I METHODS of NICHE PARTITIONING BETWEEN

Home , Andean mountain cat, Kinkajou, Margay, Pampas cat, Pantanal cat, Speothos

METHODS OF NICHE PARTITIONING BETWEEN ECUADORIAN CARNIVORES AND HABITAT PREFERENCE OF THE MARGAY ( LEOPARDUS WIEDII )

Anne-Marie C. Hodge

A Thesis Submitted to the University of North Carolina Wilmington in Partial Fulfillment for the Requirements for the Degree of Master of Science

Department of Biology and Marine Biology

University of North Carolina Wilmington

2012

Approved By:

Advisory Committee

Travis Knowles Steven Emslie

Marcel van Tuinen Brian Arbogast Chair

Accepted by

Dean, Graduate School

TABLE OF CONTENTS ABSTRACT ...... iv DEDICATION ...... v LIST OF TABLES ...... vi LIST OF FIGURES ...... vii CHAPTER 1: MARGAY ACTIVITY PATTERNS AND DENSITY...... 1 Introduction ...... 1 Methods...... 5 Study Location ...... 5 Equipment ...... 6 Transect Design ...... 7 Calculations...... 8 Results ...... 10 Discussion ...... 12 CHAPTER 2: MARGAY HABITAT CHOICE ...... 18 Introduction ...... 18 Habitat Terminology ...... 20 Margay Habitat Choice ...... 21 Conservation Implications of Habitat Choice ...... 22 Methods...... 24 Variables Measured ...... 24 Analyses Conducted...... 24 Results ...... 26 Discussion ...... 27 CHAPTER 3: INTRAGUILD NICHE PARTITIONING ...... 36 Introduction ...... 36 Methods...... 39 Trap Success ...... 39 Latency to Initial Detection ...... 39 Sørensen’s Index ...... 40 Temporal Activity Patterns ...... 40

Results ...... 41 Discussion ...... 43 CHAPTER 4: COMPARISON OF DOMESTIC CANINE AND MARGAY TRAP SUCCESS ..63 Introduction ...... 63 Methods...... 66 Results ...... 67 Discussion ...... 69 CHAPTER 5: MARGAY LITERATURE REVIEW ...... 77 Introduction ...... 77 Evolution ...... 77 Distribution ...... 79 Morphology and Physiology ...... 79 Reproduction and Development ...... 80 Feeding Behavior ...... 82 Habitat ...... 83 Home Range and Activity Patterns ...... 83 Current Conservation Status ...... 84 REFERENCES ...... 87

iii

ABSTRACT

Assessing the abundance, distribution, and spatial and temporal activity patterns of carnivores is an essential step towards understanding their basic ecology and interactions with sympatric guild members. Camera-trapping techniques are of great utility in such studies, as they are noninvasive, provide synchronized photographic and environmental data, and minimize disturbance of animals’ natural activity patterns. The present study was conducted at the Wildsumaco Wildlife Sanctuary in Sumaco, Ecuador, between July 2010 and July 2011. WWS is situated between 1300 and 1400 m on the eastern slope of the Andes mountains. This region encompasses a rare zone of elevational overlap between upland Andean and lowland Amazonian species, resulting in a highly diverse and unique assemblage of mammals. Transects of remote cameras were established in order to assess the effects of habitat structure, baiting regimen, and activity patterns of sympatric carnivores and domestic dogs on margay camera-trap success. Canopy cover was significantly correlated with margay trap success. The data also support the hypothesis that sympatric carnivores use complementary niche partitioning as a mechanism of co-existence, compensating for overlap in one or more niche dimensions with separation on one or more other dimensions. Domestic dogs were shown to have negative effects on margay trap success, and to penetrate over 500 meters beyond forest edges. The results of this study yielded insights into the mechanisms utilized to allow co-existence of carnivores in the eastern Andes, in addition to producing new information on habitat preferences and behavior of the margay. It is hoped that the results of this study will lay important groundwork for further investigation of mammal community ecology in the Sumaco region.

DEDICATION

Many thanks to my major professor, Brian Arbogast, for his mentorship, guidance, logistical support, and friendship. I would also like to thank Travis Knowles, who had the most direct involvement with the implementation of my study “on the ground” at Sumaco. My other committee members, Steve Emslie and Marcel Van Tuinen, also provided valuable advice and support. In addition, Fred Scharf and Adam Fominaya were instrumental in helping with statistical analyses. The following people were essential in assisting with the physical establishment of camera transects and habitat data collection: Debbie Folkerts, Molly Folkerts, Katie Glynn, Tom McMeans, Pamela Rivera, Aaron Corcoran, Katie Dowell, and Lauren Gentry. Many thanks to Kathy Hodge, Gerald Hodge, Andrea Arbogast, and Dae Grodin for assuming stewardship of Charlie—my best friend and sidekick—while I was on my equatorial excursions. This study would not have been possible without the generosity of the owners and managers of the Wildsumaco Wildlife Sanctuary: Jim Olson, Bonnie Olson, and Jonas Nilsson. Extra thanks to Jim for facilitating the most astounding cribbage victory of my long record of skunking adversaries. The field staff at WWS was essential in assisting with transport and monitoring of the cameras, especially Campeón, Miguel, and Byrón. I am thankful to all of the drivers from Marco Tours that safely transported us back and forth over the Andean pass multiple times without any catastrophic collisions or accidental dives from precipices. Easier said than done, and highly appreciated. Thanks to Virginia Chala for all of the candlelight Spanish tutoring sessions, jokes, and never actually making us eat sopa del mono . It is with regret but fondness that we remember our late friend Miguel Chala, a Wildsumaco employee whose infectious cheer greatly enriched the Sumaco experience for everyone with whom he interacted.

LIST OF TABLES

Table Page 1.1: Summary of margay captures for each transect ...... 15 1.2: Total number of individuals and average number of captures per individual for male and female margays ...... 16

2.1: Habitat choice—Multiple regression results...... 30 2.2: Ranking matrix of margay habitat preference ...... 31 2.3: Simplified habitat ranking matrix ...... 32 2.4: Habitat choice—Chi Square results ...... 33 3.1: Capture success for mammalian carnivores at Wildsumaco ...... 54 3.2: Latency to detection for carnivores at Wildsumaco ...... 55 3.3: Sørensen’s Index of Overlap values ...... 56 3.4: Temporal activity patterns of sympatric carnivores ...... 57 4.1: Comparison of margay and dog captures with and without bait ...... 72 4.2: Large carnivore captures with and without bait...... 73 4.3: Distance from road of margay and dog captures ...... 74 5.1: Comparison of margay body size between sexes ...... 85

LIST OF FIGURES

Figure Page 1.1: Map of study area ...... 17 2.1: Habitat choice p-plot ...... 34 2.2: Availability vs utilization of canopy cover ...... 35 3.1: Sørensen’s Index of overlap results ...... 58 3.2: Correlation between Sørensen’s Index value and difference in body mass ...... 59 3.3: Latency to detection ...... 60 3.4: Comparison of latency to detection and total trap success ...... 62 3.5: Temporal activity patterns ...... 63 4.1: Distance from road to margay and dog capture locations ...... 75 4.2: Proportion of exclusive capture locations for margays and dogs ...... 76 5.1: Current distribution of the margay ...... 86

vii

CHAPTER 1: MARGAY ACTIVITY PATTERNS AND DENSITY

INTRODUCTION

Conducting population surveys of elusive, nocturnal species—such as mammalian carnivores—can be a challenging undertaking, especially in densely forested habitats.

Traditional methods of trapping and attaching radio collars or other tracking devices to animals often may involve considerable stress and disturbance for both researchers and wildlife. In addition, sample sizes using these methods are constrained by the cost requirements of having a telemetry or GPS collar for each individual animal and the logistical complications with successfully capturing target species (Gese 2001). Surveys of tracks and other indicators of animal presence are other less invasive methods of calculating population abundance, yet the time and effort required to use these tracking methods to monitor wide-ranging carnivores may be prohibitive, especially in challenging terrain such as tropical mountainsides. In addition, comparative studies have shown camera studies to be more accurate than track surveys in determining carnivore population densities (Balme et al. 2009).

Over the past decade, advances in remotely triggered camera technology have led to a boom in the use of this of methodology for population surveys (Kays et al. 2009, Kelly 2008). Camera sampling is noninvasive, requiring no immobilization of or physical contact with the study animals, and no risk of harming non-target species that happen upon traps. Digital cameras are capable of storing vast numbers of samples, in some cases the equivalent of museum specimens, with no disturbance to the organisms themselves and no need to remove individuals from the population in order to document their occurrence. In addition, the technique is much more time efficient than traditional invasive survey methods. Kelly (2008) notes that the number of “trap

nights” possible in a mere two-month camera survey matches that of traditional inventories that took 1-21 years. This efficiency is especially critical for the study of large carnivores, which exist at low densities across expansive home ranges (Gompper et al. 2006). This strategy has been used to study a range of life history topics, including population estimation, nesting behavior, feeding ecology, habitat selection, and temporal activity patterns (Cutler and Swann

1999). Although Larrucea et al. (2007) show that camera trapping may not provide an unbiased sample of the population—with certain individuals or species having higher detectability—the method is less biased than the alternative, which involves physically trapping, subduing, and/or repeatedly following a limited number of individuals.

The combination of remote camera surveys with capture-recapture statistical analysis can be used to estimate population density and distribution (Karanth 1995, Karanth and Nichols 1998,

Silver et al. 2004). Obtaining population density estimates for elusive species is critical. Many regions are currently experiencing devastating rates of habitat loss, and once a habitat has been destroyed it may be impossible to reconstruct the baseline densities and population abundances for its biota. Without knowledge of how declining carnivores use the landscape, and in what abundances and spatial distributions, conservation efforts will have no baseline from which to begin and no final goal for which to aim.

Another advantage of remote camera surveys is that “information on non-targets often far outweighs the information obtained on target species,” due to the relatively low abundance of mammalian carnivores within their communities (Kelly and Holub 2008). This non-target information has two main benefits: 1) these “by-catch photos” provide the opportunity to collect data on prey species and other animals which share habitat with the target carnivores; and 2) even though non-target species photos may make up the vast majority of the raw dataset, those

extra data cost a minimal amount in terms of resources and no extra trapping effort, as they only demand a bit of battery power and digital storage space. In areas which are rich in biodiversity yet poorly represented in the literature, such as eastern Ecuador, the side benefit of community- wide biodiversity sampling will be extremely valuable in documenting the presence—and possibly even the first records of existence at mid-elevation altitudes—of species within this understudied faunal community.

The margay is an ideal focal animal for a camera trap study, as each individual bears a distinct and unique pattern of spots on its coat. These act as natural “tags,” and allow researchers to distinguish specific individuals, enabling estimations of abundance and movement patterns

(Silver et al. 2004). Despite its wide geographic range, the margay has been relatively understudied, and little information is available regarding its habitat preferences. In order to prevent this declining species from edging ever closer to extinction, it is crucial to understand what the habitat that allows it to thrive. Camera studies allow documentation of natural habitat preferences without the risk of changing an animal’s behavior or movement patterns after a physical trapping event (Di Bitetti et al. 2006).

Due to their elusive habits and widespread movements, even indirect observations of carnivores can be difficult to obtain. To help increase detectability of animals during camera surveys, bait and scent lures are sometimes used to bring the animals within the sensory range of the cameras (Downey et al. 2007 ; Moen and Lindquist 2006; Mowat et al. 2000). Although food bait is often used, it does not last long in the field, especially in the tropics, and is difficult to standardize due to variation in the time that it will remain at the station, due to vagaries of the weather and the appetites of passing animals (Schelexer 2008). In addition, consumable

“reward” baits may result in repeated sampling as animals become habituated to a regularly

replenished food source, which can affect analyses of abundance and movement patterns

(Brongo et al. 2005).

A more practical option for studies in remote, steep terrain are scent lures, which can be renewed at standardized times and are not consumed by the target animals (Graves and

Boddicker 1987; Dobbins 2004). Diefenbach et al. (1994) found that bobcat population size correlated to the proportion of scent stations visited, and Conner et al. (1983) found that population indices determined from scent station records correlated to changes in abundances of raccoons, bobcats, and grey foxes that were also tracked using more invasive methods, such as radio-collars. While liquid scent lures have often been used (York et al. 2001), they may be less effective in extremely wet climates. For the present study, I used scent lure tablets, which are plaster discs impregnated with several different fatty acids. Harrison (1997) showed that margays spent more time investigating several lures, including the fatty acid tablets, than they did control lures. For these reasons, in addition to their portability and easy storage maintenance, fatty acid tablets were chosen as the best form of lure for this project.

Pilot studies suggested that Sumaco is home to an unusually high density of margays, as 85 capture events were recorded over the course of 3220 trap nights within a 5 km 2 area

(Vanderhoff et al. 2012), a trap success rate of 0.026. This rate was very high relative to capture success rates documented in other studies, which ranged from 0.006 (Dillon 2005) down to 0.003

(Goulart et al. 2009). The preliminary results from the pilot testing period at Sumaco suggested that this locality was an ideal place to investigate margay activity patterns. In addition, the low margay trap success found in previous studies at other sites could have been due to different densities or detectability of margays in those areas, or it could have been due to the fact that the

sampling transects were designed for larger cat species and thus covered a larger area but were more sparsely positioned.

A main objective of this study was to use camera traps to study the activity patterns, density, and abundance of margays in a mid-elevation tropical forest. Specifically, I sought to use remote camera data to 1) estimate the local density and abundance of margay populations at Sumaco; 3) determine the degree of spatial overlap between and within sexes; and 4) test the effectiveness of fatty acid scent lures for attracting margays to camera stations.

METHODS

Study Location

Fieldwork was conducted at Wildsumaco Wildlife Sanctuary (WWS), which is located just

above the 1000-1200 m plateau that comprises the southern side of the Volcán Sumaco. The

cone of the volcano rises 3732 meters. Volcán Sumaco lies within the Gran Sumaco region of

the Napo province (Fig. 2.1). This site falls within one of the world’s 25 recognized biodiversity

hotspots, the Tropical Andes region. WWS is a 5 km2 area of comprised of both secondary and primary forest patches, interspersed with cow pastures and small homesteads.

Being the easternmost volcano found along the Ecuadorian stretch of the Andes, Sumaco bears the distinction of linking the Andean peaks to the Amazon basin. Thus, this volcano boasts high biodiversity, with up to 6000 plant species, many of which are endemic (Neill and Palacios

2011). It is a critical area of concern, as it is one of the only remaining regions in Ecuador that preserves large tracts of primary forest that corridor includes such a wide elevational range, and because the páramo habitat covering its upper reaches is thought to be the only zone of such habitat in the country that has never been disturbed by fire or grazing (Neill and Palacios 2011)

The slopes of Sumaco receive 3500-6000 mm of rain every year. Rainfall for a given month varies greatly between years, yielding no true dry season, but the months of July-August and

November-January tend to receive less rainfall than other months (Neill and Palacios 2011).

Sumaco is the source of water that flows through many of the tributaries that drain into the

Upper Napo, making it a critical regulator of the local hydrological cycle. Due to its high degree of biodiversity and critical ecosystem services, a large portion of the Gran Sumaco is protected by the Ecuadorian government. Unfortunately, enforcement is lax and deforestation and agriculture have encroached despite its protected status. The Napo province’s cattle boom wilted in the 1990s, leaving residents grasping for new ways to support themselves (Perreault 2003).

The ground cover is unsuitable for many cash crops, so the local economy depends largely on the few plants that do grow well there, primarily naranjilla ( Solanum quitoense ) and tree tomatoes

(Cyphomandra betacea ). Naranjilla exhausts the soil after about three years, requiring intermittent fallow periods before a plot of land can be recultivated. It is also highly vulnerable to damage from fungi and insects, which means that cultivation of this crop requires heavy use of pesticides and fungicides (Buck 2006).

WWS land ranges from 1250-1450 m in elevation and has one gravel road, which runs approximately north-south, with the northbound direction going up in elevation. The sanctuary also includes a network of well-maintained hiking trails. Not all of the forest within WWS is contiguous. Ground cover within the study site ranges from cleared pastures to primary premontane forest.

Equipment

I used digital Reconyx RC55 cameras, with both motion and infrared sensors enabled. The cameras were set to Rapidfire mode, automatically taking one picture every 1-2 seconds after

being triggered. Upon being triggered, the cameras shot ten continuous photos without the need for additional stimulation. The camera batteries and memory cards were checked weekly to ensure the equipment was in good working order. Photos were stored to 4 GB SD memory cards, which have enough capacity to hold tens of thousands of photos without exceeding their storage capacity, due to the compressed file size generated by the Reconyx cameras.

Transect Design

Each session consisted of two transects: one through forest, and one along forest edge. Each transect was comprised of eight forest stations and four edge stations. When possible, two cameras were set at each station in order to photograph passing animals from both sides. This was not always possible due to constraints on the number of cameras operating at a given time, as the humid conditions were very challenging for the equipment.

Cameras were mounted on tree trunks less than one meter from the ground. A second set of cameras was mounted on tree trunks facing the canopy 3-4 meters from the ground. There were four of these “tree cameras” on the forest transect and two on each edge transect.

Sampling Duration

Data were collected between July 2010 and June 2011, over the course of four separate sampling sessions. Each session was 9-12 weeks in duration. Transects and sampling durations were as follows: F.A.C.E. Trail: 7 July-10 October 2010; Benavides Trail: 8 October-3

December 2010; Waterfall Trail: 16 December 2010-17 February 2011; Puffbird Trail: 20

February-27 May 2011. The number of trap nights—the product of the number of nights sampled and the number of cameras deployed—per transect were as follows: F.A.C.E.: 1065 trap nights, Benavides: 992 trap nights, Waterfall: 1120 trap nights, and Puffbird: 864 trap nights.

Due to the failure of the tree cameras to capture any carnivores and logistical concerns with site

selection for canopy cameras, they were not included in this analysis, and those cameras were not counted towards the total number of trap nights. Double-camera stations were counted as a single trap station, and thus represented as a single trap night per 24 hour period.

Spacing

Camera stations were spaced 200-300 meters apart on each transect. Because few camera surveys have specifically targeted carnivore species as small as the margay, the correct spacing was estimated by scaling down distances used in studies of larger felids (Kelly 2003; Silver et al.

2004; Wang and McDonald 2009). I scaled reported average home range for the different species to the average of reported home range estimates for the margay to ensure that the cameras were spaced in an equivalent proportion to home range size similar to studies of larger species.

Bait

On two transects, F.A.C.E. Trail and Waterfall Trail, the first half of the sampling period (4-5 weeks) was unbaited. Beginning with the 5 th week, one fatty acid tablet was placed at each camera station each week for the remaining 5-7 weeks of the sampling session. Tablets were placed on the ground approximately 2 m in front of each camera. The fatty acid tablets were ordered from the USDA Pocatello Supply Depot in Pocatello, Idaho, and consisted of 1-inch discs of plaster of Paris impregnated with eight different fatty acid compounds.

Calculations

Trap Success

independent. I calculated TS for each carnivore species (margay, puma, jaguarundi, ocelot, grisón, tayra), and domestic dogs. Other vertebrates were grouped into several categories, based on body mass as reported in the literature (Tirira 2007): small prey mammals (< 5kg), medium prey mammals (5-15 kg), large prey mammals (> 5 kg), small birds (< 500 g), large birds, (> 500 g) and small carnivores (kinkajous, coatis, and weasels), and TS was calculated for each of these groups as well. A correlation matrix was created in SPSS to identify any other mammal groups that showed trap successes significantly correlated to that of the margay.

Sample area

Effective sample area was determined by calculating the area of the study site using Google

Earth. Waypoints and tracks of each transect were taken using a handheld GPS unit (Earthmate

P-40). These data were uploaded onto a computer and the resulting .GPX file was imported into

Google Earth. A polygon encompassing the study area was drawn using these transects. This polygon was divided in half into two triangles, and their areas were calculated and summed to find the total polygon area using a free online area calculation program (www.csgnetwork.com).

Individual Identification

Each margay photo was compared to all others in order to match individuals based on spot patterns. The entire set of locations and times of capture for each margay was recorded. This procedure was conducted separately by both A. M. C. Hodge and E. N. Vanderhoff, to achieve independent verification of identifications.

Latency to Detection

Calculation of the latency to initial detection (LTD) was used as an additional method to determine the effectiveness of noninvasive sampling techniques (Gompper et al. 2006; Foresman and Pearson 1998). This metric is calculated by counting the number of days from the setting of

the cameras until the first detection of a given species, using the time- and date-stamps that the digital cameras attach to each image. The margay LTD was calculated for each of the 4 transects sampled in this study: F.A.C.E Trail, Benavides Trail, Waterfall Trail, and Puffbird Trail.

Density

Population density for each species was calculating by dividing the estimated abundance by the effective sample area.

Effect of bait

Paired Student’s t-tests were performed in order to to compare margay trap success between baited and non-baited treatments on the two transects on which bait was used, the F.A.C.E. and

Waterfall trails. The effect of bait on domestic dog activity was also tested, in order to determine whether the bait increased domestic dog presence at the camera stations, and whether that affected margay trap success.

RESULTS

Margay Trap Success

There were 33 margay capture events on the F.A.C.E. transect, 17 on the Benavides transect,

40 on the Waterfall transect, and 13 on the Puffbird transect, for a total of 103 margay capture events across the entire study (Table 2.1). The trap success (TS) was 0.031 for the F.A.C.E trail,

0.017 for Benavides, 0.036 for Waterfall, and 0.015 for Puffbird. The TS across the entire study was 0.0255.

Individual Identifications and Abundance

Individual margays from each photo event were compared using their distinctive spot patterns, to determine the number of unique individuals captured on each transect. The number of individuals recorded within each transect are as follows: F.A.C.E.: 6, Benavides: 4, Waterfall:

4, Puffbird: 2. After cross-matching margays captured on different transects, it was determined that 12 distinct individuals were photographed over the course of the study.

On the Waterfall transect, there were 4 capture events that recorded 2 margays in a single capture event; individual identifications using spot patterns showed that the same two individuals were recorded together each time. Two capture events were not assigned to an individual identification, as the quality and/or angle did not allow for matching of coat patterns.

At least one individual, a male, was recorded on multiple transects and seemed to dominate the margay capture events. This individual was recorded 25 times on F.A.C.E., 1 time on

Benavides, 7 times on Waterfall, and 10 times on Puffbird, and accounted for 75.7%, 5.8%,

17.5%, and 76.9% of capture events on these transects, respectively. This male accounted for

41.7% of all margay capture events across the study.

There were more captures per individual for males than for females: across the entire study

(all 4 transects), there were 3 individual males, captured an average of 19.3 times each. In contrast, the 6 individual females were captured an average of 5 times each. It was impossible to assign a sex to 3 of the individuals, which could either be females or juvenile males. These 3 individuals were captured an average of 2 times each.

Density

The total area of the polygon encompassing the study site was 2.32 km 2. Given the abundance estimate of 13 individuals photographed across the study site, the data suggest a margay density at WWS of 5.6 margays/km 2.

Effect of bait

When margay trap success (TS) before bait and during the use of bait on the F.A.C.E. trail

were compared using a Student’s t-test, use of bait significantly decreased margay TS (p =

0.0402), yet bait did not have a significant effect on margay trap success on the Waterfall Trail

(p = 0.1330). When margay TS is compared across transects, however, transects that included bait for part of their sampling session had a significantly higher margay TS (p = 0.0229). [See

Chapter 5 for additional results relevant to these data].

DISCUSSION

The margay trap success at Sumaco was strikingly high compared to margay trap success in previous camera surveys of tropical carnivores (Dillon 2005; Kelly 2003) —and nearly 3 times as high as that of the most successful study prior to this Goulart 2009). There are multiple potential explanations for the relatively high trap success and abundance of margays within such a small area. Sumaco exists within a patched network of forest and cleared land, with alternating plots of intact forest, cow pastures, naranjilla fields, and human homesteads. It is possible that a

“habitat packing” effect is occurring, in which margays are seeking refuge in remaining forest patches as surrounding habitat is destroyed, creating higher densities than would be found under undisturbed circumstances. This would be consistent with previous observations that small cats sometimes have higher population densities in small habitat fragments than in larger ones

(Passamani et al. 1997 [in Whiteman et al. 2007]), and the fragmented nature of WWS and its surroundings would fit the description of such a landscape.

It is also possible that margays at Sumaco are no more abundant than at other sites, yet they are more detectable due to having more terrestrial movement patterns than do margay populations in other localities or regions. All of the margay photo capture events were made by ground cameras, despite having a set of cameras set in the canopy along each transect. The rarity of ocelots and jaguarundis and lack of jaguars may allow margays to spend more time on the forest floor than at areas where the larger, more dominant cats are more abundant. This could

appear to be a mesopredator release effect, and yet may simply be an apparent effect due to sampling bias: the margays may not be more abundant, but simply more terrestrial and thus more detectable. Further, detailed studies of margay movement patterns in multiple localities will be needed to elucidate this issue.

Bait appeared to significantly increase the chances of a margay photo capture across all transects, and yet within specific transects the effects differed: on the F.A.C.E trail bait actually significantly decreased the margay capture frequency, while there was no significant effect on the Waterfall trail. This suggests that there was some difference between the two transects that may have altered the effects of the bait on the margays. The difference also could have been related to seasons. The F.A.C.E. trail was sampled from late July-early October of 2010, during the tail end of the rainy season and beginning of the dry season. In contrast, the Waterfall Trail was sampled from mid-December to mid-February, during the most significant dry period of the year according to both measurements at WWS and to rainfall patterns reported in the literature

(Buytaert et al. 2006). It is possible that the precipitation may have had an impact on the attenuation of the bait’s scent. In addition, activity of other animals that may be attracted to the bait, such as domestic dogs, may have an aversive effect on margay presence that overrides any attraction of margays to the bait.

Male margays had a much higher number of captures per individual, although this result was skewed by the fact that the most common male was captured 49 times. It is possible that the larger home range size of males may contribute to more frequent and farther excursions across the landscape. These travels likely result in detection rates of males that are disproportionate to their abundance in the local population relative to females due to the fact that males tend to cover more ground and thus are more likely to frequent the game trails on which camera stations are

located. In addition, female movements may be limited due to avoidance of the territories of other females, as many cats exhibit more overlap between males and females than between individuals of the same sex.

Table 1.1 The number of margay captures on each transect and the number of individuals comprising those captures, in addition to the sex, latency to detection, and overall TS for each transect.

Latency to Margay Margay Individual Undetermined Transect Males Females Initial Captures/Trap Captures Margays Sex Detection Night F.A.C.E. 33 6 2 2 2 1 0.031 Benavides 17 4 1 2 1 4 0.017 Waterfall 40 4 2 2 0 1 0.036 Puffbird 13 2 1 0 1 8 0.015 Mean = TOTAL 103 12* 3* 6* 3* Mean = 0.025 3.5

*This total does not represent the sum of individuals from the 4 transects, as some individuals appeared on more than one transect

Table 1.2 Total number of individuals and average number of captures per individual for male and female margays.

Males Females Unknown Total individuals 3 6 3

Average captures per individual 19.3 5 2

Figure 1.1 Map of study area: Sumaco National Park and Volcán Sumaco. Map source: Smithsonian Institution.

Chapter 2: Margay Habitat Choice

INTRODUCTION

The selection of suitable habitat by an animal is a multi-faceted issue. Block and Brennan

(1993) pointed out that “habitat” is defined across a variety of scales, ranging from entire regions and landscapes to specific microhabitats. Habitat requirements at the broadest scale—geographic range—are largely dependent upon phylogenetic history (Hutto 1985) and historical biogeography.

Within the geographic range of its species, an individual animal must find a suitable home range—one that provides adequate resources and mating opportunities, yet is not overcrowded or overly disturbed by human activities. Jaenike and Holt (1991) point out that it is likely that the selective pressures for habitat preference within a species fall into the category of “soft selection,” allowing for behavioral polymorphisms that allow the species to adapt in the face of changing environmental conditions or shifts in the densities of either conspecifics, predators, or prey. These polymorphisms may also allow conspecifics to share a territory or adjacent home ranges while avoiding strong competition for the same core areas.

An animal’s home range must fall within a range of climatic conditions amenable to both the animal in question and its preferred prey species—conditions that often vary seasonally at a given location. Powell (2000) suggests that a suitable home range must provide benefits outweighing the costs of monitoring and defending, mapping, and maintaining the area, expressed by the equation CD + C R < B, where CD represents daily energetic costs, and CR

represents the costs of monitoring, maintaining, defending, and navigating the territory, and B represents overall benefits.

Within an acceptable home range, the animal selects habitat at another, more finely scaled level, concentrating its activities in areas that offer optimal foraging opportunities. These areas are often defined by physical structures in the environment that facilitate successful procurement of resources with minimal risk from predators and/or competitors. These physical components may include concealment structures—or, conversely, lack of refugia for prey species (Gotceitas and Colgan 1989), presence of resources that attracts prey, or proximity to a water source, or even patterns of sunlight and/or wind. Thus, it is necessary to view habitat choice as a hierarchical process, in which multiple scales and multiple collinear factors may influence spatial activity patterns and foraging choices (McLoughlin et al. 2004).

Body size is another factor that may contribute to habitat selection and preference. Smaller species require less energy from both a gross energy standpoint—because mammalian metabolisms scale with body size, smaller species require less energy per unit of body mass

(Glazier 2010). Brown and Maurer (1989) suggested that smaller animals, due to their diminutive size, must be more selective in their prey and habitat choices, because they cannot travel as far or handle as large a variety of prey as their larger counterparts. Also, smaller species are likely to rank lower in the intraguild dominance hierarchy. Thus, to avoid interference competition, this subordinate ranking requires either more efficient resource procurement and/or specialization (Rosenzweig and Abramsky 1986) or more plastic foraging behaviors that may serve to expand niche breadth and reduce the need for competition (Linnell et al. 2000). For example, Di Bitetti et al. (2009) studied two species of sympatric South American foxes, and found that the smaller of the two displayed more behavioral plasticity—in terms of spatial and temporal activity—suggesting that these methods of avoidance by a smaller, solitary species served to decrease the risk of interspecific competition with a larger, pair-foraging congener.

Some studies have come to differing conclusions, however. For example, Durant et al. (2010) conducted a study of 26 African carnivores ranging in size from 0.35-161.5 kg, and found that neither habitat selectivity nor overall niche breadth decreased with body mass.

Habitat Terminology

Given these multiple scales of interest—geographic range, home range, territory, foraging area, etc—what specifically do researchers refer to when they discuss “habitat?” Hall et al.

(1997) takes issue with the inconsistent usage of the term “habitat” in wildlife ecology, and presents an extensive meta-analysis that supports his claim, which asserts that a variety of definitions are often applied to the same word, both in theory and in practice. It is pointed out that “habitat” is often confused with “habitat type,” which was defined as referring to the vegetational community and structure at a given location. In contrast, the term habitat “relates to

“the presence of a species, population, or individual (animal or plant) to an area’s physical and biological characteristics…more than vegetation or vegetation structure, it is the sum of the specific resources that are needed by organisms” Daubenmire (1968) [cited in Hall 1997].

In addition, attention must be paid to the terminology employed to describe an organism’s patterns of habitat usage, as pointed out by Basille (2008). Habitat selection implies that a species chooses to use a specific type of habitat from amongst other, equally available options. In contrast, habitat preference refers to “the disproportional use of some resources over

others,” a seemingly vague distinction that is cited as a “consequence of the process” of habitat

selection itself (Hall 1997). Basille (2008) breaks the definition down further, to contend

“preference is the source of motivation, selection is the process involved, and choice is the final

result.”

Based upon those semantics, the patterns described in this thesis refer to “habitat choice,” although “habitat selection” is sometimes used interchangeably here as in the broader literature.

The term “habitat preference” is avoided, since it is impossible to evaluate the motivations of remotely observed wild animals.

Margay Habitat Choice

The margay is often noted for its high degree of arboreality in comparison to other small wild felids, and it is in fact thought to be the most arboreal cat present in the Americas (Reid 1997).

This spatial specialization is evidenced by the fact that the margay has evolved a proportionally longer tail than its closest relatives, the ocelot ( L. pardalis ) and the Andean cat ( Leopardus jacobita ; De Oliviera 1998; Johnson 1998), presumably to aid with balance as it traverses the canopy. The margay’s limb morphology also facilitates deft climbing and balancing: its metatarsals are flexible and its hind foot can rotate 180˚, allowing it to descend trunks head-first in a squirrel-like fashion (Nowak 2005).

Possibly because of this predilection for arboreality, margays appear to prefer dense evergreen and deciduous forest, occurring nearly exclusively in these habitats. There has been disagreement in the literature as to the proportion of margays’ movements that are made in the canopy versus on the ground, and it is possible that these patterns vary between populations. [See

Chapter 1 for detailed information on previously reported patterns of margay habitat selection].

Goulart et al. (2009) suggests that margays may use terrestrial trails primarily to move between dense forest patches in search of prey, but off-trail habitat use was not evaluated.

Arboreal habits may serve as clues to a possible mechanism of niche differentiation employed by the margay. Studies have suggested that carnivore species may use vertical niche partitioning to gain access to arboreal prey, allowing dietary specialization, and possibly to escape

interference competition on the ground (Glen and Dickman 2008; Ray and Sunquist 2001, Jones and Barmuta 2000)

Rocha-Mendes and Bianconi (2009) point to research which has suggested that margays are more sensitive to habitat disturbance than their sympatric congeners, the ocelot and oncilla, potentially due to their greater utilization of available canopy for foraging and traversing their territory. They aren’t entirely absent in disturbed areas, however, and have been reported to exist in areas suffering from a high degree of forest fragmentation (De Oliviera 1998).

Conservation Implications of Habitat Choice

Knowledge of which habitat characteristics are preferred by a species of concern is of critical importance to effectively understand the ecology and natural history of that species, as well as to strategically protect and conserve its populations. While multiple factors contribute to population declines of margays, two of the most intense pressures thought to be affecting the species are habitat loss and forest fragmentation (Payan et al. 2010).

Of the world’s 25 recognized biodiversity hotspots, the Chocó-Darién-Western Ecuador and

Tropical Andes locations—both home to critical populations of margays representing an endemic subspecies, L. w. pirrensis —are two of the eight considered to be the most threatened by deforestation (Brooks et al. 2002). This ominous trend creates a critical situation: the forests in some of the richest regions of the world are rapidly disappearing and protected areas cannot be designed with maximum efficiency when there is a dearth of detailed knowledge regarding what habitat is preferred by species of concern.

Resources available to conservation efforts are often extremely limited, forcing conservationists to prioritize—and sometimes even triage (Bottrill et al. 2008, Hobbs and

Kristjanson 2003; McDonald-Madden et al. 2008)— the protection of specific habitat patches or

subpopulations that are thought to be high quality and important to the protection of either biodiversity or specific species of interest.

In order to make such choices, scientists and policy-makers require detailed data regarding the specific aspects of a habitat that are preferred by a species and that appear to foster healthy communities and biodiversity. Hence, the study of habitat choice is of critical concern, and is an example of how the study of the natural history of organisms and their communities must not fall by the wayside (Hafner 2007).

Using previous studies (De Azevado 1996; De Oliviera 1998; Goulart et al. 2009) as a guide,

I sought to assess margay habitat choice at the microhabitat level. I measured habitat variables instead of classifying habitat as primary or secondary forest to avoid arbitrarily assigning a location to a category based on human assumptions of what, for example, a primary forest

“should” look like. In addition, sampling was conducted both within interior forest and along forest edges, to assess habitat choice in the context of disturbed forests—as these situations have become more the rule than the exception in many landscapes. I also sought to assess whether margay habitat choice shifted according to the presence or absence of sympatric carnivores. I hypothesized that: (1) margay trap success would show a positive relationship to canopy cover, tree density, tree girth, and habitat type—which, at Sumaco, relates to distance from the nearest road, as this represents a gradient of disturbance from open to primary habitat, and (2) that margay trap success would be significantly greater within forests than along forest edge, and (3) that margay habitat choice will not significantly differ in the presence or absence of sympatric carnivores, due to their ability to use the canopy as a spatial refuge.

METHODS

Variables measured

Several key habitat variables were measured, all of which pertained to structural vegetation at

the camera station sites. Canopy cover was measured using a canopy cover grid. This tool

consisted of a tubular vertical sighting scope device with a grid overlaid over the ocular opening,

which was used to calculate the proportion of the visual field of canopy versus sky (Johansson

1985, Fiala et al. 2006). The distance to the nearest tree in each cardinal direction (north/0,

east/90, south 180, west 270) from the tree the camera was mounted upon was measured as an

index of tree density. Each of these trees was also measured for their diameter at breast height

(DBH) , which was averaged to obtain an index of habitat successional stage, under the

assumption that larger trees are indicative of a more mature forest. Distance from the camera to

the center of the nearest trail was also recorded. I hypothesized that margay trap success would

be positively correlated with canopy cover, tree dispersion (as measured by distance between

trees) and DBH, as these are all indicators of a dense and developed forest.

Analyses Conducted

Regression

The significance of specific habitat features on margay habitat selection was evaluated with a stepwise multiple regression analysis, using the following variables: canopy cover, tree dispersion, distance to nearest trail, and average tree diameter at breast height (DBH). The function was analyzed using SPSS, with each camera site being the experimental unit and margay trap success as the dependent variable.

Chi Square Analysis

Using the results of the regression as a guide, each camera site was assigned to one of four categories based on canopy cover: open (0-25% cover), light (26-50%), medium (51-75%), and dense (76-100%). The proportion of sites belonging to each category was calculated to determine its relative availability. Utilization of sites was calculated as the proportion of camera sites in each category that were actually visited (n = 89 capture events for 12 individuals), and a

Chi-square test was performed to detect deviations from random use.

Compositional Analysis

Habitat preference was assessed using compositional analysis (Aebischer et al. 1993). Each camera site was assigned to one of four categories described above, based upon canopy cover.

The proportion of sites belonging to each category was calculated to determine its relative availability. Utilized sites were calculated as the proportion of camera sites in each category that were actually visited. The available and utilized camera site types were transformed to log- ratios, using di = ln( Xui /X uj ) – ln( Xai /X aj ), where Xui is the proportion of habitat used by habitat category i and Xai is the proportion of habitat category i available.

To rank the usage of sites with differing canopy cover, a matrix was created comparing

use of each habitat type in relation to the other three. When di > 0 relative to habitat j, habitat i is

used more than would be expected if preference were random. The number of cells in each row

of the matrix with positive values was used to rank the habitat categories in order of increasing

proportionate use. In instances in which a category was not visited at all for a given individual,

the 0 value was replaced with a non-zero value (0.0001) to allow for log-ratio transformations, as

recommended in Aebischer et al. (1993).

Two additional matrices were constructed, one using log-ratios (as described above) for

margays known to be male (n = 4), and the other for known females (n = 5). The Aebischer’s t-

values for each cell in these matrices (the quotient of the mean/standard error of pooled values for each element) were compared using a paired Student’s t-test to ascertain whether there was a

significant difference in habitat selection between males versus females. The t-value in the

Aebischer method is unrelated to the t-value in the Student’s t-test.

RESULTS

Cameras mounted in the canopy captured no margay images; every photo series involved margays either walking along the ground or on fallen logs that were resting on the ground.

Regression

2 The stepwise regression model resulted in F38, 244 = 35.698, p < 0.001, Adjusted R = 0.133.

The results showed that canopy coverage ( p < 0.001) and distance between trees ( p = 0.002) were significant predictors of margay trap success (Table 3.1, Figure 3.1). Neither DBH nor distance from the trail were significant, so these variables were excluded from the model in the stepwise regression.

Compositional Analysis:

A total of 89 margay captures was recorded. The overall use of habitat based on canopy cover differed significantly from random (Table 3.2, 3.4). The matrix analysis showed ranked habitat use according to habitat type, with Medium> Dense > Open > Light. There was no significant difference between Aebischer’s t-values for each element between male and females margays ( t

= 0.0589, df = 22, SE = 0.70, p = 0.9535).

Chi Squared Analysis:

The available habitat and actual captures are shown in Table 3.4 and Figure 3.2. Use of habitat with different degrees of canopy cover differed proportionate to availability, suggesting

that margays prefer areas with different degrees of canopy cover (df = 3, X2 = 3958.793, p <

0.001).

DISCUSSION

Despite the margay’s reported preference for moving through the canopy, camera traps were

universally unsuccessful in detecting margays above ground level. This result is likely due to the

fact that margays and other tropical forest cats are known to use man-made paths and trails along

the ground (Di Bitetti et al. 2009; Emmons 1987; Rocha-Mendes and Bianconi 2009), while in

the canopy it is much more difficult to predict movement paths and aim cameras accordingly.

Absence of evidence is not evidence of absence, however; the lack of camera trap success for

margays with the tree cameras does not imply that they do not move through the canopy—only

that their detectability there is vanishingly low with only four canopy cameras deployed at a

time. Future studies focusing all of their resources on sampling in the canopy may be needed to

better elucidate the arboreal movement patterns of this species.

It is also possible that the margays at Sumaco spend proportionately less time in the canopy

than do margays in populations that have been studied in other localities. This behavior could be

due to a localized prey preference or foraging strategy that may be specific to this population of

margays, or even to their entire subspecies, although this is unlikely due to the broad geographic

range of L. w. pirrensis , which occurs only in a narrow swath of Central America from Panama

to northern Peru. The exclusive detection of margays movements along the ground could also be

due to reduced pressure to use canopy refugia, due to the absence of jaguars and extreme rarity

of ocelots and jaguarundis at Sumaco.

There could also be certain features of the local vegetation that cause less arboreal activity at

this site, such as tree species composition, characteristic horizontal structure within the canopy,

or percentage of understory cover. As at least one other study has documented significant margay movements along the ground (Konecny 1989), it seems that the proportion of time that margays spend in the canopy may be somewhat plastic and dependent upon the local habitat features and/or faunal community. Future research efforts should focus on elucidating arboreal behaviors at a finer scale.

Despite the lack of camera trap events of margays moving through the canopy, the data show that canopy cover and tree dispersion were significant predictors of margay trap success. Tree

DBH and distance to nearest trail were not significant predictors of margay trap success, suggesting that overall habitat density is more important for predicting margay presence than are the sizes of trees. It is possible that distance to trail would be a more important factor if sampling were conducted in a forest tract with more foot traffic than typically is present at WWS.

The significant difference between margay habitat selection and availability of each habitat type suggests that margays strongly prefer medium (50-75%) canopy cover. The reasons that these sites were preferred over dense cover are not clear. A potential causal factor is that boundaries between sites in different categories may not always be discrete. Interestingly, open habitat was preferred over lightly canopied sites. As most of this habitat was located along forest-pasture edges, it is possible that margays move through these sites en route to hunting rodents in adjacent cow pastures, a behavior that was documented by an edge transect camera during the present study.

Overall, the patterns recorded at Sumaco are consistent with previous studies, which have noted that the margay seems to prefer forested areas (De Oliviera 1998) and to be especially vulnerable to deforestation (Rocha-Mendes and Bianconi 2009). These results highlight the strong impact that habitat type and composition have on margay activity, and thus the

importance of curbing habitat destruction and deforestation as part of efforts to reverse the margay’s declining population trend and protect habitat conducive to restoring populations.

Table 2.1 Results of a stepwise multiple regression, using canopy, tree dispersion, DBH, distance to trail as potential predictive variables, camera site as the experimental unit, and margay trap success as the dependent variable. The resulting regression equation was: y = 0.019 x1 – 0.001 x2, - 0.346, where x1 = canopy and x2 = distance.

Variable SE(ß) t p Canopy 0.401 8.330 < 0.001 Tree Dispersion -0.153 -3.180 0.002 DBH -0.039 -0.819 0.413 Trail Distance -0.72 -1.511 0.132

Table 2.2 Ranking matrix for margay habitat preference based on proportional trap success at sites with different degrees of canopy cover (Open = 0-25% cover, Light = 26-50%, Medium = 51-75%, and Dense = 76-100%). Values are given as (mean ± standard error) obtained by averaging each matrix element over all 13 margays. The higher the rank score, the more often that category of canopy coverage was preferred to the others.

Canopy Open Light Medium Dense Rank Category 3.0329 ± -2.0685 ± Open -0.9067 ± 1.55088 1 1.55537 1.6352 -3.0849 ± -3.6439 ± Light -3.3109 ± 1.63567 0 1.55088 1.67527 4.7309 ± 1.1618 ± 1.98527 3 Medium 2.071 ± 1.63567 1.44302 -1.1618 ± Dense 0.1391 ± 1.4318 3.7268 ± 1.6397 2 1.98527

Table 2.3 Simplified preference matrix, with + indicating the row category was preferred to the column category and – indicating the row category was not preferred to the column category.

Open Light Medium Dense Rank Open + - - 1 Light - - - 0 3 Medium + + +

Dense + + - 2

Table 2.4 Results of Chi Square analysis, comparing proportion of habitat available (expected value) to proportion at which captures were made (observed value). Raw numbers were adjusted from the decimal fractions.

Canopy Observed Expected Category Open 0.124 0.041 Light 0.112 0.367 Medium 0.562 0.449 Dense 0.202 0.143 X2 = 3958.793 p < 0.001

Figure 2.1

P-plot resulting from stepwise multiple regression analysis .

Figure 2.2 Comparison of the availability versus utilization of camera stations with differing degrees of canopy cover. Open = 0-25% cover, Light = 26-50%, Medium = 51-75%, and Dense = 76-100%.

CHAPTER 3: INTRAGUILD NICHE PARTITIONING

INTRODUCTION

The apex predators in many ecological communities are often mammalian carnivores and these species play critical roles in structuring the population dynamics and diversity of many organisms within their habitat (Borer et al. 2005). In addition, many habitats are home to multiple predator species, and the interactions between these sympatric carnivores create a fine balance that may be easily disrupted due to changes in the population dynamics of even a single member of the local carnivore guild (Caro and Stoner 2003; Crooks and Soule 1999). Issues of intraguild competition seem to be especially relevant to the conservation of felids. These intraguild interactions can take the form of either fatal “interference competition” or indirect

“exploitation competition”—competition for resources between guild members (Case and Gilpin

1974, Polis et al. 1989).

Palomares and Caro (1999) report 97 different pair-wise predation interactions between carnivores, involving 54 victim and 27 perpetrating species, and the family Felidae was the only one in which more species were killers than victims in interspecific confrontations. Other studies of felids have suggested that smaller felids may change their dietary and spatial preferences at a given locality if there is competitive pressure from larger cats (Hass 2009 , Durant et al. 2010 ,

Moreno et al. 2006).

Lucherini et al. (2009) predict that smaller cats should avoid pumas in Andean habitats, yet suggest that, conversely, diet segregation between small and larger felids may allow tolerance for spatial overlap. Niche differentiation and resource partitioning strategies are

critical in areas of high biodiversity as they allow multiple members of a guild to co-exist without overwhelming pressure from interspecific competition. Studies have shown that the most common forms of niche partitioning are spatial (Palomares et al. 1996), temporal (Di Bitetti et al.

2009), and dietary (Jones and Barmuta 2000; Karanth and Sunquist 2000). Species sometimes mix and match partitioning strategies so that different potential methods of niche differentiation are utilized in a complementary fashion. For example, Jones and Barmuta (2000) found that where diet overlapped significantly between dasyurid carnivores, either habitat preferences or spatial activity patterns at the microhabitat scale—degree of arboreality, in that case—were likely to differ.

The majority of the world’s felid species currently face dire circumstances. The IUCN has declared 25 out of 36 (nearly 70%) of felid species to be of conservation concern, and five of the six species currently classified as Least Concern have decreasing population trends (IUCN

2011), suggesting they may also be of conservation concern in the near future. Felids are often hypercarnivorous (Van Valkeburgh 1991; Kok and Nel 2004), and are in the dualistic position of being essential regulators of smaller vertebrates while also being critically dependent on the persistence of those prey populations and vulnerable to population imbalances at lower trophic levels. Therefore, the conservation of even the most diminutive cats is of crucial importance for the maintenance of healthy, functioning ecological communities. Intraguild competition, combined with the role of carnivores in top-down regulation of biodiversity and ecosystem function, is an essential issue that must be better understood to maximize the effectiveness of efforts to both conserve biodiversity and maintain community function.

WWS has been shown to have a remarkably high diversity of mammals within a relatively small area (Hodge et al. 2010). Over one third of Ecuador’s terrestrial carnivore species have

been documented at the site, in an area of only approximately 5 km 2. Of the ten species of mammalian carnivores that have been documented at Sumaco, four are felids. This high concentration of carnivore diversity, and the high proportion of felids that have been shown in other studies to participate in competitive intraguild interactions suggests that these interactions are likely a selective pressure at this site. This competition should concomitantly affect the activity patterns and overall densities of the local carnivore populations, both felid and otherwise. An understanding of the mechanisms being used by Sumaco’s carnivores to partition niche space—and thus co-exist—is critical for the future of conservation efforts and our understanding of guild dynamics in mid-elevation zones, in which it is possible to have unique assemblages representing confluences of upland and lowland species (Shepherd and Kelt 1999).

In this study, I sought to investigate the evidence for competition and niche partitioning between sympatric carnivores by testing for significant differences in camera trap success and spatial and temporal activity patterns between species, in addition to determining whether spatial overlap was correlated to differences in body size. I hypothesized that for the three major niche dimensions included in this study—body size, spatial occurrence patterns, and temporal activity patterns—complementary differentiation would be evinced by the data. This means that if two species overlapped in one or two dimensions, they would be significantly different in the third, and vice-versa. I hypothesize that size-structured partitioning allows a higher degree of spatial overlap than might be expected by chance, but that in that case significant differentiation will then have to occur within the temporal dimension as well. This temporal dimension is expected due to “ecology of fear” effects that have previously been documented in communities that are populated by multiple predator species (Laundré et al. 2009). I also predict significant

differences between temporal activity patterns between carnivores of different families, due to phylogenetic constraints on behavior.

METHODS

Trap Success

Trap success (TS) was calculated as the number of capture events per trap night for each transect. The number of trap nights is equal to the number of 24 hour periods sampled multiplied by the number of cameras on that transect. Capture events of individuals of the same species that occurred within 1 hour of one another were logged as a single event, in order to ensure each capture was independent. TS was calculated for each carnivore species (margay, puma, jaguarundi, ocelot, grisón, tayra), and pair-wise correlation analyses were conducted to identify any significant relationships between TS within the carnivore guild.

Other vertebrates were grouped into several categories based on body mass as reported in the literature (Tirira 2007): small prey mammals (< 5kg), medium prey mammals (5-15 kg), large prey mammals (> 5 kg), small birds (<500 g), large birds, (> 500 g) and small carnivores

(kinkajous, coatis, and weasels), and TS was calculated for these groups as well. A correlation matrix was assembled to compare correlations between carnivore species and the other vertebrate groups.

Latency to Initial Detection

The latency to initial detection (LTD)—a metric describing the number of days subsequent to the start of the sampling session that a species is first detected, [as described in Chapter 2 above], was calculated for each species for each of the four sampling transects: F.A.C.E. Trail,

Benavides Trail, Waterfall Trail, and Puffbird Trail. The average LTD across all transects was also calculated for each species. Species that were not captured on a given transect are excluded

from the histogram for that transect. To identify any potential relationship between overall trap success and average LTD, a Spearman’s correlation test comparing these variables was conducted using SPSS.

Sørenson’s Index

Sørensen’s index values were calculated to assess pair-wise patterns of spatial co-occurrence

2a between carnivores. This index of similarity is calculated using the formula 2a+b+c (Pielou

1984). In this equation, a represents the number of camera stations that photographed both species, b represents the number with only one member of the species pair present, and c is the

number of stations that captured the second member of the species pair but not the first. Due to

the absence of carnivores larger than the tayra and margay on 3 of the 4 transects, Sørensen’s

index values were obtained from the F.A.C.E. Trail transect only. To evaluate the potential for

body size to influence niche partitioning, a correlation function was conducted to compare

Sørensen’s Index of Overlap with the difference in body mass (mass of Species A – mass of

Species B) between each pair of sympatric carnivores.

Temporal Activity Patterns

Each photo was automatically labeled by the inbuilt Reconyx camera software, providing a

record of the date, time, and moon phase during which the capture event took place. Each 24

hour period was divided into four time categories, following Lucherini et al. (2009) in

designating the periods: dawn (1 hour before to 1 hour after sunrise), day, dusk (1 hour before to

1 hour after sunset), and night. I then used Chi-square tests to evaluate the null hypothesis that

each species would have an equal proportion of captures in each time category. In addition, I

used ANOVA to compare temporal activity patterns across species and identify significant

interspecific temporal partitioning patterns.

RESULTS

Trap Success

Trap success (TS) was highest for the margay, which had an overall TS of 0.025. Table 4.1 shows trap success, LTD, and percent of transects of occurrence for each species on each sampling transect. The margay was found on each transect (the F.A.C.E., Benavides, Waterfall, and Puffbird Trails). The puma, ocelot, jaguarundi, and grisón were found only on F.A.C.E.

Trail, while the tayra was found on the F.A.C.E., Waterfall, and Puffbird trails.

The use of bait significantly decreased margay TS on the F.A.C.E. transect (p = 0.0402) and did not affect margay TS on the Waterfall trail (p = 0.1330). Bait had no significant effect on larger carnivore TS on either trail (total p = 0.20).

Latency to Initial Detection

The LTD for carnivores across all four transects at Sumaco ranged from 1 to 74 days (Table

4.1). More detailed information on LTD for each individual transect is shown in Table 4.2 and

Figure 4.3 A-D. The puma and jaguarundi, which were only detected on one transect, had LTD values of 33 and 74, respectively. The margay LTD ranged from 1-9 days across transects. The margay had the shortest latency to detection on each transect on which it was captured, and the shortest average LTD across the study. Excluding the margay, the coati had the shortest average

LTD (5 days), while the jaguarundi had the longest (74 days). It should be noted that the coati, jaguarundi and ocelot were each only captured on one transect, and thus the “average” LTD is actually just the value from one sampling session. For species besides the margay that were captured on more than one transect, the dog had the lowest average LTD (13 days) and the tayra had the longest (40 days).

There was no significant relationship between average LTD and trap success for each species if TS was calculated only for transects upon which a given species was detected (p =

0.862; rho = -0.082). If trap success was calculated for each species across the entire extent of the study, however, there was a significant negative relationship between LTD and trap success

(p = 0.009; rho = -0.878; Fig. 4.4).

Sørensen’s Index

The results for the Sørensen’s index of overlap for each species pair are shown in Table

4.3and Figure 4.1. The data indicate that the Sørensen’s value was highest for the margay-puma pair and that these two species had a much higher degree of overlap than did the other pairs of cats (Table 4.1, Fig. 4.1). The Sørensen’s value for intraspecific overlap of margays was 0.23, lower than that of many of the margay’s interspecific pairings. Pairings for which the index of overlap was zero are not shown in the table or histogram.

Comparison of Sørensen’s Index value and body mass difference for each species pair showed a significant correlation ( r = 0.757) between these two parameters, indicating that species with a larger difference in body mass are more likely to overlap spatially (Fig. 4.2).

Temporal Activity Patterns

Chi-square tests for each carnivore species failed to support the null hypothesis that there was no significant preference for specific temporal periods (Table 4.3). Instead, each species did show a strong preference for activity during certain times of day over others (p < 0.05). The dog, jaguarundi, grisón, coati, and kinkajou were primarily diurnal, while the puma, margay, ocelot, and weasel were primarily diurnal. The tayra showed equal trap success in both day and dusk, yet was never detected during the night or dawn hours. The results of the ANOVA comparing

temporal patterns between species were non-significant, indicating no differences in temporal activity patterns between species when the assemblage was considered as a whole.

DISCUSSION

Trap Success

The trap success for the margay, as discussed in Chapter 2, was intriguingly high relative to the results of previous studies reported in the literature (Kelly 2003; Dillon 2005; Goulart et al.

2009), perhaps from either unusual abundance or unusually high detectability of this species at

Sumaco. The high density of margays and very low density of ocelots and jaguarundis suggest that the composition of this assemblage may be a case of mesopredator release in which smaller carnivores increase in density and/or abundance in the absence of larger competitors (Roemer et al. 2009).

There was, however, a high degree of overlap between the margay and the apex predator, the puma, which may superficially suggest that mesopredator release may not be the driving force determining margay abundance at this site. Classification as “mesopredator” is relative to community composition, however, and it is possible that it is the ocelot, not the puma, that exerts the most competitive pressure upon the margay, due to their more similar body sizes and prey bases (Jones and Barmuta 2000). In that case, it is possible that the dearth of ocelots at Sumaco

(potentially an effect of puma presence) could be the “releasing” factor on the margay that allows a relatively high abundance of margays in the presence of a much larger felid, the puma.

Extrapolated onward, it is possible that the absence of jaguars is a factor that allows the pumas to occupy the site, thus controlling ocelots and allowing for the mesopredator release of the margay. For example, a study by Ripple et al. (2011) suggests that the encroachment of coyotes in response to the extirpation of wolves can cause declines in local lynx populations due

to exploitation competition as coyotes consume loads of snowshoe hares, the primary prey of the lynx. Essentially, removal of the apex predator, the wolf, actually harms that of a smaller predator, the lynx, because it releases competitive pressure on an intermediate carnivore, the coyote. Cases of such trophic cascades are fascinating systems of critical conservation concern, but longer, much more broadly-scaled studies are required to confirm the occurrence of such phenomena in a complex ecosystem.

It is also possible that margays at WWS are not actually any more dense or abundant than at other camera survey sites, yet the lack of medium-sized cats has allowed the diminutive margay to be more habitually terrestrial, thus increasing its detectability by cameras stationed low to the ground. This “ecology of fear” effect has been shown in other systems (Linnell and Strand 2000 ,

Creel et al. 2001; Creel and Christianson 2007 ), and Gompper and Vanak (2008) point out that significant dietary overlap is not necessarily a requirement for larger carnivores to prove a limiting factor on populations and activity patterns of smaller sympatric predators. This could explain why ocelots—which have some degree of arboreality but do not match margays in their specialization for locomotion through the canopy—might exert pressure while maintaining some degree of size and dietary differentiation.

The comparatively low trap success for larger felids is likely related to transect design and the small size of the study site. The camera station distribution was designed to target margays, and the distance between cameras was calculated by reviewing camera studies targeting larger species (Kelly 2003; Silver et al. 2004; Wang and Macdonald 2009) and scaling their reported camera spacing down so that it was proportionally relative to the average of reported home range sizes for the margay. As such, it is likely that the total area sampled in this study did not encompass more than one home range for the puma, which has been shown to use home ranges

from 90-350 km 2 (Grigione et al. 2002), and had little chance of overlapping multiple home ranges for the jaguarundi (home range = 20.47 km 2; Michalski et al. 2006) and ocelot (home

range = 2.1-31 km 2; Murray and Gardner 1997). As a result, the sample sizes for these species were not sufficient to generate robust population estimates, although records of the timing and location of the activities of these species are useful for inferring potential partitioning of spatial and temporal activity amongst sympatric predators.

Latency to Detection

Latency to detection varied widely between species. The most frequently detected species, the margay, had the lowest LTD value, and rarer predator species had significantly higher LTD times. There was a significant negative relationship between LTD and trap success—the faster a species was detected on a transect, the higher the number of total capture events occurred over the course of sampling. LTD may thus be useful as a predictor of relative trap success within some mammal assemblages. Caution should be taken before approaching LTD as an index for local population density, however. Not all species have the same detection probability, and detectability within a species may vary according to sex, age, or social ranking. Gompper et al.

(2006) showed that cameras missed weasels at 100% of the sites at which they had been detected using other methods, martens at 88% and raccoons at 55%. In contrast, the cameras were shown to detect foxes at 100% of the sites where their presence had been otherwise confirmed. Thus, the use of camera trap LTD seems to be a poor strategy for predicting presence and/or density for some species, while it is highly effective for others, highlighting the value of using taxonomically targeted and/or diverse sampling methodology in field studies.

It is possible that prior visits to a station by larger species could deter later visits by smaller species, resulting in a confounding aversive effect that would render LTD as a suboptimal tool in

predicting population density or overall trap success rate. This effect may even manifest within a species, if the presence of a dominant individual proves aversive to the detection of conspecifics within their core territories or foraging routes. In cases such as these, a short LTD may not necessarily predict high future trap success.

The opposite effect could also occur, however, with the presence of an individual subsequently attracting other conspecifics to the same site. Paull et al. (2011) showed that detection of any one of several small species of mammals by a camera was a significant predictor of a member of the same genus being detected by the same camera within 24 hours. This effect may be especially strong with species that routinely communicate via scent marking. For example, DeMatteo et al. (2004) determined that bush dogs were attracted to camera traps through the deposition of urine. In such cases, even a species with a long LTD could potentially rack up a high trap success after its initial “discovery” by the cameras, albeit with a more temporally concentrated set of captures. Although capturing behaviors such as scent-marking in remote photographs is rare, there was one photo series event within the data presented here that documented a male margay urine-marking approximately 2.5 meters directly in front of a camera station. Given those limitations, LTD remains an important metric to record, as it has high utility use in planning sampling effort required in future studies targeting one or more of the species of interest.

Sørensen’s Index

The margay had a relatively high degree of overlap with the puma on the transects where they co-occurred, and the highest Sørensen’s Index value for any pair of carnivore species in this data set. This spatial overlap is likely mitigated by their dietary differences, which is a function of

their vastly different body size—an average of 3.6 kg for the margay (De Oliviera 1998) and 48 kg for the puma (Nowell and Jackson 2006)

The ocelot and jaguarundi had low degrees of spatial overlap with the puma on a camera-by- camera basis, yet all three of these medium- to large-bodied cats were present only on the

F.A.C.E. transect; the ocelot and jaguarundi were absent from the three transects on which the puma was also absent. This suggests that their low abundance at this site may be due to other factors besides intraguild competition from the puma, and that their co-occurrence on the

F.A.C.E. trail may have been due to prey abundance or connectivity with the surrounding forests bordering WWS property. That the medium to large felids co-occurred on a transect with little overlap at individual camera stations suggests that partitioning and avoidance may occur at relatively fine scales within a given habitat.

These data also indicate that size-structured niche partitioning may be influencing guild structure at WWS. Werner and Gilliam (1984) asserted that body size “is manifestly one of the most important attributes of an organism from an ecological and evolutionary point of view,” and went on to point out that size is a major determinant of an animal’s resource and energy requirements and competitive ability in relation to “natural enemies.” Additionally, Jaksic et al.

(1981) showed that mean prey size taken by a predator was one of three important niche dimensions determining guild structure—with the other two being hunting activity period and hunting habitat. Overall, the literature suggests that predators of different sizes are likely to prefer different prey (Konecny 1989; Rosenweig 1966; Carbone et al. 1999), resulting in size- structured niche partitioning manifested through the use of different methods and routes for foraging by species of different body sizes. Species with the highest indices of spatial overlap were those with the largest differences in body size.

That pattern is consistent with the observations and results from the present study at Sumaco.

The margay and puma had a much higher degree of overlap than did any of the pairs of cats that were more similar in size. This trend was supported statistically: differences in body mass between two species significantly correlated to higher Sørensen’s Index values.

Size-structured partitioning could have multiple, non-mutually exclusive causes. First, preferred prey size often correlates to predator body mass (Carbone et al. 1999), and Vézina

(1985) found that the ratio of prey to predator weight actually increases with predator size.

Carbone et al. (1999) found that a transition from primary selection of small prey—45% or less of the predator’s mass—to that of large prey occurs when a predator surpasses a body mass of

21.5 kg. The same study found that all canids and felids analyzed with body masses above 21.5 kg were “purely vertebrate feeders,” while only half of smaller species depended entirely upon vertebrates as prey.

The only species present in the current analysis that exceeded 21.5 kg is the puma, and so there was often significant spatial and/or temporal overlap between this species and smaller sympatric carnivores. This pattern is consistent with size-structured niche partitioning, as the puma’s reliance on a larger prey base reduces the potential for exploitation competition.

There may also be some degree of taxonomic partitioning: the two mustelids had a Sørensen’s

Index of overlap value of 0, and the tayra overlapped only with the margay and puma, with an index value of 0.33 relative to each of them. Phylogenetic constraints on body size and diet selection have been shown to affect the structure of both food webs (Cattin et al. 2004) and trophic guilds (Dayan and Simberloff 1998; Van Valkenburgh 2007), and shared evolutionary history may be linked to morphological specializations that constrain performance in some

habitat types relative to others (Brown 1996; Jones and Barmuta 2000), which can also result in

niche differentiation.

It is noteworthy that margays had lower intra specific rather than inter specific spatial co- occurrence. This pattern potentially represents another level of niche partitioning, as members of the same species are likely to have the highest degree of niche overlap, and thus are one another’s strongest competitors for resources, especially in non-social species (Arlettaz et al.

1997; Polis 1984). Edwards et al. (1998) found that intraspecific was greater than interspecific overlap for several niche parameters for two species of sympatric squirrels, although the differences were not statistically significant for all of the parameters. It is also possible that obligate carnivores, such as felids, have less capacity for trophic polymorphisms that allow some organisms to reduce intraspecific competition (Dayan and Simberloff 2005).

Abrams (1980) argued that niche overlap should not be viewed as an index of intensity of competition, but that its more appropriate value is as a measure of the ratio of inter- to intraspecific competition. Similarly, the Lotka-Volterra model of co-existence dictates that coexistence will occur if interspecific is lower than intraspecific competition (MacArthur 1970;

Chessón and Kuang 2008). Thus, that there appears to be a high abundance of margays at

Sumaco may tip the scales in favor of coexistence with other species, since interspecific competition may be less of a selective pressure on local margays than competition with conspecifics.

In addition, Glen and Dickman (2008) point out that spatial overlap does not preclude the possibility of competition and/or avoidance. In their study, they found that although quolls and eutherian carnivores showed high degrees of spatial overlap, there was evidence for significant, finely-scaled spatial avoidance, with quolls preferring to forage in forested areas within a given

territory while eutherians foraged more in open areas. With the margay’s reported predilection for hunting in the canopy, it could be that a similar spatial-partitioning phenomenon is occurring at Sumaco, allowing high degrees of interspecific overlap without implying that this indicates a lack of interspecific competition.

Temporal Partitioning

It has been suggested that temporal activity differentiation can decrease the degree of exploitation competition between species with similar diets (Konecny et al. 1989), making this dimension a critical factor when analyzing mechanisms of carnivore co-existence. The data from

Sumaco show that each carnivore species had a significant preference for activity within specific periods of the 24-hour cycle. The jaguarundi, tayra, grisón, and dog were primarily diurnal, while the puma, margay, and ocelot were primarily nocturnal. Once again, the margay and puma, which have the largest difference in body size of all of the possible pairs of felids, have the most similar patterns within this niche dimension. The margay is closest in body mass to the jaguarundi and there was no overlap at all in their temporal patterns.

Phylogenetic constraints may also carry significance in determining a species’ position within a temporal niche dimension. Out of the four felid species documented at Sumaco, all but the jaguarundi were strongly nocturnal. Both of the procyonids, the coati and kinkajou, were exclusively nocturnal. Amongst the three mustelid species, the weasel and grisón were captured only during the day, while tayra trap success was equivalent between day and dusk with no detections during the night or dawn hours. Thus, it appears—albeit qualitatively—that this niche dimension may not be independent of phylogeny.

It should be noted that the sample sizes for the ocelot and jaguarundi were very small, and so conclusions drawn from data regarding those species should be considered preliminary

estimations. The activity patterns displayed in the present data, however, are consistent with those documented in the literature. The patterns match those found by Konecny (1989), who surveyed temporal habits of the margay, ocelot, tayra and jaguarundi in Belize and found the margay and ocelot to be nocturnal and the tayra and jaguarundi to be diurnal, as they also appear to be at Sumaco.

The only species for which the patterns detected here were ambiguous was the grisón, which was photographed on a single occasion during the day. Thus, in the data analysis it appeared to be entirely diurnal. During pilot testing at this study site, all detections of this species were during the night. Other literature, however, has characterized grisóns as being diurnal as well

(Grigione and Mrykalo 2004). Little work has been done on the natural history of these elusive mustelids, and further investigations of its temporal patterns are needed.

Although interspecific analysis did not show significant differences in temporal activity preferences, that may be because, of the 10 species observed, four were strongly diurnal and five were strongly nocturnal, with one, the tayra, showing equal preference for day and dusk. It could be that such an equivalent set of species in each category resulted in overall statistical insignificance.

Conclusions

The present data strongly suggest that complementary niche partitioning is a mechanism by which a high degree of carnivore diversity is maintained at Sumaco. Although all species did not differ in every niche dimension analyzed here—body size (a proxy for preferred prey size), spatial activity, and temporal activity—it did appear that species that overlapped in one dimension showed significant differentiation in another. For example, the margay showed a nearly identical pattern of temporal activity to the puma, and the margay and puma had the

highest degree of spatial overlap out of all of the pair-wise species comparisons, yet they also have the widest difference in body mass of any of the felids that occur at Sumaco. In contrast, margay and jaguarundi, which are closest in body mass of all the felids, had completely different temporal patterns and little spatial overlap.

The trend of complementary niche differentiation patterns was apparent across families as well. The margay and tayra, and jaguarundi and tayra, two species pairs that are also very similar in body size, showed significantly different temporal patterns and little spatial overlap. In addition, previous studies have shown the tayra to be fairly omnivorous (Konecny 1989), which may allow for broader dietary partitioning despite its being close in body size to multiple sympatric predators.

These conclusions extend only to ecological mechanisms for niche differentiation, yet it should be noted that evolutionary mechanisms of differentiation and phylogenetic constraints on niche breadth are also factors determining co-existence between predators (Van Valen 1965; Van

Valkenburgh 1985, 1991). For example, felids often have narrower dietary niche breadths than sympatric carnivores of similar sizes due to their reduced dentition and lack of crushing surface for chewing (Kok and Nel 2004; Kruuk 1986).

The present study did not investigate prey selection or specific foraging strategies, which can be significantly influenced by evolutionary and phylogenetic factors such as skull and/or dental morphology and locomotor function (Ewer 1968, in Kok and Nel 2004). To obtain the fullest understanding of niche dynamics within a carnivore guild, it is necessary to examine such phenomena as well, and ideally such studies will be used in synergism with ecological research to obtain detailed and multidimensional knowledge about the complex interactions that occur to facilitate species diversity within a guild.

Table 3.1

Capture success for mammalian carnivores at WWS.

% of Species Transects LTD Trap Success Detected Dog 0.5 13 0.01435 Margay 1 4 0.25489 Puma 0.25 33 0.001732 Jaguarundi 0.25 74 0.00025 Ocelot 0.25 66 0.00025 Grisón 0.25 40 0.00025 Tayra 0.75 40 0.001732

Table 3.2

Latency to detection for carnivores at WWS.

Species F.A.C.E. Benavides Waterfall Puffbird Average Margay 1 5 1 9 4 Puma 33 - - - 33 Jaguarundi 74 - - - 74 Ocelot 66 - - - 66 Tayra 38 - 9 72 40 Grisón 40 - - - 40 Dog 21 5 - - 13 Coati 0 5 - - 5 Kinkajou 0 41 - - 41

Table 3.3

Pairwise comparison of spatial overlap between mammalian carnivores at WWS.

Species pair Sørenson's Index

Margay-Puma 0.6

Margay-Tayra 0.33

Margay-Dog 0.57

Margay-Ocelot 0.2

Margay-Jaguarundi 0.2

Margay-Grison 0.2

Puma-Tayra 0.33

Margay-Margay 0.23

Table 3.4 Results of Chi-Square analysis of temporal activity patterns of sympatric carnivores to determine whether the distribution of recorded activity differed from expected if no time category were preferred by each species.

Dusk Dawn (500- Day (700- Night Species (1730- Chi Square P Value 700) 1730) (1930-500) 1930) Dog 0 0.68 0.16 0.16 105.44 <<<0.001 Margay 0.08 0 0.08 0.84 154.94 <<<0.001 Puma 0.11 0 0.22 0.67 103.76 <<<0.001 Ocelot 0 0 0 1 300 <<<0.001 Jaguarundi 0 1 0 0 300 <<<0.001 Tayra 0 0.5 0.5 0 100 <<<0.001 Grisón 0 1 0 0 300 <<<0.001

Figure 3.1

Sørensen’s index of overlap for sympatric species pairs. Pairs for which the index value was 0 are not shown.

Figure 3.2 Correlation between Sørensen’s Index of Overlap and difference in body mass between pairs of sympatric carnivores. The dotted line represents the trend only; no regression analysis was conducted.

Correlation Coefficient: 0.757

Sorensen’s Index Value

Figure 3.3 Latency to detection of carnivore species on each of 4 transects.

Figure 3.4 Spearman’s correlation comparison of LTD and total trap success for each carnivore species. LTD = Latency to Detection; Total_TS = Trap success.

Figure 3.5 Temporal activity patterns of sympatric carnivores, as calculated by dividing each 24-hour day cycle into 4 categories: dawn (1 hour before to 1 hour after sunrise), day, dusk (1 hour before to 1 hour after sunset), and night.

CHAPTER 4: EFFECT OF DOMESTIC CANINES ON WILD CARNIVORE BEHAVIOR

INTRODUCTION

Challenges related to the presence and activity of domestic dogs are critical issues for wildlife popuations, and have specific implications for carnivore conservation efforts. Dogs have been shown to act as vectors for disease (Cleaveland et al. 2000; Fiorello et al. 2004; Roelke-Parker et al. 1996 ; Sillero-Zubiri et al. 1996), in addition to serving as both predators to small vertebrates and prey items for larger carnivores.

Human subsidization allows dog populations to persist and venture into habitats that might not naturally support them and the opportunistic consumption of supplementary food items by domestic carnivores can tax a system that is already at or near its carrying capacity for mammalian carnivores (Butler et al. 2004). Domestic dogs in rural communities are often provisioned with food from humans and their movement patterns are rarely restricted (Butler and

Bingham 2000). Because humans artificially sustain them above carrying capacity, dogs are unlikely to respond to dropping prey abundance as a wild animal might, and their continued density and predation may disrupt the natural predator-prey dynamics for response to changing resource availability.

Despite their domesticated status, dogs almost inevitably participate in the intraguild interactions between members of the local carnivore assemblage. This can occur either through indirect exploitation competition (consumption of resources also used by wild carnivores) or by interference competition , which can range from confrontational displays to killing and consumption of other carnivores (Palomares and Caro 1999). Several studies have shown that

foxes alter their utilization of available habitat within a landscape based upon the presence of domestic dogs (Mitchell and Banks 2005; Silva-Rodríguez et al. 2011; Vanak and Gompper

2010), suggesting that exploitation competition does indeed occur between domestic and wild canids.

In another study, Reed and Merenlander (2011) sampled reference areas as well as recreation areas that (1) excluded dogs; (2) allowed leashed dogs, or (3) allowed dogs to roam off leash.

The data showed significantly more carnivore scats in reference areas than in designated recreation areas, which could be attributable to the presence of domestic dogs accompanying human visitors. There was no difference in carnivore scat abundance between recreation area types, however. This could indicate that (1) human activity is the main factor affecting carnivore activity, regardless of dogs or that (2) dog exclusion is not well-enforced, leading to domestic dog activity even in recreational areas in which dogs are officially excluded. This case illustrates the inextricable link between the presence of dogs and human activity; it is often difficult to disentangle the two as causal factors.

A study of domestic dog-wild carnivore dynamics in the Swenga Wildlife Research Area,

Zimbabwe, found that dogs penetrated up to 6 km into the protected area (Butler et al. 2004).

This study also found dogs to be the most common carnivore sighted along the border of the research area, and data showed that dogs increased their incursions into the research area in the dry season—which is also season with the highest disease prevalence.

The ominous link between peak park incursion during a time of peak disease outbreaks could have dire implications for disease ecology and the health of wildlife populations—including jackals, wild dogs, hyenas, lions, and leopards—that are vulnerable to canine distemper. Recent studies have shown that climatic extremes—which are predicted to occur with increased

frequency due to mounting shifts in global climate patterns—can lead to devastating rates of disease and co-infections of different pathogens amongst sympatric species (Craft et al. 2008;

Munson et al. 2008). Domestic dogs were the source of a rabies outbreak in Ethiopia’s Bale park in the early 1990’s that wiped out 77% of the park’s adult and sub-adult Ethiopian wolves

(Sillero-Zubiri et al. 1996)—a devastating blow to one of the world’s most endangered carnivore species. Likewise, a 1994 outbreak of a morbillivirus—a neurological disease similar to canine distemper—in the Serengeti of Tanzania swept through the wild carnivore population, and within a short time it had had affected 85% of the lion population in that area and had also spread to lions in Kenya, as well as to other local carnivore populations (Roelke-Parker et al. 1996).

Although the majority of dog-wild carnivore conflict studies have been conducted in Africa, the phenomenon is not limited to that continent. It has been suggested that the introduction of dingoes to Australia in prehistoric times was a factor in the elimination of the Tasmanian wolf

(Thylacinus cynocephalus ) and Tasmanian devil ( Sarcophilus harrisii ) from the mainland

(Corbett 1995). In addition, domestic dogs in India deplete the supply of blackbuck ( Antelope cervicapra ) fawns, to the detriment of the threatened Indian wolf ( C. lupus pallipes ; Jhala and

Giles 1991). A study in Brazil’s Amazonian basin showed infection of crab-eating foxes

(Cerdocyon thous ) with canine leishmaniasis, was most likely contracted from sympatric domestic dogs. The dogs showed high serological prevalence of the parasite (Courtenay et al.

1994), and yet a follow-up study showed that crab-eating foxes are unable to maintain the leishmaniasis transmission cycle independently of domestic dogs, despite the fact that 78% were seropositive for the parasite (Courtenay et al. 2002). That disturbing revelation is a testament to just what a pervasive and penetrating impact dogs can have on the health of a wild carnivore population.

Interference competition between dogs and wild carnivores has also been documented.

Campos et al. (2007) found both coati and grisón remains in domestic dog scat, and other studies have found red fox remains in scats from domestic dogs. Butler et al. (2004) also documented dog predation in Zimbabwe upon wild ungulates, primates, and rodents, in addition to domestic animals such as goats and sheep. This highlights two issues of concern.

First, dogs could be competing with carnivores for food, or, as they displayed high chase rates but relatively low capture rates, they could affect the “ecology of fear” dynamics (Laundré et al.

2009) between wild predators and their natural prey, resulting in simultaneous interference and exploitation competition. Secondly, the killing of domestic livestock by dogs could be erroneously or disingenuously attributed to wild animals, fueling desire for retribution in the form of poaching and creating challenges for attempts to attain full cooperation from local people in the context of wild carnivore conservation efforts.

Domestic dogs may also play the role of prey within a guild: studies have documented attacks on and/or consumption of dogs by a range of large felids (Ralls and White 1995; Farrell et al.

2000; Kojola et al. 2004; Vanak 2008). Butler et al. (2004) attributed reported dog kills to three large carnivores within the Swenga Wildlife Research Area with over 90% of the kills attributed to felids.

METHODS

Camera trapping and baiting procedures were conducted as described in Chapter 2 above.

Effect of bait

Dogs were absent on both the Waterfall and Puffbird Trails: those trails served as natural dog- free control transects. Trap success was calculated for both dogs and margays on the F.A.C.E. and Benavides trails. Linear regression analysis was conducted to test for differences between

margay TS in the presence and absence of dogs on the F.A.C.E. and Benavides Trails. In addition, paired Student’s t-tests were used to compare dog TS with bait to dog TS without bait; the same analysis was conducted on margay TS with and without bait. Wild carnivores larger than the margay (puma, ocelot, jaguarundi, tayra) were lumped into a single group due to low sample sizes for individual species, and carnivore TS with and without bait was also calculated.

Distance from Road

The distance from the road to the location of each camera at which dogs and/or margays were captured was calculated using Google Earth software. These road-to-camera distances were averaged for each of the two species under two conditions: (1) using all cameras at which the species was captured on that transect and (2) using only the cameras at which a given species was captured to the exclusion of the other. A Student’s t-test was conducted to compare the mean distances for captures of dogs and margays under each of these four conditions (all

F.A.C.E captures, exclusive F.A.C.E. captures, all Benavides captures, and exclusive Benavides captures).

RESULTS

Effect of bait

Weekly captures of dogs and margays and the results of the Student’s t-tests comparing baited and non-baited treatments are shown in Table 5.1. On the F.A.C.E. Trail, for which dogs were present and bait was tested, margay TS showed a significant decline during the portion of the sampling session during which bait was introduced (p = 0.0402). On the dog-free transect on which bait was tested (Waterfall Trail), there was no significant difference between margay TS during the baited and unbaited portions of the sampling session (p = 0.1330). In the analysis of the effects of bait on dogs, bait had a significantly positive effect on dog TS (p = 0.0304) on the

F.A.C.E Trail. Bait had no effect on trap success for large carnivores as a group (p = 0.20; Table

5.2).

Distance from Road

Average distances to camera capture stations and the results of the Student’s t-tests for each of the four conditions are shown in Table 5.3 and Figure 5.1. On the F.A.C.E. Trail, margays and dogs were each captured at seven different camera stations. Five stations obtained captures of both dogs and margays, and margays and dogs were found exclusive of one another at two stations each. The average distance from the road of all cameras capturing dogs on F.A.C.E.

Trail was 380.980 m (SD = 232.097), while for margays on the F.A.C.E. the average was

549.011 m (SD = 174.407), a statistically significant difference (t = -3.001, p = 0.00213).

Cameras 22 and 24 captured dogs but not margays, accounting for 16% of dog captures on the

F.A.C.E. Trail; Cameras 11 and 16 captured margays but not dogs, accounting for 22.5% of margay captures on this transect. There was a statistically significant difference in the distance

(in meters) from the trailhead between cameras that captured only dogs (M = 104.63, SD =

43.5265) while those capturing only margays were farther from the road and entrance to the trail system (M = 647.623 m, SD = 67.25568); t = -16. 226, p < 0.001.

Dogs were also present on the Benavides Trail. No bait was used during this sampling session. Dogs were captured at six different camera stations; margays were captured at seven, and four stations captured both dogs and margays. Cameras 12 and 16 captured dogs exclusive of margays (accounting for 57.58% of all Benavides dog captures), and Cameras 22, 26, and 27 captured margays but not dog (accounting for 37.5% of all Benavides margay captures). The proportions of exclusive versus joint capture locations for each species are shown in Figure 5.2.

There was no significant difference in the average distance from road of all dog capturing

stations (M = 258.586, SD = 177.162), and the average distance to road from all margay capturing stations (M = 304.5, SD = 153.940); t = -0.0893, p = 0.190.

There was, however, a significant difference between the average distance from road of cameras that captured dogs but not margays (M = 155.758, SD = 102.077), and for stations capturing margays but not dogs (M = 288.607, SD = 23.018); t = -5.193, p < 0.001.

DISCUSSION

The results show that dogs have the potential to affect margay activity patterns in some circumstances: periods of high dog activity—correlated to the introduction of bait—resulted in significant declines in margay trap success. It is possible that this situation is analogous to an indirect interaction, in which the presence of one prey species has a negative effect on another species. It was not due to direct competition between the two prey, but because the presence of one boosts the presence of a shared predator. In this case, however, one of the “prey” species boosting the presence of the dogs would be the fatty acid bait, which significantly increased dog trap success on the transect.

Over the entire study, there was no significant relationship between margay and dog TS. It is possible that considering the data as a whole, or even trap success per transect, is too general. It is also possible that the absence of dogs from two of the four transects skewed these statistics.

That cameras capturing margays were significantly farther from the road than those capturing dogs is noteworthy: forest interiors may serve as a refuge for margays so that their overall abundances and activity level for a given transect were not negatively affected by domestic dog activity. It is also possible that there is a threshold activity level at which margays begin to avoid a site due to dog presence, and that the use of bait artificially created such a situation over the course of several weeks.

Although domestic dogs were demonstrably attracted to the fatty acid bait, it is interesting that they still showed a stronger tendency for activity in parts of the forest relatively close to the road, despite that a trail of fatty acid tablets led to the farthest reaches of the transect. There was a significant amount of puma activity on the F.A.C.E. trail, which could be one factor in the rarity of dog captures beyond 500 m from the beginning of the transect. Dogs also showed a tendency not to penetrate deep into the forest on the Benavides Trail, however, despite the lack of pumas, ocelots, and jaguarundis on that transect, suggesting that predation risk—if it is indeed a factor—is only one of several.

Dogs were more often captured at locations without margays on the Benavides Trail than the

F.A.C.E. Trail (Fig. 5.2). This pattern could be due to multiple factors: while the F.A.C.E. Trail is 1.6 km down the road from the WWS Lodge, the Benavides Trail begins directly behind the lodge’s staff quarters. Thus, the trail is more accessible to dogs that live and forage around the staff quarters, and where higher levels of human activity might be aversive to margays. This trail also had a lower abundance of margays overall relative to the F.A.C.E. Trail. It is also possible that the heat cycle of the resident kitchen canine was a significant attractant to courting dogs, which would have passed through the periphery of the forest surrounding Benavides Trail to pay their visits. Despite those factors, the trend for dog TS to drop with distance from the road was consistent significant on both transects (Table 5.2).

These results provide support for the theory behind “edge effects,” which predict that habitat disturbance does not stop at the tree line of a habitat patch, but can penetrate a certain distance within the patch, effectively shrinking the area of undisturbed habitat (Broadbent et al. 2008;

Harper et al. 2005; Murcia 1995). In this case, the invasive disturbance was the presence of dogs, and despite the presence of an attractant placed along an established trail, they showed a

significant pattern of limited—yet significant—penetration of the patches of forest that were sampled in this study. This information is of crucial importance, because conservation resources are almost inevitably finite and decisions must often be made between protecting several small patches of forest (Fischer and Lindenmayer 2002) or fewer large patches. Furthermore, the benefits to patch size structure across a landscape often depend upon the scale at which biodiversity is measured. Stefan (1999) showed that although small patches may have high alpha diversity and similar resource richness to larger patches, this approach to habitat preservation can still result in reduced beta and gamma diversity on a landscape scale. More research is needed to determine causal mechanisms in the preference of domestic dogs for areas closer to the road— despite the presence of a well-maintained trail system at WWS—and such information will be critical for conservation planning.

Table 4.1

Weekly number of captures for margays and dogs under non-baited and baited treatments on F.A.C.E. Trail. There were no dogs present on the Waterfall Trail.

F.A.C.E. TRAIL WATERFALL TRAIL Margay Non- Margay Margay Non- Margay Bait Bait Dog Non-Bait Dog Bait Bait Bait 3 1 0 8 4 2 4 3 0 8 4 2 8 2 0 4 4 5 2 0 2 1 3 1 3 2 1 2 10 3 T-test results: p = 0.0402 p = 0.0304 p = 0.1330

Table 4.2 Weekly captures for larger carnivores (puma, ocelot, jaguarundi, tayra) under non-baited and baited treatments on the F.A.C.E. Trail.

Non-Bait Bait 0 4 0 1 1 0 1 1 1 3 0 2 T-test results p = 0.2042

Table 4.3 Comparison of mean distance (in meters) from road to camera stations at which dogs and margays were captured.*

F.A.C.E Benavides F.A.C.E All Benavides Exclusive Exclusive Captures All Captures Captures Captures Dog Mean 380.98 104.63 258.856 155.758 Dog SD 232.097 43.5265 177.162 102.077 Margay Mean 549.011 647.623 304.5 288.607 Margay SD 174.407 67.25568 165.94 23.018 T-stat § -3.001 -16.226 -0.0983 -5.193 P-value 0.00213* <0.001* 0.19 <0.001*

*Comparisons were calculated using a Student’s t-test to evaluate the difference between means of two samples.

Figure 4.1 Average distance (in meters) from road to capture locations for dogs and margays on the F.A.C.E. and Benavides trails.

Figure 4.2 Proportion of captures of dogs and margays that were at locations at which the other species was not captured (blue) or at which both species were captured (red).

CHAPTER 5: MARGAY LITERATURE REVIEW

INTRODUCTION

The margay, Leopardus wiedii (Schinz, 1821), is a small, spotted tropical cat native to Central and South America. The margay has recently suffered significant population declines (Payan et al. 2008), and yet its conservation needs have not received the same attention as those of felids that are larger and better known, such as the ocelot ( Leopardus pardalis ), puma ( Puma concolor ), and jaguar ( Panthera onca ). In addition, attempts to conduct studies of margay behavior face significant challenges, due to the species’ low population densities, small size, and nocturnal and arboreal activity patterns. Despite its relatively broad geographic range within

Latin America, the species has long been considered rare in the wild (Leopold 1959).

Evolution

The group of small South American cats that includes the margay—the “ocelot lineage”

(Eizirik et al. 1998)—last shared a common ancestor approximately 5.1 million years ago

(Johnson and O’Brien 1997). The margay is one of 7 South American felids that are thought to share a recent common ancestor and are therefore often conceived of as the “ocelot lineage”

(Collier and O’Brien 1985; Johnson et al. 1999). Within the ocelot lineage, Johnson et al. (1999) cite the ocelot and margay as being “closely related sister taxa,” which also share a fairly common recent ancestor with the Andean mountain cat (L. jacobita). The genus Leopardus— which was until recently considered to be a subgenus within Felis (Nowak 2005)—is endemic to

South America and currently includes 9 species. Members of this genus have a reduced chromosome number (2n = 36) relative to other felids (De Oliviera 1998). Johnson et al. (1998) suggest that Leopardus includes 2 distinct lineages, based upon mitochondrial DNA analysis.

The first lineage is comprised of the ocelot, margay, and Andean mountain cat, and the second is comprised of the other 6 congeners: the colocolo ( L. colocolo ), Pantanal cat ( L. braccatus ),

Pampas cat ( L. pajeros ), Geoffroy’s cat ( L. geoffroyi ), kodkod ( L. guigna ), and oncilla— sometimes referred to as the tigrina—(L. tigrinus ).

Bininda-Edmonds et al. (1999) and Agnarssen et al. (2010) based on molecular evidence found the clade Leopardus to be non-monophyletic, suggesting that further molecular and morphological analyses are needed to elucidate the precise relationships of species within this genus, as well as that of the group’s relationship to the two other main felid lineages, the pantherids and the puma group (Johnson et al. 1998).

It is thought that the margay’s often-noted neoteny contributed to a relatively rapid speciation event from other members of the ocelot lineage (Fagen and Wiley 1978). The specific epithet,

“wiedii ,” is a tribute to Prince Maximillian zu Wied, a German naturalist and owner of the collection from which the first described specimen was obtained (Allen 1916). The common name is thought to be derived from the word “ mbaracayá ,” which means “wild cat” in the language of the Guarani, a tribe native to the western Amazon basin (Cabrera and Yeppes 1960, in De Oliviera 1998). Some of the earliest margay specimens—representing two different subspecies—were collected during Theodore Roosevelt’s legendary post-presidential expedition through Colombia, one of which is cataloged as being collected by Kermit Roosevelt himself

(Allen 1916).

The margay is currently comprised of 10 recognized subspecies. Multiple subspecies appear to co-occur in certain regions. The species and their geographic ranges are: L. w. amazonicus, from the Brazilian Amazon; L. w. boliviae, from Brazil and Bolivia; L. w. cooperi, from Nuevo

Leon, Mexico; L. w. glauculus, from Sinaloa through north Oaxaca, Mexico; L. w. nicaraguae,

from Honduras to Costa Rica; L. w. oaxacensis , from Tamaulipas to Oaxaca; Mexico; L. w.

pirrensis , from Panama to northern Peru ; L. w. salvinius, from Chiapas, Guatemala, and El

Salvador; L. w. vigens , from Orinoco to the Amazon basin ; and L. w. yucatanicus , from Chiapas to northern Guatemala and Yucatan (Downey 1994; Wozencraft 2005). Based on the accepted geographic distributions, the margays observed in eastern Ecuador for the purposes of this thesis should belong to L. w. pirrensis , although morphological measurements were not collected due to the priority of utilizing noninvasive methodology.

Distribution

According to mitochondrial DNA analysis and the fossil record, the last common ancestor of the Leopardus group originated in South America, and the margay has radiated northwards subsequent to its divergence as a species. A single set of subfossil remains—thought to be around 4,400 years old—of the margay have been found in North America, from a site in Orange

County, Texas (Eddleman and Akersten 1966). The current range of the margay (Fig. 1.1) extends from just south of the Mexican state of Sonora throughout Central America and down into much of South America, occurring in Peru, Ecuador, Bolivia, Columbia, Venezuela, the

Guianas, Paraguay, Brazil, and reaching the southernmost limits of its range in northern

Argentina and northwestern Uruguay (De Oliviera 1998). The margay is found in forested habitats and has a broad elevational tolerance, occurring in areas ranging from sea level up to

3000 m (De Oliviera 1998).

Morphology and Physiology

The margay is distinct from several other Leopardus species in that it lacks small solid dot patterns on its fur, instead displaying rosettes along the lateral portions of the body and large, solid spots on the mid- to upper back. In addition, it has a convex braincase that creates the

impression that their eyes tend to bulge more than those of closely related cat species (De

Oliviera 1998), thus creating a seemingly neotenic visage (Fagen and Wiley 1978). There is little difference in body size between the sexes (Table 1.1).

The results of study of felid limb morphology showed the margay to be one of only 3 of the world’s felid species to exhibit distinctly arboreal limb morphology (Meachen-Samuels and Van

Valkenburgh 2009). In addition, the margay’s tail:body length ratio—around 7:10—is greater than that of terrestrial species, an adaptation for deft balance commonly found amongst arboreal mammals (Hayssen 2008). The only other arboreal specialists within the Felidae are the clouded leopard ( Neofelis nebulosa ) and the marbled cat ( Pardofelis marmorata ), two southeast Asian species that are only distantly related to the margay (Bininda-Edmonds et al. 1999; Agnarssen et al. 2010). This finding shows that arboreal adaptations have arisen independently at least twice during the felid radiation, but are rare adaptations found in few extant felid species.

McNab (2000) reported that the margay has a relatively low basal metabolic rate, about 75-

85% of that expected for a mammal of its size, potentially due to having proportionally lower muscle mass than their more terrestrial counterparts. Its basal rate of metabolism was 937.2 cm 3

3 O2/g·h, compared to 1296.9 2 cm O2/g·h for the slightly smaller domestic cat ( Felis sylvestris ).

This finding fits a general trend, however, in which arboreal mammals weighing over 1 kg tend to have lower than expected metabolic rates (McNab 1978).

Reproduction and Development

A comparative study of the effect of season on testicular function of 3 small cats, the ocelot

(L. pardalis ), margay ( L. wiedii ), and oncilla ( L. tigrinus ) species in captivity, found that the margay had the lowest numbers of motile spermatozoa (23.4 ± 2.8 x 10 6) and the lowest

percentage of morphologically normal spermatozoa (57.4 ± 2.8%) (Morais et al. 2002). Serum

testosterone levels for the margay were not significantly different from its congeners, and fecal androgen levels were not affected by season. The authors concluded that male margays are capable of breeding throughout the year. As with any study of captive adults, however, it is difficult to determine whether the same conditions would be present in wild populations.

Swanson and Wildt (1997) analyzed ejaculate of several species of wild-caught felids, and found the margay to have low sperm count and ejaculate volume. It was noted that these values were lower for the sampled wild-caught margays than for those of age-matched males sampled from North American zoos. The effects of captivity on wild-caught animals, however, have the potential to significantly alter patterns of reproductive success, sometimes reducing births because of poor vitamin and mineral supplementation or a diet based upon incorrect prey items, and sometimes reducing post-parturition survival due either to stress levels or naivety of captive mother animals.

Female oestrus cycles last from 32-36 days (Paintiff and Anderson 1980, Mansard 1997). The heat period has been reported as ranging from 4-10 days (De Oliviera 1998) and from 7-13 days

(Mansard 1997). Mansard (1997) includes observations of a captive pair of margays, and reported very low receptivity of the female towards the male’s courtship advances. The authors recommended total privacy during pregnancy and birthing in order to increase the odds of survival for offspring. Challenges with reproduction in captive settings are not uncommon, however, which make it difficult to draw broad conclusions about margay courtship in general from dynamics observed in zoos or breeding facilities.

Although some have reported litter sizes of one or two kittens (Fagen and Wiley 1978;

Paintiff and Anderson 1980), the margay only has one pair of nipples, suggesting that they are most likely to produce a single offspring at a time (De Oliviera 1998). The first set of canine

teeth appears at around 20 days, while the adult teeth erupt between 99-165 days (Fagen and

Wiley 1978). Margays begin to emerge from the den approximately 5 weeks after birth, and are weaned at about 8 weeks of age. Margays reach adult size between 8-10 months of age, with females tending to complete their development earlier than males (Petersen and Petersen 1978).

Young margays reach sexual maturity between 9-12 months of age (De Oliviera 1998).

Feeding Behavior

The diet of the margay is thought to be comprised primarily of arboreal mammals and birds

(De Oliviera 1998), and observations have been made of margays preying upon wedge-capped capuchins ( Cercopithecus nigrivittatus ; Beebe 1925), golden-handed tamarins ( Saguinus midas niger ; De Oliviera 1998) and even a great fruit-eating bat ( Artibeus lituratus ; Rocha-Mendes and

Bianconi 2009). Results of a scat analysis conducted by Wang et al. (2002), however, showed that 52.9% of the margay’s prey consisted of terrestrial mammals. Margays appear to opportunistically take many small vertebrates, and De Azevado (1996) observed the predation upon and consumption of an amphibian by a margay during the daylight hours, despite this species’ reportedly strict predilection for nocturnal activity. Scat analysis by Konecny (1989) found no primate remains, yet identified arthropods in 33.3 % of scats and fruit in 14.4 %, in addition to birds (29.2 %) and arboreal rodents such as Ototylomys phyllotis (48.1 %), Sciurus deppei (22.2 %) and Reithrodontomys (18.5 %) and the small marsupial genus Marmosa (18.5

%). These data—especially the presence of fruit remains in such a marked proportion of samples—suggest that the margay may have more catholic prey tastes than is typical of other felids.

Habitat

Margays appear to prefer dense evergreen and deciduous forest, and occur nearly exclusively in these habitats. They range from tropical lowland forests all the way up to cloud forests at the height of their elevational range (De Oliviera 1998). In forests that have been subject to human modifications, margays were found to prefer narrow trails in dense forest patches over wider trails closer to streams (Goulart et al. 2009). Konecny (1989) reported findings that all of their radio-collared margays’ movements were made along the forest floor, despite their reputation for moving primarily through the canopy. Rocha-Mendes and Bianconi (2009) point to research which has suggested that margays are more sensitive to habitat disturbance than their sympatric congeners, the ocelot ( L. pardalis ) and oncilla ( L. tigrinus ), although margays have also been reported to exist in areas suffering from a high degree of forest fragmentation (De Oliviera

1998).

Home Range and Activity Patterns

Home range estimates for the margay range from 1-20 km 2 (Payan et al. 2008). Konecny

(1989) reported that margays rest in tangles of vegetation 7-10 meters above the forest floor, and seem to switch around between different resting sites within a home range. In the same study, margay activity peaked between 0100-0500 h, and rates of travel ranged from a high of around

550 m/hr between 0300-0400 h to a low of around 50 m/hr on average between 1100-1200 h.

Crawshaw et al. (1995; in Di Bitetti et al. 2008), however, found no difference in activity levels between day and night hours, suggesting that temporal patterns may be somewhat plastic and dependent upon environmental conditions. Although margays are thought to rest and do a great deal of foraging in the canopy, De Oliviera (1998) asserts that “all traveling is done on the ground.”

Current Conservation Status

The margay is currently listed as “Near Threatened” by the IUCN, and its population trend is classified as Decreasing (Payan et al. 2008). Margays have been heavily persecuted due to both demand from the international trade in wildlife parts and their image as a threat to poultry stocks(Payan et al. 2008). Their decline was accelerated by increased restrictions on trade in ocelot skins (Payan et al. 2008), which made the smaller spotted cats attractive alternative targets in a regulatory system in which species labels were rarely challenged or subjected to objective analysis and verification.

Population fragmentation due to deforestation and habitat conversion is a critical concern for many species in South America, and margay populations have suffered greatly due to habitat destruction. In addition, forest fragmentation leads to small and isolated populations, which are cut off from source-sink population regulation patterns (Pulliam 1988) and are at increased risk of disease outbreaks and inbreeding depression.

Table 5.1

Comparison of margay body size between sexes. Adapted from De Oliviera (1998).

Sex Body Length (mm) Tail length (mm) Body mass (kg) Male 552 (470-720) 384.4 (305-490) 3.6 (2.3-4.9) Female 544 (501-690) 391.5 (300-483) 3.0 (2.3-3.5)

Figure 5.1 Current distribution of the margay. Adapted from Payan et al. 2008.

REFERENCES

Abrams, P. 1980. Some comments on measuring niche overlap. Ecology. 61:44-49. Aebischer, N. J., P. A. Robertson, and R. E. Kenward. 1993. Compositional analysis of habitat use from animal radio-tracking data. Ecology 74:1313-1325. Agnarsson, I., M. Kuntner, and L. J. May-Collado. 2010. Dogs, cats, and kin: a molecular species-level phylogeny of Carnivora. Molecular Phylogenetics and Evolution 54:726- 745. Allen, J. A. 1916. Mammals collected on the Roosevelt Brazilian expedition, with field notes. Bulletin of the American Museum of Natural History. 35:559-610. Arlettaz, R., N. Perrin, and J. Hausser. 1997. Trophic resource partitioning and competition between two sibling bat species Myotis myotis and Myotis blythii. Journal of Animal Ecology 66:897-911. Balme, G. A., L. T. B. Hunter, and R. Slotow 2009. Evaluating methods for counting cryptic carnivores. Journal of Wildlife Management 73:433-441. Basille, M. 2008. Habitat selection by lynx (Lynx lynx) in a human-dominated landscape--From theory to application. PhD Dissertation, Biométrie et Biologie Évolutive. Beebe, W. 1925. Studies of a tropical jungle: one quarter of a square mile at Kartabo, British Guiana. Zoologica 6:1-193. Bininda-Edmonds, O. R. P., J. L. Gittleman, and A. Purvis. 1999. Building large trees by combining phylogenetic information: a complete phylogeny of the extant Carnivora (Mammalia). Biological Reviews 74:143-175. Block, W. M. and L. A. Brennan. 1993. The habitat concept in ornithology: theory and applications. In: Current Ornithology (D. M. Power, ed.). Plenum Press, New York, New York. pp. 35-91. Borer, E. T., E. W. Seabloom, J. B. Shurin, K. E. Anderson, C. A. Blanchette, B. Broitman, S. D. Cooper, and B. S. Halpern. 2005. What determines the strength of a trophic cascade? Ecology. 86:528-537. Bottrill, M. C., L. N. Joseph, J. Carwardine, M. Bode, C. Cook, E. T. Game, H. Grantham, S. Kark, S. Linke, E. McDonald-Madden, R. L. Pressey, S. Walker, K. A. Wilson, and H. P. Possingham 2008. Is conservation triage just smart decision making? Trends in Ecology and Evolution 23:649-654. Broadbent, E. N., G. P. Asner, M. Keller, D. E. Knapp, P. J. C. Oliveira, and J. N. Silva 2008. Forest fragmentation and edge effects from deforestation and selective logging in the Brazilian Amazon. Biological Conservation 141:1745-1757. Brongo, L. L., M. S. Mitchell, and J. B. Grand. 2005. Effects of trapping with bait on bait- station indices to black bear abundance. Wildlife Society Bulletin 33:1357-1361. Brooks, T. M., R. A. Mittermeier, C. G. Mittermeier, G. A. B. Da Fonseca, A. B. Rylands, W. R. Konstant, P. Flick, J. Pilgrim, S. Oldfield, G. Magin, and C. Hilton-Taylor 2002. Habitat loss and extinction in the hotspots of biodiversity. Conservation Biology 16:909-023. Brown, J. H. and B. A. Maurer. 1989. Macroecology—the division of food and space among species on continents. Science 243:1145-1150. Brown, J. S. 1996. Coevolution and community organization in three habitats. Oikos 75:41-48. Buck, S. 2006. The role of trust in knowledge acquisition, technology

adoption and access to bank loans: results from field experiments in the Ecuadorian Amazon. Master’s Thesis Dissertation, Virginia Polytechnic Institute and State University. Department of Agriculture and Applied Economics. Butler, J. R. A. and J. Bingham. 2000. Demography and dog-human relationships of the dog population in Zimbabwean communal lands. Veterinary Record 147:442-446. Butler, J. R. A, J. T. du Toit, and J. Bingham. 2004. Free-ranging domestic dogs ( Canis familiaris ) as predators and prey in rural Zimbabwe: threats of competition and disease to large wild carnivores. Biological Conservation 115:369-378. Buytaert, W., R. Celleri, P. Willems, B. De Bievre, G. Wyseure. 2006. Spatial and temporal rainfall variability in mountainous areas: A case study from the south Ecuadorian Andes. Journal of Hydrology 329:413-421. Cabrera, A. and J. Yeppes. 1960. Mamîferos Sud Americanos (vida, costumbres y descripsion). Histori Natural Ediar, Compa˜ia Argentino de Ciencias Naturales “Bernardino Rivadavia,” Ciencias Zoológicas 6:161-247. Carbone, C., G. M. Mace, S. C. Roberts, and D. W. Macdonald 1999. Energetic constraints on the diet of terrestrial carnivores. Nature 402:286-288. Campos, C. B., C. F. Esteves, K. Ferraz, P. G. Crawshaw, and L. M. Verdade. 2007. Diet of free-ranging cats and dogs in a suburban and rural environment, south-eastern Brazil. Journal of Zoology 273:14-20. Caro, T. M. and C. J. Stoner. 2003. The potential for interspecific competition among African carnivores. Biological Conservation 110:67-75. Case, T. J. and M. E. Gilpin. 1974. Interference competition and niche theory. Proceedings of the National Academy of Sciences 71:3073-3077. Cattin, M. F., L. F. Bersier, C. Banasek-Richter, R. Baltensperger and J. P. Gabriel. 2004. Phylogenetic constraints and adaptation explain food-web structure. Nature 427:835-839. Chessón, P. and J. J. Kuang. 2008. The interaction between predation and competition. Nature 456:235-238. Cleaveland, S., M. G. J. Appel, W. S. K. Chalmers, C. Chillingworth, M. Kaare, and C. Dye. 2000. Serological and demographic evidence for domestic dogs as a source of canine distemper virus infection for Serengeti wildlife. Veterinary Microbiology 72:217-227. Collier, G. E. and. S. J. O’Brien. 1985. A molecular phylogeny of the Felidae: immunological distance. Evolution 39:473-487. Conner, M. C., R. F. Labisky, and D. R. Progulske. 1983. Scent-station indices as measures of population abundance for bobcats, raccoons, gray foxes and opossums. Wildlife Society Bulletin 11:146-152. Corbett, L. K. 1995. The dingo in Australia and Asia . Australian Natural History Series, University of New South Wales Press, Sydney. Creel, S., G. Spong, and N. M. Creel. 2001. Interspecific competition and the population biology of extinction-prone carnivores. In : Conservation of Carnivores (J. Gittleman, D. Macdonald, S. Funk, and R. Wayne, eds). Cambridge University Press, pp 35-60. Crooks, K. R. and M. E. Soulé. 1999. Mesopredator release and avifaunal extinctions in a fragmented system. Nature 400:563-567. Courtenay, O., D. W. MacDonald, R. Lainson, J. J. Shaw, and C. Dyes. 1994. Epidemiology of canine leishmaniasis: a comparative serological study of dogs and foxes in Amazon Brazil. Parasitology 109:273-279.

Courtenay, O., R. J. Quinnell, L. M. Garcez, and C. Dye. 2002. Low infectiousness of a wildlife host of Leishmania infantum : the crab-eating fox is not important for transmission. Parasitology 125:407-414. Craft, M. E., P. L. Hawthorne, C. Packer and A. P. Dobson 2008. Dynamics of a multihost pathogen in a carnivore community. Journal of Animal Ecology 77:1257-1264. Crawshaw, P. G. Jr. 1995. Comparative ecology of ocelot ( Felis pardalis ) and jaguar ( Panthera onca ) in a protected subtropical forest in Brazil and Argentina. PhD Dissertation, University of Florida. Culter, T. L. and D. E. Swann. 1999. Using Remote Photography in Wildlife Ecology: A Review. Wildlife Society Bulletin 27:571-581. Daubenmire, R. 1968. Plant communities: a textbook of plant synecology . Harper and Row, New York, N. Y. 300 pp. Dayan, T. and D. Simberloff. 1998. Size patterns among competitors: ecological character displacement and character release in mammals, with special reference to island populations. Mammal Review 28:99-124. Dayan, T. and D. Simberloff. 2005. Ecological and community-wide character displacement: the next generation. Ecology Letters 8:875-894. De Azevado, F. C. C. 1996. Notes on the behavior of the margay Felis wiedii (Schinz, 1821), (Carnivora, Felidae), in the Brazilian Atlantic Forest. Mammalia 60:325-328. DeMatteo, K. E., O. Carrillo, G. L. Zuercher, S. Ramírez, K. Smith, and I. J. Porton. 2004. A technique for attracting bush dogs Speothos venaticus in the wild. Canid News 7:6. De Oliveira, T. G. 1998. Leopardus wiedii . Mammalian Species. 579:1-6. Di Bitetti, M. S., Y. E. Di Blanco, J. A. Pereira, A. Paviolo, and I. Jiménez Pérez. 2006. Density, habitat use and activity patterns of ocelots ( Leopardus pardalis ) in the Atlantic Forest of Misiones, Argentina. Journal of Zoology 270:153-163. Di Bitetti, M. S., A. Paviolo, C. D. De Angelo, and Y. E. Di Blanco 2008. Local and continental correlates of the abundance of a neotropical cat, the ocelot ( Leopardus pardalis ). Journal of Tropical Ecology 24:189-200. Di Bitetti, M. S., Y. E. Di Blanco, J. A. Pereira, A. Paviolo, and I. Jiménez Pérez. 2009. Time partitioning favors the coexistence of sympatric crab-eating foxes ( Cerdocyon thous ) and pampas foxes ( Lycalopex gymnocercus ). Journal of Mammalogy 90:479-490. Diefenbach, D. R., M. J. Conroy, R. J. Warren, W. E. James, and L. A. Baker. 1994. A test of the scent-station survey technique for bobcats. Journal of Wildlife Management 58:10- 17. Dillon, A. 2005. Ocelot density and home range in Belize, Central America: camera-trapping and radio telemetry. Master’s Thesis. Virginia Polytechnic and State University, Fisheries and Wildlife Sciences. 1-152. Dobbins, C. L. 2004. Evaluation of lures, baits, and urines. Bullet Printing, Murfreesboro, NC. Downey, P. J. 1994. Hair-snare survey to assess distribution of margay (Leopardus wiedii) inhabiting El Cielo Biosphere Reserve, Tamaulipas, Mexico. Bachelor’s Thesis, Oklahoma State University Department of Geology. pp1-50. Downey, P. J., E. C. Hellgren, A. Caso, S. Carvajal, and K. Frangioso. 2007. Hair snares for noninvasive sampling of felids in North America: do gray foes affect success? Journal of Wildlife Management 71:2090-2094. Durant, S. M., M. E. Craft, C. Foley, K. Hampson, A. L. Lobora, M. Msuha, E. Eblate, J. Bukombe, J. Mchetto, and N. Pettorelli. 2010. Does size matter? An investigation of

habitat use across a carnivore assemblage in the Serengeti, Tanzania. Journal of Animal Ecology 79:1012-1022. Eddleman, C. D. and W. A. Akersten. 1966. Margay from the post-Wisconsonian of southeastern Texas. The Texas Journal of Science 18: 378-385. Edwards, J. W., D. G. Heckel, and D. C. Guynn. 1998. Niche overlap in sympatric populations of fox and gray squirrels. Journal of Wildlife Management 62:354-363. Eizirik, E., et al. 1998. Phylogeographic patterns and evolution of the mitochondrial DNA region in two neotropical cats (Mammalia, Felidae). Journal of Molecular Evolution 47:613-624. Emmons, L. H. 1987. Comparative feeding ecology of felids in a Neotropical rainforest. Behavioral Ecology and Sociobiology. 20, 4, 271-283. Ewer, R. F. 1968. Ethology of Mammals. London: Logos Press. Fagen, R. M. and K. S. Wiley. 1978. Felid paedomorphosis, with special reference to Leopardus . Carnivore 1:72-81. Farrel, L. A., J. Roman, and M. E. Sunquist. 2000. Dietary separation of sympatric carnivores identified by molecular analysis of scats. Molecular Ecology 9:1583-1590. Fiala, A. C. S., S. L. Garman, and A. N. Gray. 2006. Comparison of five canopy cover estimation techniques in the western Oregon Cascades. Forest Ecology and Management 232:188-197. Fiorello, C. V., S. L. Deem, M. E. Gompper, and E. J. Dobovi. 2004. Seroprevalence of pathogens in domestic carnivores on the border of Madidi National Park, Bolivia. Animal Conservation.7:45-54. Fischer, J. and D. B. Lindenmayer. 2002. Small patches can be valuable for biodiversity conservation: two case studies on birds in southeastern Australia. Biological Conservation 106:129-136. Foresman, K. R. and D. E. Pearson. 1998. Comparison of proposed survey procedures for detection of forest carnivores. Journal of Wildlife Management 62, 4, 1217-1226. Gese, E. M. 2001. Monitoring of terrestrial carnivore populations . USDA National Wildlife Research Center. pp 1-27. Glazier, D. S. 2010. A unifying explanation for the diverse metabolic scaling in animals and plants. Biological Reviews 85:111-138. Glen, A. S. and C. R. Dickman. 2008. Niche overlap between marsupial and eutherian carnivores: does competition threaten the endangered spotted-tailed quoll? Journal of Applied Ecology 45:700-707. Gompper, M. E., R. W. Kays, S. D. Lapoint, D. A. Bogan, and C. R. Cryan. 2006. A comparison of noninvasive techniques to survey carnivore communities in northeastern North America. Wildlife Society Bulletin 34:1142-1151. Gompper, M. E. and A. T. Vanak. 2008. Subsidized predators, landscapes of fear and disarticulated carnivore communities. Animal Conservation 11:13-14. Gotceitas, V. and P. Colgan. 1989. Predator foraging success and habitat complexity: quantitative test of the threshold hypothesis. Oecologia 80: 158-166. Goulart, F. V. B., N. C. Caceres, M. E. Graipel, M. A. Tortato, I. R. Ghizoni Jr, L. Gustavo, and R. Oliviera-Santos. 2009. Habitat selection by large mammals in southern Brazilian Atlantic forest. Mammalian Biology. 74, 182-190. Graves, G. E. and M. L. Bodicker. 1987. Field evaluation of olfactory attractants and strategies used to capture depredating coyotes. Great Plains Wildlife Damage Control . 195-294.

Grigione, M. M., P. Beier, R. A. Hopkins, D. Neal, W. D. Padley, C. M. Schonewald, and M. L. Johnson. 2002. Ecological and allometric determinants of home-range size for mountain lions ( Puma concolor ). Animal Conservation 5:317-324. Grigione, M. M. and R. Mrykalo. 2004. Effects of artificial night lighting on endangered ocelots (Leopardus pardalis ) and nocturnal prey along the United States-Mexico border: a literature review and hypotheses of potential impacts. Urban Ecosystems 7:65-77. Hafner, Mark S. 2007. Field research in mammalogy: An enterprise in peril. Journal of Mammalogy 88:1119-1128. Hall, L. S., P. R. Krausman, and M. L. Morrison 1997. The habitat concept and a plea for standard terminology. Wildlife Society Bulletin 25:173-182. Harper, K. A., S. E. Macdonald, P. J. Burton, J. Chen, K. D. Brosofske, S. C. Saunders, E. S. Euskirchen, D. Roberts, M. S. Jaiteh, and P. Esseen. 2005. Edge influence on forest structure and composition in fragmented landscapes. Conservation Biology 19:768-782. Harrison, R. L. 1997. Chemical attractants for Central American felids. Wildlife Society Bulletin 25:93-97. Hass, C. C. 2009. Competition and coexistence in sympatric bobcats and pumas. Journal of Zoology 278:174-180. Hayssen, V. 2008. Patterns of body and tail length and body mass in Sciuridae. Journal of Mammalogy 89:852-873. Hobbs, R. J. and L. J. Kristjanson. 2003. Triage: how do we prioritize health care for landscapes? Ecological Restoration & Management 4:S39-S45. Hodge, A. M. C., B. S. Arbogast, E. N. Vanderhoff, J. Camper, J. R. Burger, and T. W. Knowles. 2010. Inventory of a mid-elevation assemblage of mammals in Ecuador’s Tropical Andes hotspot. Poster presentation. Annual meeting of the Society for Integrative and Comparative Biology, Seattle, Washington. Husson, A. M. 1978. The mammals of Suriname . E. J. Brill, Leiden, The Netherlands. Hutto, R. L. 1985. Habitat selection by nonbreeding, migratory land birds. In: Habitat Selection in Birds (M. L. Cody, Ed.) Academic Press, Orlando, FL. pp 455-476. IUCN. 2011. Red List of Threatened Species. Online. Accessed 12 October 2011. . Jaksic, F. M., H. W. Greence, and J. L. Yanez. 1981. The guild structure of a community of predatory vertebrates in central Chile. Oecologica 49:21-28. Jaenike, J. and R. D. Holt. 1991. Genetic variation for habitat preference: evidence and explanations. American Naturalist 137:S67-S90. Jhala, Y. V. and R. H. Giles. 1991. The status and conservation of the wolf in Gujarat and Rajasthan, India. Conservation Biology 5:476-483. Johansson, T. 1985. Estimating canopy density by the vertical tube method. Forest Ecology and Management 11:139-144. Johnson, W. E. and S. J. O’Brien. 1997. Phylogenetic reconstruction of the Felidae using 16S rRNA and NADH-5 mitochondrial genes. Journal of Molecular Evolution 44:S98-S116. Johnson, W. E., M. Culver, J. A. Iriarte, E. Eizirik, K. L. Seymour, and S. J. O’Brien 1998. Tracking the evolution of the elusive Andean mountain cat ( Oreailurus jacobita ) from mitochondrial DNA. American Genetic Association 89:227-232. Johnson, W. E., J. P. Slattery, E. Eizirik, J-H. Kim, M. M. Raymond, C. Bonacic, R. Cambre, P. Crawshaw, A. Nunes, H. N. Seuanez, M. A. M. Moreiera, K. L. Seymour, F. Simon, W. Swanson, and S. J. O’Brien. 1999. Disparate phylogeographic patterns of molecular

genetic variation in four closely related South American small cat species. Molecular Ecology 8:S79-S94. Jones, M. E. and L. A. Barmuta. 2000. Niche differentiation among sympatric dasyurid carnivores. Journal of Mammalogy 81: 434-447. Karanth, K. U. and M. E. Sunquist. 2000. Behavioral correlates of predation by tiger ( Panthera tigris ), leopard ( Panthera pardalis ) and dhole ( Cuon alpinus ) in Nagaradahole, India . Journal of Zoological Society of London 250:255-265. Kays, R., B. Kranstauber, P. A. Jansen, C. Carbone, M. Rowcliffe, T. Fountain, and S. Tilak. 2009. Camera traps as sensor networks for monitoring animal communities. 34th IEEE Conference on Local Computer Networks. Kelly, M. J. 2003. Jaguar monitoring in the Chiquibul forest. Caribbean Geography 13:19-32. Kelly, M. J. 2008. Design, evaluate, refine: camera trap studies for elusive species. Animal Conservation. 11:182-184. Kelly, M. J. and E. L. Holub. 2008. Camera trapping of carnivores: trap success among camera types and across species, and habitat selection by species, on Salt Pond Mountain, Giles County, Virginia. Northeastern Naturalist 15:249-262. Kojola, I., S. Ronkainen, A. Hakala, S. Heikkinen and S. Kokko 2004. Interactions between wolves Canis lupus and dogs C. familiaris in Finland. Wildlife Biology 10:101-105. Kok, O. B. and J. A. Nel. 2004. Convergence and divergence in prey of sympatric canids and felids: opportunism or phylogenetic constraint? Biological Journal of the Linnean Society 83:527-538. Konecny, M. J. 1989. Movement patterns and food habits of four sympatric carnivore species in Belize, Central America. Advances in Neotropical Mammalogy 1989:243-264. Kruuk, H. 1986. Interactions between Felidae and their prey species: a review. In: Cats of the World: Ecology, Conservation, and Management (S. D. Miller and D. D. Everett, Eds.) Washington D.C.: National Wildlife Foundation. pp. 353-374. Larrucea, E. S., P. F. Brussard, M.M Jaeger, R. H. Barrett 2007. Cameras, coyotes, and the assumption of equal detectability. Journal of Wildlife Management. 71:1682-1689. Laundré, J. W., J. M. M. Calderas, and L. Hernández. 2009. Foraging in the landscape of fear, the predator’s dilemma: where should I hunt? The Open Ecology Journal 2:1-6. Leopold, A. S. 1959. Wildlife of Mexico . University of California Press. Linnell, J. D. C., J. E. Swenson, and R. Andersen 2000. Conservation of biodiversity in Scandinavian boreal forests: large carnivores as flagships, umbrellas, indicators, or keystones? Biodiversity and Conservation 9:857-868. Linnell, J. D. C. and O. Strand. 2000. Interference interactions, co-existence and conservation of mammalian carnivores. Diversity and Distributions 6:169-176. Lucherini, M., J. I. Reppucci, R. S. Walker, M. L. Villalba, A. Wurstten, Giovana Gallardo, R. Villalobos, and P. Perovic. 2009. Activity pattern segregation of carnivores in the high Andes. Journal of Mammalogy 90:1404-1409. Mansard, P. 1997. Breeding and husbandry of the Margay Leopardus wiedii yucatanica at the Ridgeway Trust for Endangered Cats, Hastings. International Zoo Yearbook 35:94-100. McArthur, R. 1970. Species packing and competitive equilibrium for many species. Theoretical Population Biology 1:1-11. McDonald-Madden, P. W. J. B., and H. P. Possingham. 2008. Subpopulation triage: how to allocate conservation effort among populations. Conservation Biology. 22:656-665.

McLoughlin, P. D., L. R. Walton, H. Dean Cluff, P. C. Paquet, and M. A. Ramsay. 2004. Hierarchical habitat selection by tundra wolves. Journal of Mammalogy 85:576-580. McNab, B. K. 1978. Energetics of arboreal folivores: physiological problems and ecological consequences of feeding on a ubiquitous food supply. In: Montgomery, G. G. (Ed .) The Ecology of Arboreal Folivores . Smithsonian Institution Press, Washington, D. C. pp. 153-162. McNab, B. 2000. The standard energetics of mammalian carnivores: Felidae and Hyaenidae. Canadian Journal of Zoology 78:2227-2239. Meachen-Samuels, J. a. B. Van Valkenburgh. 2009. Craniodental indicators of prey size preference in the Felidae. Biological Journal of the Linnean Society 96:785-799. Michalski, F., P. G. Crawshaw, T. G. De Oliveira, and M. E. Fabián. 2006. Notes on home range and habitat use of three small carnivores in a disturbed vegetation mosaic of southeastern Brazil. Mammalia 70:52-57. Mitchell, B. D. and P. B. Banks. 2005. Do wild dogs exclude foxes? Evidence for competition from dietary and spatial overlaps. Austral Ecology 30:585-591. Moen, R. and E. L. Lindquist. 2006. Testing a remote camera protocol to detect animals in the Superior National Forest. NRRI Technical Report No. NRRI/TR-2006-28. 1-20. Morais, R. N., R. G. Mucciolo, M. L. F. Gomes, O. Lacerda, W. Moraes, N. Moreira, L. H. Graham, W. F. Swanson, and J. L. Brown. 2002. Seasonal analysis of semen characteristics, serum testosterone and fecal androgens in the ocelot ( Leopardus pardalis ), margay ( L. wiedii ) and tigrina ( L. tigrinus ). Theriogenology 57:2027-2041. Moreno, R. S., R. W. Kays, and R. Samudio. 2006. Competitive release in diets of ocelot (Leopardus pardalis ) and puma ( Puma concolor ) after jaguar ( Panthera onca ) decline. Journal of Mammalogy 87:808-816. Mowat, G., C. Shurgot, and K. Poole. 2000. Using track plates and remote cameras to detect marten and short-tailed weasels in coastal cedar hemlock forests. Northwestern Naturalist 81:113-121. Munson, L., K. A. Terio, R. Kock, T. Mlengeya, M. E. Roelke-Parker, E. Dubovi, B. Summers, A. R. E. Sinclair, and C. Packer. 2008. Climate extremes promote fatal co-infections during canine distemper epidemics in Africa lions. PLoS ONE 3,6: e2545. Murcia, C. 1995. Edge effects in fragmented forests: implications for conservation. Trends in Ecology and Evolution 10:58-62. Murray, J. L. and. G. L. Gardner. 1997. Leopardus pardalis . Mammalian Species 548:1-10. Neill, D. and W. Palacios. 2011. Gran Sumaco and Upper Napo Region, Ecuador. Smithsonian Institution. Access online 19 September 2011: http://botany.si.edu/projects/cpd/sa/sa38.htm . Nowak, R. M. 2005. Walker’s Carnivores of the World . Johns Hopkins University Press; Baltimore, MD. p. 253. Nowell, K. and P. Jackson. 2006. Wild cats. Status Survey and Conservation Action Plan. IUCN/SSC Cat Specialist Group. IUCN, Gland, Switzerland. Paintiff, J. A. and D. E. Anderson. 1980. Breeding the margay Felis wiedii at the New Orleans Zoo. International Zoo Yearbook 20:223-224. Palomares, F., P. Ferrareas, J. Fedriani, and M. Delibes. 1996. Spatial relationships between Iberian lynx and other carnivores in an area of south-western Spain. Journal of Applied Ecology 33:5-13.

Palomares, F. and. T. M. Caro. 1999. Interspecific killing among mammalian carnivores. American Naturalist 153:492-508. Passamani, M., S. I. Mendes, A. G. Chiarello, J. A. Passamani, and R. R. Laps. 1997. Reintroucao do sagui-de-cara-branca ( Callithrix geoffroyi ) em fragmentos de mata Atlantica no sudests do Brasil. In: Ferrari, S. F. and H. A.. Schneider, (Eds.). Primatologia no Brasil, Anais do VI Congresso Brasileiro de Primatologia, Rio de Janeiro, R. J. vol. 5. Universidade Federal do Pará, Belém, pp 119-128. Paull, D. J., A. W. Claridge, and S. C. Barry. 2011. There’s no accounting for taste: bait attractants and infrared digital cameras for detecting small to medium ground-dwelling mammals. Wildlife Research 38:188-195. Payan, E., E. Eizirik, T. de Oliveira, R. Leite-PItman, M. Kelly, and C. Valderrama. Leopardus wiedii . IUCN 2008. IUCN Red List of Threatened Species. Version 2010.4. . Accessed 7 May 2011. Perreault, T. 2003. Making space: community organization, agrarian change, and the poitics of scale in the Ecuadorian Amazon. Latin American Perspectives 30:96-121. Petersen, M. K. and Mk. Petersen. 1978. Growth rate and other postnatal developmental changes in margays. Carnivore 1:87-92. Pielou, E. C. 1984. The Interpretation of Ecological Data . Wiley-Interscience, Hoboken, NJ. Polis, G. A. 1984. Age structure component of niche width and intraspecific resource partitioning: can age groups function as ecological species? The American Naturalist 123:541-564. Polis, G. A. and C. A. Myers. 1989. The ecology and evolution of intraguild predation: potential competitors that eat each other. Annual Review of Ecology and Systematics 20:297-330. Powell, R. A. 2000. Animal home ranges and territories and home range estimators. In: Research Techniques in Animal Ecology (L. Boitani, Ed.) Columbia University Press. New York City, New York. pp. 65-103. Pulliam, H. R. 1988. Sources, sinks, and population regulation. The American Naturalist 132:652-661. Ralls, K. and P. J. White. 1995. Predation on San Joaquin kit foxes by larger canids. Journal of Mammalogy 76:723-729. Ray, J. C. and M. E. Sunquist. 2001. Trophic relations in a community of African rainforest carnivores. Oecologica. 127:395-408. Reed, S. E. and A. M. Merenlender. 2011. Effects of management of domestic dogs and recreation on carnivores in protected areas in northern California. Conservation Biology 25:504-513. Reid, F. A. 1997 . Field guide to the mammals of Central America and Southeast Mexico. Oxford University Press, New York. Ripple, W. J., A. J. Wirsing, R. L. Beschta, and S. W. Buskirk. 2011. Can restoring wolves aid in lynx recovery? The Wildlife Society Bulletin 35:514-518. Rocha-Mendes, F. and G. V. Bianconi. 2009. Opportunistic predatory behavior of margay, Leopardus wiedii (Schinz, 1821), in Brazil. Mammalia 73:151-152. Roelke-Parker, M. E., L. Munson, C. Packer, R. Kock, S. Cleaveland, M. Carpenter, S. J. O’Brien, A. Pospischill, R. Hofmann-Lehmann, H. Lutz, G. L. M. Mwamengele, M. N. Mgasa, G. A. Machango, B. A. Summers, and M. J. O. Appel 1996. A canine distemper virus epidemic in Serengeti lions ( Panthera leo ). Nature. 379:441-445.

Roemer, G. W., M. E. Gompper, and B. Van Valkenburgh 2009. The ecological role of the mammalian mesocarnivore. Bioscience 59:165-173. Rosenweig, M. L. 1966. Community structure in sympatric carnivora. Journal of Mammalogy 47:602-612. Rosenzweig, M. L. and Z. Abramsky. 1986. Centrifugal community organization. Oikos 46:339-348. Schlexer, F. V. 2008 Attracting animals to detection devices. In: Noninvasive Survey Methods for Carnivores ( R. A. Long et al., Eds.). Island Press. Shepherd, U. L. and D. A. Kelt. 1999. Mammalian species richness and morphological complexity along an elevational gradient in the arid south-west. Journal of Biogeography 26:843-855. Sillero-Zubiri, C., A. A. King, and D. W. Macdonald. 1996. Rabies and mortality in Ethiopian wolves ( Canis simensis ). Journal of Wildlife Diseases. 32:80-86. Silva-Rodríguez, E. A. and K. E. Sieving. 2011. Influence of care of domestic carnivores on their predation of vertebrates. Conservation Biology 25:808-815. Silver, S. C., L. E. T. Ostro, L. K. Marsh, L. Maffei, A. J. Noss, M. J. Kelly, R. B. Wallace, H. Gomez, and G. Ayala. 2004. The use of camera traps for estimating jaguar Panthera onca abundance and density using capture/recapture analysis. Oryx 38:1-7. Stefan, A. 1999. Invasion of matrix species in small habitat patches. Conservation Ecology. [Online]. Accessed 2 November 2011. 3:1 . Swanson, W. F. and D. E. Wildt. 1997. Strategies and progress in reproductive research involving small cat species. International Zoo Yearbook 35:152-159. Tirira, D. 2007. Mamíferos del Ecuador: Guia de Campo . Wang, S. W. and D. W. Mcdonald. 2009. The use of camera traps for estimating tiger and leopard populations in the high altitude mountains of Bhutan. Biological Conservation 142:606-613. Wayne, R., D. Benveniste, D. Janczewski, and S. J. O’Brien. 1989. Molecular and beiochemical evolution of the Carnivora. In: Carnivore Behaviour, Ecology, and Evolution (J. L. Gittleman, Ed.). London: Chapman & Hall. pp 465-494. Vanak, A. T. 2008. Intraguild interactions between native and domestic carnivores in central India. PhD Dissertation, University of Missouri Department of Natural Resources. 1-197. Van Valen, L. 1965. Morphological variation and width of the ecological niche. The American Naturalist 99:377-390. Van Valkenburgh, B. 1985. Locomotor diversity within past and present guilds of large predatory mammals Paleobiology 11:406-428. Van Valkenburgh, B. 1991. Iterative evolution of hypercarnivory in canids (Mammalia: Carnivora): evolutionary interactions among sympatric predators. Paleobiology 17:340- 362. Van Valkenburgh, B. 2007. Déja vu: the evolution of feeding morphologies in the Carnivora. Society for Integrative and Comparative Biology 47:147-163. Vanak, A. T. and M. E. Gompper. 2010. Interference competition at the landscape level: the effect of free-ranging dogs on a native mesocarnivore. Journal of Applied Ecology 47:1225-1232. Vanderhoff, E. N., A. M. C. Hodge, B. S. Arbogast, J. Nilsson, T. W. Knowles. 2012. Preliminary camera-trap survey of margays ( Leopardus wiedii ) in the eastern Andean foothills of Ecuador. Mastazoologia Neotropical. [In press].

Vézina, A. F. 1985. Empirical relationships between predator and prey size among terrestrial vertebrate predators. Oecologica 67:555-565. Wang, E. 2002. Diets of ocelots ( Leopardus pardalis ), margays ( L. wiedii ), and oncillas ( L. tigrinus ) in the Atlantic Rainforest in Southeast Brazil. Studies on Neotropical Fauna and Environment 37:207-212. Wayne, R. K., R. E. Benveniste, D. N. Janczewski, and S. O’Brien. 1989. Biomolecular and biochemical evolution of the Carnivora. In: Carnivore behavior, ecology, and evolution (J. L. Gittleman, Ed.). Cornell University Press, Ithaca, New York. Pp 465-494. Werner, E. E. and J. F. Gilliam. 1984. The ontogenetic niche and species interactions in size- structured populations. Annual Review of Ecology and Systematics 15:393-425. Whiteman, C. W., E. R Matushima, U. E. C. Confalonieri, M. D. C. Palha, A. S. L. Da Silva, and V. C. Monteiro. 2007. Human and domestic animal populations as a potential threat to wild carnivore conservation in a fragmented landscape from the Eastern Brazilian Amazon. Biological Conservation 138:290-296. Wozencraft, C. W. 2005 Order: Carnivora. In: Mammal Species of the World: A Taxonomic and Geographic Reference (D. E. Wilson and D. M. Reeder, Eds). Johns Hopkins University Press. York, E. C., T. L. Moruzzi, T. K. Fuller, J. F. Organ, R. M. Sauvajot, and R. M. DeGraaf. 2001. Description and evaluation of a remote camera and triggering system to monitor carnivores. Wildlife Society Bulletin 29:1228-1237.