragmentation: implications for ecosystem function and conservation. Biological Conservation, 169, 372-383.

Boesch, C., & Boesch, H. (1989). Hunting behavior of wild chimpanzees in the Taï National Park. American Journal of Physical Anthropology, 78(4), 547-573.

Börger L., Danziel, B.D., & Fryxell, J. M. (2008). Are there general mechanisms of animal home range behaviour? A review and prospects for future research. Ecology Letters, 11(6), 637-650.

Boulinier, T., Nichols, J.D., Sauer, J.R., Hines, J.E. & Pollock, K.H. (1998). Estimating species richness: the importance of heterogeneity in species detectability. Ecology, 79(3), 1018-1028.

Brandon, K., Redford, K.H. & Sanderson, S. editors. (1998). Parks in peril: people, politics, and protected areas. Washington, DC, USA: Island Press.

Brodie, J.F., Helmy, O.E., Brockelan, W.Y., & Maron, J.L. (2009). Bushmeat poaching reduces the seed dispersal and population growth rate of a mammal-dispersed tree. Ecological Applications, 19(4), 854-863.

Brown, J. (1995). Macroecology. Chicago, Illinois, USA: University of Chicago Press.

118

Buckland, S. T., Anderson, D. R., Burnham, K. P., & Laake, J. L. (1993). Distance sampling: estimating abundance of biological populations. London, UK: Chapman & Hall.

Buij, R., McShea, W.J., Campbell, P., Lee, M.E., Dallmeier, F., Guimondou, S., … Alonso, A. (2007). Patch-occupancy models indicate human activity as major determinant of forest elephant Loxodonta cyclotis seasonal distribution in an industrial corridor in Gabon. Biological Conservation, 135(2), 189-201.

Burnham, K.P., & Anderson, D.R. (2002). Model selection and multimodel inference: a practical information-theoretic approach. New York, New York, USA: Springer- Verlag.

Burnham, K.P., & Overton, W.S. (1979). Robust estimation of population size when capture probabilities vary among animals. Ecology, 60(5), 927-936.

Butchart, S.H.M., Walpole, M., Collen, B., van Strien, A., Scharlemann, J.P.W., Almond, R.E.A., … Watson, R. (2010). Global biodiversity: Indicators of recent declines. Science, 328(5982), 1164-1168.

Butynski, T.M. (1997). African primate conservation – The species and the IUCN/SSC Primate Specialist Group network. Primate Conservation, 17, 87-100.

Calenge, C. (2006). The package "adehabitat" for the R software: A tool for the analysis of space and habitat use by animals. Ecological Modelling, 197(3), 516-519.

Central Intelligence Agency (CIA). (2018). The World Factbook: Rwanda. Washington, D.C., USA. Retrieved 30 April 2018.

Challender, D.W.S. & MacMillan, D.C. (2014). Poaching is more than an enforcement problem. Conservation Letters, 7(5), 484-494.

Chapman, C. A., & Wrangham, R. W. (1993). Range use of the forest chimpanzees of Kibale: Implications for the understanding of chimpanzee social organization. American Journal of Primatology, 31(4), 263-273.

Child, B. (1996). The practice and principles of community-based wildlife management in Zimbabwe: the CAMPFIRE programme. Biodiversity & Conservation, 5(3), 369.

Clutton-Brock, T. H., & Harvey, P. H. (1977). Primate ecology and social organization. Journal of Zoology, 183(1), 1-30.

Clutton-Brock, T. H., & Harvey, P. H. (1979). Home range size, population density and phytogeny in primates. In I.S. Bernstein & E.O. Smith (Eds.), Primate Ecology and Human Origins (pp. 201-214). New York, New York, USA: Garland Press.

119

Cooney, R., Roe, D., Dublin, H., Phelps, J., Wilkie, D., Keane, A., … Biggs, D. (2017). From poachers to protectors: engaging local communities in solutions to illegal wildlife trade. Conservation Letters, 10(3), 367-374.

Craigie, I.D., Baillie, J.E.M., Balmford, A., Carbone, C., Collen, B., Green, R.E., Huton, J.M. (2010). Large mammal population declines in Africa's protected areas. Biological Conservation, 143(9), 2221-2228.

Critchlow, R., Plumptre, A.J., Driciru, M., Rwetsiba, A., Stokes, E.J., Tumwesigye, C., … Beale, C.M. (2015). Spatiotemporal trends of illegal activities from ranger‐ collected data in a Ugandan national park. Conservation Biology, 29(5), 1458- 1470.

Cunha, A.A. (2010). Negative effects of tourism in a Brazilian Atlantic forest National Park. Journal for Nature Conservation, 18(4), 291-295.

Desai, A.A. & Riddle, H.S. (2015) Human-elephant conflict in Asia. United States Fish and Wildlife Service. Available from https://www.fws.gov/international/pdf/Human-Elephant-Conflict-in-Asia- June2015.pdf (accessed April 2017).

Dorazio, R.M., Royle, J.A., Soderstrom, B., and Glimskar, A. (2006). Estimating species richness and accumulation by modeling species occurrence and detectability. Ecology, 87(4), 842-854.

Dudley, N. & Stolton, S. editors. (2008) Defining protected areas: an international conference in Almeria, Spain. Gland, Switzerland: IUCN. 220 pp.

Dutton, P., & Chapman, H. (2014). Dietary preferences of a submontane population of the rare Nigerian-Cameroon chimpanzee (Pan troglodytes ellioti) in Ngel Nyaki Forest Reserve, Nigeria. American Journal of Primatology, 77(1), 86-97.

Dyer, E.E., Cassey, P., Redding, D.W., Collen, B., Franks, V., Gaston, K.J., … Blackburn, T.M. (2017). The global distribution and drivers of alien bird species richness. PLoS Biol, 15(1), e2000942.

Edderai, D. & Dame, M. (2006). A census of the commercial bushmeat market in Yaoundé, Cameroon. Oryx, 40(4), 472-475.

Effiom, E.O., Nuñez-Iturri, G., Smith, H.G., Ottosson, U., & Olsson, O. (2013). Bushmeat hunting changes regeneration of African rainforests. Proceedings of the Royal Society B, 280(1759), 20130246.

120

Engel, S., Pagiola, S. & Wunder, S. (2008). Designing payments for environmental services in theory and practice: an overview of the issues. Ecological Economics, 65(4), 663-674.

Ernest, S.K.M., and Brown, J.H. (2001). Homeostasis and compensation: the role of species and resources in ecosystem stability. Ecology, 82(8), 2118-2132.

Eskildsen, A., Carvalheiro, L.G., Kissling, W.D., Biesmeijer, J.C., Schweiger, O., and Hoye, T.T. (2015). Ecological specialization matters: long-term trends in butterfly species richness and assemblage composition depend on multiple functional traits. Diversity and Distributions, 21(7), 792-802.

ESRI. (2011). ArcGIS Desktop: Release 10. Redlands, CA: Environmental Systems Research Institute.

Estes, J.A., Terborgh, J., Brashares, J.S., Power, M.E., Berger, J., Bond, W.J., … Wardle, D.A. (2011). Trophic downgrading of planet earth. Science, 333(6040), 301-306.

Fa, J.E., Seymour, S., Dupain, J., Amin, R., Albrechtsen, L., & Macdonald, D. (2006). Getting to grips with the magnitude of exploitation: bushmeat in the Cross- Sanaga rivers region, Nigeria and Cameroon. Biological Conservation, 129(4), 497-510.

Fagan, W.F., Cantrell, R.S., & Cosner, C. (1999). How habitat edges change species interactions. The American Naturalist, 153(2), 165-182.

Fischer, J., Lindenmayer, D.B., & Cowling, A. (2004). The challenge of managing multiple species at multiple scales: reptiles in an Australian grazing landscape. Journal of Applied Ecology, 41(1), 32-44.

Fitzgibbon, C., Mogaka, H., and Fanshawe, J. (1995). Subsistence hunting in Arabuko- Sokoke Forest, Kenya, and its effects on mammal populations. Conservation Biology, 9(5), 1116-1126.

Fossey, D. (1974). Observations on the home range of one group of mountain gorillas (Gorilla gorilla beringei). Animal Behaviour, 22(3), 568-581.

Furuichi, T., Hashimoto, C., & Yahiro, Y. (2001). Fruit availability and habitat use by chimpanzees in the Kalinzu Forest, Uganda: examination of fallback foods. International Journal of Primatology, 22(6), 929-945.

Gallou, A., Baillet, Y., Ficetola, G.F., and Despres, L. (2017). Elevational gradient and human effects on butterfly species richness in the French Alps. Ecology and Evolution, 7(11), 3672-3681.

121

Gibson, C.C. & Marks, S.A. (1995). Transforming rural hunters into conservationists: an assessment of community-based wildlife management programs in Africa. World Development, 23(6), 941-957. Goodall, J. (1986). The chimpanzees of Gombe: Patterns of behavior. Cambridge, Massachusetts, USA: Harvard University Press.

Goswami, V.R., Medhi, K., Nichols, J.D., & Oli, M.K. (2015). Dynamic occupancy models provide a mechanistic understanding of human-wildlife conflict. Conservation Biology, 29(4), 1100-1110.

Gotelli, N.J., & Ellison, A.M. (2004). A primer of ecological statistics. Sunderland, Massachusetts, USA: Sinauer Associates.

Gratwicke, B., and Speight, M.R. (2005). The relationship between fish species richness, abundance, and habitat complexity in a range of shallow tropical marine habitats. Journal of Fish Biology, 66(3), 650-667.

Gray, M. & Kalpers, J. (2005). Ranger based monitoring in the Virunga-Bwindi region of East-Central Africa: a simple data collection tool for park management. Biodiversity & Conservation, 14(11), 2723-2741.

Gross-Camp, N.D., Masozera, M., & Kaplin, B.A. (2009). Chimpanzee seed dispersal quantity in a tropical montane forest of Rwanda. American Journal of Primatology, 71(11), 901-911.

Hall, J.S., Saltonstall, K., Inogwabini, B.I., & Omari, I. (1998). Distribution, abundance, and conservation status of Grauers gorilla. Oryx, 32(2), 122-130.

Harris, S., Cresswell, W.J., Forde, P.G., Trewhella, W.J., Woollard, T., & Wray, S. (1990). Home-range analysis using radio-tracking data – a review of problems and techniques particularly as applied to the study of mammals. Mammal Review, 20(2-3), 97-123.

Hashimoto, C., Tashiro, Y., Kimura, D., Enomoto, T., Ingmanson, E.J., Idani, G., and Furuichi, T. (1998). Habitat use and ranging of wild bonobos (Pan paniscus) at Wamba. International Journal of Primatology, 19(6), 1045-1060.

Hayne, D.W. (1949). Calculation of size of home range. Journal of Mammalogy, 30(1), 1-18.

Herbinger, I., Boesch, C., & Rothe, H. (2001). Territory characteristics among three neighboring chimpanzee communities in the Taï National Park, Côte d’Ivoire. International Journal of Primatology, 22(2), 143-167.

122

Hofer, H., Campbell, K.L.I, East, M.L., & Huish, S.A. (2000). Modeling the spatial distribution of the economic costs and benefits of illegal game meat hunting in the Serengeti. Natural Resource Modeling, 13(1), 151-177.

Houston, A.I., Higginson, A.D., & McNamara, J.M. (2011). Optimal foraging for multiple nutrients in an unpredictable environment. Ecology Letters, 14(11), 1101-1107.

Ijeomah, H.M., Ogogo, A.U. & Ogbara, D. (2013). Analysis of poaching activities in Kainji Lake National Park of Nigeria. Environment and Natural Resources Research, 3(1), 51-61.

Jachmann, H., Blanc, J., Nateg, C., Balangtaa, C., Debrah, E., Damma, F., Atta-Kusi, E., & Kipo, A. (2011). Protected area performance and tourism in Ghana. South African Journal of Wildlife Research, 41(1), 95-109.

Kano, T. (1972). Distribution and adaptation of the chimpanzee on the eastern shore of Lake Tanganyika. Kyoto University African Studies, 7, 37-129.

Keane, A., Jones, J.P.G. & Milner-Gulland, E.J. (2011). Encounter data in resource management and ecology: pitfalls and possibilities. Journal of Applied Ecology, 48(5), 1164-1173.

Kertson, B. N., & Marzluff, J. M. (2010). Improving studies of resource selection by understanding resource use. Environmental Conservation, 38(1), 18-27.

Kéry, M., Guillera-Arroita, G., and Lahoz-Monfort, J.J. (2013). Analysing and mapping species range dynamics using occupancy models. Journal of Biogeography, 40(8), 1463-1474.

Kéry, M., & Royle, J.A. (2008). Hierarchical Bayes estimation of species richness and occupancy in spatially replicated surveys. Journal of Applied Ecology, 45(2), 589- 598.

Kinnaird, M.F., and O’Brien, T.G. (2012). Effects of private-land use, livestock management, and human tolerance on diversity, distribution, and abundance of large African mammals. Conservation Biology, 26(6), 1026-1039.

Kissui, B.M. (2008). Livestock predation by lions, leopards, spotted hyenas, and their vulnerability to retaliatory killing in the Maasai steppe, Tanzania. Animal Conservation, 11(5), 422-432.

Knapp, E.J., Rentsch, D., Schmitt, J., Lewis, C., & Polasky, S. (2010). A tale of three villages: choosing an effective method for assessing poaching levels in western Serengeti, Tanzania. Oryx, 44(2), 178-184.

123

Kӧndgen, S., Kühl, H., N’Goran, P.K., Walsh, P.D., Schenk, S., Ernst, N., … Leendertz, F.H. (2008). Pandemic human viruses cause decline of endangered great apes. Current Biology, 18, 260-264.

Krebs, C.J. (1999). Ecological Methodology. Menlo Park, California, USA: Benjamin Cummings.

Krief, S., Huffman, M. A., Sévenet, T., Hladik, C. M., Grellier, P., Loiseau, P. M., & Wrangham, R. W. (2006). Bioactive properties of plant species ingested by chimpanzees (Pan troglodytes schweinfurthii) in the Kibale National Park, Uganda. American Journal of Primatology, 68(1), 51-71.

Laake, J.L. (2013). RMark: An R interface for analysis of capture-recapture data with MARK. AFSC Processed Rep 2013-01, 25p. Alaska Fisheries Science Center, NOAA, National Marine Fish Service, 7600 Sand Point Way NE, Seattle WA 98115.

Lambert, J. E., & Garber, P. A. (1998). Evolutionary and ecological implications of primate seed dispersal. American Journal of Primatology, 45(1), 9-28.

Laurance, W.F., Useche, D.C., Rendeiro, J., Kalka, M., Bradshaw, C.J.A., Sloan, S.P., … Zamzani, F. (2012). Averting biodiversity collapse in tropical forest protected areas. Nature, 489, 290-294.

Lawlor, L.R. (1979). Direct and indirect effects of n-species competition. Oecologia, 43(3), 355-364.

Lenoir, J., Gégout, J.C., Marquet, P.A., de Ruffray, P., and Brisse, H. (2008). A significant upward shift in plant species optimum elevation during the 20th century. Science, 320(5884), 1768-1771.

Leopold, A. (1933). Game Management. New York City, New York, USA: Charles Scribner’s Sons.

Lepczyk, C.A., Flather, C.H., Radeloff, V.C., Pidgeon, A.M., Hammer, R.B., and Liu, J. (2008). Human impacts on regional avian diversity and abundance. Conservation Biology, 22(2), 405-416.

Lindsey, P.A., Balme, G., Becker, M., Begg, C., Bento, C., Bocchino, C., … Zisadza- Gandiwa, P. (2013). The bushmeat trade in African savannas: impacts, drivers, and possible solutions. Biological Conservation, 160, 80-96.

Lindsey, P.A., Romañach, S.S., Matema, S., Matema, C., Mupamhadzi, I., & Muvengwi, J. (2011). Dynamics and underlying causes of illegal bushmeat trade in Zimbabwe. Oryx, 45(1), 84-95.

124

Linkie, M., Martyr, D.J., Harihar, A., Risdianto, D., Nugraha, R.T., Maryati, … Wong, W. (2015). Safeguarding Sumatran tigers: evaluating effectiveness of law enforcement patrols and local informant networks. Journal of Applied Ecology, 52(4), 851-860.

Lyons, J.E., Runge, M.C., Laskowski, H.P., and Kendall, W.L. (2008). Monitoring in the context of structured decision-making and adaptive management. Journal of Wildlife Management, 72(8), 1683-1692.

MacArthur, R.H. (1972). Geographical ecology: patterns in the distribution of species. Princeton, New Jersey, USA: Princeton University Press.

MacArthur, R.H., and MacArthur, J.W. (1961). On bird species diversity. Ecology, 42(3), 594-598.

Mackenzie, C.A., Chapman, C.A. & Sengupta, R. (2011). Spatial patterns of illegal resource extraction in Kibale National Park, Uganda. Environmental Conservation, 39(3), 38-50.

MacKenzie, D.I. (2006). Modelling the probability of resource use: the effect of, and dealing with, detecting a species imperfectly. Journal of Wildlife Management, 70, 367-374.

MacKenzie, D.I., Nichols, J.D., Hines, J.E., Knutson, M.G., & Franklin, A.B. (2003). Estimating site occupancy, colonization, and local extinction when a species is detected imperfectly. Ecology, 84(8), 2200-2207.

MacKenzie, D.I., Nichols, J.D., Lachman, G.B., Droege, S., Royle, J.A., & Langtimm, C.A. (2002). Estimating site occupancy rates when detection probabilities are less than one. Ecology, 83(8), 2248-2255.

MacKenzie, D.I., Nichols, J.D., Royle, J.A., Pollock, K.H., Bailey, L.L., and Hines, J.E. (2018). Occupancy estimation and modeling: inferring patterns and dynamics of species occurrence. 2nd edition. London, UK: Academic Press.

MacKenzie, D.I. & Royle, J.A. (2005). Designing occupancy studies: general advice and allocating survey effort. Journal of Applied Ecology, 42(6), 1105-1114.

MacKenzie, D.I., Royle, J.A., Nichols, J.D., Hines, J.E., Pollock, K.H., & Bailey, L.L. (2006). Occupancy estimation and modeling: Inferring patterns and dynamics of species occurrence. New York, New York, USA: Academic Press.

Manly, B. F. J., McDonald, L. L., Thomas, D. L., McDonald, T. L., & Erickson, W. P. (2002). Resource Selection by Animals: Statistical Design and Analysis for Field Studies. Boston, Massachusetts, USA: Kluwer Academic Press.

125

Marchesi, P., Marchesi, N., Fruth, B., & Boesch, C. (1995). Census and distribution of chimpanzees in Cote d’Ivoire. Primates, 36(4), 591-607.

Martin, A. & Caro, T. (2013). Illegal hunting in the Katavi-Rukwa ecosystem. African Journal of Ecology, 51(1), 172-175.

Martin, J., McIntyre, C.L, Hines, J.E., Nichols, J.D., Schmutz, J.A., and MacCluskie, M.C. (2009). Dynamic multistate site occupancy models to evaluate hypotheses relevant to conservation of golden eagles in Denali national park, Alaska. Biological Conservation, 142(11), 2726-2731.

McIntyre, N. E., & Wiens, J. A. (1999). Interactions between landscape structure and animal behavior: the roles of heterogeneously distributed resources and food deprivation on movement patterns. Landscape Ecology, 14(5), 437-447.

McNeely, J.A., Harrison, J. & Dingwall, P.R. (1994). Protecting nature: regional reviews of protected areas. IUCN. Gland, Switzerland.

McShane, T.O. & Wells, M.P. (2004). Getting biodiversity projects to work: towards more effective conservation and development. New York, New York, USA: Columbia University Press.

Milder, J.C., Scherr, S.J. & Bracer, C. (2010). Trends and future potential of payment for ecosystem services to alleviate rural poverty in developing countries. Ecology and Society, 15(2), 4.

Mills, T.J., Quigley, T.M., & Everest, F.J. (2001). Science-based natural resource management decisions: what are they? Renewable Resources Journal, 19(2), 10-15.

Moore, J.F., Mulindahabi, F., Masozera, M.K., Nichols, J.D., Hines, J.E., Turikunkiko, E., & Oli, M.K. (2018). Are ranger patrols effective in reducing poaching-related threats within protected areas? Journal of Applied Ecology, 55(1), 99-107.

Moro, M., Fischer, A., Czajkowski, M., Brennan, D., Lowassa, A., Naiman, L.C., & Hanley, N. (2012). An investigation using the choice experiment method into options for reducing illegal bushmeat hunting in western Serengeti. Conservation Letters, 6(1), 37-45.

Morrison, M.L., Marcot, B.G., Mannan, R.W. (2006). Wildlife-Habitat relationships: concepts and applications. Washington, D.C., USA: Island Press.

Mouchet, M., Levers, C., Zupan, L., Kuemmerle, T., Plutzar, C., Erb, K., … Haberl, H. (2015). Testing the effectiveness of environmental variables to explain European terrestrial vertebrate species richness across biogeographical scales. PLoS ONE, 10(7), e0131924.

126

Myers, N. (1988). Threatened biotas: “hot spots” in tropical forests. Environmentalist, 8(3), 187-208.

Myers, N., Mittermeier, R. A., Mittermeier, C. G., da Fonseca, G. A. B., & Kent, J. (2000). Biodiversity hotspots for conservation priorities. Nature, 403(6772), 853- 858.

Newmark, W.D. (2008). Isolation of African protected areas. Frontiers in Ecology and the Environment, 6(6), 321-328.

Newton-Fisher, N. E. (2003). The home range of the Sonso community of chimpanzees from the Budongo Forest, Uganda. Journal of African Ecology, 41(2), 150-156.

Nguyen, T.H., Sinha, A., Gholami, S., Plumptre, A., Joppa, L., Tambe, M., … Beale, C. (2016). Protecting wildlife under imperfect observation. AAA Workshops, North America. Available from http://www.aaai.org/ocs/index.php/WS/AAAIW16/paper/view/12657 (accessed November 2016).

Nichols, J.D., Boulinier, T., Hines, J.E., Pollock, K.H., & Sauer, J.R. (1998). Estimating Rates of Local Species Extinction, Colonization, and Turnover in Animal Communities. Ecological Applications, 8(4), 1213-1225. doi:10.2307/2640974

Noon, B.R. (2003). Monitoring ecosystems: interdisciplinary approaches for evaluation ecoregional initiatives. Washington, D.C., USA: Island Press.

O’Brien, T.G., & Kinnaird, M.G. (1997). Behavior, diet, and movements of the Sulawesi Crested black macaque (Macaca nigra). International Journal of Primatology, 18(3), 321-351.

Oli, M.K., Taylor, I.R. & Rogers, M.E. (1994). Snow leopard (Panthera uncia) predation of livestock: an assessment of local perceptions in the Annapurna Conservation Area, Nepal. Biological Conservation, 68(1), 63-68.

Olupot, W., Chapman, C. A., Brown, C. H, & Waser, P. M. (1994). Mangabey (Cercocebus albigena) population density, group size and ranging: A twenty-year comparison. American Journal of Primatology, 32(3), 197-205.

Paine, R. (1980). Food Webs: Linkage, Interaction Strength and Community Infrastructure. Journal of Animal Ecology, 49(3), 666-685.

Pagiola, S., Arcenas, A. & Platais, G. (2005). Can payments for environmental services help reduce poverty? An exploration of the issues and evidence to date from Latin America. World Development, 33(2), 237-253.

127

Parody, J.M., Cuthbert, F.J., and Decker, E.H. (2001). The effect of 50 years of landscape change on species richness and community composition. Global Ecology and Biogeography, 10(3), 305-313.

Plumptre, A. J. (1995). The effects of trampling damage on herbivores on the vegetation of the Parc National des Volcans, Rwanda. African Journal of Ecology, 32(3), 115-129.

Plumptre, A.J. (2000). Monitoring mammal populations with line transect techniques in African forests. Journal of Applied Ecology, 37(2), 356-368.

Plumptre, A.J., Davenport, T.R.B., Behangana, M., Kityo, R., Eilu, G., Ssegawa, P., … Moyer, D. (2007). The biodiversity of the Albertine Rift. Biological Conservation, 134(2), 178-194.

Plumptre, A.J., Fuller, R.A., Rwetsiba, A., Wanyama, F., Kuirakwinja, D., Driciru, M., … Possingham, H.P. (2014). Efficiently targeting resources to deter illegal activities in protected areas. Journal of Applied Ecology, 51(3), 714-725.

Plumptre, A., Hart, J.A., Hicks, T.C., Nixon, S., Piel, A.K. & Pintea, L. (2016). Pan troglodytes ssp. schweinfurthii. (errata version published in 2016) The IUCN Red List of Threatened Species 2016: e.T15937A102329417.

Plumptre, A.J., Masozera, M., Fashing, P.J., McNeilage, A., Ewango, C., Kaplin, B.A., & Liengola, I. (2002). Biodiversity surveys of the Nyungwe Forest Reserve in S.W. Rwanda. Wildlife Conservation Society Working Paper, 19, 1-96.

Plumptre, A.J. & Reynolds, V. (1996). Censusing chimpanzees in the Budongo forest, Uganda. International Journal of Primatology, 17(1), 85-99.

Powell, R. A. (2012). Movements, home ranges, activity, and dispersal. In L. Boitani & R.A. Powell (Eds.), Carnivore ecology and conservation: a handbook of techniques (pp. 188–217). London, UK: Oxford University Press.

Purvis, A. & Hector, A. (2000). Getting the measure of biodiversity. Nature, 405, 212– 219.

Pusey, A. E., Pintea, L., Wilson, M. L., Kamenya, S., & Goodall, J. (2007). The contribution of long-term research at Gombe National Park to chimpanzee conservation. Conservation Biology, 21(3), 623-634.

R Development Core Team. (2015). R: A language and environment for statistical computing. R Foundation for Statistical Computing. Vienna. Available from http://www.R-project.org (accessed November 2015).

128

Rahbek, C. (1995). The elevational gradient of species richness: a uniform pattern? Ecography, 18(2), 200-205.

Reynolds, V., Plumptre, A. J., Greenham, J., & Harborne, J. (1998). Condensed tannins and sugars in the diet of chimpanzees (Pan troglodytes schweinfurthii) in the Budongo Forest, Uganda. Oecologia, 115(3), 331-336.

Ricketts, T. & Imhoff, M. (2003). Biodiversity, urban areas, and agriculture: locating priority ecoregions for conservation. Conservation Ecology, 8(2), http://www.consecol.org/vol8/iss2/art1/.

Risdianto, D., Martyr, D.J., Nugraha, R.T., Harihar, A., Wibisono, H.T., Haidir, I.A., … Linkie, M. (2016). Examining the shifting patterns of poaching from a long-term enforcement intervention in Sumatra. Biological Conservation, 204(Part B), 306- 312.

Rovero, F., Martin, E., Rosa, M., Ahumada, J.A., and Spitale, D. (2014). Estimating species richness and modelling habitat preferences of tropical forest mammals from camera trap data. PLos ONE, 9(7), e103300.

Royle, J.A., & Nichols, J.D. (2003). Estimating abundance from repeated presence/absence data or point counts. Ecology, 84(3), 777-790.

Seaman, D. E., & Powell, R. A. (1996). An evaluation of the accuracy of kernel density estimators for home range analysis. Ecology, 7(7), 2075-2085.

Sharma, K., Wright, B., Joseph, T., & Desai, N. (2014). Tiger poaching and trafficking in India: Estimating rates of occurrence and detection over four decades. Biological Conservation, 179, 33-39.

Smith, D.R. & Anderson, D.R. (1987). Effects of lengthy ringing periods on estimators of annual survival. Acta Ornithologica, 23(1), 69-76.

South, A. (1999). Extrapolating from individual movement behavior to population spacing patterns in a ranging mammal. Ecological Modelling, 117(2), 343-360.

St John, F.A.V., Keane, A.M., Edwards-Jones, G., Jones, L., Yarnell, R.W., & Jones, J.P.G. (2012). Identifying indicators of illegal behaviour: carnivore killing in human-managed landscapes. Proceedings of the Royal Society B, 279(1729), 804-812.

Stanford, C. B., & Nkurunungi, J. B. (2003). Behavioral ecology of sympatric chimpanzees and gorillas in Bwindi Impenetrable National Park, Uganda: diet. International Journal of Primatology, 24(4), 901-918.

129

Stefanescu, C., Herrando, S., and Paramo, F. (2004). Butterfly species richness in the north-west Mediterranean Basin: the role of natural and human-induced factors. Journal of Biogeography, 31(6), 905-915.

Stein, B.A., Staudt, A., Cross, M.S., Dubois, N.S., Enquist, C., Griffis, R., … Paris, A. (2013). Preparing for and managing change: climate adaptation for biodiversity and ecosystems. Frontiers in Ecology and the Environment, 11, 502-510.

Steinmetz, R., Srirattanaporn, S., Mor-Tip, J., & Seuaturien, N. (2014). Can community outreach alleviate poaching pressure and recover wildlife in South-East Asian protected areas? Journal of Applied Ecology, 51(6), 1469-1478.

Stokes, E.J. (2010). Improving effectiveness of protection efforts in tiger source sites: developing a framework for law enforcement monitoring using MIST. Integrative Zoology, 5(4), 363-377.

Stoner, C., Caro, T., Mduma, S., Mlingwa, C., Sabuni, G., Borner, M., & Schelten, C. (2007). Changes in large herbivore populations across large areas of Tanzania. African Journal of Ecology, 45(2), 202-215.

Struhsaker, T.T. (1997). Ecology of an African rainforest. Gainesville, Florida, USA: University Press of Florida.

Sun, C., Kaplin, B., Kristensen, K. A., Munyaligoga, V., Mvuklyumwami, J., Kajondo, K., & Moermond, T. C. (1996). Tree phenology in a tropical montane forest in Rwanda. Biotropica, 28(4), 668-681.

Sutter, R.D., Wainscott, S.B., Boetsch, J.R., Palmer, C.J., and Rugg, D.J. (2015). Practical guidance for integrating data management into long-term ecological monitoring projects. Wildlife Society Bulletin, 39, 451-463.

Swihart, R.K., & Slade, N.A. (1985). Influence of sampling interval on estimates of home-range size. The Journal of Wildlife Management, 49(4), 1019-1025.

Tanner, E.V.J., Vitousek, P.M., & Cuevas, E. (1998). Experimental investigation of nutrient limitation of forest growth on wet tropical mountains. Ecology, 79(1), 10- 22.

Tobler, M.W., Hartley, A.Z., Carrillo-Percastegui, S.E., Powell, G.V.N. (2015). Spatiotemporal hierarchical modelling of species richness and occupancy using camera trap data. Journal of Applied Ecology, 52(2), 413-421.

Turner, L.A. (2000). Chimpanzee ranging and vegetation at Kasoje in Mahale Mountains National Park, Tanzania. Kyoto University (unpublished doctoral dissertation).

130

Tweheyo, M., & Babweteera, F. (2007). Production, seasonality, and management of chimpanzee food trees in Budongo Forest, Uganda. African Journal of Ecology, 45(4), 535-544.

Tweheyo, M., Lye, K. A., & Weladji, R. B. (2004). Chimpanzee diet and habitat selection in the Budongo Forest Reserve, Uganda. Forest Ecology and Management, 188(1), 267-278.

United Nations Development Program (UNDP). (2014). Country Information: Rwanda. Retrieved from http://www.rw.undp.org/content/rwanda/en/home/countryinfo.html.

Van Orsdol, K. G., Hanby, J. P., & Bygott, J. D. (1985). Ecological correlates of lion social organization (Panthera leo). Journal of the Zoological Society of London, 206(1), 97-112. van Schaik, C. P., Madden, R. H., & Ganzhorn, J. U. (2005). Seasonality and primate communities. In Brockman D. K. and van Schaik, C. P. (Eds.), Seasonality in Primates: Studies of Living and Extinct Human and Non-Human Primates (pp. 445-463). Cambridge, UK: Cambridge University Press.

Vasconcelos, R.P., Henriques, S., Franca, S., Pasquaud, S., Cardoso, I., Laborde, M., and Cabral, H.N. (2015). Global patterns and predictors of fish species richness in estuaries. Journal of Animal Ecology, 84(5), 1331-1341.

Vedder, A. L. (1984). Movement patterns of a group of free-ranging mountain gorillas (Gorilla gorilla beringei) and their relation to food availability. American Journal of Primatology, 7(2), 73-88.

Veneklaas, E. J. (1991). Litterfall and nutrient fluxes in two montane tropical rain forests, Colombia. Journal of Tropical Ecology, 7(3), 319-336.

Watson, F., Becker, M.S., McRobb, R., & Kanyembo, B. (2013). Spatial patterns of wire- snare poaching: Implications for community conservation in buffer zones around National Parks. Biological Conservation, 168, 1-9.

Wearn, O.R., Carbone, C., Rowcliffe, J.M., Bernard, H., and Ewers, R.M. (2016). Grain- dependent responses of mammalian diversity to land use and the implications for conservation set-aside. Ecological Applications, 26(5), 1409-1420.

White, G.C. & Burnham, K.P. (1999). Program MARK: Survival estimation from populations of marked animals. Bird Study, 46, 120-139.

Wilcox, B.A. (1984). In situ conservation of genetic resources: determinants of minimum area requirements. In McNeely, J.A. & Miller, K.R. (Eds.), National parks: conservation and development (pp. 639-647). Washington, D.C., USA: Smithsonian Institution Press.

131

Wildlife Conservation Society (WCS). (2009). Biodiversity survey of Nyungwe National Park, Rwanda. Unpublished progress report, Wildlife Conservation Society, Bronx, New York, USA.

Wilkie, D., Painter, M. & Jacob, I. (2016). Rewards and Risks associated with community engagement in anti-poaching and anti-trafficking. Biodiversity Technical Brief. United States Agency for International Development, Washington.

Williams, B.K., Nichols, J.D., & Conroy, M.J. (2002). Analysis and management of animal populations. San Diego, California, USA: Academic Press.

The World Bank. (2017). World Development Indicators. Retrieved 30 April 2018.

Worton, B. J. (1989). Kernel methods for estimating the utilization distribution in home- range studies. Ecology, 70(1), 164-168.

Wright, J.H., Hill, N.A.O., Roe, D., Rowcliffe, J.M., Kümpel, N.F., Day, M., … Milner- Gulland, E.J. (2016). Reframing the concept of alternative livelihoods. Conservation Biology, 30(1), 7-13.

Wunder, S. (2007). The efficiency of payments for environmental services in tropical conservation. Conservation Biology, 21(1), 48-58.

Yoccoz, N.G., Nichols, J.D. & Boulinier, T. (2001). Monitoring of biological diversity in space and time. Trends in Ecology and Evolution, 16, 446–453.

Zhou, Y., Buesching, C.D., Newman, C., Kaneko, Y., Xie, Z., & Macdonald, D.W. (2013). Balancing the benefits of ecotourism and development: The effects of visitor trail-use on mammals in a Protected Area in rapidly developing China. Biological Conservation, 165, 18-24.

Zipkin, E.F., DeWan, A., and Royle, J.A. (2009). Impacts of forest fragmentation on species richness: a hierarchical approach to community modelling. Journal of Applied Ecology, 46(4), 815-822.

Zipkin, E.F., Royle, J.A., Dawson, D.K., and Bates, S. (2010). Multi-species occurrence models to evaluate the effects of conservation and management actions. Biological Conservation, 143(2), 479-484.

132

BIOGRAPHICAL SKETCH

Jennifer Moore received her Bachelor of Arts in statistics and geography with a minor in environmental policy & culture and her Bachelor of Music in oboe performance in 2011 from Northwestern University. She received her Master of Environmental

Management with a certificate in geospatial analysis in 2013 from the Nicholas School of the Environment at Duke University. Jennifer has been conducting research in Africa since 2012 and has worked in Madagascar and Gabon, as well as Rwanda. Her interests are in the use of statistics and population models in answering questions related to wildlife conservation and protected area management, with a focus on African wildlife and national parks. Jennifer began her PhD in the Department of Wildlife

Ecology and Conservation at the University of Florida in August of 2014 and completed her degree in August of 2018.

133