Dissertation Working Plan
Zach J. Farris Candidate for Degree of Doctor of Philosophy Fisheries and Wildlife Sciences 100 Cheatham Hall Viginia Tech Blacksburg, VA 24061
Title: “Carnivore ecology across the Masoala-Makira Landscape”
Goal & Objectives: My goal is to address gaps in our understanding of carnivore ecology and the impacts of forest fragmentation, human encroachment, and poaching on carnivores thereby aiding in conservation of this complex, endemic ecosystem. I have four objectives to achieve this goal:
1.) Estimate density, activity patterns, and occupancy rates of fosa (C. ferox), fanaloka (Fossa fossana), falanouc (Eupleres goudotii), ring-tailed mongoose (Galidia elegans), broad-striped mongoose (Galidictis fasciata), brown-tailed mongoose (Salanoia concolor) and invasive species, including the small Indian civet (Viverricula indica), domestic dogs, and feral cats, within and among 3 fragmented and 3 non-fragmented sites.
2.) Determine factors influencing carnivore densities, activity, and occupancy within, and across these study sites, including landscape characteristics (using GIS), microhabitat features, climatic conditions, prey species (namely lemurs), and human presence and activity.
3.) Capture and radio collar C. ferox and F. fossana to determine seasonal home range and activity patterns for both males and females, as well as collect anatomical measurements.
4.) Quantify the seasonal diet of C. ferox and F. fossana through collection and analyses of scat, as well as conduct genetic and disease analyses from blood and scat samples. Table of Contents
Introduction:…………………………………………………………………………………… 4 Project Overview………………………………………………………………………. 4 Goals and Objectives…………………………………………………………………... 6 Study Site: Masoala-Makira Landscape……………………………………………….. 7 Effects of Poaching: Makira Protected Area…………………………………... 8 Selection of Study Sites………………………………………………………... 9 Methods:……………………………………………………………………………………….. 11 Objective I:…………………………………………………………………………….. 11 Camera Trapping Methods and Grid Arrangement……………………………. 11 Population Variables: Comparison of Density Estimation Techniques……….. 13 Population Variables: Occupancy and Trap Success………………………….. 18 Pilot Camera Study (2008-2009)……………………………………...…........ 19 Objective I: Questions & Hypotheses ………………………………………… 20 Objective II:………………………………………………………………………….... 20 Within Study Sites: Habitat Analyses………………………………………... 20 Landscape Scale………………………………………………………. 21 Microhabitat Scale…………………………………………………….. 22 Within Study Sites: Lemur Surveys…………………………………………......23 Within Study Sites: Species Interactions ……………………………………… 25 Across Study Sites: Comparison of Fragmented vs. Non-Fragmented Sites……25 Objective II: Questions & Hypotheses………………………………………… 27 Objective III:……………………………………………………………………………28 Capture Methods………………………………………………………………..28 Radio Telemetry and Home Range Analyses………………………………….. 29 Objective III: Questions & Hypotheses…………………………………….….. 31 Objective IV:………………………………………………………………………… 31 Blood Samples…………………………………………………………………. 31 Scat Samples…………………………………………………………………… 32 Objective IV: Questions & Hypotheses…………………………….……….… 33
Expected Outcomes………………………………………………………………………….. 35
Schedule……………………………………………………………………………….……… 37
Literature Cited……………………………………………………………………………… 38
Figures and Tables…………………………………………………………………………… 45 Introduction Project Overview Madagascar is consistently ranked as one of the top ten global biodiversity hotspots and a top conservation priority (Mittermeir et al. 2004). Of the unique wildlife in Madagascar, top carnivores are critically important as they may exert significant influence on ecosystem structure and serve as “umbrella species” due to their large home ranges (Noss 1990; Gittleman et al. 2001). Unfortunately, our current knowledge of Malagasy carnivores is poor, severely limiting efforts to conserve them or the diverse species that fall under their ‘umbrella.’ Due to anthropogenic disturbances, forest fragmentation is increasingly widespread and biodiversity loss continues to mount. Currently, less than 10% of the original forest remains in Madagascar and that which remains is increasingly threatened by logging and other threats related to ongoing political instability (Green and Sussman 1990; Butler 2010). Studies investigating the effects of fragmentation on carnivores and their primary prey (lemurs) are lacking. Forest fragmentation decreases habitat quality (Harper et al. 2007), impedes gene flow (Craul et al. 2009), and even exacerbates predation events on lemurs (Irwin et al. 2009), thus studies investigating habitat fragmentation from a broad ecological standpoint are desperately needed. To date, a handful of studies have been conducted on the food habits (Hawkins and Racey 2008; Dollar et al. 2006; Goodman et al. 1997; Wright 1997; Rasoloarison et al. 1995), activity patterns (Dollar et al. 1999; Hawkins 1998; Wright 1997), and population density (Hawkins 1998, Gerber et al. 2009) of the largest endemic carnivore, the fosa (Cryptoprocta ferox). Most of these studies come from western deciduous forest with little data collected concerning C. ferox population density, sex ratio, activity patterns, and home range within eastern rainforest habitat where the most dramatic rates of deforestation are occurring. C. ferox, currently classified by IUCN as Vulnerable (Hawkins and Dollar 2009), live in low population densities throughout Madagascar, with an estimated 0.26 individuals per km² in deciduous forest (Hawkins 1998) and 0.17 per km² in eastern rainforest (Gerber et al. 2009). Pilot study work in the eastern rainforests of Ranomafana National Park indicate not only low densities of this top predator within the protected area, but also a complete absence of the C. ferox from fragmented forest areas < 25 km from the park (Gerber and Karpanty, unpublished data). However, a recent camera-trapping study in rainforest fragments < 2.5 km from contiguous forest did find one individual C. ferox indicating some use of fragments is possible but that distance to contiguous forest may be important (Gerber, unpublished data). Similarly, interviews of villagers living 0-20 km from the border of Ranomafana National Park demonstrated that C. ferox are sited outside of protected areas, but never > 6.1 km from the park boundary (Kotschwar, unpublished data). As a result of these low population densities and preliminary work indicating a negative relationship between C. ferox presence and both fragmentation and distance to protected areas, C. ferox may be the most vulnerable of all carnivores to the current ecological changes in Madagascar (Hawkins 2003). As the largest extant carnivore in Madagascar, C. ferox plays a dynamic role in this ecosystem as a significant lemur predator. Currently, there is strong evidence that C. ferox impacts both lemur behavior and population dynamics in forests across Madagascar, particularly in rainforest habitat (Irwin et al. 2009; Wright et al. 1997; Karpanty and Wright 2007). As C. ferox and lemurs are increasingly forced into isolated fragments of forest, natural or exacerbated predation rates by fosa may negatively impact lemur populations which are simultaneously being limited by declining habitat quality and human encroachment. For example, Irwin et al. (2009) showed how C. ferox killed an entire group of diademed sifakas (Propithecus diadema) at the Tsinjoarivo forest, extirpating this species from a forest fragment. In addition to elucidating the ecology of the carnivore community, my study will investigate how the dynamics of C. ferox-lemur interactions may change across fragmented and non-fragmented forests. The Masoala-Makira landscape represents one of the last remaining large tracts of primary rainforest (732,750ha) that is critical for the long-term conservation of Madagascar’s severely threatened flora and fauna (Golden 2009; Kremen 2003). This area is a top conservation priority as it contains the highest levels of biodiversity in all of Madagascar, including 6 of 8 endemic carnivores (Garbutt 2007). Further, these forests also protect regional watersheds providing water to more than 100,000 people (WCS 2004). Unfortunately, recent studies within this region have highlighted strong human-wildlife conflicts, including human encroachment of forest habitat, predation of livestock, and extensive use of carnivores, lemurs, and small-mammals as bushmeat (Golden 2009). Additionally, due to the recent political turmoil in Madagascar, an increase in illegal and now government-sanctioned logging has been reported for regions of Masoala (Butler 2010). The effects of these ongoing and new disturbances on carnivores, lemurs, and other wildlife are not known and desperately need investigation. The Wildlife Conservation Society (WCS) has been actively working in this region since 1994 to investigate human-wildlife conflicts, particularly use of bushmeat, including locations and rate of take. In addition, I have been working with WCS to conduct pilot camera trapping studies for the past year to provide baseline data on carnivore presence and distribution, as well as training guides in preparation for this field project. Information from this pilot study has been used to design this proposed long-term study. Goals and Objectives: My ultimate goal is to address gaps in our understanding of carnivore ecology and the impacts of forest fragmentation, human encroachment, and poaching on carnivores thereby aiding in conservation of this complex, endemic ecosystem. I have four objectives to achieve this goal:
1.) Estimate density, activity patterns, and occupancy rates of fosa (C. ferox), fanaloka (Fossa fossana), falanouc (Eupleres goudotii), ring-tailed mongoose (Galidia elegans), broad-striped mongoose (Galidictis fasciata), brown-tailed mongoose (Salanoia concolor) and invasive species, including the small Indian civet (Viverricula indica), domestic dogs, and feral cats, within and among 3 fragmented and 3 non-fragmented sites.
2.) Determine factors influencing carnivore densities, activity, and occupancy within, and across these study sites, including landscape characteristics (using GIS), microhabitat features, climatic conditions, prey species (namely lemurs), and human presence and activity.
3.) Capture and radio collar C. ferox and F. fossana to determine seasonal home range and activity patterns for both males and females, as well as collect anatomical measurements.
4.) Quantify the seasonal diet of C. ferox and F. fossana through collection and analyses of scat, as well as conduct genetic and disease analyses from blood and scat samples.
Study Site: Masoala-Makira Landscape Madagascar currently has a total of 1,698,639ha in protected areas, distributed among 46 nature reserves, national parks, and special reserves (Randrianandianina et al. 2003). Of this total protected area, 43% is found within Masoala National Park (210,000ha) and Makira Protected Area (522,750ha). As the biggest protected complex in Madagascar the Masoala-Makira landscape (Figure 1) provides the largest area of intact rainforest remaining in Madagascar and is critical for the long-term conservation of Madagascar’s endemic and severely threatened flora and fauna (Golden 2009; Kremen 2003; WCS 2004). The Masoala-Makira landscape protects more than half of Madagascar’s floral diversity (Holmes 2007), including a significant portion of Madagascar’s unique lowland, cloud, and littoral forests (Kremen 2003). Moreover, this landscape safeguards numerous wildlife species including several critically endangered species including the silky sifaka (Propithecus candidus) and the Madagascar serpent eagle (Eutriorchis astur). The Masoala-Makira landscape is believed to have the highest levels of biodiversity in all of Madagascar (Holmes 2007). With more than 22 lemur species found in these forests and six of eight endemic carnivores, this landscape is critical for the long term protection of Madagascar’s wildlife. With approximately 270,000 Malagasy people living within or near the protected area complex, these forests are inextricably tied to the livelihoods of the Malagasy people across this region (Holmes 2007). To that end, it is estimated that the forests of the Masoala-Makira landscape alone protect regional watersheds providing water to more than 100,000 people (WCS 2004). Unfortunately, human demands on the landscape continue to increase with a steady annual population growth of approximately 3% each year (CIA 2010). As is the case for the majority of Madagascar, the greatest threat to these forests comes from slash and burn agriculture. This unsustainable farming practice of the Malagasy, known as tavy, has had detrimental effects on the endemic and diverse flora and fauna throughout the island. It is estimated that the Antongil Bay landscape is estimated to lose 1,500 ha of primary forest per year due to these unsustainable farming practices (Meyers 2001). In addition to these untenable farming practices, recent studies within this region have also highlighted strong human-wildlife conflicts, including human encroachment of forest habitat, predation of livestock, and extensive use of carnivores, lemurs, and small mammals as bushment (Golden 2009). The wide-ranging effects of these human-wildlife conflicts have been little studied and, therefore, their role in altering the population dynamics of these taxa is unknown. Additionally, due to the recent political turmoil in Madagascar, an increase in illegal, and now government-sanctioned, logging has been reported for regions of Masoala (Butler 2010). The effects of these ongoing and new disturbances on carnivores, lemurs, and other wildlife are not known and desperately need investigation. Effects of Poaching across Makira Protected Area A paucity of data exists on poaching and its effects on the wildlife of Madagascar. The reports that have been made available (see Golden 2009 for review) are often species specific (Goodman 2006), region specific for areas outside of rainforest habitat (Goodman and Raselimanana 2003; Bollen and Donati 2006), or are anecdotal in nature (Garcia and Goodman 2003). Studies investigating poaching and bushmeat use throughout Africa have shown the detrimental effects these unsustainable practices can have on wildlife populations (Bowen-Jones and Pendry 1999; Cowlishaw et al. 2005; Fa et al. 2005; Brashares et al. 2010). Despite the known existence of bushmeat use throughout Madagascar, little effort has been made to better understand the risk of hunting as it applies to wildlife conservation in Madagascar. Recently C. Golden (2009) completed the first long term study of bushmeat hunting and use in Madagascar. A report from Golden’s pilot surveys detailed the annual harvest rates for 14 villages across the Makira Protected Area and estimated sustainability based on these rates. Golden reported that 23 species of mammal were hunted for consumption, including four species of lemur which are being hunted unsustainably (Eulemur albifrons, Hapalemur griseus, Varecia variegate, and Indri indri). Additionally, Golden found that C. ferox is hunted well beyond sustainability with a maximum harvest rate estimated at 0.26 individuals/km/yr. The results of this pilot survey are alarming and highlight the growing need for studies investigating the effects of poaching, in addition to human encroachment and fragmentation, on Madagascar’s wildlife populations. This study aims to address these invasive pressures and provide the first comparison of density/occupancy estimates for populations across sites with high poaching rates and those with reduced poaching rates. Selection of Study Sites The factors affecting carnivore population densities can be wide ranging and are often difficult to quantify. These factors may include the aforementioned poaching, human encroachment and human related diseases, reduction in prey biomass, and, most importantly, habitat degradation and fragmentation. The combination of these invasive pressures, resulting in demographic and environmental stochasticity, often causes detrimental effects to carnivore populations (Sunquist and Sunquist 2001). Unfortunately, investigating each invasive pressure individually, so as to understand the role each plays on carnivore population dynamics, is not possible in the case of this research project. As a result, this study will investigate the effects of these invasive pressures simultaneously across “fragmented sites” within the Masoala-Makira landscape. This study will investigate the role these pressures have on carnivore population dynamics by selecting sites that have consistently high to moderate rates of poaching, human encroachment, and habitat fragmentation for comparison to relatively non-impacted sites. As habitat degradation and fragmentation are known to negatively affect wildlife populations, a need exists to accurately measure/quantify the level of threat for a given area based on landscape metrics and vegetative structures. However, a great difficulty exists in attempting to quantify or estimate biological diversity with landscape and/or vegetative features. For years conservationists have used vegetative characteristics (cover type, stand age, core area, etc.) as a surrogate to estimate biodiversity and design reserves (Hunter et al. 1988; Cushman et al. 2008). A recent study by Cushman et al. (2008) addressed the assumptions required for the use of vegetative surrogates to define habitat and measure wildlife abundance and/or viability. They determined that landscape pattern is a poor indicator of wildlife abundance as habitat is often perceived differently by various species and the scale of these measurements can produce dramatically different results (Cushman et al. 2008). As a result, to adequately investigate the effects of habitat fragmentation on Madagascar’s carnivores this study will classify habitat for each study site on a fine scale and incorporate vegetative and landscape characteristics that are known to have a strong impact on carnivore populations (e.g. core area, connectivity, edge). The Masoala-Makira landscape provides a unique backdrop for this study. These two protected areas connect across the Antongil Bay to form the largest protected complex in Madagascar. A great emphasis has been expressed concerning the level of connectivity between these two areas, particularly the region just north of Maroantsetra (Figure 1). The loss of connectivity between these two areas would certainly have significant long term consequences to carnivore populations and metapopulations occupying this region, as well as countless other endemic wildlife populations and metapopulations. To investigate the level of connectivity and how fragmentation in these expansive corridor areas is affecting carnivore populations, I will select sites that form an arc across this landscape (see Figure 2 for example of site layout). This particular layout with allow us to determine if carnivores are migrating and/or occupying these forests across these expansive corridor areas. To compare carnivore populations across fragmented and non-fragmented forests I will compare camera trapping data from 3 fragmented and 3 non-fragmented sites. However, to provide consistency among sites and ensure replicable sites are used I will select six locations across the Masoala-Makira landscape to analyze for consideration as “fragmented” study sites and six sites across the landscape for consideration as “non-fragmented” sites. I will use satellite imagery of the Masoala-Makira landscape provided by WCS to select the twelve sites (6 “fragmented” and 6 “non-fragmented”) for consideration. The criteria for selecting these twelve sites include: 1) data from Golden (2009) on poaching rates; 2) distance to nearest village; 3) forests with low levels of connectivity; 4) pilot study data from WCS; and 5) logistical constraints due to travel to field sites. Once the sites are chosen for consideration I will analyze each site in Multispec Windows application and classify all features in the study area as rainforest, degraded (matrix), cultivated land, or water. Once classified I will analyze each study site in program FRAGSTATS to measure habitat features and quantify the following metrics: Landscape, Edge, Core, and Connectivity. Finally, once all features for each of the twelve study sites have been classified and all variables measured, I will run a fuzzy cluster analysis (given the number of variables with continuous data) to determine the three most similar “fragmented” sites and the three most similar “non-fragmented” sites. Additionally, I will conduct a Discriminant Function Analysis on these data to check for misclassifications and to determine which variables were most significant for assigning classes. These methods for the selection of sites will provide replicable study sites for my final comparison of fragmented vs. non-fragmented forests. I will complete these analyses and select sites before leaving for the field this May 2010. METHODS Objective I Estimate density, activity patterns, and occupancy rates of fosa (C. ferox), fanaloka (Fossa fossana), falanouc (Eupleres goudotii), ring-tailed mongoose (Galidia elegans), broad-striped mongoose (Galidictis fasciata), brown-tailed mongoose (Salanoia concolor) and invasive species, including the small Indian civet (Viverricula indica), domestic dogs, and feral cats, within and among 3 fragmented and 3 non-fragmented sites.
Camera Trapping Methods and Grid Spacing Numerous studies have utilized camera traps to estimate population densities for tigers (Karanth 1995; Karanth and Nichols 1998; Karanth et al. 2004; Linkie et al. 2006), jaguars (Kelly 2003; Wallace et al. 2003; Silver et al. 2004; Maffei et al. 2004; Soisalo and Cavalcanti 2006), ocelots (Trolle and Kery 2003; Maffei et al. 2005; Di Bitetti et al. 2006; Dillon and Kelly 2007, Dillon and Kelly 2008), pumas (Kelly et al. 2008), and Madagascar’s carnivores (Gerber et al. 2009). Camera trapping methods used for studying rare and elusive species is highly advantageous at it provides a non-invasive approach for gathering data on activity, species richness, occurrence, and/or abundance for countless species in a variety of habitat types, including grasslands (Silveira et al. 2003, Heilbrun et al. 2006), deciduous forests (Hackett et al. 2007, Kelly and Holub 2008), alpine forest (Xu et al. 2008, Jackson et al. 2006), tropical dry forest (Karanth et al. 2004), and rainforest (O’Brien 2008, Gerber et al. 2009). I will use the standardized protocol employed by these remote camera surveys to examine carnivore population ecology within fragmented and non-fragmented forests across the Masoala-Makira landscape. At each study site I will set up remote sensing cameras within a fixed grid across the landscape. These cameras utilize motion- and heat-sensing infrared technology which allows them to detect both movement and heat from passing wildlife. To maximize the number of carnivore captures, cameras will be placed along established trails at each study site. However, if existing trails are not present (as will likely be the case for non-fragmented sites) new trails will be cut for camera placement. Studies have highlighted the importance of using established or permanent trail systems for surveys (Maffei et al. 2004; Dillon and Kelly 2007; Sanderson 2004); therefore, efforts will be made to minimize new trail use when possible. Any new trails that are cut will not be monitored for one week to reduce sampling bias due to disturbance. No bait or lures will be used to attract animals to camera stations. Each study site will have 25 camera stations, operational 24h/day, consisting of two cameras mounted on opposing sides of the trail to ensure capture of both flanks of the animal for positive identification. Likewise, cameras will be off-set to reduce mutual flash interference. Cameras will be mounted ~20-30cm off the ground to maximize capture probability based on C. ferox, F. fossana, and co-occurring carnivore measurements (Hawkins 2003; Garbutt 2007; Kerridge et al. 2003). This study will deploy three different camera types for surveys: Reconyx PC85, DeerCam DC300, and Moultrie D40. The Reconyx and Moultrie cameras use digital technology, whereas the DeerCam cameras require standard film development. Each camera station (consisting of two opposing cameras) will utilize a combination of these three camera types so as to offset the number of malfunctions and poor capture quality of the less efficient cameras. Each camera will be checked every 10-14 days to change batteries and ensure proper functioning. Additionally, during routine camera checks, we will replace memory cards for digital cameras (Recoynx and Moultrie) and rolls of film (DeerCam) as needed. All photographs will record the date and time of day for each capture for data analysis. Cameras will operate for a minimum of 60 days at each study site to maximize number of captures for statistical analysis while avoiding violation of the closed population assumption of the mark – recapture methods (Besbeas et al. 2002; White et al. 1982). Studies have highlighted the importance of camera spacing for the estimation of population variables, particularly when animal movement is used for density estimation (e.g. ½ MMDM and MMDM) (Stickel 1954, Tanaka 1980, Wilson and Anderson 1985, Wegge et al. 2004, Dillon and Kelly 2007). The appropriate choice for camera trap placement can be particularly difficult to determine because a tradeoff exists between using close camera spacing to improve estimates of animal movement and using larger camera spacing to cover sufficient area for density estimates. Appropriate camera spacing is imperative so as to not provide erroneous estimates of population density (Wegge et al. 2004; Dillon and Kelly 2007). The appropriate camera spacing, and ensuing density estimation, should be evaluated based on the home range of an individual focal species (Dillon and Kelly 2007, Sanderson 2004). As a result, camera spacing for this study will result from the estimated home range of F. fossana, as this is the only Madagascar carnivore species with distinct markings allowing for individual recognition. I estimated F. fossana home range to be 0.75km from MMDM estimates of Gerber et al. (2009) and Farris et al. (in prep), as well as results from a telemetry study by Kerridge et al. (2003). As a result, I will establish a 5x5 point grid with cameras stations spaced at 0.75km across the landscape (Figure 3). This camera spacing is used to ensure no individual F. fossana within the camera grid has a zero probability of detection. This camera grid arrangement will provide a total survey area of just over 20km² (Figure 3). Population Variables: Comparison of Density Estimation Techniques For analyses of camera trapping data I will provide density estimates for F. fossana and V. indica (the only two individually identifiable species), minimum abundance estimates for C. ferox, occupancy estimates for all captured carnivores and invasive species (domestic dogs and feral cats), and trap success for carnivores, small mammals, invasive species, and humans. The methods to be used for estimating these population variables are given below. I. Traditional Density Estimation (Buffer Estimates): Individual F. fossana will be identified based on their distinctive coat markings as has been proven effective in pilot studies by Gerber et al. (2009) and Farris et al. (in prep). Additionally, individual V. indica will be identified for fragmented sites providing a sufficient number of captures are available for analysis. Attempts will be made to identify individual C. ferox using the methods of Kelly et al. (2008) for subtly marked carnivores. Markings used for identification of C. ferox will be organized into three categories: obvious marks (scars, tail kinks, and ear nicks), less obvious marks (healed scars, tail-tip coloration, and rings), and subtle markings (undercoat and underside of leg coloring and body shape). To prevent bias and increase the precision of the results I will use multiple investigators for identification of individual F. fossana, V. indica, and C. ferox. To provide abundance estimates for each individually identifiable F. fossana, V. indica, and C. ferox, I will create a capture history for each individual. A “capture” is defined as a distinct individual captured on camera within a 30 minute time period, regardless of number of photos. Each capture history will consist of a value of 1 for each sampling occasion (24 hr period) in which the individual is captured and a 0 for each occasion where the individual is missed. As abundance/density estimation assumes demographic and geographic closure for the population, I will test for closure in program CAPTURE (Otis et al. 1978; White et al. 1982; Rexsted and Burnham 1991) and/or program MARK (White and Burnham 1999). The closure test in program MARK will have preference (if sample sizes are sufficient) as this approach tests the assumption by comparing the data’s fit to an open population model versus a closed population model (White and Burnham 1999). This often requires manipulation of data in which capture histories are collapsed into two or three day events (Dillon and Kelly 2007). Once closure has been verified statistically I will use program CAPTURE (Otis et al. 1978; White et al. 1982; Rexsted and Burnham 1991) or program MARK (White and Burnham 1999; Cooch and White 2006) to provide abundance estimates. Program CAPTURE will be used if sample sizes are low as this program uses a discriminant function model selection algorithm that is not as robust as other estimation procedures. Program MARK uses model selection with Aikaike Index Criterion values and allows model averaging to include model uncertainty (White and Burnham 1999; Cooch and White 2006). This abundance estimation technique reduces bias and is generally viewed as the best approach if sample sizes are sufficient. Programs CAPTURE and MARK incorporate sources of variation and generate models to estimate abundance based on the structure of the mark-recapture data (White et al. 1982; Rexsted and Burnham 1991; White and Burnham 1999). The models (sources of variation) include: Mo the null model (constant capture probability), Mt time variation in capture probability, Mh individual heterogeneity in capture probabilities, Mb behavioral response after capture, and combinations of multiple factors (Mth, Mhb, and Mtb) (Rexsted and Burnham 1991; Karanth 1995; White et al. 1982). Based on the model selection algorithm calculated, I will select the appropriate model for abundance estimates (Otis et al. 1978; Rexsted and Burnham 1991; White and Burnham 1999). To estimate density (D = N/A; where N is abundance and A is area where animals are found) for each study site I will use the abundance estimates calculated above and divide by the effective area sampled (km²). As camera spacing has been arranged based on home range sizes of F. fossana (and the estimated home range of V. indica), I will only be able to provide density estimates for these two species. The effective sampling area for this grid arrangement will not allow density estimates for C. ferox, however we can still calculate minimum numbers for this species for comparison across sites. The effective trap area for this density estimation technique will be calculated based on the ad hoc approach of area covered by the cameras within the study site plus a buffer around the area equal to one half the mean maximum distance moved (½ MMDM) by each recaptured carnivore (Wilson and Anderson 1985; Karanth and Nichols 1998; Kelly 2003). I will create circular buffers around each camera trap with a radius equal ½ MMDM and dissolve these buffers and calculate total area of the buffered cameras (Kelly 2003; Silver et al 2004; Dillon and Kelly 2007). Further, as has been done for small mammal sampling (Wilson and Anderson 1985) and other large bodied carnivores (Parmenter et al. 2003; Trolle et al. 2007; Gerber et al. 2009; Tioli et al. 2009), I will also provide estimates using the full mean maximum density moved (MMDM) as a surrogate for effective sampling area. Standard errors for each density estimate will be calculated following Nichols and Karanth (2002). The estimates produced by these two approaches (½ MMDM and full MMDM) will be compared to the results presented by Gerber et al. (2009) and Gerber (in prep) for F. fossana at Ranomafana National Park. As F. fossana, V. indica, and C. ferox are not identifiable without error, I will test for variation in the numbers and population estimates, among investigators and among sites, using the Scheirer-Ray-Hare extension of the Kruskal-Wallis test (Kelly et al. 2008). Likewise, I will use a 2-factor analysis of variance (ANOVA) to determine if the ½ MMDM and full MMDM estimates varied more among investigators within sites than among sites (Kelly et al. 2008). I will then use a Tukey-Kramer honestly significant test to determine if F. fossana, V. indica, and C. ferox move more within one site compared to others. To determine if densities varied among investigators and sites I will use a 2-factor analysis of variance (ANOVA) to test the log-transformed density estimates (Kelly et al. 2008). II. Maximum Likelihood Methods and Bayesian Hierarchical Approaches to Density Estimation Much debate still exists concerning the use of MMDM techniques for the estimation of population density (Soilsalo and Cavalcanti 2006; Maffei and Noss 2008; Dillon and Kelly 2007, 2008). As a result, density estimation techniques are now available that do not rely on buffer values estimated from individual animal movement to determine effective sampling area. Program DENSITY (Borcher and Efford 2007; Efford 2007; Efford et al. 2004) contrasts with the MMDM technique in that it uses a spatially explicit method, based on camera trap locations, to estimate effective trap area. This approach uses the trap detection of each individual to relax the biases associated with trapping grid edge effects, as well as the controversies associated with the MMDM technique (Efford 2007). Density estimation by MMDM methods is also problematic because of individual heterogeneity in capture probabilities, lack of geographic closure, movement of camera traps, and presence of potential ‘holes’ in the camera grid especially when camera malfunctions occur. To address these concerns effectively, Royle et al. (2009) have developed an approach based on Bayesian analysis of the hierarchical model using Markov chain Monte Carlo (MCMC) simulation. This method uses data augmentation to treat a “missing data” problem where home range centers are the missing data (Royle and Young 2008) and is implemented in Program SpaceCap. I will use these newer approaches to density estimation as well as the older approach for comparison to past work. III. 2-Dimensional Gas Model A relatively new approach exists for estimating density of captured animals. This approach does not require individual recognition for analyses and, as a result, will be used on all captured animals in this study. This technique, known as the 2-dimensional gas model (Rowcliff et al. 2008), uses photographic rates to compare with density estimates for the species being analyzed. Rowcliffe et al. (2008) treat contact rates between cameras and animals as a physics problem and model the underlying process using a gas model (or encounter model) to scale trapping rates with density. This approach eliminates the requirement for individual identification in camera trap density estimation surveys, but does require accurate measurement of model parameters (e.g. day range and group size). The application of this estimation technique will be contingent upon radio telemetry data collected on F. fossana and C. ferox (see objective 3). IV. Comparing Estimates with Ranging Patterns from Telemetry Data Camera trapping has become an invaluable technique in recent years to estimate densities of numerous elusive animals (mainly felids); however, debates still exist concerning proper camera trapping protocol. For example, recent studies have shown that the correlation in camera spacing, length of trapping period (Dillon and Kelly 2007; Wegge et al. 2004), small survey areas (Maffei and Noss 2008), and lack of knowledge on actual home ranges (Soisalo and Cavalcanti 2006) may cause an overestimation of density. Overestimating the density of a vulnerable carnivore with limited distribution, living with increasing environmental pressures could pose serious issues for effective conservation strategies. Individual F. fossana and C. ferox will be captured and radio collared to investigate ranging patterns and construct home ranges (see Objective 3). The availability of these data will allow the incorporation of radio telemetry data into traditional density estimation techniques. The ½ MMDM and full MMDM calculations (see Traditional Density Estimation) are meant to be a surrogate for home range radius. However, without information on home range size it is impossible to know if this MMDM approximates this radius. The determination of home range and daily movement patterns of C. ferox and F. fossana through the use of radio telemetry will enable me to evaluate whether the ½ MMDM or full MMDM buffer provides an adequate surrogate for the home range radius of these two carnivores within rainforest habitat. As evidence, other studies using a combination of mark-recapture and radio telemetry data have found more accurate estimates of animal movement and recapture rates; however, a minimum of 25 individuals is highly recommended for proper analysis and comparison (Powell et al. 2000). Additionally, with the data collected from this portion of the study I can explore the approach of Rowcliffe and Carbone (2008) to estimate densities from trapping rates. While camera trapping mark-recapture has now grown in popularity, the technique is somewhat limited because of its requirement of recognition of individuals with identifiable coat markings. Rowcliffe and Carbone (2008) have developed a method that negates the requirement of individual identification by adding a component that “linearly scales trapping rate with density, depending on two key biological variables (average animal group size and day range).” In other words, using the data gathered on average daily ranges for C. ferox and F. fossana from the radio telemetry study, I can create alternate density estimates to compare with the mark-recapture analysis. To date, no other study has used both protocols (standard mark-recapture and non-recognition method) on a naturally occurring population to compare density estimates. This will provide alternate density estimation methods that could then be used on other non-marked species captured by remote cameras. The implications of these findings will have a major impact on future studies utilizing camera traps to estimate population densities of carnivores throughout Madagascar. The results of this study will, therefore, provide guidelines for future camera trap studies conducted on endemic Malagasy carnivores. Population Variables: Occupancy and Trap Success While we can determine density through traditional camera-trapping for individually marked species (F. fossana, V. indica), density cannot be determined for prey animals or non-target species that lack distinct markings (except perhaps through newer methods currently being developed as mentioned above). Therefore, I will use occupancy estimation and trap-success estimates to provide critical information on species occurrence and activity levels. Occupancy: Occupancy estimation uses a likelihood based model approach to estimate detection/non-detection of a given species across a site, rather than using capture histories for individuals (MacKenzie et al. 2002, Nichols and Karanth 2002, MacKenzie et al. 2003, MacKenzie 2005). Detection probabilities are modeled to evaluate species presence/absence. This approach provides a more accurate assessment of true absence, as opposed to simple non-detection across the study site. However, the results are heavily dependent upon survey effort and, therefore, surveys must be consistent so as not to create biases in estimates of patch use when comparing across sites. Occupancy, or presence/absence data, can be analyzed statistically within program PRESENCE (MacKenzie et al. 2002; MacKenzie et al. 2003) to select the appropriate model for parameter estimates. Absence of a particular species will be determined if no individuals are captured within that survey area (camera site) during the full survey period (60 days). Based on the data I will be collecting it is possible to model occupancy as a response to landscape metrics, microhabitat variables, and activity of co-carnivores following Davis et al. (In press). I will model occupancy in program PRESENCE for all carnivore species (C. ferox, F. fossana, G. elegans, G. fasciata, E. goudotii, and S. concolor), small mammal species, and invasive species (domestic dog, feral cat, and V. indica) to determine what is influencing presence/absence of each of these species. I will evaluate which landscape and habitat variables most influence presence/absence of Madagascar’s endemic carnivores across fragmented and non-fragmented forests. Further, I will evaluate how presence or absence of C. ferox affects co-occurring carnivores, small mammals, and invasive species. Trap Success: I will determine trap success for carnivores, small mammals, invasive species, and humans by dividing the number of captures for each species by the total number of trap nights multiplied by 100 (e.g. captures/100TN) (Dillon and Kelly 2007; Kelly et al 2008; Kelly and Holub 2008). A single capture event will be all photographs taken of an individual within a 30 minute period (Di Bitetti et al. 2006). Trap nights refer to a 24 hour period in which at least one camera within a given camera station is functioning. Trap success will be used to assess the activity level of each species at each camera station and across forest types. In addition, I will model the trapping rate as a function of site specific habitat variables and trapping rates of other species to determine what influences trapping rates of species (Davis et al. in press). Pilot Camera Trapping Study (2008-2009) The Wildlife Conservation Society initiated a pilot camera study across Makira Protected Area in 2008. Their work provided camera trapping data from three sites, including Anjanaharibe, Soavera, and Vinanibe. I have recently applied these population estimation techniques described above to the capture data from these three sites, including ½ MMDM and full MMDM for density estimation in program CAPTURE, maximum likelihood estimation in program DENSITY, occupancy estimation in program PRESENCE, and trap success from number of captures per trap night. Using these data I have been able to provide density and occupancy estimates for F. fossana, as well as trap success for carnivores, small mammals, invasive species, and humans. The successful application of these estimators has confirmed their efficacy for the capture data I have described in this study. Moreover, the results of these preliminary analyses concur with the findings of Gerber et al. (2009) for F. fossana density and occupancy across rainforest habitat. Finally, the data collected from these three sites (Anjanaharibe, Soavera, and Vinanibe) will likely be incorporated into some of the analyses for my dissertation. The addition of these data will increase the sample size by adding three new replicates (camera study sites); however, no habitat sampling or lemur surveys have been conducted for these three sites. Therefore, the application of these data for all analyses I plan to conduct is not possible. Objective I: Questions and Hypotheses Are the four density estimation techniques congruent in their estimates of F. fossana capture data? H0: All four density estimation approaches are congruent. Ha: The density estimation approaches are not congruent.
Objective II Determine factors influencing carnivore densities, activity, and occupancy within, and across these study sites, including landscape characteristics (using GIS), microhabitat features, climatic conditions, prey species (namely lemurs), and human presence.
Within Study Sites: Habitat Analyses In addition to above analyses carnivore density, occupancy, and/or trap success will be analyzed as a continuous response variable to multiple habitat characteristics at both the landscape and the microhabitat scales. An understanding of how carnivore activity is correlated to specific habitat variables is vital for protection of these species and their environment. For example, in Virginia, USA, bobcat presence at camera stations is highly correlated with percent of deciduous forest type while bobcat trap success across camera sites is positively linked to distance away from high-use roads (Kelly and Holub 2008). Minimal information is available concerning Madagascar’s carnivores and habitat associations. A brief account of the data currently available is given here. C. ferox are known to occupy all types of intact forests throughout Madagascar (Hawkins 2003); however, it is not known to which type of habitat they prefer or what habitat features they are most sensitive to. F. fossana are known to feed on a variety of foods and may show preference for aquatic taxa (Kerridge et al. 2003; Albignac 1971). In fact, in a study conducted by Albignac (1971) 15 of the 18 individuals captured were found next to a water source. G. elegans is quite common and found in all intact forests, preferring primary forest (Goodman 2003; Garbutt 2007). Unfortunately, the recent work by Gerber (in prep) has shown a strongly negative relationship with G. elegans and feral cats. G. fasciata occurs at low densities within lowland and mid-altitude forests, typically below 700m (Garbutt 2007). S. concolor is an extremely rare species with a very limited distribution (Hawkins et al. 2008); a paucity of information exists concerning the natural history for this species, however, it is believed to be restricted to primary forest (Hawkins et al. 2008). Finally, the introduced species V. indica is known to exclusively inhabit degraded habitat and prefer forests around villages (Garbutt 2007). I will explore these habitat associations further and determine how activity level and density vary and co-vary both within and across study sites. This information is vital to effective management plans designed to protect the little known carnivore community of Madagascar. I. Landscape scale: To compare trapping rates of carnivores to landscape metrics within study sites, I will record the location of each camera station using a handheld GPS unit and plot these locations on available GIS maps for Masoala, Makira, and corridor areas. I will use available GIS layers to compare carnivore trap success at each camera station to, for example: nearest roads, populated areas, habitat type, nearest water sources, slope, and elevation. The GIS data I have available for comparison includes: 1) vegetation classification for entire island of Madagascar; 2) DEM for Masoala-Makira landscape; 3) satellite imagery of Masoala-Makira landscape, 4) park boundary for Masoala and Makira; 4) location and population sizes for villages across Masoala and Makira regions; and 5) location of high poaching activity. In addition to these comparisons of carnivore activity to landscape metrics, I will use the metrics constructed in FRAGSTATS for each of my survey areas (as discussed in Selection of Sites) to characterize the landscape for 100, 250, and 500m radii surrounding each camera station following Kelly and Holub (2008), (Figure 4). Such variables will include % habitat cover type, % water, length of river, length of road, etc. Discrete values for each of these variables will be calculated for each station at each of the 3 radii. I will then use linear regression techniques following Davis et al. (In Press) to build models based on apriori hypotheses comparing carnivore trapping rates to habitat variables across the camera stations within each site. I will use AIC model selection procedures to select the best models and model-average as needed. For example, I would expect C. ferox, F. fossana, G. elegans, S. concolor, and G. fasciata trapping rates to increase with canopy cover, percent water, distance to human habitation and to decline with proximity to human habitation. I will also use logistic regression to compare these habitat variables at sites where target carnivores are present to sites where target carnivores are absent. To describe general climatic conditions for each study site I will measure rainfall and temperature (minimum and maximum) every day at three camera traps across each study site (center most station and two stations at furthermost corners of survey area) to determine if camera trapping rates are influenced by weather conditions. II. Microhabitat scale: The landscape scale analyses may be too coarse to determine finer scale habitat selection by Malagasy carnivores. To characterize habitat features on a finer scale I will collect microhabitat data surrounding each camera station. Microhabitat data collection will follow Davis (2007) and Gerber (2008) by walking a 100 meter transect in three directions starting at the camera station and will classify the canopy cover and type every 10 meters (starting 10 meters from the camera station) resulting in a total of 30 canopy points (10 in each of 3 directions) for each camera station (Figure 5). I will take additional measurements at six points, located along these three “canopy transects” where we will also estimate canopy height with a clinometer and sample the understory using a point intercept technique. I will estimate tree density using number of trees counted and area surveyed through the point-quarter technique which will include: distance to nearest tree, diameter breast height (DBH), live or dead, and its taxonomy (to species where possible). I will also measure canopy height directly above the camera station itself. To sample understory thickness, I will set up 40 meter long transects centered on and running perpendicular to the canopy transect (Figure 5). I will sample every two meters by dropping a two meter long pole, recording all vegetation in contact, and classifying all recorded individuals into one of three height intervals (0-0.5m; 0.5-1m; 1-2m). Percent understory cover for each height interval, on each transect, will be calculated as the number of points in contact with vegetation divided by total points recorded, multiplied by 100. I will compare mean values for each height interval and measure variation for heterogeneity of understory cover. To conduct analyses on landscape and habitat variable data I will follow the methods of Davis et al. (In press), as described above. Using occupancy and/or trap success of each carnivore as a measure of activity, I will model captures to landscape and habitat variables to characterize habitat use for each species. I will examine which habitat variables are most important for each species by use of “global habitat” models (Davis et al., In press). I will rank each model by AIC scores to determine which habitat variables are most meaningful to each species. Within Study Sites: Lemur Surveys Researchers have shown that lemurs can make up more than 50% of C. ferox diet (Rasolonandrasana 1994; Hawkins 1998). Moreover, lemurs play an integral role in the endemic ecosystems throughout Madagascar and are vulnerable to widespread forest degradation. As a result, I will compare lemur density/abundance with C. ferox density and trap success to better understand the predator-prey dynamics across fragmented and non-fragmented forests. As the largest extant carnivore in Madagascar, C. ferox plays a dynamic role in this ecosystem as a significant lemur predator. Currently, there is strong evidence that C. ferox impacts both lemur behavior and population dynamics in forests across Madagascar, particularly in rainforest habitat (Irwin et al. 2009; Wright et al. 1997; Karpanty and Wright 2007). As C. ferox and lemurs are increasingly forced into isolated fragments of forest, natural or exacerbated predation rates by C. ferox may negatively impact lemur populations which are simultaneously being limited by declining habitat quality and human encroachment. For example, Irwin et al. (2009) showed how C. ferox killed an entire group of diademed sifakas (Propithecus diadema) at the Tsinjoarivo forest, extirpating this species from a forest fragment. In addition to elucidate the ecology of the carnivore community, my study will investigate how the dynamics of C. ferox-lemur interactions may change across fragmented and non-fragmented forests. To determine lemur densities within each study site, I will use the methodology for estimating primate densities from previous studies (Peres 1999; Buckland et al. 1993; Burnham et al. 1980; Sterling and Rakotoarison 1998; Whitesides et al. 1988; Feistner and Schmid 1999; Merenlender et al. 1998; Fashing and Cords 2000). I will use straight-line compass line transect method (NRC 1981) and distance estimation techniques (Buckland et al. 2001) to measure lemur densities using two camera stations within each camera trapping grid as starting points for those transects (see Figure 6). I will use previous WCS established survey transects or will establish line transects, 4km in length, from the central most camera station of each camera grid and at one other randomly selected camera stations within the camera grid. Two line transects will be established in randomly selected directions at both camera stations to conduct lemur surveys for a total of 4 transects per camera grid (Peres 1999; Whitesides et al. 1988). However, to reduce impact on forest and minimize time constraints, I will attempt to use any available existing trails when available. I will measure and mark each lemur transect in 50m intervals. All newly cut transects will remain un-sampled for at least one day to minimize bias due to disruption caused by establishing each transect (Peres 1999). I will move slowly (rate of approx. 1.5km/hr) stopping periodically (100m) to watch and listen for lemurs. I will spend no more than 15 minutes at each sighting. Surveys will not be conducted during periods of heavy rain as this minimizes visual detection and hinders ability to detect calls elicited by lemurs (Sterling and Rakotoarison 1998; Feistner and Schmid 1999; Peres 1999). Finally, I will also search for secondary signs of lemur presence such as nest sites (Cheirogaleus and Daubentonia) and feeding sites (Daubentonia and Hapalemur). Further, I will investigate tree holes for signs of habitation or presence of nocturnal lemurs. Each transect will be surveyed 10 times (5 diurnal and 5 nocturnal) for a total distance on each one-way transect of 20km (providing the division in diurnal and nocturnal surveys). This corresponds to a total of 40km for the two one-way transects at each camera station and a total of 80km for two-way sampling of both transects at each camera station (160km total including diurnal and nocturnal). Additionally, this would correspond to a total of 160km for each camera trapping grid (2 transects at 2 stations per grid). Systematic observations will be made during both day and night. Diurnal censuses will take place during peak activity times for lemurs, during morning (0600-1100hr.) and afternoon (1500-1730hr). Likewise, nocturnal surveys will be conducted at night (1830-2300hr) during peak activity times (Feistner and Schmid 1999). During nocturnal surveys we will use dim headlamps to search for eyeshine from the tapetum lucidum. Each trail will only be surveyed once for any given day (either diurnal or nocturnal). In other words, multiple surveys will not be conducted for a given transect within a 24 hour period. For all observations we will note species, time of day, group size, behavior, position on trail/transect, elevation, distance from observer, angle, height from ground, detection cue, and habitat type (Whitesides et al. 1988; Peres 1999; Feistner and Schmid 1999). Sight distance (SD) and angle will be converted into perpendicular distance (PD) for data analysis (Burnham et al. 1980; Buckland et al. 1993, 2001; Peres 1999). I will use range finders to measure distance. Observers will only move minimal distances off transects (10m max) if determined necessary (Peres 1999). Perpendicular distances will be grouped for analysis. Transect data will be analyzed using program DISTANCE (Laake et al. 1991; Buckland et al. 1993). However, program DISTANCE requires a minimum of 30 observations per lemur species for analysis and recommends 60-80 observations for more accurate estimates (Buckland et al. 2001). As a result, I plan to use alternative abundance estimates for lemur species that I do not have sufficient numbers to meet the minimum recommendation for program DISTANCE. For those lemur species that have fewer than 30 observations I will divide the total number of observations by the total survey area for a minimum density estimate within that site. Survey area, in this case, will be calculated by multiplying total survey length by trail width (for example see Johnson and Overdorff 1999). Trail width will be calculated according to Whitesides et al. (1988) using the perpendicular distance estimates from group to transect. However, if constrained by time then sighting rates will be used for analysis as opposed to perpendicular distance estimates. Within Study Sites: Species Interactions The calculation of density, occupancy, and/or trap success (activity) estimates for carnivores, small mammals, invasive species, and humans will allow investigation of interactions among carnivore species, predator-prey, native-invasive, and human-wildlife dynamics. Understanding the relationship among each of these comparisons will have a significant impact on our attempts to better understand and protect this complex and endemic ecosystem. For example, Davis et al. (In press) showed ocelot trap success across camera stations is positively associated with jaguar trap success but not puma trap success in Belize Central America. Recently, Gerber (In prep) found a strong negative relationship between G. elegans trap success and feral cat trap success. For the carnivores of Madagascar almost no information is available concerning interactions among carnivores, their prey, and humans. As described above, I will model occupancy in program PRESENCE for all carnivore species to investigate influences on carnivore presence/absence. More specifically, I will determine how human activity (trap success) influences carnivore presence/absence. Additionally, I will investigate how carnivore presence/absence influences small mammal and invasive species activity. To explore the possibility of mesopredator release, I will determine how presence/absence of C. ferox influences co-occurring carnivore activity. Across Study Sites: Comparing Fragmented and Non-Fragmented Sites Habitat fragmentation can be evaluated based on four criteria: loss of habitat, loss of quality, loss of connectivity, and loss of continuity (Hanski 2005). These indicators/values of fragmentation vary across species and habitat type. They are defined here according to C. ferox populations across the Masoala-Makira landscape. Habitat loss refers to the amount of core forested area, vital for the reproductive fitness of C. ferox, lost over a given amount of time. Loss of quality refers to the impact on C. ferox’s ability to produce and maintain a viable population within a core forested area. Further, quality includes resource requirement for maintaining this level of viability. Loss of connectivity refers to the impact of habitat loss on C. ferox migration, gene flow, and heterozygosity levels across populations/metapopulation. Finally, loss of continuity refers to the effects and implications of fragmentation on a core forested area over time. While connectivity focuses primarily on a spatial scale, continuity focuses on the temporal effects of fragmentation (Hanski 2005). An understanding of how each of these factors is affecting C. ferox and co-occurring carnivore populations is critical for the long term protection of these species and this dynamic ecosystem. For example, large sections of primary forest are being lost where Masoala National Park and Makira Protected Area connect, near the village of Ambohimarina (50o 1’ 24” E, 15o 16’ 47” S), resulting in loss of connectivity between these two vast protected areas. As a result, this loss of forested area may be diminishing the reproductive fitness for C. ferox populations within this particular area. Furthermore, this loss of connectivity may be impacting migration among populations for this protected region. This negative impact on migration can eventually lead to reduction in gene flow and a decrease in heterozygosity as inbreeding becomes more likely. Overtime this could lead to extinction of C. ferox populations/metapopulations within this core area between Masoala and Makira and potentially result in mesopredator release, trophic cascading, equilibrium simplification, and even further extinctions (Palomares et al. 1995; Rogers and Caro 1998; Terbough et al. 1999; Terbough et al. 2001). A great need exists for understanding the effects of fragmentation, particularly the loss of connectivity, on carnivore populations throughout Madagascar. To provide a comparison of the effects of fragmentation on Madagascar’s carnivores and their prey I will average trap success for each species across the three fragmented sites and the three non-fragmented sites, thus providing trap success for each species across fragmented and non-fragmented forests. I will use a Student’s t-test to compare the average trap success for each carnivore species, small mammal, and invasive species across fragmented and non-fragmented forests. Additionally, I will use GIS layers to construct metrics of habitat fragmentation for each of the six study sites, such as: amount of edge habitat (study site size/circumference), level of protection, distance from human habitation, population size of nearest village, land use patterns surrounding each study site, amount of human/wildlife conflict, density of roads and trails, etc. I will examine whether high fragmentation and human activity leads to lower densities and trap success of target species across study sites. This analysis will also shed light on which areas have the greatest human impact, whether that negatively impacts all species equally, and which areas might function as core protected areas. I will locate corridor areas near or within each study site and compare carnivore trap success across each. Further, I will identify those metrics of habitat fragmentation which are most prevalent in corridors with very low carnivore trap success. In addition, I will use the activity pattern and home range data collected on telemetered C. ferox and F. fossana (see Objective 3) to understand how fragmentation (loss of habitat, loss of quality, and loss of connectivity) is affecting the behavior and movement patterns of these two threatened carnivore species. Finally, as described in “Selection of Sites”, I will sample sites in an arc like pattern across the landscape to assess the following issues concerning connectivity across the Masoala-Makira landscape: 1) degree of connectivity among forest fragments between the two protected areas; 2) carnivore population variables within the corridors between the two protected areas; 3) lemur, small mammal, invasive species, and human population variables within the corridors between the two protected areas; and 4) C. ferox and F. fossana migration and activity patterns within corridors between the two protected areas. Objective II: Questions and Hypotheses
Is carnivore trap success associated with landscape metrics? H0: Landscape metrics are not associated with carnivore trap success. Ha: Landscape metrics are associated with carnivore trap success.
Is carnivore trap success associated with micro-habitat variables? H0: Carnivore trap success is not associated with micro-habitat variables. Ha: Carnivore trap success is associated with micro-habitat variables.
Does a relationship exist between C. ferox abundance/trap success and lemur density? H0: No measurable relationship exists between C. ferox density and lemur density. Ha: A measurable relationship exists between C. ferox density and lemur density.
Does a relationship exist between C. ferox trap success and co-occurring carnivore/small mammal/invasive species/human trap success? H0: No measurable relationship exists between C. ferox trap success and co- occurring carnivore/small mammal/invasive species/human trap success. Ha: A measurable relationship exists between C. ferox trap success and co- occurring carnivore/small mammal/invasive species/human trap success.
Does a relationship exist among carnivore trap success and small mammal/invasive species/human trap success? H0: No measurable relationship exists among carnivore trap success and small mammal/invasive species/human trap success. Ha: A measurable relationship exists among carnivore trap success and small mammal/invasive species/human trap success.
Is C. ferox and co-occurring carnivore trap success greater in non-fragmented forests compared to fragmented forests? H0: C. ferox and co-occurring carnivore trap success not affected by fragmentation. Ha: C. ferox and co-occurring carnivore trap success is affected by fragmentation.
Objective III Capture and radio collar C. ferox and F. fossana to determine seasonal home range and activity patterns for males and females, as well as collect anatomical measurements.
Capture Methods To trap, anesthetize, and radio track individual C. ferox and F. fossana I will use the methods carried out by other studies on C. ferox (Dollar 1999; Hawkins and Racey 2005), F. fossana (Kerridge, unpublished data), and ocelots (Dillon and Kelly 2008). I will use collapsible, single door Tomahawk box traps (Tomahawk Trap Co., Tomahawk, WI, USA) to capture individuals. Traps will be baited with either live chickens or chicken pieces and will be checked at least twice daily. C. ferox will be immobilized while in the trap with either a mixture of telazol (25mg), xylazine (15mg), and butorphanol (1mg) based on prior work conducted by Dillon and Kelly (2008) and Dollar (1999) or 5mg ketamine and 1mg xylazine per kg body weight based on prior work conducted by Hawkins and Racey (2005). For F. fossana I will communicate with F. Kerridge to determine the appropriate methods for immobilization. Trapping and immobilization will take place with a trained veterinarian on site. Anesthetized individuals will be weighed, measured, photographed, and fitted with an ATS medium neoprene belting radio collar (Advanced Telemetry Systems, Isanti, MN, USA). Radio collars will have spotted and striped patterns for future definitive identification of individuals captured by remote cameras after release. After individuals are measured and fitted with radio collars they will be returned to their trap and location of capture and monitored until release. Anatomical measurements to be taken include: body length, tail length, hindlimb length, hindfoot length, hindlimb circumference, forelimb length, forefoot length, forelimb circumference, chest circumference, neck circumference, canine lengths, and genital measurements (for description of measurements see Dollar, 1999). Photographs will be taken of each side of the captured individual for identification and comparisons for camera trap data. Target number of individuals will be 10 C. ferox, five within fragmented and five within non-fragmented, and 10 F. fossana, five within fragmented forest and five within non-fragmented. This issue of site fidelity may have implications on the averaging of individuals across fragmented and non-fragmented forests (Swihart and Slade 1997; Laver and Kelly 2008). As a result, I will compare results of telemetered individuals within fragmented forests with individuals in non-fragmented forests prior to averaging all C. ferox or F. fossana across all sites. Radio Telemetry and Home Range Analyses I will conduct error testing on collars from a known location to determine the average bearing error and standard deviation across the study site (Dillon and Kelly 2008; White and Garrott 1990; Millspaugh and Marzluff 2001). I will locate each radio collared individual using ATS R4000 receivers and Yagi 3-element directional antennas. I will take simultaneous triangulation locations one to two times daily and at least 4 hours apart to avoid autocorrelation (Dillon and Kelly 2008; White and Garrot 1990). The issue of serial autocorrelation and time to statistical and biological independence is expected to be species specific and will be assessed for both C. ferox and F. fossana (Laver and Kelly 2008). I will enter the standard deviation of the bearing error into program LOAS (Location of a Signal) or Program LOCATE to estimate each telemetered individual location and error ellipse (Dillon and Kelly 2008). To determine home range for each C. ferox and F. fossana I will separate each individual location into either hot-wet season (Nov-April) or cool-dry season (May-Oct). I will use the Home Range Extension (Rodgers and Carr 1998) in ArcView to determine a 95% fixed kernel (FK) and a 100% minimum convex polygon (MCP) home range for each telemetered individual during each season. MCP home ranges will be determined using 100% of the locations for each individual, excluding outliers resulting from large error ellipses (Dillon and Kelly 2008). Additionally, I will incorporate recommendations from Laver and Kelly (2008) for home range analyses including, but not limited to, check of home range asymptotes, reporting range and distribution of home range estimates, and use of appropriate measures of dispersion. To compare male and female home ranges to see if they differ significantly I will use Students T-test. Likewise, to compare home ranges across seasons I will use a paired t-test (Sokal and Rohlf 1995). In addition to calculating home ranges, I will use consecutive radio telemetry locations 12 and 36 hours apart to determine the daily distance moved for each C. ferox and F. fossana. I will average the linear distance between each pair of consecutive readings across all locations to estimate an average daily distance moved per individual (Dillon and Kelly 2008). I will use Students T-test to compare male and female daily distance moved. Depending on number of C. ferox and F. fossana radio collared, I will also determine the home-range overlap between C. ferox individuals, F. fossana individuals, and between C. ferox and F. fossana using MCP home ranges. I will average percent overlap across seasons to determine average male-male, female-female, and male-female per cent overlap. Telemetry will be conducted on a schedule that includes daytime and nighttime fixes for later comparison of nocturnal and diurnal locations and ranging behavior.
Objective III: Questions and Hypotheses
Does C. ferox or F. fossana activity pattern and home range vary across fragmented and non-fragmented forest sites?
H0: There is no significant difference in C. ferox and F. fossana activity pattern and/or home range across fragmented and non-fragmented sites. Ha: There is a significant difference in C. ferox and F. fossana activity pattern and home range across fragmented and non-fragmented sites.
Does C. ferox or F. fossana activity pattern and home range vary across sex or season? H0: There is no significant difference in C. ferox and F. fossana activity pattern and/or home range across sex or season. Ha: There is a significant difference in C. ferox and F. fossana activity pattern and home range across sex and/or season.
Objective IV Quantify the seasonal diet of C. ferox and F. fossana through collection and analyses of scat, as well as conduct genetic and disease analyses from blood and scat samples.
Collection and Analysis of Blood Samples Given that population viability is impacted by more than just animal numbers, I will collect blood samples from all captured C. ferox and F. fossana to determine genetic diversity for the population sampled. As a result of Madagascar’s habitat destruction, a sub-population that has contracted into a small isolated pocket, may suffer from loss of genetic diversity through genetic drift, and may become inbred if individuals cannot disperse across the matrix of human-dominated landscape. While sample sizes are likely to be low, this study will provide a baseline indication of genetic diversity for C. ferox and F. fossana through calculating allelic frequencies and degrees of heterozygosity. Heterozygosity will be compared to Hardy Weinberg expectations to reveal signs of inbreeding and to determine whether populations are mating randomly. I will collect blood samples (approx. 1cc per 500g) from anesthetized individuals to be analyzed after field work is completed. Blood samples will be stored following proper protocol (e.g. recommended temperature and length of time). The genetic analysis for this portion of the study will be conducted by a conservation genetics lab having prior experience working with samples collected from endangered wildlife (such as Dr. Ed Louis at Omaha’s Henry Doorly Zoo or Dr. Lisette Waits at the University of Idaho). Blood samples will also be analyzed to screen for wildlife diseases and blood-borne pathogens. Additionally, blood samples will be compared for carnivores across fragmented and non-fragmented forest sites. Carnivores occupying fragmented landscapes are more likely to be exposed to human, feline, and canid diseases such as rabies, parvovirus, canine distemper, and mange. It is possible to screen for these diseases and compare across fragmented and non-fragmented sites. Again, samples sizes are small, but this will give preliminary data on wildlife health of C. ferox and F. fossana within this region of Madagascar. Analyses will be conducted by a carnivore disease ecologist such as Dr. Kathleen Alexander – Virginia Tech. Collection and Analysis of Scat Samples The scat of C. ferox is said to be uniquely identifiable and easily distinguishable from other carnivore scat (Hawkins 1998, Dollar 2006, Gerber pers. comm.). The scat (Figure 7) is characterized as having a strong musky smell, black or gray cylinders with twisted ends, and typically 10-14 cm long and 1.5-2.5 cm wide (Hawkins and Racey 2008; Garbutt 2007; Hawkins 2003; Dollar 2006; Gerber pers. comm.). There are often large quantities of C. ferox hair within each sample. I have been unsuccessful in attempting to find a description of the scat of F. fossana in the literature. As a result, I will attempt to find an accurate description through interviews with individuals knowledgeable on F. fossana and other Madagascar’s carnivores, such as F. Kerridge. C. ferox and F. fossana scat will be collected from traps when individuals are captured, as individuals are known to defecate inside traps when captured and anesthetized (Dollar et al. 2006). In addition, scat will be collected opportunistically in the field when conducting radio telemetry, habitat sampling, lemur surveys, and camera maintenance at each study site. Each sample will be wrapped in aluminum foil and sealed in plastic bags with desiccant (Dollar et al. 2006). I will also record the GPS location of each sample and the date it was collected. Upon returning from the field I will dry each sample in an oven and store it in a cool dry area. For analysis, all samples will be autoclaved and hair and bone fragments separated for identification. Minimum number of individuals within a scat sample will be determined by pairing right and left osteological elements. Additionally, the minimum number of individuals will be corrected for relative frequency of prey following Hawkins and Racey (2008). I plan to collaborate with the Department of Paleontology at the University of Antananariavo for the identification of specimens given their thorough reference collection and experience in identifying prey species from carnivore scat (Hawkins and Racey 2008; Dollar et al. 2006; Rasoloarison et al. 1995; Goodman et al. 1997). Pieces of each scat sample will also be stored for three separate analyses to be conducted later. I will use DET buffer for later genetic analysis of fecal samples (Wultsch 2008) utilizing methods for isolating DNA from scats of numerous carnivore species developed by Dr. Waits. Preliminary scat samples will enable Dr. Waits to develop similar DNA extraction methods for C. ferox and F. fossana as this will later open the door for non-invasive, genetic mark-recapture population monitoring in the future. A piece of each sample will also be stored in ethanol or formalin for an analysis of macroparasites, as these parasites can provide an indication of animal health. In combination with the blood samples above, this analysis should give preliminary baseline information on C. ferox and F. fossana parasite loads within fragmented and non-fragmented forest sites. Finally, from scat samples that are opportunistically collected, we will also store small pieces in ethanol for later analysis of hormone levels, namely glucocorticoids (GLCs). GLCs have been shown to become elevated in carnivores when they are exposed to high-stress environments (Creel 2005). They can also stay permanently elevated when carnivores are exposed to chronic stress and this can negatively impact reproduction and immune function (Sapolsky et al. 2000). These samples will be analyzed in the lab of Dr. Janine Brown at the Smithsonian Institute who specializes in hormone analysis from fecal samples. This baseline data on C. ferox and F. fossana fecal DNA amplification and genetic diversity could form the basis of future research examining genetic diversity and connectivity across the Masoala-Makira landscape. For example, C. ferox could be the focal species for a landscape genetics project examining the impact of fragmentation on genetic diversity and gene flow across disconnected reserves. Combined with GIS, this approach could directly identify corridors of gene flow across sub-populations and determine how natural and anthropogenic disturbances impact gene flow. Baseline data on parasites and hormones can also add to our knowledge of how fragmentation impacts C. ferox or F. fossana health by examining whether increased fragmentation leads to high parasite and GLC loads.
Objective IV: Questions and Hypotheses Do increasing fragmentation rates negatively impact C. ferox and F. fossana populations across the Masoala-Makira landscape?
H0: Blood-borne pathogen and macroparasite loads for C. ferox and F. fossana populations are congruent across fragmented and non-fragmented sites. Ha: Blood-borne pathogen and macroparasite loads for C. ferox and F. fossana populations are not congruent across fragmented and non-fragmented forest sites.
H0: Presence of rabies, distemper, parvovirus, and mange in C. ferox and F. fossana will be consistent across fragmented and non-fragmented sites. Ha: Presence of rabies, distemper, parvovirus, and mange in C. ferox and F. fossana will be greater in fragmented sites compared to non-fragmented sites.
Does C. ferox and F. fossana diet differ across the Masoala-Makira landscape?
H0: C. ferox and F. fossana diet is congruent across fragmented and non- fragmented forest sites. Ha: C. ferox and F. fossana diet is not congruent across fragmented and non- fragmented forest sites.
Expected Outcomes I expect to see similarity among density comparisons for F. fossana across all study sites. However, recent analyses for this species have shown an overestimation of density using the ½ MMDM technique as opposed to full MMDM, maximum likelihood estimation, and Bayesian hierarchical approach (Gerber et al. 2009, Gerber In prep, and Farris In prep). I expect to find higher rates of occupancy/trap success for F. fossana, C. ferox, E. goudotii, S. concolor, and G. fasciata across non-fragmented forests. Further, I expect trap success of S. concolor and G. fasciata to be low across all sites as these two species are known to be rare despite levels of connectivity and fragmentation. I expect an absence of F. fossana, E. goudotii, S. concolor, and G. fasciata across fragmented forests. However, I expect occupancy/trap success of G. elegans and V. indica to be higher across fragmented forests as these two species are known to successfully inhabit and exploit resources across fragmented landscapes (Garbutt 2007, Dunham 1998). In relation to landscape metrics I expect to see a positive correlation in rates of occupancy/trap success for F. fossana, C. ferox, E. goudotii, S. concolor, and G. fasciata and landscape metrics associated with non-fragmented forests (e.g. increased distance to edge, increased core area, increased forest cover). Conversely, I expect to see a positive correlation in occupancy/trap success rates of G. elegans and V. indica associated with fragmented forests. For those species that are highly sensitive to fragmentation (C. ferox, F. fossana, S. concolor, and G. fasciata) I expect to see a positive relationship with the following micro-habitat variables: tree density, basal area, DBH, percent understory cover and heterogeneity, and canopy cover. Further, I expect to see strong correlation with increased understory cover for F. fossana as this variable is associated with increases in small mammal numbers. I expect to see strong correlations in tree density, basal area, and DBH for C. ferox as this predatory species considers lemurs as a significant prey source (Wright et al. 1997). As C. ferox is the primary predator for lemurs, I expect to find a strong relationship between these two groups. I expect to see a greater abundance of lemurs in areas where C. ferox occupancy/trap success is diminished. I expect to see higher rates of trap success for small mammals across fragmented study sites because of the lower occupancy/trap success rates of carnivores. Additionally, I expect to find higher rates of trap success for invasive species (V. indica, domestic dog, and feral cat) and humans across fragmented sites. I expect to see higher rates of occupancy/trap success for co-occurring carnivores, excluding F. fossana, where C. ferox rates are low, as a result of mesopredator release. However, I do not expect to see a relationship between F. fossana population variables and C. ferox variables as these two species are found to co-occur throughout primary rainforest and exist in low numbers across fragmented forests (Gerber et al. 2009, Gerber In prep, Farris In prep). I expect to see larger home ranges for C. ferox and F. fossana across fragmented forests resulting from limited and/or widely dispersed resource availability across these sites. Additionally, I expect to see greater inter-species and intra-species home range overlap for these two carnivores across fragmented sites. I expect to find larger home ranges and increased activity for males compared to females for both C. ferox and F. fossana. I expect to find more temporal variability in activity for C. ferox as this top predator has shown wide ranging activity patterns in previous studies (Gerber Pers. obs., Farris Pers. obs., Dollar 2006, Hawkins 1998). Conversely, I expect to see strictly nocturnal activity for F. fossana and G. fasciata and strictly diurnal activity for G. elegans and S. concolor. I expect to see little seasonal affect on activity for all carnivore species across the Masoala-Makira landscape as seasonal affects are much less significant compared to more southern rainforest areas. In relation to blood and scat analyses, I expect to find higher levels of macroparasite and blood borne pathogens for C. ferox and F. fossana occupying fragmented forests. Additionally, for C. ferox and F. fossana in forests with increased fragmentation and low levels of connectivity, I expect to find lower levels of heterogeneity from genetic analyses. I expect to find a greater number of C. ferox and F. fossana scats across non-fragmented forests. Moreover, I expect to find a greater composition of lemur remains in C. ferox scat from non-fragmented forests. Finally, I expect to find more homogeneity in prey consumption (primarily consisting of invasive rodent species) for both C. ferox and F. fossana across fragmented forests.
Schedule
Literature Cited:
Albignac, R. 1971. Notes ethologiques sur quelques carnivores Malgaches: le Fossa Fossa (Schreber). Revue d’Ecologie Appliquee, 24: 383-394. Besbeas, P., S. N. Freeman, B. J. T. Morgan, and E. A. Catchpole. 2002. Integrating mark- recapture-recovery and census data to estimate animal abundance and demographic parameters. Biometrics 58:540-547. Bollen, A. and Donati, G. 2006. Conservation status of the littoral forest of south-eastern Madagascar: A review. Oryx, 40, 1–10. Bowen-Jones, E. and S. Pendry. 1999. The threat to primates and other mammals from the bushmeat trade in Africa, and how this threat could be diminished. Oryx, 33(3): 233-246. Brashares, J., P. Arcese, M. Sam, P. Coppolillo, A. Sinclair, and A. Balmford. 2004. Bushmeat hunting, wildlife declines, and fish supply in West Africa. Science 306, 1180-1183. Borchers, D. L., and M. G. Efford. 2007. Spatially Explicit Maximum Likelihood Methods for Capture–Recapture Studies. Biometrics. Buckland, S. T., Anderson, D. R., Burnham, K. P. and J. L. Laake. 1993. Distance Sampling: Estimating Abundance of Biological Populations. Chapman & Hall, London Buckland, S.T., Anderson, D. R., Burnham, K. P., Laake, J. L., Borchers, D. L., and L. Thomas. 2001. Introduction to Distance Sampling, Oxford: Oxford University Press. Burnham, K. P., Anderson, D. R. and J. L. Laake. 1980. Estimation of density from line transect sampling of biological populations, Wildlife Monographs, 72: 1-202. Butler, R. 2010. Madagascar sanctions logging of national parks. Wildmadagascar.org, http://news.mongabay.com/2010/0111-madagascar.html. January 11, 2010. CIA (CENTRAL INTELLIGENCE AGENCY) WORLD FACTBOOK. 2010. Africa: Madagascar. Last updated: February 4, 2010. ISSN 1553-8133 Cowlishaw, G. S. Medelson, and J. Rowcliffe. 2005. Evidence for post-depletion sustainability in a mature bushmeat market. Journal of Applied Ecology, 42: 460-468. Craul, M., Chikhi, L., Sousa, V., Olivieri, G., Rabesandratana, A., Zimmermann, E., and U. Radespiel. 2009. Influence of forest fragmentation on an endangered large-bodied lemur in northwestern Madagascar. Biological Conservation, 142(12):2862-2871. Creel, S. 2005. Dominance, Aggression, and Glucocorticoid Levels in Social Carnivores, Journal of Mammalogy, 86(2): 255-264. Cooch, E. and G. White. 2006. Program MARK: A Gentle Introduction, 4th edition. Available at http://www.phidot.org/software/mark/docs/book/. Cushman,S., Mckelvey, K., Flather, C., and K. Mcgaigal. 2008. Do forest community types provide a sufficient basis to evaluate biological diversity? Frontiers in Ecology and the Environment 6(1):13-17. Davis, M.L. 2007. Co-Varying Predator Densities and Mesopredator Release of the Ocelot across Five Study Sites in Belize - Working Plan, Virginia Tech. Davis, M., Kelly, M.J., and D. Stauffer. In press. Carnivore coexistence and habitat use in the Mountain Pine Ridge Forest Reserve, Belize. Animal Conservation. Di Bitetti, M. S., A. Paviolo, and C. De Angelo. 2006. Density, habitat use and activity patterns of ocelots (Leopardus pardalis) in the Atlantic Forest of Misiones, Argentina, Journal of Zoology 270:153-163. Dillon, A. and M.J. Kelly. 2007. Ocelot Leopardus pardalis in Belize: the impact of trap spacing and distance moved on density estimates, Oryx 41: 469–477. ______. 2008. Ocelot home range, overlap and density: Comparing radio telemetry with camera trapping, Journal of Zoology, 275: 391-398. Dollar, L. 1999. Preliminary report on the status, activity cycle, and ranging of Cryptoprocta ferox in the Malagasy rainforest, implications for conservation, Small Carnivore Conservation, 20: 7-10. ______. 2006. Morphometrics, diet, and conservation of Cryptoprocta ferox. Phd Dissertation. Duke University. Durham, NC. Dunham, A. E. 1998. Notes on the behavior of the Ring-tailed mongoose, Galidia elegans at Ranomafana National Park, Madagascar. Small Carnivore Conservation 19:21-24. Efford MG, Dawson DK, and Robbins CS. 2004. DENSITY: Software for analyzing capture– recapture data from passive detector arrays. Animal Biodiversity and Conservation 27:217–228 Efford, M. G. 2007. Density 4.1: Software for spatially explicit capture–recapture. Pages Department of Zoology in University of Otago, Dunedin, New Zealand. ,Dunedin. Fa, J.E., Peres, C.A. and Meeuwig, J. 2002. Bushmeat exploitation in tropical forests: an Intercontinental comparison. Conservation Biology, 16, 232–237. Fashing, P. and M. Cords. 2001. Diurnal primate densities and biomass in the Kakamega Forest: An evaluation of census methods and a comparison with other Forests, American Journal of Primatology, 50: 139-152. Feistner, A.T.C. and J. Schmid. 1999. Lemurs of the Reserve Naturelle Integrale d’Andohahela, Madagascar, Fieldiana: Zoology, 269-283.GARBUTT, N. 2007. Mammals of Madagascar. A Complete Guide. Yale University Press, New Haven, US, and London, UK. Garcia, G. and Goodman, S.M. 2003. Hunting of protected animals in the Parc National d’Ankarafantsika, north-western Madagascar. Oryx, 37, 115–118. Gerber, B. 2008. Population abundance, density, and occupancy/use of carnivores in the rainforests of southeastern Madagascar. Thesis Working Plan. Virginia Tech University. Gerber, B, Karpanty, S.M, Crawford, C., Kotschwar, M., and J. Randrianantenaina. 2009. An assessment of carnivore relative abundance and density in the easter rainforests of Madagascar using remotely-triggered camera traps. Oryx p. 1-7. Gittleman, J.L., Funk, S.M., Macdonald, D. and R.K. Wayne. 2001. Why Carnivore Conservation? In Gittleman, J.L., Funk, S.M., macdonald, D. And R.K. Wayne (eds). Carnivore Conservation, Cambridge: Cambridge University Press, p. 1-7. Golden, C. 2009. Bushmeat hunting and use in the Makira Forest, north-eastern Madagascar: a conservation and livelihoods issue. Oryx, 43(3), 386–392 Goodman, S.M. 2003. Carnivora: Galidia elegans, Ring-tailed mongoose, Vontsiar Mena. Pp. 1351-1354 in The natural history of Madagascar (S. M. Goodman and J. P. Benstead, eds.). University of Chicago Press, Chicago, Illinois. ______. 2006. Hunting of Microchiroptera in southwestern Madagascar. Oryx, 40, 1–4. Goodman, S.M. and Raselimanana, A. 2003. Hunting of wild animals by Sakalava of the Menabe region: a field report from Kirindy-Mite. Lemur News, 8, 4–6. Goodman, S. M., Langrand O. and B. P. N. Rasolonandrasana. 1997. The food habits of Cryptoprocta ferox in the high mountain zone of the Andringitra Massif, Madagascar (Carnivora, Viverridae), Mammalia 61:185–192. Green, G.M. and R.W. Sussman. 1990. Deforestation history of the eastern rain forests of Madagascar from satellite images, Science, 248: 212-215. Hackett. M., Lesmeister, D., Desanty-Combes, J., Montague, W., Millspaugh, J., and M. Gompper. 2007. Detection rates of Eastern Spotted Skunks (Spilogale putorius) in Missouri and Arkansas using live capture and non-invasive techniques. American Midland Naturalist 158:123-131. Hanski, I. 2005. The shrinking world: Ecological consequences of habitat loss. Excellence in Ecology. International Ecology Institute, Oldendorf, Germany. Harper, G., Steininger, M., Tucker, C., Juhn, D., and F. Hawkins. 2007. Fifty Years of deforestation and forest fragmentation in Madagascar. Environmental Conservation, 34:325-333. Hawkins, C.E. 1998. The behaviour and ecology of the Fossa, Cryptoprocta ferox (Carnivora: Viverridae) in a dry deciduous forest in western Madagascar. Phd thesis, University of Aberdeen. ______. 2003. Cryptoprocta ferox, fossa, fosa. Pp. 1360–1363 in The natural history of Madagascar (S. M. Goodman and J. P. Benstead, eds.). University of Chicago Press, Chicago, Illinois. Hawkins, A.F.A. and L. Dollar. 2008. Cryptoprocta ferox. In: IUCN 2008. 2008 IUCN Red List of Threatened Species. <www.iucnredlist.org>. Downloaded on 15 March 2010. Hawkins, C.E. and P.A. Racey. 2008. Food Habits of an Endangered Carnivore Cryptoprocta ferox, in the Dry Deciduous Forests of Western Madagascar, Journal of Mammalogy, 89(1): 64-74. Hawkins, A.F.A., Durbin, J. and L. Dollar. 2008. Salanoia concolor. In: IUCN 2010. IUCN Red List of Threatened Species. Version 2010.1. www.iucnredlist.org. Downloaded on 18 March 2010. Heilburn, R., Silvy, A. Peterson, M., and M. Tewes. 2006. Estimating bobcat abundance using automatically triggered cameras. Wildlife Bulletin 34(1):69-73. Holmes, C. 2007. Linking Livelihoods, Land Stewardship, and Resource Conservation in the Antongil Bay Landscape, Madagascar. Wildlife Conservation Society: Protected Areas and Human Livelihoods. Edited by K. Redford and E. Fearn. Working Paper No. 32, December 2007. Hunter, M. L., G. L. Jacobson, and T. Webb. 1988. Paleoecology And The Coarse-Filter Approach To Maintaining Biological Diversity. Conservation Biology 2:375-385. Irwin, M.T., Raharison, J.L., and P.C. Wright. 2009. Spatial temporal variablibility in predation on rainforest primates: do forest fragmentation and predation act synergistically? Animal Conservation 12(3): 220-230. Jackson, R., Roe, J., Wangchuk, R., and D. Hunter. 2006. Estimating snow leopard population abundance using photography and capture-recapture techniques. Wildlife Society Bulletin 34(3):772-781. Johnson, S. E., and D. J. Overdorff. 1999. Census of brown lemurs (Eulemur fulvus sspp.) In southeastern Madagascar: Methods-testing and conservation implications. American Journal of Primatology 47:51-60. Karanth, K. U. 1995. Estimating tiger Panthera tigris populations from camera-trap data using capture-recapture models, Biological Conservation 71:333-338. Karanth, K. U., R. S. Chundawat, J. D. Nichol, and N. S. Kumar. 2004. Estimation of tiger densities in the tropical dry forests of Panna, Central India, using photographic capture-recapture sampling, Animal Conservation 7:285-290. Karanth, K. U., and J. D. Nichols. 1998. Estimation of tiger densities in India using photographic captures and recaptures, Ecology 79: 2852-2862. Karpanty, S. and Wright, P. 2007. Predation on lemurs in the rainforest of Madagascar by multiple predator species: Observations and Experiments. Pp. 77-99 in Primate Anti-predator Strategies (Gursky and Nekaris, eds.). Springer US. Kelly, M. J. 2003. Jaguar monitoring in the Chiquibul Forest, Belize, Caribbean Geography 13: 19-32. Kelly, M.J, Noss, A.J., Dibetitti, M.S., Maffei L., Aripse, R., Paviolo, A., Deangelo, C.D. and Y.E. Diblano. 2008. Estimating puma densities from camera trapping across three study sites: Bolivia, Argentina, Belize, Journal of Mammalogy, 89: 408–415. 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, Northeast Naturalist, 15(2): 249-262. Kerridge, F.J., Ralisoamalala, R.C., Goodman, S.M. and S.D. Pasnick. 2003. Fossa fossana, Malagasy Striped Civet, Fanaloka. Pp. 1363-1365 in The natural history of Madagascar (S. M. Goodman and J. P. Benstead, eds.). University of Chicago Press, Chicago, Illinois. Kremen, C. 2003. The Masoala Peninsula, Pp. 1459-1466 in The natural history of Madagascar (S. M. Goodman and J. P. Benstead, eds.). University of Chicago Press, Chicago, Illinois. Laake, J. L., Buckland, S. T., Anderson, D. R. And K. P. Burnham. 1991. DISTANCE User’s Guide. Colorado State University, Fort Collins, Colorado Linkie, M., Chapron, G., Martyr, D., Holden, J. and N. Williams. 2006. Assessing the Viability of Tiger Subpopulations in a Fragmented Landscape, Journal of Applied Ecology, 43(3): 576-586. Laver, P. N., and M. J. Kelly. 2008. A critical review of home range studies. Journal of Wildlife Management 72:290-298. Mackenzie, D. I. 2005. Was it there? Dealing with imperfect detection for species presence/absence data. Australian & New Zealand Journal of Statistics 47:65-74. Mackenzie, D. I., J. D. Nichols, J. E. Hines, M. G. Knutson, and A. B. Franklin. 2003. Estimating site occupancy, colonization, and local extinction when a species is detected imperfectly. Ecology 84:2200-2207. Mackenzie, D. I., J. D. Nichols, G. B. Lachman, S. Droege, J. A. Royle, and C. A. Langtimm. 2002. Estimating site occupancy rates when detection probabilities are less than one. Ecology 83:2248-2255. Maffei, L., Noss, A. J., Cuellar, E. and D. I. Rumiz. 2005. Ocelot (Felis pardalis) population densities, activity, and ranging behaviour in the dry forests of eastern Bolivia: data from camera trapping, Journal of Tropical Ecology 21: 349-353. Maffei, L., E. Cuellar, and A. Noss. 2004. One thousand jaguars (Panthera onca) in Bolivia's Chaco? Camera trapping in the Kaa-Iya National Park, Journal of Zoology 262: 295-304. Maffei, L.A. and A.J. Noss. 2008. How small is too small? Camera trap survey areas and density estimates for ocelots in the Bolivian Chaco, Biotropica 40: 71–75. Merenlender, A., Kremen, C., Rokotondratsima, M. and A. Weiss. 1998. Monitoring Impacts of Natural Resource Extraction on Lemurs of the Masoala Peninsula, Madagascar, Conservation Ecology, 2(2): 5. Meyers, D. 2001. Makira Forest Project, Madagascar. Report to the Ministry of Environment. MEF-IRG/PAGE-USAID. Millspaugh, J.J. and J.M. Marzluff. 2001. Radio tracking and animal populations. San Diego, CA: Academic Press. Mittermeir, R.A., Gil, P.R., Hoffman, M., Pilgrim, J., Brooks, T., Mittermeir, C., Lamoreux, J. and G. Fonseca. 2004. Madagascar And The Indian Ocean Islands. Pp. 138-144, In Mittermeir, R.A., Gil, P.R., Hoffman, M., Pilgrim, J., Brooks, T., Mittermeir, C., Lamoreux, J. and G. Fonseca (eds.), Hotspots Revisited: earth’s biologically richest and most threatened terrestrial ecoregions. CEMEX, Mexico City. Nichols, J.D. and K.U. Karanth. 2002. Statistical concepts: estimating absolute densities of tigers using capture-recapture sampling. In Field monitoring tigers and their prey: a manual for researchers, managers and conservationists in tropical Asia: 121–137. Karanth, K.U. & Nichols, J.D. (Eds). Bangalore, India: Centre for Wildlife Studies. Noss, R.F. 1990. Indicators for Monitoring Biodiversity: a hierarchical approach, Conservation Biology, 4: 355-364. NRC. 1981. Techniques for the Study of Primate Population Ecology. Subcommittee on Conservation of Natural Populations. National Research Council. National Academy Press, Washington, D.C. O’brien, T.G. 2008. On the use of automated cameras to estimate species richness for large and medium sized rainforest mammals. Animal Conservation (11): 179-181. Otis, D. L., Burnham, K. P., White, G. C. and D. R. Anderson. 1978. Statistical-inference from capture data on closed animal populations, Wildlife Monographs: 7-135. Palomares, F., P. Gaona, P. Ferreras, and M. Delibes. 1995. Positive effects on game species of top predators by controlling smaller predator populations: an example with lynx, mongooses, and rabbits. Conservation Biology 9:295-305. Peres, C.A. 1999. General guidelines for standardizing line transect surveys for tropical forest primates, Neotropical Primates, 7(1): 11-16. Parmenter, R. R., T. L. Yates, D. R. Anderson, K. P. Burnham, J. L. Dunnum, A. B. Franklin, M. T. Friggens, B. C. Lubow, M. Miller, G. S. Olson, C. A. Parmenter, J. Pollard, E. Rexstad, T. M. Shenk, T. R. Stanley, and G. C. White. 2003. Small-mammal density estimation: A field comparison of grid-based vs. Web-based density estimators. Ecological Monographs 73:1-26. Powell, L. A., M. J. Conroy, J. E. Hines, J. D. Nichols, and D. G. Krementz. 2000. Simultaneous use of mark-recapture and radiotelemetry to estimate survival, movement, and capture rates. Journal of Wildlife Management 64:302-313. Randriaanandianina, B. L., Andriamahaly, F. Harisoa, and M. Nicoll. 2003. The Role of the Protected Areas in the Management of the Island’s Biodiversity. Pp. 1423–1432 in: The Natural History of Madagascar (S. M. Goodman and J. P. Benstead, eds.). University of Chicago Press, Chicago, Illinois. Rasoloarison, R. M., Rasolonandrasana, B. P. N., Ganzhorn, J. U.and S. M. Goodman. 1995. Predation on vertebrates in the Kirindy Forest, western Madagascar, Ecotropica 1:59–65. Rasolonandrasana, B.P.N. 1994. Contibution à l’étude de l’alimentation de Cryptoprocta ferox Bennett (1833) dans son milieu naturel. D.E.A. thesis, Université d’Antananarivo, Antananarivo, Madagascar. Rexstad, E., and K. P. Burnham. 1991. User's guide for interactive program CAPTURE: abundance estimation of closed animal populations. Colorado Cooperative Fish and Wildlife Research Unit, Fort Collins, CO; Colorado, United States. Rogers, C. M., and M. J. Caro. 1998. Song sparrows, top carnivores, and nest predation: a test of the mesopredator release hypothesis. Oecologia 116:227-233. Rodgers, A.R. and A.P. Carr. 1998. HRE: The home range extension for arcview. Ontario, Ministry of Natural Resources, Centre of Northern Forest Ecosystem Research, Thunder Bay, Ontario, Canadian Management, 49, 675–678. Rowcliffe, J. M., and C. Carbone. 2008. Surveys using camera traps: are we looking to a brighter future? Animal Conservation 11:185-186. Rowcliffe, J. M., J. Field, S. T. Turvey, and C. Carbone. 2008. Estimating animal density using camera traps without the need for individual recognition. Journal of Applied Ecology 45:1228-1236. Royle, J. A., K. U. Karanth, A. M. Gopalaswamy, and N. S. Kumar. 2009. Bayesian inference in camera trapping studies for a class of spatial capture-recapture models. Ecology 90:3233-3244. Royle, J. A., and K. V. Young. 2008. A hierarchical model for spatial capture-recapture data. Ecology 89:2281-2289. Sanderson, J.G. 2004. Tropical Ecology, Assessment and Monitoring Initiative. Camera Phototrapping Monitoring Protocol. Center for Applied Biodiversity Science, Conservation International. Sapolsky, R.M., Romero, L.M. and A.U. Munck. 2000. How Do Glucocorticoids Influence Stress Responses? Integrating Permissive, Suppressive, Stimulatory, and Preparative Actions, Endocrine Reviews, 21 (1):55-89. Silver, S. C., Ostro, E., Marsh, L., Maffei, L., Noss, A., Kelly, M., Wallace, R., Gomez, H., and G. Ayala. 2004. The use of camera traps for estimating jaguar Panthera onca abundance and density using capture/recapture analysis, Oryx 38: 148-154. Soisalo ,M.K. and M.C. Calvacanti. 2006. Estimating the density of a jaguar population in the Brazilian Pantanal using camera traps and capture–recapture sampling in combination with GPS radio-telemetry, Biological Conservation 129: 487–496. Sokal, R.R. and F.J. Rohlf. 1995. Biometry, the principles and practice of statistics in biological research, 3rd ed. New York: Freeman and Company Sterling, E.J. and N. Rakotoarison. 1998. Rapid assessment of richness and density of primate species on the Masoala Peninsula, Eastern Madagascar, Folia Primatologica, 69(1): 109-116. Stickel, L. F. 1954. A comparison of certain methods of measuring home range of small mammals. Journal of Mammalogy:1-15. Sunquist and Sunquist. 2001. Changing landscapes: consequences for carnivores. In Gittleman, J.L., Funk, S.M., Macdonald, D. and R.K. Wayne (Eds). Carnivore Conservation, Cambridge: Cambridge University Press, p. 1-7. Swihart, R. and N. Slade. 1997. On testing for independence of animal movements. American Statistical Association and the International Biometric Society. Journal of Agricultural, Biological, and Environmental Statistics 2(1): 48-63. Tanaka, R. 1980. Controversial problems in advanced research on estimating population densities of small rodents. Population Ecology:1-66. Terborgh, J., J. Estes, P. Paquet, K. Ralls, D. Boyd-Heger, B. Miller, and R. Noss. 1999. The Role of Top Carnivores in Regulating Terrestrial Ecosystems. In M. E. Soulé, and J. Terborgh, editors. Continental Conservation: Scientific Foundations of Regional Reserve Networks. Island Press,Washington. Terborgh, J., L. Lopez, V. P. Nunez, M. Rao, G. Shahabuddin, G. Orihuela, M. Riveros, R. Ascanio, G. H. Adler, T. D. Lambert, and L. Balbas. 2001. Ecological meltdown in predator-free forest fragments. Science 294:1923-1926. Tioli, S., F. Cagnacci, A. Stradiotto, and A. Rizzoli. 2009. Edge Effect on Density Estimates of a Radiotracked Population of Yellow-Necked Mice. Journal of Wildlife Management 73:184-190. Trolle, M. and M. Kery. 2003. Estimation of ocelot density in the pantanal using capture- recapture analysis of camera-trapping data, Journal of Mammalogy 84: 607-614. Wallace, R. B., Ayala, G. and F. Espinoza. 2003. Camera trapping for jaguar F(Panthera onca) in the Tuichi Valley, Bolivia, Journal of Neotropical Mammology, 10: 133-139. WCS (Wildlife Conservation Society). 2004. Contexte environnemental du site de Conservation de Makira. Unpublished Report for WCS, Madagascar. Wegge, P., C. Pokheral, and S. Jnawali. 2004. Effects of trapping effort and trap shyness on estimates of tiger abundance from camera trap studies. Animal Conservation:251–256. White, G. C., Anderson, D. R. Burnham, K. P. and D. L. Otis. 1982. Capture-recapture and Removal Methods for Sampling Closed Populations Los Alamos National Laboratory, Los Alamos, USA. White, G.C. and R.A. Garrot. 1990. Analysis of wildlife radio tracking data. San Diego, CA: Academic Press. White, G.C. and K. P. Burnham. 1999. Program MARK: Survival estimation from populations of marked animals. Bird Study 46 Supplement: 120-138. Whitesides G.H., Oates, J.F., Green, S.M., and R.P. Kluberdanz. 1988. Estimating primate densities from transects in a West African rain forest: a comparison of techniques, Journal of Animal Ecology, 57: 345–367. Wilson, K. R., and D. R. Anderson. 1985. Evaluation of two density estimators of small mammal population size. Journal of Mammalogy 66:13-21. Wright, P. C., Heckscher, S. K. and A. E. Dunham. 1997. Predation on Milne-Edward’s sifaka (Propithecus diadema edwardsi) by the fossa (Cryptoprocta ferox) in the rain forest of southern Madagascar. Folia Primatologica 68:34–43. Wultsch, C. 2008. Noninvasive tracking of jaguars (Panthera onca) and co-occurring feline species in Belize by combining molecular scatology, remote camera trapping and GIS: the impact of fragmentation. Project report from Kaplan Grant submitted to PANTHERA, USA, October 2008. Xu, A., Jiang, Z., Li, C., Guo, J., Da, S. Cui, Q., Yu, S., and G. Wu. 2008. Status and conservation of the snow leopard Panthera uncial in the Gouli Region, Kunlun Mountains, China. Oryx, 42(3): 460-463.
Figure 1. Masoala-Makira Landscape. The Masoala National Park (right, outlined in yellow) represents the largest protected area in Madagascar, while Makira Conservation site (left, outlined in purple) is the longest contiguous stretch of primary rainforest remaining in all of Madagascar. (Image modified from Wildlife Conservation Society, www.wcs.org)
 Figure 2. Vegetation classification for the Masoala-Makira landscape showing park boundaries for Makira (left in black) and Masoala (right in blue) and 3 potential fragmented sites (white) and 3 potential non-fragmented sites (red) in an arc shape pattern across the region. Final site selection will be conducted before commencement of field season using methods described in “Selection of Sites,” p. 7. Figure 3: Grid design for carnivore camera-trapping study. This grid will be placed in six study sites across the Makira Protected area and Masoala National Park from June 2010-May 2012. Each camera station (25 total) will be separated by 750 m for a possible total grid size of 4.5 *4.5 km (20.25km²). Each grid will be operational in each site for 60 days. Figure modified from Gerber (2008). Figure 4. Example of 250m buffer surrounding camera traps from which habitat types were extracted (e.g. % cover deciduous forest, % cover coniferous forest, % agriculture, % water). Figure taken from Holub and Kelly (2006).
Figure 5. Habitat sampling scheme to be conducted for each camera site. Modified from Davis (2007).
Figure 6. Grid design for lemur surveys with two, 4km long transects extending from the center most camera station and two additional transects extending from a randomly chosen camera station within the grid.  Figure 7. Scat collected from Cryptoprocta ferox at Ranomafana National Park, Madagascar. Sample collected and photographed by B. Gerber. Image used with permission from B. Gerber. | ||||||