Methods

With the assistance of Earthwatch volunteers, the lead PI, Luke Dollar, conducted his first field season at the Ampijoroa Research Station in 1999 to investigate the research site and lay the groundwork for this ongoing project. The PI led a trapping and radiotelemetry study in the Lac Tsimaloto region of Ankarafantsika (22 km east of the Ampijoroa Research Station) the year before, in 1998. At the Lac Tsimaloto site, six different Cryptoprocta were trapped in 12 trapping events during 400 trap-days. More than 300 hours of radiotelemetry data were also collected during this 3.5 week survey. While the biomass of predators found in the Lac Tsimaloto (1998 and 2001 surveys) and Ampijoroa (1999-present) regions of Ankarafantsika National Park was relatively constant, the composition of Ampijoroa’s top trophic level was completely unexpected.

In 1999, we made an unexpected discovery. Wild cats, large competitive felids (possibly the African Wild Cat or a Malagasy variant), are present in Ampijoroa’s savannahs and transition forests. This elucidates an alarming problem faced by endangered endemic carnivores worldwide. While the cats are relatively prevalent in the open and scrubby natural spaces found on Ampijoroa’s topography, they do not enter or use resources from deep within Ampijoroa’s woodlands, where the fossa retains a competitive predatory advantage. Although the fossa is still the only predator capable of preying on the largest of prey species – Ampijoroa’s larger primates – the presence of the wild cats leads to a notable difference in the density and ranging of Cryptoprocta. At Ampijoroa, fossa occur at lower density than in the nearby Tsimaloto region. In addition, their home ranges are larger. Although the diet of fossa in the presence of this new-found felid competitor has not yet been fully examined, preliminary results indicate that Cryptoprocta specialize more heavily on prey species that the wildcats are incapable of taking than they would in habitats free from competition. We have collected more than 1000 scat samples over the last three years. Analysis is complete on 800 of these as of September, 2001.

In 2001, we have noted a significant increase in density of Cryptoprocta nomadically passing through our study areas. While we would like to believe that this increased trap success belies an increased level of conservation for fossa in the Ampijoroa region, other factors are also likely affecting this trend. Significant habitat disturbance in the east end of Ankarafantsika (as identified by satellite analysis and ground truthing surveys) has caused an upsurgence of pressure leading to an apparent westward migration into the remaining, less-disturbed habitats within the Park.

Primary Objectives and Their Methods

1) Conduct trapping surveys to further locate and identify the carnivore populations and trends in the Ampijoroa Research Station area of Ankarafantsika National Park, Madagascar, collect anatomical data on each species of carnivore at this site, and to collect carnivore scat samples for analysis of diet composition/contents; and

2) Establish baseline indices or indicative measures of relative abundance of Cryptoprocta ferox and additional carnivore populations in the dry, deciduous forests of Ankarafantsika;

Trapping surveys

Initially, large grids and transects were mapped onto several portions of the study area. Large Hav-A-Hart and Tomahawk live-capture cage traps are placed at regular intervals on the transects and grids, for a total of more than 40 active traps. As there is no means of travel within the research area other than by foot, our trap-line size (up to 25 km in a single hike) is the largest that is physically feasible to practically pursue the goals of this study. Considering the relatively large home range of Cryptoprocta (Dollar, 1997; Hawkins, 1998), a much larger area would be required to attain actual abundance measures (Wilson et al, 1996). Therefore, attaining indices of relative abundance (as in Caughley, 1977, among others) is a more realistic focus for this study (objectives 1 and 2). The location of the trap grids remains constant from survey period to survey period to lower the likelihood of bias in achieving objective number 2. In addition to following our specific mark-recapture protocol for multiple 10 day trapping sessions (in adherence to previous studies), other traps are often opportunistically moved to additional, remote parts of the surrounding forests so as to potentially increase the sample size of study animals.

Ten day trap surveys are conducted by each team of Earthwatch volunteers. This affords a total of at least 400 traps/day/team session at the research site. Traps are checked twice daily while active trapping is in effect. Two trap-checking groups “split up” to each check separate trap lines. During each check of the traps, a walking census of all animals encountered is also conducted, providing baseline ecological monitoring data for analysis of long-term biodiversity trends in Ankarafantsika. All fossa scats (>1000 to date) are collected during grid checks (objective 1). Fresh and incompletely dessicated fossa scats are easily differentiated from wildcat scats by their characteristic smell. Additional distinguishing features include the shape and deposition/distribution of the scats themselves. In addition, fossa scats have a characteristic long tail that is not seen in the wildcat samples we have collected to date.

Trap check teams stay in contact with one another, the radiotelemetry teams, and base camp via the use of hand-held radio. The number of captures per unit effort provides a baseline measure or index for assessment of relative abundance of species captured. This measure is important for the long-term monitoring of carnivore populations within the research area, helping to meet objectives 1, 2, 6 and 10 as outlined in section 2.

In addition to the cage-trapping portion of the survey, phototraps are also used to help evaluate the variety and presence of carnivores and their prey in Ankarafantsika National Park. Phototrapping involves a wide-angle, weatherproof photographic apparatus attached to either an infrared or motion sensor - or even a pressure pad - which triggers the shutter release when activated, exposing the animal and time of passage on film. This type of mechanism can ‘trap’ more than one animal before reactivation, does not depend upon physical encounter, investigation, and entry. This procedure involves much less trauma on part of the animal in question -- an animal need merely pass by the camera to be ‘trapped,’ after which it continues on its way unaware of its contribution to the assessment of its environment. This technique has proven successful for photocapturing terrestrial, arboreal, and avian fauna and is responsible for the discovery of more than one previously unidentified species and has also provided incontrovertible evidence and identification of poachers and illegal tree cutters in many areas (Wilson et al., 1996).

While photo-traps are not used as replacements for cage traps, some cameras are placed in proximity to cage traps to facilitate an evaluation of the number of trap visits by study animals relative to actual capture events at those traps. Phototraps placed in additional areas of the research site help to confirm or refute that all possible carnivore species are sampled at Ampijoroa by affording a non-invasive capture technique. After the cage trapping portion of each field season is complete, camera traps are placed along the grid system to provide “recaptures” of the individually marked fossa as well as uncaptured and immigrant animals. This bait station analysis also provides us a means of forming an index of relative abundance for fossa population monitoring (Bookhout, 1996) that does not rely on an animal entering into an enclosed cage. Nor does it depend upon the reliability of multiple observers to find and assess numbers of feces or tracks found on the grid. This camera technique has been successfully implemented on a number of North American carnivores including bears (D. Martorello & M. Pelton, pers comm), marten, fisher, lynx, and wolverine (Zielinski and Kucera, 1995).

Animal handling, processing, and measurement

Captured Viverricula will be released from the trap into a handling bag. Drug is administered via intramuscular injection with the animal still in the handling bag. Captured Cryptoprocta and Wildcats are tranquilized while still in the trap using the Pneu-dart drug delivery system. Using the Pneu-dart system, a trained staff member blowpipes the trapped larger carnivores using procedures outlined by Glander, et al. (1992). Anesthesia is delivered via dart in the hindquarters and only if the animal is facing away from the shooter so as to reduce the risk of damage resulting from shots in the face, abdomen, shoulder, or neck.

Once the animal appears to be adequately tranquilized, the darter and an assistant remove the tranquilized animal from the trap or handling bag. Anatomical measurements are taken prior to affixing and activation of the radiocollar device to captured fossa. Anatomical measurements collected include weight, total body length, tail length, hind limb length, hind foot length, hind limb girth, forelimb length, forefoot length, forelimb girth, chest girth, neck circumference, height at shoulder, ear length, canine anterior-posterior and lateral diameter, carnassial molar lengths, and genital measurements. All anatomical data are collected by the lead PI so as to avoid bias in interobserver measurement techniques.

The morphometrics selected represent a conglomeration of anatomical measurements used in several different realms of mammal ecology. Definitions for most of these measures are derived from Dayan & Simberloff (1994) and Eason Smith & Pelton (1996). Body length is measured from the tip of the nose to the base of the tail. Tail length is measured from the base of the tail to the tip of the most distal bony tail segment (tip of the last tailbone). Hind limb and forelimb length is measured from the medial fold of the limb to the tip of the longest portion of the foot pad. Hind foot and forefoot length is measured from the most proximal to the most distal portion of the foot pad (from the maximum points on the foot). Forelimb girth is measured around the widest portion of the brachium. Hind limb girth is measured around the widest portion of the thigh region. Chest girth is measured just inferior to the forelimbs. Neck circumference is measured at its widest point. Collection of this information fulfills objective 1.

After anatomical measurements, ear-tagging, blood and tissue collection (for population genetic analysis - objective 9 – and for epidemiological study – objective 13), and attachment of a unique color-coded radiocollar are complete, the study animal is returned to the traps at the location of capture, monitored until free from drug effects, and released.

3) Track captured fossa and wildcats using radiotelemetry to determine their home ranges and activity patterns;

Activity and ranging assessments using radiotelemetry

There are two main foci in the use of radiotelemetry in this project. They are: to assess ranging and the pattern of activity in Cryptoprocta and other carnivores, including wild cats, in the study area.

Once the first new radiocollars (1 year life span) are attached and study animals are released each new field season, the radiotracking portion of this project begins. Subgroups of volunteers and staff members collect ranging data on study animals using two-receiver triangulation techniques from recently constructed, permanent radiotelemetry towers at far ends of the research area. This technique involves radiotelemetry teams placing themselves at known, regular monitoring stations/locations (towers) and simultaneously recording the bearing of a given radio-signal from that location. Locations for each study animal are measured twice per day (roughly every 12 hours) when within range. The previous range of our radiotelemetry equipment was limited to less than 2 kilometers. With the completion of 5 m tall radiotelemetry towers (permanent, large antenna at 7 m), that range is increased by up to an order of magnitude. Teams coordinate their measurements using hand-held 5-watt (high output) 2-way radios. The locations of and bearings between the monitoring stations (towers) is known to within 5 m, measured using Global Positioning System (GPS) data collected when the station locations were originally identified.

As the need for additional monitoring sites presents itself, they are mapped onto more remote portions of the study area for future tower placement. Two additional towers are slated for construction within the research area within the next six months, for a total of four monitoring stations. The monitoring stations and all their inter-site proximities are mapped onto x, y axis, and assigned x, y coordinates. The simultaneous signal bearings taken at these stations will be used to calculate an x, y coordinate of the animal’s location at that time using the program TRIANG (White and Garrott, 1990). Using the program WILDTRACK, these location coordinates, taken over time, will be used to calculate movement patterns. Home ranges will be calculated using the minimal convex polygon method of home range estimation.

Only two radio towers are used for regular data collection at present, as only two towers have been constructed. The project vehicle is opportunistically used as a third triangulation point. We will be able to more regularly utilize three position triangulation techniques with the upcoming completion of construction of additional towers.

In addition, measurements of activity patterns are facilitated by the use of activity sensors built into the radiocollars. Volunteers not only note bearing to the study animal from their monitoring towers, but also note the activity level given by different activity-level signals emitted by the radiocollar. This facilitates analyses of daily and seasonal activity patterns based on “resting,” “locally active,” and “traveling” measures.

his schedule of radiotelemetry will continue throughout the ensuing year’s life of each collar, providing information on seasonal changes in ranging, activity pattern, and habitat usage and preferences, helping to meet research objectives 3, 6, 8, 10, & 11.

4) Conduct extensive groundtruthing studies for ongoing remote sensing projects examining trends in fossa habitat;

We examine trends in fossa habitat by using Landsat 5 images as a baseline for forest cover at the beginning of the last decade and conducting further analyses on more recent images. The image or images corresponding to each area of habitat are rectified to recent, geometrically correct Landsat 7 images of the same path and row to an error of less than 12 meters per 100 kilometers. The corrected Landsat 5 baseline image is then overlaid upon the corresponding newer Landsat 7 image to create a single twelve-layer file. This file is then classified using supervised classifications for forest maintained, original nonforest, forest lost, water, and various image masking characters such as shadow, cloud, or smoke. Erdas Imagine 8.4 software is used for image analysis.

The image or images corresponding to each protected area under scrutiny are reclassified multiple times using ground truth data. Ground truthing consists of visiting a point of interest, usually within the protected area, ascertained from preliminary satellite analysis. At this point, precise location is confirmed and reconfirmed via GPS positioning. Visual confirmation of whatever land cover or geological clues may also exist at that point is also taken into account. Thereafter, detailed observations of the terrain and cover from N, S, E, and W venues are collected (and digitally photographed). This information is then used in additional classification iterations of the satellite images at hand before the images are then analyzed for deforestation counts.

5) Continue conservation, development, and capacity building projects in & around Ankarafantsika.

Generations of scientists have been trained just to observe and strictly avoid any forms of interference within the systems where they work. To enclose one’s self into an ivory tower at the front lines of the global biodiversity crisis and merely observe is a luxury that we can no longer afford. The process of conservation occurs at a number of levels, from global to grassroots. It is at the finest of scales – in this case, the Ankarafantsika National Park - that the process of conservation both begins and ends. In Madagascar, Earthwatch scientists have implemented the first hand conservation reconnaissance they have gathered. Working at opposite ends of the island, Dr. Patricia Wright and I have put our impressions to work for the benefit of the habitats we study. We have formed ecological monitoring teams of local people who work first-hand to save “their own backyards.” Both Ankarafantsika and Ranomafana National Parks will have centers for conservation, research, and training finished on-site within the next two years. We are making significant advances in learning from and preserving our research sites by taking first-hand knowledge of our respective areas and infusing the local populations with our same enthusiasm for protecting Madagascar’s biodiversity, while simultaneously going face to face with park managers to make our voices heard. We are not simply gathering data for scientific publication. We are examining a conservation crisis with scientific scrutiny and mounting battles to protect Madagascar’s biodiversity.

The presence of our core team, accentuated by the presence of Earthwatch volunteers facilitates extraordinary conservation and development initiatives at Ankarafantsika. The following will discuss applications of our presence from multiple perspectives, including those of policy and research. The bullet points and discussion following them are all points of action conducted by our team and its individual members.

While our research on the fossa indirectly contributes to conservation and development, it is the impetus for our on-site presence. This presence facilitates our research and development initiatives. Year-round data collection leads team members to be on site year round when tracking animals. Infrastructural support for our teams – as supplied by local communities - gives a sustained, viable alternative to traditional environmentally-unsound practices. Without our research base, this additional work for the conservation and development of Ankarafantsika would not be possible.

the project | research objectives | application of results