Content uploaded by Brian D. Gerber
Author content
All content in this area was uploaded by Brian D. Gerber on Jun 02, 2015
Content may be subject to copyright.
COMPARING DENSITY ANALYSES AND CARNIVORE ECOLOGY IN
MADAGASCAR’S SOUTHEASTERN RAINFOREST
Brian Daniel Gerber
Thesis submitted for the Faculty of the
Virginia Polytechnic Institute and State University
in partial fulfillment of the requirements for the degree of
Master of Science
in
Fisheries and Wildlife Sciences
Sarah M. Karpanty, Chair
Marcella J. Kelly
Dean F. Stauffer
December 7, 2010
Blacksburg, Virginia
Keywords: density, Madagascar, carnivore, habitat
Copyright 2010, Brian D. Gerber
COMPARING DENSITY ANALYSES AND CARNIVORE ECOLOGY IN
MADAGASCAR’S SOUTHEASTERN RAINFOREST
Brian Daniel Gerber
Abstract
Madagascar is renowned for its biodiversity, but also for forest loss, fragmentation, and
degradation, making it a global conservation priority. With few studies dedicated to
Madagascar‟s carnivores, little is known about their ecology. My objectives were to 1) compare
density estimation techniques applicable to enumerating rare and/or elusive carnivores, 2)
investigate Malagasy carnivore distributions, abundance and density, and occupancy/use across
four sites that vary in forest disturbance, and 3) explore temporal activity patterns of rainforest
carnivores. I found the spatially-explicit-capture-recapture models were empirically superior, as
they are flexible and account for spatial variation in detection probability and area estimation. I
found both endemic and exotic carnivore composition varied among four rainforest sites:
Primary, Selectively-logged, Fragments <2.5 km and Fragments >15 km from contiguous-
primary rainforest. All endemic carnivores were present in the Primary and Selectively-logged
rainforest, while endemic carnivore species richness decreased and exotic carnivore species
richness increased in the fragmented forests. Malagasy civet (Fossa fossana) density ± SE was
significantly less in the Selectively-logged compared to the Primary rainforest (1.38 ± 0.22, 3.19
± 0.55 civets/km2, respectively); they were absent from both fragmented forests. Fossa
(Cryptoprocta ferox) density ± SE was not different between the Primary and Selectively-logged
rainforests (0.12 ± 0.05, 0.09 ± 0.04 adults/km2, respectively); a single animal was detected in
the Fragments <2.5 km, while none were detected in the Fragments >15 km. Malagasy
carnivores had varied temporal activity overlap (5.8-88.8%). C. ferox preferred crepuscular
activity, but overall exhibited a cathemeral activity pattern.
iii
Intended authorship and target journals for manuscripts included in this thesis:
Chapter Two:
Empirical evaluation of closure violations, effects of lure, and abundance and density estimation
for mesocarnivores using photographic-sampling data of the Malagasy civet
Brian D. Gerber1, Sarah M. Karpanty1, Marcella J. Kelly1
1 Department of Fisheries and Wildlife Sciences, Virginia Polytechnic Institute and State
University, Blacksburg, Virginia, USA 24061-0321
*In review: Submitted on September 29th, 2010 to Population Ecology
Chapter Three:
The impact of forest logging and fragmentation on carnivore species composition, density, and
occupancy in Madagascar‟s southeastern rainforests
Brian D. Gerber1, Sarah M. Karpanty1, Johnny Randrianantenaina2
1 Department of Fisheries and Wildlife Sciences, Virginia Polytechnic Institute and State
University, Blacksburg, Virginia, USA 24061-0321
2 Centre ValBio, Ranomafana, BP 33, 312 Ifanadiana, MADAGASCAR
* In review: Submitted on October 31st, 2010 to Biological Conservation
iv
Chapter Four:
Temporal activity patterns of Malagasy rainforest carnivores
Brian D. Gerber1, Sarah M. Karpanty1, Johnny Randrianantenaina2
1 Department of Fisheries and Wildlife Sciences, Virginia Polytechnic Institute and State
University, Blacksburg, Virginia, USA 24061-0321
2 Centre ValBio, Ranomafana, BP 33, 312 Ifanadiana, MADAGASCAR
*In review: Submitted on November 1st, 2010 to Journal of Mammalogy
v
Acknowledgments
First and foremost, I would like to thank Belita for being wonderful.
I want to thank my advisor, Sarah Karpanty, for providing the best graduate experience I
could ever have asked for; your consistent encouragement, enthusiasm and guidance has been
instrumental in the fruition of this project as well as my personal advancement as an ecologist
and conservationist. I can‟t thank you enough for your support as a mentor, role model, and
friend. I am also grateful to my committee members, Dean Stauffer and Marcella Kelly, for their
open doors which I made ample use of. Thank you both for your incredible classes and time
spent with me discussing all topics, from ecological theory through applied statistical analyses.
Your dedication to the pursuit of knowledge and conservation is inspiring.
Thanks to all the graduate students in Dr. Karpanty‟s lab (M. Kotschwar, K. Ray, M.
Hillman, Z. Farris) and Dr. Jim Fraser‟s lab (J. Cohen, D. Catlin, J. Dwyer, A. DeRose-Wilson,
and J. Felio). The incredible synergy of interactions and partnerships these labs foster was truly
a unique experience which I am incredibly grateful to have been a part. M. Kotschwar has been
the best colleague to have had in the office and out in the rainforests; thanks for your constant
support and friendship. I am indebted to J. Cohen for his enormous efforts and time spent
teaching me the intricacies of mark-recapture and our more broad discussions of epistemology.
Jim, thanks for sharing the duct tape, eastern shore‟s barrier islands, and your wisdom. Your
dedication and curiosity about life and its mysteries are truly inspiring. I am grateful to everyone
in the Virginia Tech Department of Fisheries and Wildlife Science including all the professors,
graduate students, and staff who have been so incredibly supportive. The dialogue among FIW‟s
community was incredible; thanks to P. Laver, T. Gorman, C. Tredrick, Sunarto, K. Thapa, D.
Bridgers, C. Latimer, R. McManamay, H. Yu, and many more. Many research colleagues have
vi
graciously corresponded with me over the years, thanks to J. Nichols, D. Murray, G. White, L.
Bailey, J. Hines, and M. Linkie. A special thanks to J. Ivan for spending an enormous amount of
time discussing the mark-recapture world with me.
I would also like to thank all the many people who have helped me on the ground in
Madagascar with whom it was my privilege to have worked with: J. Randrianantenaina, M.
Velonabison, D. Andrianoely, F.H. Andriamiharisoa, J. C. Razafimahaimodison, D.
Randrianarisataa, Rafidiarison, B. Marine, M. Kotschwar, C. Latimer, S. Arrigo-Nelson, K.
Banner-Martin, E. Larney, K. Muldoon, Manana, P. Wright, and all of Centre Valbio, Sadabe,
Ialatsara Forest Station and MICET staff. Thanks to Daniel and Berenice Rajaona for their
permission and assistance to study the carnivores at the rainforests of Ialatsara Forest Station.
I am grateful to Sadabe for their permission and guidance working at the beautiful area of
Mahatsinjo, Tsinjoarivo. Thanks to M. Irwin, J.L. Raharison, E. Razanadrakoto, Jules, C.
Razanadrakoto, and all those living and working at Mahatsinjo that worked so hard and treated
me with such kindness. This thesis would have not been possible without the hard work of all
the above remarkable people. I am grateful for the financial support provided by Virginia Tech,
National Geographic Society Committee on Research and Exploration, the National Science
Foundation Graduate Research Fellowship Program, Sigma Xi Master‟s Degree Award, Sigma
Xi Grants-in-Aid, and the Burd Sheldon McGinnes Graduate Fellowship. I thank the
Government of Madagascar, Madagascar National Parks, the Direction des Eaux et Forêts, and
CAFF/CORE for permission to conduct this research.
My entire family has been so supportive with my wanderings and interests in life. Thank
you so much! Lastly, thanks Brittany and Ketan for being there.
vii
Table of Contents
Acknowledgments v
List of Figures x
List of Tables xiii
Chapter 1 – Madagascar Conservation and Carnivore Ecology 1
INTRODUCTION: 1
OBJECTIVE 1: 4
BACKGROUND: 4
CHAPTER 2 OBJECTIVES: 6
OBJECTIVES 2 AND 3: 6
BACKGROUND: 6
MADAGASCAR CONSERVATION: 9
CHAPTER’S 3 AND 4 OBJECTIVES: 11
REFERENCES: 12
Chapter 2 - Empirical evaluation of closure violations, effects of lure, and abundance and
density estimation for mesocarnivores using photographic-sampling data of the Malagasy
Civet 29
ABSTRACT: 29
INTRODUCTION: 30
MATERIALS AND METHODS: 33
STUDY AREA AND SPECIES 33
FIELD METHODOLOGY 34
ANIMAL IDENTIFICATION AND CAPTURE HISTORIES 35
ASSESSING CLOSURE VIOLATION 35
ABUNDANCE AND DENSITY ESTIMATION 36
EFFECT OF LURE ON MOVEMENT AND TEMPORAL ACTIVITY PATTERNS 39
RESULTS: 40
ANIMAL IDENTIFICATION AND CAPTURE HISTORIES 40
ASSESSING CLOSURE VIOLATION 40
EFFECT OF LURE ON ABUNDANCE, DENSITY, MOVEMENTS, AND ACTIVITY 41
COMPARISON OF DENSITY ESTIMATION ANALYSES 42
DISCUSSION: 43
ACKNOWLEDGMENTS: 47
viii
REFERENCES: 48
Chapter 3 - The impact of forest logging and fragmentation on carnivore species
composition, density, and occupancy in Madagascar’s southeastern rainforests 64
ABSTRACT: 64
1. INTRODUCTION: 65
2. MATERIALS AND METHODS: 67
2.1. STUDY AREA 67
2.2. RAINFOREST HABITAT SAMPLING 69
2.3. PHOTOGRAPHIC-SAMPLING 70
2.3.1. CARNIVORE SPECIES COMPOSITION 71
2.3.2. CARNIVORE DENSITY 72
2.3.3. CARNIVORE OCCUPANCY 73
2.3.4. RNP AND MADAGASCAR CARNIVORE POPULATION ESTIMATES 74
3. RESULTS: 75
3.1. VEGETATION STRUCTURE OF RAINFOREST GRIDS 75
3.2. CARNIVORE SPECIES COMPOSITION 76
3.3. CARNIVORE DENSITY 77
3.4. CARNIVORE OCCUPANCY 78
3.5. RNP AND MADAGASCAR CARNIVORE POPULATIONS 80
4. DISCUSSION: 80
5. CONSERVATION IMPLICATIONS: 85
ACKNOWLEDGMENTS: 87
REFERENCES: 88
Chapter 4 - Temporal activity patterns of Malagasy rainforest carnivores 115
ABSTRACT: 115
INTRODUCTION: 116
MATERIALS AND METHODS: 118
STUDY AREAS: 118
SAMPLING: 120
DATA ANALYSES: 120
Temporal Activity Pattern Distribution and Overlap 120
Relative Preference and Selection/Avoidance of Time Periods in the Diel-Cycle 122
RESULTS: 124
DISCUSSION: 127
ACKNOWLEDGMENTS: 133
REFERENCES: 134
ix
Appendix A: Native/Exotic Carnivore Species Accounts for Southeastern Madagascar 158
NATIVE CARNIVORES 158
FOSSA (CRYPTOPROCTA FEROX): 158
MALAGASY SMALL-TOOTHED CIVET (EUPLERES GOUDOTII): 161
MALAGASY CIVET (FOSSA FOSSANA): 163
BROAD-STRIPED MONGOOSE (GALIDICTIS FASCIATA): 165
EXOTIC CARNIVORES: 169
DOMESTIC DOG (CANIS FAMILIARIS): 169
WILD/FERAL CAT (FELIS SP.): 171
REFERENCES: 174
Appendix B: Relevant literature from the ISI Web of Science used to investigate common
mark-recapture analyses. 179
x
List of Figures
Chapter 1:
Figure 1. Madagascar‟s extant carnivore phylogeny.
27
Figure 2. Madagascar‟s primary forest cover (2005) and protected areas. Forest cover was
provided by Conservation International 2010.
28
Chapter 2:
Figure 1. We placed 26 camera stations over a 6.53 km2 area along the Sahamalaotra trail
system within the rainforests of Ranomafana National Park, Fianarantsoa province in
southeastern Madagascar from 9 June–8 August, 2008.
62
Figure 2. Density and 95% confidence limits using four methods of estimation on the complete
dataset for the Malagasy civet (Fossa fossana) with statistical significance (95%
confidence interval of two means overlap less than half the average margin of error)
indicated with different capital letters (α = 0.05). Sampling was carried out using 26
camera stations at the Sahamalaotra trail system within the rainforests of Ranomafana
National Park, Madagascar from 9 June-8 August, 2008.
63
Chapter 3:
Figure 1. Carnivores were sampled at four photographic-sampling sites within Madagascar‟s
eastern rainforests from May-December, 2008 and October-December, 2009. Top insert
map shows political provinces and location on the island of Madagascar, while the
bottom insert map shows the camera station sampling layout among rainforest
fragments at Mahatsinjo, Tsinjoarivo as an example. Rainforest extent from
Conservation International, 2010.
111
Figure 2. Broad-striped mongoose (Galidictis fasciata) occupancy variation by basal area in
contiguous rainforests within Ranomafana National Park and occupancy variation by
fragment size in fragmented rainforests (<2.5 km and >15 km from intact contiguous
rainforest), sampled within Madagascar‟s eastern forests from May-December, 2008
and October-December, 2009.
112
Figure 3. Ring-tailed mongoose (Galidia elegans elegans) occupancy variation within
rainforest fragments and surrounding matrix (open human-dominated habitat) and the
effect of exotic-wild cats (Felis catus/silvestris). Sampling occurred within
113
xi
Madagascar‟s eastern forests from May-December, 2008 and October-December, 2009.
Figure 4. Madagascar‟s primary forest cover (2005) and estimated fossa (Cryptoprocta ferox)
populations >300 adults. Ninety-five percent of the total rainforest population is
contained in two forest complexes, while 80% of the total dry forest population is
contained in two forest complexes. Forest fragments were joined into a single
population when forested areas large enough to hold ≥ 1 fossa were <4.9 km from each
other.
114
Chapter 4:
Figure 1. Carnivores were sampled at four photographic-sampling sites within Madagascar‟s
eastern rainforests from May-December, 2008 and October-December, 2009. Top insert
map shows political provinces and location on the island of Madagascar, while the
bottom insert map shows the camera station sampling layout among rainforest
fragments at Mahatsinjo, Tsinjoarivo as an example. Rainforest extent from
Conservation International, 2010.
150
Figure 2. Temporal activity of Malagasy rainforest carnivores, exotic carnivores, and local
people. Species activity samples were collected at four rainforest sites within
Madagascar‟s eastern forests; data were pooled across sites when activity distributions
were not different. Solid ( ) lines indicate the use of all data from all relevant sites for
each species, circles (○) indicate data from the Primary site, dotted (…) lines indicate
data from the Selectively-logged site, dashed (---) lines indicate data from Fragments
<2.5 km site, and plus signs (+) indicate data from the Fragments >15 km site. Tick
marks on the x-axis represent all activity samples for each species. Y-axis is not
consistent among all graphs.
151
Figure 3. Proportional activity of Malagasy carnivores used for compositional analyses and
average availability of the temporal classes. Species activity samples were collected at
four rainforest sites within Madagascar‟s eastern forests, which were pooled when the
proportional use of activity time classes across sites were not different. Photographic
activity-sampling occurred from May-December, 2008 and October-December, 2009.
154
Figure 4. Selection and/or avoidance of dawn, day, dusk, and night by Galidictis fasciata,
exotic carnivores, and local people. Species activity samples were sampled at four
rainforest sites across Madagascar, which were pooled when the activity proportions of
time classes were not different across sites. The bars indicate the available time for
each time class, while the circular-points indicate the proportion of activity-use and
confidence interval in each time class; dark grey bars indicate the time class was
selected (lower confidence limit of activity-use is above availability) or avoided (upper
155
xii
confidence limit of activity-use is below availability), while white bars indicate use was
in proportion to availability (confidence limits include availability). When there were
no observations in a temporal class and the availability was large, we assumed the
species avoided this time period.
Appendix A:
Figure 1. Rebioma (http://www.rebioma.net) fossa (Cryptoprocta ferox) extent of occurrence
and area of occupancy.
158
Figure 2. Rebioma (http://www.rebioma.net) Small-toothed civet (Eupleres goudotii) extent of
occurrence and area of occupancy.
161
Figure 3. Rebioma (http://www.rebioma.net) Malagasy civet (Fossa fossana) extent of
occurrence and area of occupancy.
163
Figure 4. Broad-striped mongoose (Galidictis fasciata) extent of occurrence based on forest
cover (2005).
165
Figure 5. Ring-tailed mongoose (Galidia elegans elegans) extent of occurrence based on forest
cover (2005).
167
Figure 6. Domestic Dog (Canis familiaris).
169
Figure 7. Wild/Feral Cat (Felis sp.)
171
Figure 8. Small Indian Civet (Vivirricula indica)
173
xiii
List of Tables
Chapter 1:
Table 1. Photographic-capture studies designed to estimate abundance or density and their use
of mark-recapture software programs, statistical models used in program CAPTURE,
closure evaluation tests, and effective sampling area estimation method. Summary
statistics from relevant literature found using the Institute for Scientific Information
Web of Science. Search date 1/16/10, using “camera” and “trap”.
25
Chapter 2:
Table 1. Photographic-sampling summary of the Malagasy civet (Fossa fossana),
sampled with and without lure from 26 camera stations at the Sahamalaotra trail system
within the rainforests of Ranomafana National Park, Madagascar from 9 June-8 August,
2008.
54
Table 2. Model selection results evaluating the effects of lure on geographic closure for the
Malagasy civet (Fossa fossana), sampled with and without lure from 26 camera stations
at the Sahamalaotra trail system within the rainforests of Ranomafana National Park,
Madagascar from 9 June-8 August, 2008.
55
Table 3. Top abundance and density estimation models incorporating detection probability
variation of the Malagasy civet (Fossa fossana), sampled with and without lure from 26
camera stations at the Sahamalaotra trail system within the rainforests of Ranomafana
National Park, Madagascar from 9 June-8 August, 2008.
57
Table 4. Population estimates, the coefficient of variation (CV), and trap success of the
Malagasy Civet (Fossa fossana), sampled with and without lure from 26 camera stations
at the Sahamalaotra trail system within the rainforests of Ranomafana National Park,
Madagascar from 9 June-8 August, 2008.
59
Table 5. Four density estimates and associated coefficient of variation (CV) of the Malagasy
civet (Fossa fossana), sampled with and without lure from 26 camera stations at the
Sahamalaotra trail system within the rainforests of Ranomafana National Park,
Madagascar from 9 June-8 August, 2008.
60
Chapter 3:
Table 1. Photographic-sampling grids at four rainforest sites in Madagascar and summary
98
xiv
capture/recapture data of the individually-identifiable Cryptoprocta ferox and Fossa
fossana. Sampling occurred from May-December, 2008 and October-December, 2009.
Table 2. Grid-level vegetation structure ± SE at four rainforest sites with increasing levels of
disturbance in southeastern Madagascar. Sampling occurred from May-December, 2008
and October-December, 2009. Different letters associated with vegetation variables
across sites indicate statistical significance (experiment-wise α=0.05).
99
Table 3. Model selection (model likelihood >0.125) using a maximum-likelihood spatially-
explicit capture-recapture model to estimate density of the fossa (Cryptoprocta ferox)
and Malagasy civet (Fossa fossana) in a Primary and Selectively-logged contiguous
rainforest grid within Madagascar‟s eastern rainforests. Sampling occurred from May-
December, 2008 and October-December, 2009.
100
Table 4. Density estimates of the fossa (Cryptoprocta ferox) and Malagasy civet (Fossa
fossana) in four rainforest areas within Madagascar with statistical significance (95%
confidence interval of two means overlap less than half the average margin of error)
indicated with different capital letters (α = 0.05). Sampling occurred from May-
December, 2008 and October-December, 2009.
102
Table 5. Model selection (model likelihood >0.125) of occupancy and detection probability for
native and exotic rainforest carnivores in southeastern Madagascar at two contiguous
(Primary and Selective-logged) and two fragmented (< 2.5 and > 15 km from contiguous
forest) sampling grids. Sampling occurred from May-December, 2008 and October-
December, 2009.
103
Table 6. Occupancy analyses of detection probability (p) regression coefficients (β (SE)) for the
top models of endemic and exotic carnivores at two fragmented and two contiguous
rainforest grids within Madagascar‟s eastern forests. Sampling occurred from May-
December, 2008 and October-December, 2009.
106
Table 7. Occupancy analyses (ψ) regression coefficients (β (SE)) for the top models of each
carnivore species at two fragmented and contiguous rainforest grids within Madagascar‟s
eastern forests. Sampling occurred from May-December, 2008 and October-December,
2009.
108
Table 8. Model-averaged occupancy (SE) of native and exotic carnivores in two contiguous and
two fragmented rainforest sites within Madagascar‟s eastern forests. Sampling occurred
from May-December, 2008 and October-December, 2009.
110
xv
Chapter 4:
Table 1. The number of photographic-activity samples for Malagasy carnivores, exotic
carnivores, and local people from four rainforest sites, sampled in two seasons within
Madagascar‟s eastern rainforests. Also, a non-parametric circular test of differences
among activity distributions (continuous time) across all sites and a pair-wise analysis
when the overall test indicated a significant difference; sites were not included when
sample sizes were < 10.
141
Table 2. Comparison of activity distributions of Malagasy and exotic carnivores before and
after the use of scent-lure to detect carnivores at photographic sampling stations.
Sampling occurred within Madagascar‟s eastern rainforest from May-August, 2008 and
October-December, 2009 at a Selectively-logged and fragmented rainforest <2.5 km
from intact primary rainforest, respectively.
143
Table 3. Test of uniform activity distributions of Malagasy carnivores, exotic carnivores, and
local people throughout the diel-cycle. Photographic-activity samples were collected at
four rainforest sites within Madagascar‟s eastern rainforests, sampled from May-
December, 2008 and October-December, 2009; data were pooled if activity distributions
among sites were not statistically different.
144
Table 4. Percentage of temporal overlap (% ± SE) across four rainforest sites within
Madagascar‟s eastern rainforests by species pairs using kernel density analyses.
Photographic activity-sampling occurred from May-December, 2008 and October-
December, 2009. A) Primary (top triangle) and Selectively-logged (bottom triangle)
rainforest sites, B) Fragments <2.5 km from intact forest (top triangle) and Fragments
>15 km from intact forest (bottom triangle).
146
Table 5. Chi-square test results for differences between activity periods (Dawn, Day, Dusk,
Night) of Malagasy carnivores and local people across four rainforest sites and sexes.
Photographic activity-sampling occurred within Madagascar‟s eastern rainforests from
May-December, 2008 and October-December, 2009.
148
Table 6. Compositional analyses of three Malagasy carnivores sampled across four rainforest
sites, sample size, test of random use of dawn, day, night, and dusk, and the ranking of
selection in order of relative preference. Photographic activity-sampling occurred within
Madagascar‟s eastern rainforests from May-December, 2008 and October-December,
2009.
149
1
Chapter 1 – Madagascar Conservation and Carnivore Ecology
Introduction:
The island of Madagascar is one of the most biologically rich areas on the planet and one
of the world‟s leading conservation priorities because of the serious threats to its unique
biodiversity (Myers et al. 2000). It is home to 771 endemic vertebrates (2.8% of global total;
Myers et al. 2000), including 99 species of lemur and nine species of carnivore (Goodman and
Helgen 2010; Irwin et al 2010). While Madagascar is renowned for its wealth of endemic flora
and fauna (Vences et al. 2009), it is also well known for its forest loss, fragmentation, and
degradation.
The main challenge to biodiversity conservation in Madagascar is forest conversion to a
human dominated landscape through slash and burn agriculture, mining, and logging
(Mittermeier et al. 2005). Humbert and Cours Darnes (1965) estimated that in 1953 forest
covered 107,430 km2, while the Joint Research Centre Project estimated that only 55,328 km2
were still remaining by 1999. These analyses showed a mean rate of 9.5% forest loss per year
between 1950 and 1993 and a 1.6% mean rate of forest loss per year between 1993 and 1999
(Dufils 2003; Humbert and Cours Darne 1965). Most recently, Madagascar is estimated to have
only 16% primary forest cover remaining (Harper et al. 2007). The remaining eastern rainforest,
which in 2000 was estimated at 41,668 km2, is confined to a slim band running longitudinally
along the eastern escarpment of the island and is highly fragmented (Harper et al. 2007). Among
all forest types, greater than 80% of forests is within 1 km of a non-forest edge (Harper et al.
2007), such that edge effects (Laurance et al. 2002) have likely altered forest structure and
function significantly across much of Madagascar.
2
Based on these estimates of deforestation from 1950 to 2000, it has been predicted that
9.1% of Madagascar‟s species have become extinct or are assured for extinction within this time
period. In addition to the 32.9% of species predicted to have gone extinct prior to 1950,
Madagascar may have lost 42% of its biodiversity since human colonization (Allnutt et al. 2008).
The extinction potential may be even worse considering that most climate predictions suggest
almost all Madagascar vegetation types will shrink by 2080 (Hannah et al. 2008).
One taxonomic group of special concern in conservation planning activities due to its
uniqueness and potential importance in ecosystem structure and function is the Malagasy
Carnivora clade, which includes nine extant species and seven genera belonging to the endemic
Family Eupleridae (Gaubert et al. 2005; Yoder et al. 2003; Goodman and Helgen 2010).
Eupleridae species previously belonged to a monophyletic Viverridae clade based on
comparative morphometric analyses, but were relatively recently reorganized within the
Feliformia sub-order (Fig. 1; Species accounts provided in Appendix A). Genetic evidence
suggests that the Malagasy carnivores are most closely related to the dominantly African Family
Herpestidae. Roughly 19 to 26 million years ago, an over-water dispersal event brought
carnivores to Madagascar (Krause 2010). The lack of Carnivora competition allowed the
original colonizers to radiate into a broad range of niches. Only a single carnivore species,
Cryptoprocta spelea, is known to have gone extinct. While considerable study of many
Malagasy taxa has directly benefited conservation planning, the dearth of studies on
Madagascar‟s threatened carnivore species (IUCN 2010) has excluded them from consideration
(Kremen et al. 2008). All nine species of extant Malagasy carnivores are categorized on the
IUCN Red List (IUCN, 2010; Endangered: giant-striped mongoose Galidictis grandidieri;
Vulnerable: fossa Cryptoprocta ferox, narrow-striped mongoose Mungotictis decemlineata,
3
brown-tailed mongoose Salanoia concolor; Near Threatened: Malagasy small-toothed civet
Eupleres goudotii and Eupleres major, Malagasy civet Fossa fossana, broad-striped mongoose
Galidictis fasciata; Least Concern: ring-trailed mongoose Galidia elegans). Knowledge is
lacking on the basic ecology and anthropogenic impacts on the populations and habitat use of all
Malagasy carnivores (Irwin et al. 2010). Further, given the known impact of carnivores on lemur
prey in fragmented forests (Irwin et al. 2009), there is concern for lemur persistence as
Madagascar‟s forests shrink, fragment, and degrade due to increasing demands from a growing
human population (Achard et al. 2002; Green and Sussman 1990).
The potential influence of carnivores on ecosystem structure and function also make them
a crucial component of any conservation plan (Noss et al. 1996). Not only do carnivores often
have the largest area requirements (Miquelle et al. 2005; Sergio et al. 2008), they can also exert
regulatory effects on threatened and endangered prey, potentially causing extirpation in
fragmented habitats (Buskirk 1999, 2003; Crooks and Soule 1999; Terborgh et al. 2001).
Carnivores may impact prey species through direct mortality (Paine 1969) and indirect
behavioral effects, such as altering prey distribution and habitat selection. To best plan
landscape conservation strategies within the context of Madagascar‟s Durban vision, which aims
to triple the total protected area in Madagascar and create corridors between new and existing
protected areas (Norris 2006), it is critical to understand the population ecology of the IUCN-
listed Malagasy carnivores (IUCN 2010).
To meet this overarching need for information on carnivore population ecology in
Madagascar, my study had three main objectives:
4
1) Evaluate density estimation techniques useful in enumerating rare and/or elusive
carnivore populations (Chapter 2).
2) Examine population state variables of endemic and exotic rainforest carnivores to
gain insight into Malagasy carnivore ecology and the impact of forest logging and
fragmentation (Chapter 3).
3) Provide a quantitative evaluation of the temporal activity patterns of Madagascar’s
carnivores in the eastern rainforests (Chapter 4).
In the sections that follow, I provide background information and justification for each of
these objectives in greater depth than I am able to do so in each individual chapter which are
formatted as publications.
Objective 1:
Evaluate density estimation techniques useful in enumerating rare and/or elusive carnivore
populations.
Background:
Estimating population abundance and density is fundamental to the study of ecology as
well as critical for appropriate conservation action. A now common approach to estimating the
abundance and density of rare and/or elusive species is the photographic-capture of individuals
and the use of the mark-recapture analytical framework (Karanth et al. 2004; Kays and Slauson
2008). Identifying uniquely marked individuals over two or more sampling occasions (usually
days or weeks) provides the necessary closed capture-recapture data to estimate the probability
5
of detection, , and thus abundance, . The technique for correcting simple animal counts (Mt+1)
or minimum number known alive (MKNA) by incorporating the sampling detection process has
long been recognized as the underpinning to accurately estimating abundance (McKelvey and
Pearson 2001; Nichols 1992; Williams et al. 2001). However, to compare populations across
areas, it necessary to estimate animal density ( = ), which requires an accurate estimate of
the sampling area ( ) and is fraught with many challenges (Parmenter et al. 2003).
Since the initial use of the photographic-capture methodology (Karanth 1995; Karanth
and Nichols 1998; Karanth and Nichols 2002), termed camera-trapping, it has been quickly
adopted as a tool for estimating abundance/density of a diverse array of species, with particular
utility for studies of large and medium terrestrial carnivores (Dillon and Kelly 2007; Gerber et al.
2010; Heilbrun et al. 2006; Jackson et al. 2006; Kelly et al. 2008; O'Brien et al. 2003; Silveira et
al. 2009; Silver et al. 2004). Over the years, substantial effort has been put towards the
refinement and standardization of using camera-traps to provide better data for comparisons
across studies (Balme et al. 2009; Dillon and Kelly 2007; Dillon and Kelly 2008; Kelly 2008;
Maffei and Noss 2007; Silveira et al. 2003; Soisalo and Cavalcanti 2006; Wegge et al. 2004).
Concurrently, there have been significant developments in the analyses of closed capture-
recapture data (Borchers and Efford 2008; Royle et al. 2009; White 2005; White 2008) for the
purposes of estimating abundance or density. Despite the availability of newer, flexible analyses
implemented in the software programs MARK (White and Burnham 1999), DENSITY (Efford et
al. 2009), and SPACECAP (Singh et al. 2010), carnivore photographic-capture studies often still
rely on a very limited set of analytical tools, namely program CAPTURE (Rexstad and Burnham
1991), coupled with ad hoc boundary-strip methods to convert abundance to density (Table 1).
6
Chapter 2 Objectives:
In my comparative examination of the strengths and weaknesses of mark-recapture
analytical techniques for enumerating carnivore abundance/density using photographic-capture
data, I identified two major challenges: the potential biases of edge effects while grid sampling
and/or the use of attractants during sampling. In Chapter 2, I use data collected on the Malagasy
civet, F. fossana to 1) compare and make recommendations regarding methods to account for
geographic closure violation in estimating density, 2) evaluate the effect of lure on closure,
abundance/density estimation, maximum movement distances, and temporal activity patterns
while photographic-sampling, as well as make recommendations for use of lure/bait in future
mesocarnivore studies, and 3) empirically compare the performance of four density estimators
when it is necessary to use closed capture-recapture models with a geographically open and ill-
defined study area and make recommendations for future studies.
Objectives 2 and 3:
Examine population state variables of endemic and exotic rainforest carnivores to
gain insight into Malagasy carnivore ecology and the impact of forest logging and
fragmentation.
Provide a quantitative evaluation of the temporal activity patterns of Madagascar’s
carnivores in the eastern rainforests.
Background:
Carnivores are generally sensitive to habitat loss, fragmentation, and degradation (Crooks
2002; Sergio et al. 2008; Weaver et al. 1996) due to their low densities, high area and energy
7
requirements, and persecution by humans (Cardillo et al. 2004; Noss et al. 1996). The decline or
extirpation of a large- or meso-carnivore can dramatically alter ecosystem dynamics and result in
trophic cascades, which could lead to species extirpations (Berger et al. 2001; Hebblewhite et al.
2005; Ripple and Beschta 2006; Roemer et al. 2009; Terborgh et al. 2001). Many carnivore
species can be useful indicators of habitat disturbance (Soulé and Terborgh 1999) and/or
biodiversity (Sergio et al. 2005; Sergio et al. 2006), making them effective focal species for
conservation planning (Carroll et al. 2001; Paul et al. 2008; Thorne et al. 2006). As such,
carnivores can be critical for biodiversity conservation and ecosystem integrity (Noss et al. 1996;
Sergio et al. 2008; Terborgh et al. 1999).
Understanding the ecology of top and mesocarnivores in Madagascar is especially
important, as 1) most Malagasy carnivores and many of their lemur prey are listed as threatened
or endangered, 2) little is known about the predator-prey dynamics linking the two groups, 3)
threats to Malagasy carnivore and lemur populations continue to be exacerbated by ongoing
habitat loss and fragmentation, and 4) as land is protected and corridors restored as part of the
Durban Vision, a lack of knowledge inhibits the inclusion of habitat requirements and movement
dynamics of the top predators in the planning process (IUCN 2010; WorldBank 2005). The
Durban Vision aims to identify new protected areas to protect Madagascar‟s rich biodiversity
and establish corridors linking protected areas (Norris 2006).
While there have been decades of intensive research on Madagascar‟s lemurs, very little
is understood regarding carnivore predation pressure and impacts on lemurs across habitats or
seasons. As primates represent the largest mammalian prey biomass of the forests of
Madagascar, carnivore predation is likely to be a significant factor in lemur social structure,
behavior, ecology and evolution. Carnivores can affect prey through direct mortality of prey
8
(Paine 1969), as well as indirectly through behavioral effects such as altering prey distribution
and habitat selection (Lima and Dill 1990; Schmitz et al. 1997). To accomplish the conservation
goals of the Durban Vision, it is essential to understand variations in carnivore demographic
parameters, habitat use, and lemur predation across Madagascar‟s diverse biomes from
undisturbed to highly disturbed systems.
The 2010 IUCN Red List of Threatened Species identifies all Malagasy carnivore‟s
populations as declining or currently unknown (IUCN 2010). Major threats to carnivores include
habitat loss and fragmentation, and harvesting by local people (Golden 2009; IUCN 2010;
Kerridge et al. 2003). Of the lemur species in the southeastern rainforests of Madagascar that are
endangered, IUCN lists two as critically endangered (Prolemur simus, Varecia variegata) and
three as endangered (Hapalemur aureus, Propithecus edwardsi Propithecus diadema; IUCN,
2010).
To date, only short-term or rapid-assessment studies of Malagasy carnivores have been
conducted in the eastern rainforests (Dollar 1999a; Dollar 1999b; Dollar et al. 1997; Dunham
1998; Gerber et al. 2010; Goodman 1996; Goodman et al. 2003; Goodman and Pidgeon 1999;
Rasolonandrasana 1994). Long-term research studies have thus far centered in the western dry-
deciduous forests and have only focused on C. ferox (Dollar 2006; Hawkins 1998). There are
limited peer-reviewed research articles available that address Malagasy carnivore ecology (Britt
et al. 2004; Goodman 1996; Goodman et al. 2003; Goodman et al. 1997; Goodman and Pidgeon
1999; Hawkins and Racey 2005, 2008, 2009; Rand 1935; Rasoloarison et al. 1995; Woolaver et
al. 2006; Wright et al. 1997). Natural history accounts, book chapters and non-peer reviewed
publications are also available, but are still relatively few (Albignac 1975; Albignac 1969;
Albignac 1971, 1973; Britt et al. 2001; Dollar 1999a; Dollar 1999b, 2000; Dollar et al. 2007;
9
Dunham 1998; Goodman and Benstead 2003; Karpanty and Wright 2007; Rasolonandrasana
1994). Only a single study has addressed population parameters of Malagasy carnivores in the
eastern rainforests, which was a pilot study leading to this thesis work (Gerber et al. 2010).
Besides this pilot study, Dollar (1999b), Dunham (1998), and Kerridge et al. (2003) have
conducted limited carnivore trapping efforts and subsequent radiotelemetry studies (C. ferox: 1-
yr, 2 adults tracked; G. elegans: 11 trapped, 7 tracked; F. fossana: 22 trapped, 4 tracked) in the
eastern rainforests. The overall paucity of published information on Malagasy carnivores is an
accurate representation of the state of knowledge of these species.
Madagascar Conservation:
Using available GIS information (Kremen et al. 2008), we know that current protected
areas across Madagascar total 40,733 km2 (Fig. 2) When classified by forest type, there is a total
protection of 23,021 km2 of rainforest, 10,440 km2 dry forest, and 1,439 km2 spiny forest.
However, taking into account the dispersion and isolation of most protected areas, there are only
five rainforests, four dry forests, and no spiny forests with greater than 1,000 km2 protected.
Similar to other developing countries, Madagascar has been successful in restraining
deforestation within their protected areas, but forest loss surrounding protected areas has been
considerable (DeFries et al. 2005). At Ranomafana National Park, one of Madagascar‟s most-
prized protected areas, endemic flora diversity is higher and exotics lower within the park than
compared to adjacent unprotected forest (Brown et al. 2009). While park boundaries may afford
some species adequate protection, human disturbance within the Ranomafana National Park can
have a significant effect on lemurs (DeFries et al. 2009) and could be a potentially important
10
factor determining whether carnivores can persist in isolated protected areas (Woodroffe and
Ginsberg, 1998).
Conservation activities in Madagascar are currently moving very slowly and/or on hold
due to ongoing political turmoil which derailed active nationwide conservation progress since
March, 2009. Prior to the conflict, now former President Marc Ravolomanana outlined an
ambitious conservation goal in the Durban Vision, which planned to protect a total of 60,000
km2 of new protected areas and establish corridors linking the new and existing protected areas.
The Durban vision specifically targeted wildlife corridors connecting existing parks in order to
protect rare habitats, watersheds, and the continuity of species‟ ranges (Norris 2006; WorldBank
2005). The application of spatial data and knowledge of a diverse set of endemic taxa has helped
to prioritize conservation by identifying optimal expansions of Madagascar‟s protected areas
(Kremen et al. 2008). The proposed areas would potentially add 2,522 km2 of dry forest and
11,009 km2 rainforest to the current protected areas, albeit, some of which has suffered from
fragmentation and degradation. Two shortcomings of the proposed conservation plan are the
lack of consideration of vegetation zones shifting due to climate change and the exclusion of the
Malagasy carnivores, likely due to the paucity of information on their population ecology and
sensitivity to anthropogenic disturbances (Gerber et al. 2010; Irwin et al. 2010). Given the
isolation and generally limited size of Madagascar‟s protected areas, there may not be a single
protected area that can maintain a viable population of the largest extant carnivore, the fossa
(Cryptoprocta ferox; Hawkins and Racey 2005) over a long period. Since many of the proposed
new conservation areas do not significantly add to already established protected areas, these
proposed areas may still not ensure any one area is large enough for a viable population of C.
ferox. Conservation activities that identify forest corridors to maximize connectivity for both
11
carnivores and other species movements in the face of climate change induced vegetation shifts
would be of great value.
Chapter’s 3 and 4 Objectives:
My study seeks to provide much needed data on the population and behavioral ecology of
carnivores in the rainforests of Madagascar to help address the substantial need for such
information in conservation planning efforts. In Chapter 3, I use non-invasive photographic-
sampling to 1) quantify carnivore species composition, density, and occupancy across a
continuum of disturbed rainforest habitats to begin to understand the impacts of forest logging
and fragmentation on Malagasy carnivores, and 2) investigate the influence of exotic carnivore
species, human activity, landscape characteristics, and micro-habitat on the occupancy of
Malagasy rainforest carnivores. In Chapter 4, I use the temporal activity information derived
from photographic captures to 1) evaluate the temporal activity patterns of Madagascar‟s native
and exotic rainforest carnivores across a continuum of disturbed rainforest, and 2) compare
activity pattern overlap among carnivores to examine temporal niche partitioning and potential
conflict between native and exotic species.
12
References:
Achard, F., Eva, H.D., Stibig, H.-J., Mayaux, P., Gallego, J., Richards, T., Malingreau, J.-P.,
2002. Determination of deforestation rates of the world's humid tropical forests. Science
297, 999-1002.
Albignac, 1975. Breeding the fossa (Cryptoprocta ferox) at Montpellier Zoo. Int Zoo Yearb 15,
147-150.
Albignac, R., 1969. Notes ethologiques sur quelques carnivores malgaches: Le Galidia elegans I.
Geoffroy. Terre et Vie 1969, 202-215.
Albignac, R., 1971. Notes ethologiques sur quelques carnivores Malagaches: Le Mungotictis
lineate Pocock. Terre et Vie 118, 328-343.
Albignac, R., 1973. Faune de Madagascar In Mammiferes Carnivores.
Allnutt, T.F., Ferrier, S., Manion, G., Powell, G.V.N., Ricketts, T.H., Fisher, B.L., Harper, G.J.,
Irwin, M.E., Kremen, C., Labat, J.-N., Lees, D.C., Pearce, T.A., Rakotondrainibe, F.,
2008. A method for quantifying biodiversity loss and its application to a 50-year record
of deforestation across Madagascar. Conservation Letters 1, 173-181.
Balme, G.A., Hunter, L.T.B., Slotow, R., 2009. Evaluating methods for counting cryptic
carnivores. Journal of Wildlife Management 73, 433-441.
Berger, J., Stacey, P.B., Bellis, L., Johnson, M.P., 2001. A Mammalian Predator-Prey Imbalance:
Grizzly Bear and Wolf Extinction Affect Avian Neotropical Migrants. Ecological
Applications 11, 947-960.
Borchers, D.L., Efford, M.G., 2008. Spatially explicit maximum likelihood methods for capture-
recapture studies. Biometrics 64, 377-385.
13
Britt, A., Welch, C., Katz, A., 2001. The impact of Cryptoprocta ferox on the Varecia v.
variegata reinforcement project at Betampona. Lemur News 6, 35-37.
Britt, A., Welch, C., Katz, A., 2004. Can small, isolated primate populations be effectively
reinforced through the release of individuals from a captive population? Biological
Conservation 115, 319-327.
Brown, K., Carter Ingram, J., Flynn, D., Razafindrazaka, R., Jeannoda, V., 2009. Protected Area
Safeguard Tree and Shrub Communities from Degradation and Invasion: A Case Study in
Eastern Madagascar. Environmental Management 44, 136-148.
Buskirk, S.W., 1999. Mesocarnivores of Yellowstone, In Carnivores in ecosystems: The
Yellowstone experience. eds T.W. Clark, A.P. Curlee, S.C. Minta, P.M. Kareiva, pp. 167-
187. Yale University Press, New Haven, CT.
Buskirk, S.W., 2003. Small and mid-sized carnivores, In Mammalian community dynamics:
Management and conservation in the coniferous forests of western North America. eds
C.J. Zabel, R.G. Anthony, pp. 207-249. Cambridge University Press, Cambridge.
Cardillo, M., Purvis, A., Sechrest, W., Gittleman, J.L., Bielby, J., Mace, G.M., 2004. Human
Population Density and Extinction Risk in the World's Carnivores. PLoS Biology 2,
e197.
Carroll, C., Noss, R.F., Paquet, P.C., 2001. Carnivores as focal species for conservation planning
in the Rocky Mountain region Ecological Applications 11, 961-980.
Conservation International, 2010. Center for Applied Biodiversity Science:
https://learning.conservation.org/spatial_monitoring/Forest/Pages/default.aspx. Data
Accessed Last: May 10, 2010.
14
Crooks, K.R., 2002. Relative sensitivities of mammalian carnivores to habitat fragmentation.
Conservation Biology 16, 488-502.
Crooks, K.R., Soule, M.E., 1999. Mesopredator release and avifaunal extinctions in a fragmented
system. Nature 400, 563-566.
DeFries, R., Hansen, A., Newton, A.C., Hansen, M.C., 2005. Increasing Isolation of Protected
Areas in Tropical Forests over the past Twenty Years. Ecological Applications 15, 19-26.
DeFries, R., Rovero, F., Wright, P., Ahumada, J., Andelman, S., Brandon, K., Dempewolf, J.,
Hansen, A., Hewson, J., Liu, J., 2009. From plot to landscape scale: linking tropical
biodiversity measurements across spatial scales. Frontiers in Ecology and the
Environment 8, 153-160.
Dillon, A., Kelly, M.J., 2007. Ocelot Leopardus pardalis in Belize: The impact of trap spacing
and distance moved on density estimates. Oryx 41, 469-477.
Dillon, A., Kelly, M.J., 2008. Ocelot home range, overlap and density: Comparing radio
telemetry with camera trapping. Journal of Zoology 275, 391-398.
Dollar, L., 1999a. Notice of Eupleres goudoutii in the rainforest of southeastern Madagascar.
Small Carnivore Conservation 20, 30-31.
Dollar, L., 1999b. Preliminary Report on the Status, Activity Cycle, and Ranging of
Cryptoprocta ferox in the Malagasy Rainforest, with Implications for Conservation.
Small Carnivore Conservation 20.
Dollar, L., 2000. Assessing IUCN Classifications of Poorly-Known Species: Madagascar's
Carnivores as a Case Study. Small Carnivore Conservation 22, 17-20.
Dollar, L., 2006. Morphometrics, diet, and conservation of Cryptoprocta ferox. Duke University.
15
Dollar, L., Forward, Z., Wright, P.C., 1997. First study of Cryptoprocta ferox in the rainforests of
Madagascar. American Journal of Physical Anthropology Supplment 24.
Dollar, L., Ganzhorn, U., Goodman, S.M., 2007. Primates and Other Prey in the Seasonally
Variable Diet of Cryptoprocta ferox in the Dry Deciduous Forest of Western Madagascar,
In Primate Anti-Predator Strategies. eds S.L. Gursky, K.A.I. Nekaris, pp. 63-76.
University of Chicago Press, Chicago.
Dufils, J.M., 2003. Remaining forest cover, In The Natural History of Madagascar. eds S.M.
Goodman, J.P. Benstead, pp. 88-96. The University of Chicago Press, Chicago.
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, M.G., Borchers, D.L., Byrom, A.E., 2009. Density Estimation by Spatially Explicit
Capture–Recapture: Likelihood-Based Methods, In Modeling Demographic Processes in
Marked Populations. eds D.L. Thomson, E.G. Cooch, M.J. Conroy, pp. 255-269.
Springer, New York.
Gaubert, P., Wozencraft, W.C., Cordeiro-Estrela, Veron, G.E., 2005. Mosaics of Convergences
and Noise in Morphological Phylogenies: What's in a Viverrid-Like Carnivoran?
Systematic Biology 54, 865-894.
Gerber, B., Karpanty, S.M., Crawford, C., Kotschwar, M., Randrianantenaina, J., 2010. An
assessment of carnivore relative abundance and density in the eastern rainforests of
Madagascar using remotely-triggered camera traps. Oryx 44, 219-222.
Golden, C.D., 2009. Bushmeat hunting and use in the Makira Forest, north-eastern Madagascar:
a conservation and livelihoods issue. Oryx 43, 386-392.
16
Goodman, S.M., 1996. The carnivores of the Reserve Naturelle Integrale d'Andringitra,
Madagascar. Fieldiana: Zoology, 289-292.
Goodman, S.M., Benstead, J.P., 2003. Natural History of Madagascar. The Univeristy of
Chicago Press, Chicago.
Goodman, S.M., Helgen, K.M., 2010. Species limits and distribution of the Malagasy carnivoran
genus Eupleres (Family Eupleridae). Mammalia 74, 177-185.
Goodman, S.M., Kerridge, F.J., Ralisoamalala, R.C., 2003. A note on the diet of Fossa fossana
(Carnivora) in the central eastern humid forests of Madagascar. Mammalia 67, 595-597.
Goodman, S.M., Langrand, O., Rasolonandrasana, B.P.N., 1997. The food habits of
Cryptoprocta ferox in the high mountain zone of the Andringitra Massif, Madagascar
(Carnivora, Viverridae). Mammalia: journal de morphologie, biologie, systematique des
mammiferes 61, 185-192.
Goodman, S.M., Pidgeon, M., 1999. Carnivora of the Reserve Naturelle Integrale d'Andohahela,
Madagascar. Fieldiana: Zoology, 259-268.
Green, G.M., Sussman, R.W., 1990. Deforestation history of the eastern rain forests of
Madagascar from satellite images. Science 248, 212-215.
Hannah, L., Dave, R., Lowry, P.P., Andelman, S., Andrianarisata, M., Andriamaro, L., Cameron,
A., Hijmans, R., Kremen, C., MacKinnon, J., Randrianasolo, H.H., Andriambololonera,
S., Razafimpahanana, A., Randriamahazo, H., Randrianarisoa, J., Razafinjatovo, P.,
Raxworthy, C., Schatz, G.E., Tadross, M., Wilmé, L., 2008. Climate change adaptation
for conservation in Madagascar. Biology Letters 4, 590-594.
17
Harper, G.J., Steininger, M.K., Tucker, C.J., Juhn, D., Hawkins, F., 2007. Fifty years of
deforestation and forest fragmentation in Madagascar. Environmental Conservation 34,
325-333.
Hawkins, C.E., 1998. Behaviour and ecology of the fossa, Cryptoprocta ferox (Carnivora,
Viverridae) in a dry deciduous forest, western Madagascar. University of Aberdeen,
Aberdeen.
Hawkins, C.E., Racey, P.A., 2005. Low population density of a tropical forest carnivore,
Cryptoprocta ferox: implications for protected area management. Oryx 39, 35-43.
Hawkins, C.E., Racey, P.A., 2008. Food Habits of an Endangered Carnivore, Cryptoprocta ferox,
in the Dry Deciduous Forests of Western Madagascar. Journal of Mammalogy 89, 64–74.
Hawkins, C.E., Racey, P.A., 2009. A novel mating system in a solitary carnivore: the fossa.
Journal of Zoology 277, 196-204.
Hebblewhite, M., White, C.A., Nietvelt, C.G., McKenzie, J.A., Hurd, T.E., Fryxell, J.M., Bayley,
S.E., Paquet, P.C., 2005. Human activity mediates a trophic cascade caused by wolves.
Ecology 86, 2135-2144.
Heilbrun, R.D., Silvy, N.J., Peterson, M.J., Tewes, M.E., 2006. Estimating Bobcat abundance
using automatically triggered cameras. Wildlife Society Bulletin 34, 69-73.
Humbert, H., Cours Darne, G., 1965. Carte internationale du tapis vegetal et des conditions
ecologiques a 1,000,000. Notice de la carte Madagascar. Travaux de la Section
Scientifique et Technique de l'Institut Francais de Pondichery (hors serie).
Irwin, M.T., Raharison, J.-L., Wright, P.C., 2009. Spatial and temporal variability in predation
on rainforest primates: do forest fragmentation and predation act synergistically? Animal
Conservation 12, 220-230.
18
Irwin, M.T., Wright, P.C., Birkinshaw, C., Fisher, B.L., Gardner, C.J., Glos, J., Goodman, S.M.,
Loiselle, P., Rabeson, P., Raharison, J.-L., Raherilalao, M.J., Rakotondravony, D.,
Raselimanana, A., Ratsimbazafy, J., Sparks, J.S., Wilmé, L., Ganzhorn, J.U., 2010.
Patterns of species change in anthropogenically disturbed forests of Madagascar.
Biological Conservation 143, 2351-2362
IUCN, 2010. IUCN Red List of Threatened Species. Accessed 22 April, 2010.
Jackson, R.M., Roe, J.D., Wangchuk, R., Hunter, D.O., 2006. Estimating snow leopard
population abundance using photography and capture-recapture techniques. Wildlife
Society Bulletin 34, 772-781.
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., Nichols, J.D., 1998. Estimation of tiger densities in India using photographic
captures and recaptures. Ecology 79, 2852-2862 | 2852.
Karanth, K.U., Nichols, J.D., 2002. Monitoring Tigers and Their Prey: A Manual for
Researchers, Managers and Conservationists in Tropical Asia. Centre for Wildlife
Studies, Banglamore.
Karanth, U., Nichols, J., Kumar, S., 2004. Photographic Sampling of Elusive Mammals in
Tropical Forest, In Sampling rare or elusive species. ed. W. Thompson, pp. 229-247.
Island Press, Washington.
Karpanty, S., Wright, P.C., 2007. Predation on lemurs in the rainforest of madagascar by
multiplenpredator species: Observations and experiments, In Primate Anti-Predator
Strategies. eds S.L. Gursky, K.A.I. Nekaris, pp. 77-99. University of Chicago Press,
Chicago.
19
Kays, R.W., Slauson, K.M., 2008. Remote Cameras, In Noninvasive Survey Methods for
Carnivores. eds R.A. Long, P. MacKay, W.J. Zielinski, J.C. Ray, pp. 110-140. Island
Press, Washington.
Kelly, M.J., 2008. Design, evaluate, refine: Camera trap studies for elusive species. Animal
Conservation 11, 182-184.
Kelly, M.J., Noss, A.J., Di Bitetti, M.S., Maffei, L., Arispe, R.L., Paviolo, A., De Angelo, C.D.,
Di Blanco, Y.E., 2008. Estimating puma densities from camera trapping across three
study sites: Bolivia, Argentina, and Belize. Journal of Mammalogy 89, 408-418.
Kerridge, F.J., Ralisoamalala, R.C., Goodman, S.M., Pasnick, S.D., 2003. Fossa fossana,
Malagasy Striped Civet, Fanaloka, In Natural History of Madagascar. eds S.M.
Goodman, J.P. Benstead, pp. 1363-1365. The University of Chicago Press, Chicago.
Krause, D.W., 2010. Biogeography: Washed up in Madagascar. Nature 463, 613-614.
Kremen, C., Cameron, A., Moilanen, A., Phillips, S.J., Thomas, C.D., Beentje, H., Dransfield, J.,
Fisher, B.L., Glaw, F., Good, T.C., Harper, G.J., Hijmans, R.J., Lees, D.C., Jr., E.L.,
Nussbaum, R.A., Raxworthy, C.J., Razafimpahanana, A., Schatz, G.E., Vences, M.,
Vieites, D.R., Wright, P.C., Zjhra, M.L., 2008. Aligning conservation priorities across
taxa in Madagascar with high-resolution planning tools Science 320, 222-225.
Laurance, W.F., Lovejoy, T.E., Vasconcelos, H.L., Bruna, E.M., Didham, R.K., Stouffer, P.C.,
Gascon, C., Bierregaard, R.O., Laurance, S.G., Sampaio, E., 2002. Ecosystem Decay of
Amazonian Forest Fragments: a 22-Year Investigation. Conservation Biology 16, 605-
618.
Lima, S.L., Dill, M., 1990. Behavioral decisions made under the risk of predation: a review and
prospectus. Canadian Journal of Zoology 68, 619-640.
20
Maffei, L., Noss, A.J., 2007. How small is too small? Camera trap survey areas and density
estimates for ocelots in the Bolivian Chaco. Biotropica 40, 71–75.
McKelvey, K.S., Pearson, D.E., 2001. Population estimation with sparse data: the role of
estimators versus indices revisited Canadian Journal of Zoology 79, 754-1765.
Miquelle, D., Stephens, P., Smirnov, E., Goodrich, J., Zaumyslova, O., Myslenkov, A., 2005.
Tigers and wolves in the Russian Far East: Competitive exclusion, functional redundancy
and conservation implications, In Large Carnivores and the Conservation of Biodiversity.
eds R. JC, B. J, R. KH, S. R., pp. 179–207. Island Press, Washington (DC).
Mittermeier, R.A., Hawkins, F., Rajaobelina, S., Langrand, O., 2005. Wilderness conservation in
a biodiversity hotspot. International Journal of Wildernss 11, 42-46.
Myers, N., Mittermeier, R.A., Mittermeier, C.G., da Fonseca, G.A.B., Kent, J., 2000.
Biodiversity hotspots for conservation priorities. Nature 403, 853-858.
Nichols, J.D., 1992. Capture-recapture models. Bioscience 42, 94-102
Norris, S., 2006. Madagascar Defiant. Bioscience 56, 960-965.
Noss, R.F., Quigley, H.B., Hornocker, M.G., Merrill, T., Paquet, P.C., 1996. Conservation
biology and carnivore conservation in the Rocky Mountains. Conservation Biology 10,
949-963.
O'Brien, T., Kinnaird, M., Wibisono, H., 2003. Crouching tigers, hidden prey: Sumatran tiger
and prey populations in a tropical forest landscape. Animal Conservation 6, 131-139.
Paine, R.T., 1969. A note on trophic complexity and community stability. American Naturalist
103, 91-93.
Parmenter, R.R., Yates, T.L., Anderson, D.R., Burnham, K.P., Dunnum, J.L., Franklin, A.B.,
Friggens, M.T., Lubow, B.C., Miller, M., Olson, G.S., Parmenter, C.A., Pollard, J.,
21
Rexstad, E., Shenk, T.M., Stanley, T.R., White, G.C., 2003. Small-mammal density
estimation: A field comparison of grid-based vs. web-based density estimators.
Ecological Monographs 73, 1-26.
Paul, B., Daniel R, M., Wayne D, S., 2008. Forks in the Road: Choices in Procedures for
Designing Wildland Linkages. Conservation Biology 22, 836-851.
Rand, A.L., 1935. On the habits of some Madagascar mammals. Journal of Mammalogy 16, 89-
104.
Rasoloarison, R.M., Rasolonandrasana, B.P.N., Ganzhorn, J.U., Goodman, S.M., 1995.
Predation on vertebrates in the Kirindy Forest, Western Madagascar. Ecotropica 1, 59-65.
Rasolonandrasana, B.P.N., 1994. Contribution a' l'etude de l'alimentation de Cryptoprocta ferox
Bennett (1833) dans son milieu naturel. University of Antananarivo, Antananarivo,
Madagascar, Antananarivo.
Rexstad, E., Burnham, K.P., 1991. Users Guide for Interactive Program CAPTURE Abundance
Estimation of Closed Population Animal Populations. Colorado State University, Fort
Collins.
Ripple, W.J., Beschta, R.L., 2006. Linking a cougar decline, trophic cascade, and catastrophic
regime shift in Zion National Park. Biological Conservation 133, 397-408.
Roemer, G.W., Gompper, M.E., Valkengurgh, B.V., 2009. The Ecological Role of the
Mammalian Mesocarnivore. Bioscience 59, 165-173.
Royle, J.A., Nichols, J.D., Karanth, K.U., Gopalaswamy, A.M., 2009. A hierarchical model for
estimating density in camera-trap studies. Journal of Applied Ecology 46, 118-127.
Schmitz, O.J., Beckerman, A.P., O‟Brien, K.M., 1997. Behaviorally mediated trophic cascades:
effects of predation risk on food web interactions. Ecology 78, 1388–1399.
22
Sergio, F., Caro, T., Brown, D., Clucas, B., Hunter, J., Ketchum, J., McHugh, K., Hiraldo, F.,
2008. Top Predators as Conservation Tools: Ecological Rationale, Assumptions, and
Efficacy. Annual Review of Ecology, Evolution, and Systematics 39, 1-19.
Sergio, F., Newton, I., Marchesi, L., 2005. Conservation Top predators and biodiversity. Nature
436, 192-192.
Sergio, F., Newton, I., Marchesi, L., Pedrini, P., 2006. Ecologically justified charisma:
preservation of top predators delivers biodiversity conservation. Journal of Applied
Ecology 43, 1049-1055.
Silveira, L., Furtado, M.M., Tôrres, N.M., Sollmann, R., Uhl, G., Jácomo, A.T.D.A., 2009.
Maned Wolf density in a central Brazilian grassland reserve. Journal of Wildlife
Management 73, 68-71.
Silveira, L., Jacomo, A.T.A., Diniz-Filho, J.A.F., 2003. Camera trap, line transect census and
track surveys: a comparative evaluation. Biological Conservation 114, 351-355.
Silver, S.C., Ostro, L.E.T., Marsh, L.K., Maffei, L., Noss, A.J., Kelly, M.J., Wallace, R.B.,
Gomez, H., Ayala, G., 2004. The use of camera traps for estimating jaguar Panthera
onca abundance and density using capture/recapture analysis. Oryx 38, 148-154.
Singh, P., Gopalaswamy, A.M., Royle, A.J., Kumar, N.S., Karanth, K.U., 2010. SPACECAP (v
1.0): A Program to Estimate Animal Abundance and Density using Bayesian Spatially-
Explicit Capture-Recapture Models. Wildlife Conservation Society - India Program,
Centre for Wildlife Studies, Bangalure, India.
Soisalo, M.K., Cavalcanti, S.M.C., 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.
23
Soulé, M.E., Terborgh, J., 1999. Continental conservation: scientific foundations of regional
reserve networks. Island Press, Washington D.C., USA.
Terborgh, J., Estes, J., Paquet, P., Ralls, K., Boyd-Heger, D., Miller, B., Noss, R., 1999. The role
of top carnivores in regulating terrestrial ecosystems, In Continental Conservation:
Scientific Foundations of Regional Reserve Networks. eds M.E. Soulé, J. Terborgh.
Island Press, Washington.
Terborgh, J., Lopez, L., Nunez, V.P., Rao, M., Shahabuddin, G., Orihuela, G., Riveros, M.,
Ascanio, R., Adler, G.H., Lambert, T.D., Balbas, L., 2001. Ecological meltdown in
predator-free forest fragments. Science 294, 1923-1926.
Thorne, J.H., Cameron, D., Quinn, J.F., 2006. A conservation design for the central coast of
California and the evaluation of mountain lion as an umbrella species. Natural Areas
Journal 26, 137-148.
Vences, M., Wollenberg, K.C., Vieites, D.R., Lees, D.C., 2009. Madagascar as a model region of
species diversification. 24, 456-465.
Weaver, J.L., Paquet, P.C., Ruggiero, L.F., 1996. Resilience and conservation of large carnivores
in the Rocky Mountains. Conservation Biology 10, 964-976.
Wegge, P., Pokheral, C., Jnawali, S., 2004. Effects of trapping effort and trap shyness on
estimates of tiger abundance from camera trap studies. Animal Conservation, 251–256.
White, C., 2005. Correcting wildlife counts using detection probabilities. Wildlife Research 32,
211-216.
White, G.C., 2008. Closed population estimation models and their extensions in Program MARK
Environmental and Ecological Statistics 15, 89-99.
24
White, G.C., Burnham, K.P., 1999. Program MARK: survival estimation from populations of
marked animals. Bird Study 46, S120-138.
Williams, B.K., Nichols, J.D., Conroy, M.J., 2001. Analysis and Management of Animal
Populations. Academic Press, San Diego.
Woodroffe, R., Ginsberg, J.R., 1998. Edge effects and the extinction of populations inside
protected areas. Science 280, 2126-2128.
Woolaver, L., Nichols, R., Rakotombololona, W.F., Volahy, A.T., Durbin, J., 2006. Population
status, distribution and conservation needs of the narrow-striped mongoose Mungotictis
decemlineata of Madagascar. Oryx 40, 67-75.
WorldBank, 2005. Third Environment Program Support Project: The World Bank. Forest
Resources in Madagascar, In Report for USAID/Madagascar. ed. I.R. Group. The World
Bank, Washington, D.C.
Wright, P.C., Heckscher, S.K., Dunham, A.E., 1997. Predation on Milne-Edward's sifaka
(Propithecus diadema edwardsi) by the fossa (Cryptoprocta ferox) in the rain forest of
southeastern Madagascar. Folia Primatologica 68, 34-43.
Yoder, A.D., Burns, M.M., Zehr, S., Delefosse, T., Veron, G., Goodman, S.M., Flynn, J.J., 2003.
Single origin of Malagasy Carnivora from an African ancestor. Nature 421, 734-737.
25
Table 1. Photographic-capture studies designed to estimate abundance or density and their use of
mark-recapture software programs, statistical models used in program CAPTURE,
closure evaluation tests, and effective sampling area estimation method. Summary
statistics from relevant literature found using the Institute for Scientific Information Web
of Science. Search date 1/16/10, using “camera” and “trap”.
Criterion
Number of Studies
Applicable to
Criteriona
Totalb
%
Study Objective
Density
29
30
97
Abundance
1
30
3
Closure Evaluation Test
Otis 1978
21
30
70
Not Mentioned
7
30
23
Stanley Test
2
30
7
Pradel
2
30
7
Mark-Recapture Software Program
CAPTURE
28
30
93
MARK
2
30
7
Neither
1
30
3
CAPTURE Model Use
Mh Jacknife Only
23
28
82
26
Mh Jacknife and Other Models
2
28
7
Not Mh Jacknife Model
3
28
11
CAPTURE‟s Model Selection selected
M0, but researcher used Mh Jackknife
16
23
70
Area Estimation Methodc
HMMDM Only
15
28
54
Multiple Methods (HMMDM+MMDM)
9
28
32
No Buffer
2
28
7
Home Range Radius
2
28
7
MMDM Only
0
28
0
aTotal number of studies may not sum to total, as some studies uses multiple methods or models.
bSee Appendix B for complete literature citations
cHMMDM and MMDM are the half and complete mean maximum distance moved of animals
detected, respectively
27
Figure 1. Madagascar‟s extant carnivore phylogeny.
Phylum Chordata
Class Mammalia
Order Carnivora
Suborder Feliformia
Family Eupleridae
Subfamily Euplerinae
Cryptoprocta
ferox
Fossa
fossana
Eupleres
goudotii
Eupleres
major
Subfamily Galidiinae
Galidia
elegans
Galidictis
fasciata
Galidictis
grandideri
Mungotictis
decemlineata
Salanoia
concolor
28
Figure 2. Madagascar primary forest cover (2005) and protected areas. Forest cover was
provided by Conservation International (2010).
29
Chapter 2 - Empirical evaluation of closure violations, effects of lure, and abundance and
density estimation for mesocarnivores using photographic-sampling data of the Malagasy
Civet
Abstract:
1. Estimating abundance and density of rare and/or elusive carnivores with capture-recapture
analyses is increasingly common. However, providing unbiased and precise estimates is still
a challenge due to 1) strict assumptions of geographic closure, which large-ranging
carnivores often violate, 2) uncertainties in area estimation when using ad hoc boundary-strip
methods to compensate for edge effects, and 3) the use of bait or lure to attract animals to the
detection device, which is rarely tested for potential biases.
2. We use photographic-sampling data of the Malagasy civet Fossa fossana collected with and
without lure to evaluate tests of geographic closure and to determine effects of lure on
geographic closure, abundance and density estimation, maximum movement distances, and
temporal activity patterns. In addition, we compare the performance of four density
estimators: 1) dividing the superpopulation ( s) by the effective sampling area (ESA),
calculated with different ad hoc boundary-strip methods ( = s/ESA), 2) dividing Core-only
abundance ( ) by the sampling area ( = /A), 3) estimating density directly with a
maximum-likelihood (ML-SECR) and 4) Bayesian spatially-explicit capture-recapture model
(B-SECR).
3. SECR model density estimates ± SE were the least precise as they incorporate spatial
variation (CV = 0.11-0.13), but consistent with each other (ML-SECR = 1.38 ± 0.18, B-
30
SECR = 1.24 ± 0.17 civets/km2), while estimates relying on boundary-strip methods do not
incorporate spatial variation, varied greatly and were generally larger than SECR model
estimates (1.26 - 2.85 civets/km2). The use of lure did not influence geographic closure,
abundance and density estimation, maximum movement distances, or temporal activity
patterns, but did provide more precise population estimates than when not using lure.
4. Synthesis and Applications. Utilizing lure to sample carnivores can provide significant
benefits, including improved precision with less effort, which can permit either greater
sampling coverage or reduced project costs. Estimating carnivore density with ad hoc
boundary-strip methods often leads to overestimation and/or increased uncertainty in density
estimates as they do not incorporate spatial variation. This may lead to inaction or poor
decisions by management agencies that can jeopardize populations, especially those already
at risk. In contrast, SECR models free researchers from making subjective decisions with
boundary-strip methods and estimate density directly, providing more comparable and
valuable population estimates.
Introduction:
Unbiased and precise estimates of abundance and density are fundamental to the study of
population ecology and essential for effective conservation and management decisions. A
common approach to estimating the abundance and density of a species is to capture, mark, and
recapture animals to apply capture-recapture (hereafter „C-R‟) analyses (White et al. 1982).
While previously less common, utilizing C-R to quantify the populations of carnivores is
increasingly widespread. This is due to the successful implementation of remote sampling
techniques, such as hair snares or scat collection which allow the isolation of individually-
31
identifiable DNA markers and photographic-sampling of species with uniquely identifying
physical marks (Long et al. 2008).
Given the small sample sizes encountered in most carnivore studies and the nearly
universal finding that detection probability is affected by heterogeneity among animals and
occasional trap responses (Noyce et al. 2001; Boulanger et al. 2004a; Boulanger et al. 2004b),
carnivore biologists primarily implement closed, versus open, C-R models to estimate abundance
( ), but see Karanth et al. (2006). To compare populations across areas it is necessary to convert
abundance to density ( ), yet traditional C-R analyses provide no direct estimate of . must
be divided by the sampling area (A) to estimate density ( = /A). However, unless the
sampling area is confined by natural barriers (Mace et al. 1994), at least some sampled
individuals will have home ranges that extend beyond the edges of the sampling area, thus
violating the basic assumption of geographic closure and positively biasing due to this “edge
effect” (White et al. 1982; Boulanger and McLellan, 2001). Given that 1) many carnivores have
large home ranges and 2) financial and logistical constraints generally prohibit sampling areas of
necessary size (Bondrup-Nielsen, 1983) or simultaneously tracking animals across these edges
(White and Shenk, 2001), the edge effect is likely to be substantial when sampling carnivores
using grids (Greenwood et al. 1985; Mowat and Strobeck, 2000; Boulanger and McLellan,
2001).
If we assume movements across the sampling area edge are random, will likely not be
biased, but will correspond to the superpopulation ( s), or those animals that occupy the
sampling area and an unknown amount of the surrounding area (Kendall, 1999). To accurately
estimate density of what is actually a geographically open population using closed C-R models,
it is necessary to estimate the effective sampling area (ESA; Wilson and Anderson, 1985), or the
32
area that pertains to the s estimate ( = s/ESA). Despite this frequent need to estimate the
ESA, there is still much debate on a robust solution; most recommendations suggest variations on
ad hoc boundary-strip methods (Soisalo and Cavalcanti, 2006; Maffei and Noss, 2007; Dillon
and Kelly, 2008; Balme et al. 2009).
Spatially-explicit C-R models (Efford et al. 2009b; Royle et al. 2009) incorporate the
spatial component of the sampling array in the C-R framework, thereby estimating density
directly without the need of an ad hoc ESA estimate. Field studies have recently provided
empirical support for the use of a maximum likelihood spatially-explicit C-R model (ML-SECR;
Obbard et al. 2010) and a Bayesian spatially-explicit C-R model (B-SECR; Gardner et al. 2010)
to estimate density of geographically open populations of a large ranging carnivore, Ursus
americanus.
In addition to the challenge of dealing with geographically open populations, carnivores
often have low detection rates, even with intense sampling efforts, which inhibit the application
of even closed C-R analyses (White et al. 1982; McCarthy et al. 2009). Thus, carnivore C-R
studies, especially those using hair snares, often utilize bait (food reward) or lure (non-food
reward) to attract animals to the detection device to more effectively (re)capture individuals
(Obbard et al. 2010; Gardner et al. 2010). In contrast, photographic-sampling studies less
frequently use bait or lure (Trolle et al. 2007; Gerber et al. 2010), but rather often place cameras
on trails to increase detection (Dillon & Kelly 2007). Few studies have examined the influence
that these attractants may have on C-R population estimation. Using attractants can potentially
increase the sample size of detected and/or repeated detections of individuals, and thereby
increase detection probability for closed C-R analyses. The advantages include more efficient
model selection, increased estimate precision, and the need for less sampling length/effort, thus
33
reducing project costs (White et al. 1982). However, attractants may also introduce bias to the
density estimate, irrespective of, or in combination with, the edge effect (Mowat and Strobeck,
2000; Gardner et al. 2010), by disrupting natural spatial and temporal movement patterns within
the sampling area, “pulling” animals onto the sampling area, and/or deterring a proportion of the
population (e.g. by sex or age) from being detected (Noyce et al. 2001).
To appropriately estimate carnivore density given the potential biases of edge effects
and/or attractants, it is necessary to assess and account for violations of the closure assumption in
C-R abundance and density estimation. In this paper, we 1) compare and make
recommendations regarding methods to account for geographic closure violation in estimating
density of the Malagasy civet Fossa fossana, Müller 1776, 2) evaluate the effect of lure on
closure, abundance and density estimation, maximum movement distances, and temporal activity
patterns while photographic-sampling, as well as make recommendations for use of attractants in
future mesocarnivore studies, and 3) empirically compare the performance of four density
estimators when it is necessary to use closed C-R models with a geographically open and ill-
defined study area and make recommendations for future studies.
Materials and Methods:
Study Area and Species
We studied Madagascar‟s third largest endemic carnivore, the near-threatened F. fossana,
at the Sahamalaotra trail-system within Ranomafana National Park from 9 June-8 August, 2008
(Fig. 1; IUCN, 2010). Sahamalaotra is montane rainforest, characterized by a 20-25 m tree
canopy dominated by Tambourissa and Weinmannia (Turk, 1997). F. fossana is a
34
mesocarnivore averaging 1.6 kg and 0.91 m in length; populations are declining due to habitat
loss and local hunting (Kerridge et al. 2003; IUCN, 2010).
Field Methodology
We deployed 26 passive-infrared camera sampling stations on trails in a systematic grid
with a random starting point using Deercam DC300‟s (DeerCam, Park Falls, USA) and Reconyx
PC85‟s (Reconyx, Inc., Holmen, Wisconsin). The photographic-sampling grid was designed
based on a preliminary study (Gerber et al. 2010) and had 3.98 stations km-2 with an average
distance and standard deviation of 566 ± 93 m between adjacent stations. Sampling stations
consisted of two independently-operating camera-traps mounted on opposite sides of a trail to
provide a photographic-capture of both flanks of each animal, thus improving individual
identification in recaptures. Cameras were approximately 20 cm above the ground and set to be
active for 24 hours/day.
We sampled for 61 nights; during the first 36 nights we did not deploy attractants.
Starting on the 37th night, 1-2 kg of chicken meat was secured within three layers of metal-wire-
mesh at all sampling stations for an additional 25 nights of sampling. Chicken was inaccessible
for consumption and acted as a scent-lure. We hung most of the chicken lure 2 m directly above
the sampling station on a line tied between two trees. We also staked a small piece of chicken
wrapped tightly in three layers of metal-wire-mesh on the ground. We checked sampling
stations every five days to ensure continued operation, replacing batteries, film, and memory
cards when necessary. We replaced chicken at least every other visit to ensure a maximum-
volatile olfactory signal. By maintaining a strict schedule, we ensured that there was no time
35
when lure was absent from any sampling station, thus reducing among-station heterogeneity
(Zielinski and Kucera, 1995).
Animal Identification and Capture Histories
Using F. fossana’s individually-identifiable spot pattern (Gerber et al. 2010), two
researchers independently agreed on the individual-identity of 96% of all capture events (N =
469) used to construct the capture histories necessary for closed C-R analyses. A capture event
was all photographs of an individual within a 0.5 hour period (O'Brien et al. 2003). We created
three datasets for comparison, 1) capture/recaptures from the complete sampling period (61
nights), 2) capture/recaptures from only the non-lure period (36 nights), and 3)
capture/recaptures from only the lure period (25 nights). A sampling occasion was a 24-hour
period from 12:00 PM to 11:59 AM.
Assessing Closure Violation
We assumed demographic closure and used three methods to evaluate geographic
closure. First, we used the closure hypothesis test of Otis et al. (1978), which assumes only
heterogeneity in the recapture probability. Second, we emulated the Stanley and Burnham
(1999a) closure test that assumes only time variation in recapture probability using the Pradel
model (Pradel, 1996) in Program MARK (v 5.1; White, 2008). Third, we used the full
capabilities of the Pradel model to evaluate geographic closure by estimating site fidelity (ф),
recruitment (f), recapture probability (p), and the composite variable of sampling area population
growth rate (λ; Boulanger and McLellan, 2001). We included a priori biologically plausible
models in this full Pradel analysis. Models included the effect of lure (lure) as a simple time
36
effect between the non-lure and lure periods, males versus females (sex), and general location of
animals on the camera grid (location). We classified location for each individual as either
“Core”, individuals that were on-average detected within the interior of the sampling area, or
“Edge”, animals that were only detected at camera stations on the edge of the sampling area.
We evaluated models using Akaike‟s Information Criterion with a small sample size bias
correction (AICc) and considered all models with ΔAICc < 2 equally parsimonious; we model-
averaged estimates to incorporate uncertainty (Burnham and Anderson, 2002). We calculated
the relative importance of a parameter (Ri) as the sum of AICc weights of all models containing
the variable. We estimated overdispersion (ĉ) with a bootstrap goodness-of-fit test using the
Cormack-Jolly-Seber model (Boulanger and McLellan, 2001). Interactive models were
prohibitive, thus using our global model ф(location + lure + sex) p(location + lure + sex), we
estimated ĉ equal to 1.17. A ĉ correction was incorporated into model selection, so we present
QAICc values. If geographic closure is met using the Pradel analysis, we expect site fidelity (ф)
to be one, immigration (f) to be zero, and thus the study area population growth rate (λ) to be
one.
Abundance and Density Estimation
We used four methods to estimate density using the complete, non-lure, and lure datasets
for F. fossana. We defined a significant difference between methods when the 95% confidence
intervals of two means overlap no more than half the average margin of error; this is equivalent
to a conservative hypothesis test at α = 0.05 (Cumming and Finch, 2005).
First, we assumed random movement across the sampling edge (Kendall, 1999) and
estimated s for all three datasets using the Huggins closed C-R model (Huggins, 1991) in
37
Program MARK. We constructed models using heterogeneity (h, 2-point mixture model), time
(time), behavior (b), sex, mean-capture distance to the sampling area edge (distedge), and mixed
combinations. A lure effect was included in the complete dataset. We evaluated model fit using
AICc. We calculated density by dividing the model-averaged s by the ESA and calculated
variance using the delta method (Karanth and Nichols, 2002).
We derived four variations of the ESA by calculating the expected half (½MMDM*) and
expected full (MMDM*) mean maximum distance moved as the MMDM is known to increase
with increasing recaptures (Tanaka, 1972). We calculated MMDM* for observed animals as,
E = W*(1-e-(i-1)b) (eqn 1)
where is the MMDM for animals captured (i) times, W* is the expected maximum distance
moved for the given population, and b represents a model parameter (Jett and Nichols, 1987).
We evaluated W* using a likelihood function, hereafter referred to as MMDM*, in two ways, 1)
using all animals detected at least twice (MMDM*) and 2) using the Core subset of animals
(MMDM-Core*). We assumed Core animals are less likely to have truncated maximum
movement distances. In contrast, Edge animals are very likely to have a maximum distance
moved of zero (having not been detected at multiple stations) or a truncated distance as their
home range is mostly outside the sampling area. MMDM*, ½MMDM*, MMDM-Core*, and
½MMDM-Core* values were applied as circular buffers to each sampling station, dissolving
overlapping areas to calculate the ESA. We removed villages, roads, and agricultural land (non-
habitat) from these buffered areas and restricted area estimation north of the Namorana river
(Fig. 1) as it likely restricts regular movement (Gerber et al. 2010).
Second, we used the Huggins model to estimate of only the Core animals. We assume
Core animals‟ home ranges are contained in the study area, thus pertains directly to the
38
sampling area (A) and no ad hoc buffer value was needed. We used the same candidate models
from the s analysis to evaluate capture histories. We model-averaged to obtain Core-only
and divided by A to calculate density; the variance was derived by dividing variance by the
square of A (Weinberg and Abramowitz, 2008).
Third, we use Program DENSITY‟s ML-SECR model (v. 4.4; Efford, 2009a) to directly
estimate density. The likelihood function was evaluated with a 2-dimensional numerical
integration using 4096 evenly distributed points within a rectangular area extending 1 km beyond
the sampling area edge. We assumed home range centers were Poisson distributed, removed
non-habitat, and again restricted area estimation north of the Namorana river. We compared the
fit of three detection functions (half-normal, hazard-rate, and negative-exponential) to model
detection probability variation away from an animal‟s home range center. We used a
conditional-likelihood function to model both g0 (detection process when a single detector is
located at the center of an animal‟s home range) and σ (spatial scale detection process away from
the center of the home range) using a priori biologically plausible models. The same variables
modeling detection probability in the s were used, except we excluded the distedge covariate.
Model fit was evaluated using AICc and we model-averaged results to derive and associated
variance.
Fourth, we used the R package SPACECAP (v. 1.0) to apply the B-SECR model to
estimate density. To compare with the ML-SECR estimates, we used the same 2-dimensional
numerical integration, removed non-habitat and restricted area estimation north of the Namorana
river. We allowed incorporation of a trap response in the model for all three datasets and ran
60000 Markov chain Monte Carlo iterations. SPACECAP is limited to the half-normal detection
function.
39
Effect of Lure on Movement and Temporal Activity Patterns
To test the effect of lure on individual‟s movements, we calculated the maximum
distance moved (MaxDM) for all individuals and Core animals only, before and after lure was
applied. We tested whether individuals detected during both sampling periods change their
MaxDM using the Wilcoxon Signed Rank test (Zar, 1998). We also tested whether MMDM of
individuals captured ≥ two times in each the non-lure and lure sampling periods is different using
all individuals (MMDM) and Core animals only (MMDM-Core) using the Wilcoxon Ranked
Sum test (Zar, 1998). Lastly, we contrast the MMDM* and MMDM-Core* for the non-lure and
lure sampling periods.
We evaluated the effect of lure on the temporal activity of F. fossana by testing if activity
distributions from data collected with and without lure were different using the non-parametric
circular Mardia-Watson-Wheeler statistical test (MWW; Batschelet 1981). In addition, we
estimated the mean temporal overlap between activity distributions using a kernel density
analysis (Ridout and Linkie, 2009). We defined a sample as the median time of all photographs
of the same individual within a 0.5 hour period, thus avoiding the issues of non-independence of
consecutive photographs (O'Brien et al. 2003). We applied a kernel estimator from Ridout and
Linkie (2009; see eqn 3.3, smoothing parameter of 1.00). We tested for a shift in the proportion
of activity in four temporal classes based on sunrise/sunset times during this study: dawn (5:25-
7:24), day (7:25-16:27), dusk (16:28-18:27), and night (18:28-5:24). We derived the proportion
of activity for each temporal class from the kernel probability distribution and used a
contingency table analysis with a likelihood ratio test to examine if animals spent a different
amount of time in any temporal class after lure was applied at the sampling stations. We
40
considered a difference (α = 0.05) in the activity distributions between the non-lure and lure
datasets and/or a shift of activity among the four temporal classes to indicate a change in activity
pattern due to lure.
Results:
Animal Identification and Capture Histories
We photographically captured 22 individual F. fossana from 61 sampling nights (Table
1). Eighteen of 22 individuals were detected in both the non-lure and lure periods; two unique
individuals were detected only in the non-lure period and two unique individuals only in the lure
period. We observed F. fossana attempt to remove the staked-ground lure in only 6% of digital-
camera capture events and did not observe any chicken being removed in 915 film images or
2296 digital images.
Assessing Closure Violation
We found the F. fossana population to be geographically closed or open depending on
the method employed. The Otis et al. (1978) test did not reject the closure assumption during the
non-lure period (Z = -1.153, P = 0.12), but did for both the lure period (Z = -2.771, P = 0.002)
and the complete dataset (Z = -2.982, P = 0.001). The Stanley and Burnham (1999a) test
similarly rejected the closure assumption, as the model constraining site fidelity (ф) to one and
immigration (f) to zero was given no support using only the Stanley and Burnham models
(QAICc Weight = 0.00; Table 2).
We found no evidence of permanent closure violation using the full Pradel analysis as the
top model included side fidelity (ф) and immigration (f) as constant, and recapture probability
41
(p) varying by location and the use of lure (Table 2). Model-averaged λ ± SE for the complete,
non-lure, and lure datasets were estimated at 1.00 ± 0.004, 1.00 ± 0.006, and 0.995 ± 0.008,
respectively. While there was no evidence of permanent closure violation, recapture probability
± SE was significantly higher for Core animals (Non-lure = 0.35 ± 0.04, Lure = 0.48 ± 0.04) than
Edge animals (Non-lure = 0.13 ± 0.02, Lure = 0.21 ± 0.03), indicating potential closure violation
by Edge animals temporarily emigrating from the sampling area, thus producing an edge effect.
Effect of Lure on Abundance, Density, Movements, and Activity
Detection probability was affected by heterogeneity, behavior, sex, and lure in most of
our selected models for s, Core-only , and of F. fossana (Table 3). We found that
complete-dataset models estimating s for F. fossana included effects of h (Ri = 100%), distedge
(Ri = 100%), b (trap-happy; Ri = 100%), sex (Ri =100%), and lure (Ri = 93%) on the probability
of detection. All models included h (Ri = 100%) in detection probabilities to estimate Core-only
. Additionally, a trap-happy b effect on the detection probability was clear in the Core-only
complete dataset (Ri = 100%) and the non-lure (Ri = 97%), but not in the lure dataset (Ri = 49%).
Males were detected more often than females when using the complete dataset for Core-only
(Ri = 98%), but an effect of sex was less evident for the non-lure (Ri = 68%), and lure (Ri = 15%)
datasets. In the ML-SECR model, we found the negative-exponential function fit all three
datasets best and variation in g0 and σ was best explained by h and/or sex (Table 3). Model
selection for the B-SECR analysis is unavailable in SPACECAP (v. 1.0), thus estimates are a
priori “best” models.
We found no effect of lure on estimates of s, (Table 4), and for each density
estimation technique (Table 5). We found higher average detection probabilities ± SE in our
42
analyses of s when using lure (capture probability = 0.33 ± 0.08, recapture probability = 0.45 ±
0.05) than while not using lure (capture probability = 0.07 ± 0.03, recapture probability = 0.31 ±
0.09). This increase in (re)capture probability increased the population estimate precision, as the
coefficients of variation decreased when using lure, except for the SECR model estimates.
We found no effect of lure on the MaxDM of all individuals (W = 9.0, P = 0.156), nor Core
individuals only (W = 0.00, P =1.0). Similarly, we found no effect of lure on the MMDM of all
individuals (Z = 1.125, P = 0.260), nor Core individuals only (Z = -0.317, P = 0.752). The use of
lure only changed MMDM* by 18 m and MMDM-Core* by 6 m (Table 5); this later increase
translates into a negligible increase of 0.2% in the ESA. The large difference between MMDM*
and MMDM-Core* reflects the exclusion of animals with poorly sampled MaxDM.
We did not observe any shift in temporal activity pattern after lure was applied (W= 0.38,
P= 0.83). The mean overlap of activity ± SE between the non-lure and lure datasets was 95.51 ±
0.02%. We found no significant difference in the proportion of activity during the dawn, day,
dusk, and night periods for the non-lure and lure sampling periods (X2 = 0.78, P = 0.68). F.
fossana were predominantly active at night (85%) as compared to dusk (9%), dawn (6%), and
day (<1%).
Comparison of Density Estimation Analyses
Given that we found no effect of lure on (Table 5), we used the complete dataset to
compare density estimation methodologies. We found derived as s/ESA varied considerably
depending on the buffer value used to calculate the ESA; the ½MMDM* buffer produced the
smallest ESA (7.99 km2) and thus the highest density estimate (Fig. 2). We found no differences
in as estimated using 1) s/MMDM-Core*, 2) Core-only , 3) ML-SECR, and 4) B-SECR
43
(α = 0.05, Fig. 2). Our estimate precision was lowest with both SECR methods as these analyses
include uncertainty and process variation in abundance and area estimation that is often
underestimated with other density estimators.
Discussion:
It is critical to test the assumption of geographic closure when using closed C-R models
to estimate abundance and density, and to fully understand if the use of attractants biases these
estimates. We recommend the Pradel model to test geographic closure in carnivore studies,
because it 1) is flexible in modeling recapture variation, especially to account for the common
occurrence of heterogeneity, 2) uses model selection procedures to estimate the components of
geographic closure, and 3) is not affected by high Type 1 errors, as are the other tests when there
is a behavioral effect (White et al. 1982; Stanley and Burnham, 1999b), or heterogeneity (Stanley
and Burnham, 1999b). However, when sample sizes are inadequate to use the Pradel model, as
with many large carnivore studies, Otis et al. (1978) and Stanley and Burnham (1999a) can be
useful when model assumptions are met (which were not in this study); otherwise, no test of
closure is appropriate.
Despite concern that attractants might compromise geographic closure, we found no
indication that F. fossana were permanently immigrating to or emigrating from our study area.
This is likely a combination of the distance the lure could be detected and the territorial behavior
of F. fossana. If the maximum distance F. fossana could detect the lure was small compared to
its home range, only animals already overlapping sampling stations would be affected. Thus,
lure could increase the detection of an animal within a small area around the sampling station.
Alternatively, if the detection distance of the lure was large, animals would not be “pulled” onto
44
the sampling area because of territoriality. Like many terrestrial carnivores, F. fossana likely
defends a territory, thus preventing individuals from moving into an area they do not normally
occur. For example, the Malay Civet (Viverra tangalunga, Gray 1832), which similarly
occupies secondary rainforest, is territorial (Jennings et al. 2006). Further evidence that F.
fossana may be territorial was that the negative-exponential function best fit our data in the ML-
SECR analyses, describing a sharp decline in movement away from the home range center.
Given that habitat extended well beyond the majority of our sampling area, we would expect that
weak territorial behavior would have resulted in the selection of a more gradual distance function
(e.g. half-normal) for these data (Obbard et al. 2010).
We found the use of lure did not alter abundance or density estimates of F. fossana,
regardless of estimation method. Similarly, we found no effect of lure on maximum movement
distances or temporal activity pattern of F. fossana. The latter is an important finding for
photographic-sampling studies, which often evaluate temporal activity.
Given the challenges of detecting carnivores frequently enough to effectively apply
closed C-R analyses, we suggest lure can be used while remotely sampling territorial animals
without risking closure violations, alterations of abundance, density, or temporal activity pattern.
Our findings are particularly relevant to methodologies such as hair snares that often employ
attractants to detect carnivores. Higher detection rates from using lure can increase estimate
precision and reduce needed sampling effort and costs. Although not employed in this study,
post-hoc collapsing of sampling occasions can also increase detection probabilities and thus
increase precision (Dillon and Kelly, 2007); however, sampling efforts may still need to be quite
large. We do suggest that if lure is used, a rigid schedule be maintained for reapplying the lure,
as to reduce sampling heterogeneity.
45
Carnivore C-R studies using a grid design also face the dual challenges of the effects of
sampling layout on (re)capture probabilities and the determination of the appropriate area for
density analyses. We found the ML-SECR and B-SECR models estimated density significantly
lower than all but one of our estimates using an ad hoc buffer value to determine the effective
sampling area. In agreement with Obbard et al. (2010), we found that using a buffer of
½MMDM* on s overestimated density compared to SECR model estimates, while our MMDM-
Core* density estimate was similar and not statistically different than either SECR density
estimate. The MMDM buffer has been supported by several studies (Parmenter et al. 2003;
Soisalo and Cavalcanti, 2006; Trolle et al. 2007; Dillon and Kelly, 2008); however, there is no
theoretical framework for why this value should provide consistent and reliable density
estimates. Obbard et al. (2010) argued that empirical support for s/MMDM may reflect the
underestimation of ½MMDM due to few recaptures per individual (Tanaka, 1972), the truncation
of movement distances due to the sampling area edge (Soisalo and Cavalcanti, 2006), and the
inclusion of zero distances moved (Dillon and Kelly, 2007). In our study, we still found that the
s/½MMDM-Core* density estimate was significantly higher than SECR model estimates even
though 73% of our F. fossana individuals were recaptured ≥ 5 times, we modeled recapture rate,
and we strategically ameliorated the issues of the sampling area edge by using Core animals,
which had no zero distances moved. Further, in contrast to Obbard et al. (2010), we use a
distance to sampling area edge covariate (distedge) to incorporate closure violation bias on
variation in detection probability to more robustly estimate s (Boulanger et al. 2004a). While
our corrected s and ½MMDM-Core* buffer still produced a higher density than either SECR
model, our estimate was less dramatically different (71-76%) than Obbard et al. (2010) found in
some cases using the ½MMDM buffer (20-200%). Ultimately, the appropriate buffer value will
46
depend on the characteristics of the sampling array layout (size, shape, and trap spacing) and the
unknown home ranges of the sampled animals that may differ in size, shape, overlap, and
proportion contained within the sampling area (Parmenter et al. 2003).
Given the uncertainties of using ½MMDM and MMDM to buffer s in density estimation,
carnivore studies often use both values, reporting two density estimates (Trolle et al. 2007). This
is unsatisfying for conservation organizations attempting to identify populations and species at
risk, as ½MMDM densities are almost twice that of using MMDM. Given the known constraints
on measuring MMDM and the uncertainties in the appropriateness of any buffer value to
calculate the ESA, it is best to abdicate ad hoc boundary-strip methods given the availability of
newer statistical methods that ameliorate these issues (Efford et al. 2009b; Royle et al. 2009).
Of all four density estimators considered, the Core-only analysis ( = ) produced the
most precise density estimate and was congruent with both SECR model estimates. We assume
animals with a mean capture distance > zero from the sampling area edge, which on average
were captured 86% of the time at sampling stations away from the edge, were completely
contained within the sampling area. Without tracking Core animals to account for the true
proportion of time Core animals spend on and off the sampling area (White and Shenk, 2001),
we cannot validate this assumption. Also, by assuming area is known exactly, we deflate the
density variance by neglecting to account for uncertainty, leading to potentially erroneous
confidence in our estimate.
Determining the correct area of a sampled population to ameliorate the edge effect is the
limiting factor in producing robust estimates of density in the C-R framework. We agree with
Obbard et al. (2010) and Gardner et al. (2010) that SECR models are preferable to either
traditional ad hoc boundary strip methods or Core-only analyses to estimate density. The SECR
47
models incorporate the very real likelihood that the sampling layout has an effect on the
detection process and area estimation (Boulanger et al. 2004b; Dillon and Kelly, 2007). We
encourage other carnivore C-R studies to employ SECR models, as they 1) have a sound
theoretical and statistical framework, 2) free researchers from making subjective decisions on
how to calculate the ESA, thus making density estimates across studies more comparable, 3)
relax the geographic closure assumption and account for the edge effect, and 4) provide
conservation agencies with important population information in a single answer from one
underlying methodology, rather than a range of answers from multiple methodologies.
Acknowledgments:
Funding was provided by Virginia Tech, National Geographic Society Committee on
Research and Exploration, National Science Foundation Graduate Research Fellowship Program,
Sigma Xi Master‟s Degree and Grants-in-Aid Awards, and the Burd Sheldon McGinnes
Graduate Fellowship. We thank Madagascar National Parks and Direction des Eaux et Forêts for
permission to conduct this research. We were assisted by ICTE/MICET, J. Randrianantenaina,
D. Andrianoely, B. Marine, M. Kotschwar, C. Latimer, J.C. Razafimahaimodison, D. Stauffer, J.
Cohen, Z. Farris and J. Ivan. We thank M. Ridout for providing the R code to analyze temporal
activity.
48
References:
Balme, G. A., Hunter, L. T. B. & Slotow, R. (2009) Evaluating methods for counting cryptic
carnivores. Journal of Wildlife Management, 73, 433-441.
Batschelet, E., 1981. Circular Statistics in Biology. Academic Press, New York.
Bondrup-Nielsen, S. (1983) Density estimation as a function of live-trapping grid and home
range size. Canadian Journal of Zoology, 61, 2361–2365.
Boulanger, J. & McLellan, B. N. (2001) Closure violation bias in DNA based mark–recapture
population estimates of grizzly bears. Canadian Journal of Zoology, 79, 642-651.
Boulanger, J., McLellan, B. N., Woods, J. G., Proctor, M. F. & Strobeck, C. (2004a) Sampling
design and bias in DNA-based capture-mark-recapture population and density estimates
of grizzly bears. Journal of Wildlife Management, 68, 457–469.
Boulanger, J., Stenhouse, G. & Munro, R. (2004b) Sources of heterogeneity bias when DNA
mark-recapture sampling methods are applied to grizzly bear (Ursus arctos) populations.
Journal of Mammalogy, 85, 618-624.
Burnham, K. P. & Anderson, D. R. (2002) Model selection and multimodel inference: a practical
information-theoretic approach. Springer-Verlag, New York.
Cumming, G. & Finch, S. (2005) Inference by eye: Confidence intervals and how to read
pictures of data. American Psychologist, 60, 170-180.
Dillon, A. & Kelly, M. J. (2007) Ocelot Leopardus pardalis in Belize: The impact of trap
spacing and distance moved on density estimates. Oryx, 41, 469-477.
Dillon, A. & Kelly, M. J. (2008) Ocelot home range, overlap and density: Comparing radio
telemetry with camera trapping. Journal of Zoology, 275, 391-398.
49
Efford, M. G. (2009a) Density 4.4: software for spatially explicit capture–recapture. Available
at: http://www.otago.ac.nz/density. Accessed 1 June 2010.
Efford, M. G., Dawson, D. K. & Borchers, D. L. (2009b) Population density estimated from
locations of individuals on a passive detector array. Ecology, 90, 2676-2682.
Gardner, B., Royle, J. A., Wegan, M. T., Rainbolt, R. E. & Curtis, P. D. (2010) Estimating black
bear density using DNA data from hair snares. Journal of Wildlife Management, 74, 318-
325.
Gerber, B., Karpanty, S. M., Crawford, C., Kotschwar, M. & Randrianantenaina, J. (2010) An
assessment of carnivore relative abundance and density in the eastern rainforests of
Madagascar using remotely-triggered camera traps. Oryx, 44, 219-222.
Greenwood, R. J., Sargeant, A. B. & Johnson, D. H. (1985) Evaluation of mark-recapture for
estimating striped skunk abundance. Journal of Wildlife Management, 49, 332-340
Huggins, R. M. (1991) Some practical aspects of a conditional likelihood approach to capture
experiments. Biometrics, 47, 725-732.
IUCN (2010) IUCN Red List of Threatened Species. Accessed 1 June 2010.
Jennings, A. P., Seymour, A. S. & Dunstone, N. (2006) Ranging behaviour, spatial organization
and activity of the Malay civet (Viverra tangalunga) on Buton Island, Sulawesi. Journal
of Zoology, 268, 63-71.
Jett, D. A. & Nichols, J. D. (1987) A field comparison of nested grid and trapping web density
estimators. Journal of Mammalogy, 68, 888-892.
Karanth, K. U. & Nichols, J. D. (2002) Monitoring Tigers and Their Prey: A Manual for
Researchers, Managers and Conservationists in Tropical Asia. Centre for Wildlife
Studies, Banglamore.
50
Karanth, K.U., Nichols, J.D., Kumar, N.S., Hines, J.E., (2006). Assessing tiger population
dynamics using photograph capture-recapture sampling. Ecology, 87, 2925-2937.
Kendall, W. L. (1999) Robustness of closed capture-recapture methods to violations of the
closure assumption. Ecology, 80, 2517-2525.
Kerridge, F. J., Ralisoamalala, R. C., Goodman, S. M. & Pasnick, S. D. (2003) Fossa fossana,
Malagasy Striped Civet, Fanaloka. Natural History of Madagascar (eds S. M. Goodman
& J. P. Benstead), pp. 1363-1365. The University of Chicago Press, Chicago.
Long, R., MacKay, P., Ray, J. & Zielinski, W. (2008) Noninvasive survey methods for
carnivores. Island Press, Washington D.C.
Mace, R. D., Minta, S. C., Manley, T. L. & Aune, K. E. (1994) Estimating grizzly bear
population size using camera sightings. Wildlife Society Bulletin, 22, 74-83.
Maffei, L. & Noss, A. J. (2007) How small is too small? Camera trap survey areas and density
estimates for ocelots in the Bolivian Chaco. Biotropica, 40, 71–75.
McCarthy, K. P., Fuller, T. K., Ming, M., McCarthy, T. M., Waits, L. & Jumabaev, K. (2009)
Assessing Estimators of Snow Leopard Abundance. Journal of Wildlife Management, 72,
1826-1833.
Mowat, G. & Strobeck, C. (2000) Estimating population size of grizzly bears using hair capture,
DNA profiling, and mark-recapture analysis. Journal of Wildlife Management, 64, 183-
193
Noyce, K. V., Garshelis, D. L. & Coy, P. L. (2001) Differential vulnerability of black bears to
trap and camera sampling and resulting biases in mark-recapture estimates. Ursus, 12,
211-225.
51
O'Brien, T., Kinnaird, M. & Wibisono, H. (2003) Crouching tigers, hidden prey: Sumatran tiger
and prey populations in a tropical forest landscape. Animal Conservation, 6, 131-139.
Obbard, M. E., Howe, E. J. & Kyle, C. J. (2010) Empirical comparison of density estimators for
large carnivores. Journal of Applied Ecology, 47, 76-84.
Otis, D. L., Burnham, K. P., White, G. C. & Anderson, D. R. (1978) Statistical inference from
capture data on closed animal populations. Wildlife Monograph, 62. 3-135.
Parmenter, R. R., Yates, T. L., Anderson, D. R., Burnham, K. P., Dunnum, J. L., Franklin, A. B.,
Friggens, M. T., Lubow, B. C., Miller, M., Olson, G. S., Parmenter, C. A., Pollard, J.,
Rexstad, E., Shenk, T. M., Stanley, T. R. & White, G. C. (2003) Small-mammal density
estimation: A field comparison of grid-based vs. web-based density estimators.
Ecological Monographs, 73, 1-26.
Pradel, R. (1996) Utilization of capture-mark-recapture for the study of recruitment and
population growth rate. Biometrics, 52, 703-709
Ridout, M. S. & Linkie, M. (2009) Estimating overlap of daily activity patterns from camera trap
data. Journal of Agricultural, Biological, and Environmental Statistics, 14, 322-337.
Royle, J. A., Nichols, J. D., Karanth, K. U. & Gopalaswamy, A. M. (2009) A hierarchical model
for estimating density in camera-trap studies. Journal of Applied Ecology, 46, 118-127.
Soisalo, M. K. & Cavalcanti, S. M. C. (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.
Stanley, T. R. & Burnham, K. P. (1999a) A closure test for time-specific capture-recapture data.
Environmental and Ecological Statistics, 6, 197-209.
52
Stanley, T. R. & Burnham, K. P. (1999b) A goodness-of-fit test for capture-recapture model Mt
under closure. Biometrics, 55, 366-375.
Tanaka, R. (1972) Investigation into the edge effect by use of capture-recapture data in a vole
population. Researches on Population Ecology, 13, 127-151.
Trolle, M., Noss, A. J., Lima, E. D. S. & Dalponte, J. C. (2007) Camera-trap studies of maned
wolf density in the Cerrado and the Pantanal of Brazil. Biodiversity and Conservation,
16, 1197–1204.
Turk, R. D. (1997) Survey and species-screening trials of indigenous trees from the vicinity of
Ranomafana National Park, Madagascar. PhD thesis, North Carolina State University.
Weinberg, S. L. & Abramowitz, S. K. (2008) Statistics Using SPSS: An Integrative Approach
Cambridge University Press, Cambridge.
White, G. C. (2008) Closed population estimation models and their extensions in Program
MARK. Environmental and Ecological Statistics, 15, 89-99.
White, G. C., Anderson, D. R., Burnham, K. P. & Otis, D. L. (1982) Capture–recapture and
removal methods for sampling closed populations. Los Alamos National Laboratory, Los
Alamos.
White, G. C. & Shenk, T. M. (2001) Population estimation with radio-marked animals. Radio
tracking and animal populations (eds J. J. Millspaugh & J. M. Marzluff), pp. 329-350.
Academic Press, San Diego, CA.
Wilson, K. R. & Anderson, D. R. (1985) Evaluation of two density estimators of small mammal
population size. Journal of Mammalogy, 66, 13-21.
Zar, J. H. (1998) Biostatistical Analysis. Prentice Hall, Upper Saddle River.
53
Zielinski, W. J. & Kucera, T. E. (1995) American marten, fisher, lynx, and wolverine: survey
methods for their detection. USDA Forest Service, Pacific Southwest Research Station,
General Technical Report PSW-GTR-157, Albany, CA.
54
Table 1. Photographic-sampling summary of the Malagasy civet (Fossa fossana), sampled with
and without lure from 26 camera stations at the Sahamalaotra trail system within the rainforests
of Ranomafana National Park, Madagascar from 9 June-8 August, 2008.
Sampling
Period
Sampling Occasions
(Nights)
Individuals
Detected
Recapturesa
(C-R, SECR)
Male
Female
Complete
61
22
264, 428
11
11
Non-Lure
36
20
128, 185
11
9
Lure
25
20
136, 243
11
9
aRecaptures for closed capture-recapture (C-R) analyses and for spatially-explicit C-R
analyses which allows animals to be caught at multiple stations on the same occasion
55
Table 2. Model selection results evaluating the effects of lure on geographic closure for the Malagasy civet (Fossa fossana), sampled
with and without lure from 26 camera stations at the Sahamalaotra trail system within the rainforests of Ranomafana National Park,
Madagascar from 9 June-8 August, 2008.
Modelsa
QAICc
ΔQAICc
wi
Model
Likelihood
No.
Parameters
QDeviance
Full Pradel Analysisb
ф(.) p(location + lure) f(.)
1030.45
0.00
0.68
1.00
5
929.24
ф(location) p(location + lure) f(location)
1033.32
2.87
0.16
0.24
7
927.92
Stanley and Burnham Modelsc
ф(.) p(time) f(.)
1177.34
148.08
0.00
0.00
62
933.22
ф(1) p(time) f(.)
1199.88
170.62
0.00
0.00
61
958.99
ф(.) p(time) f(0)
1206.45
177.21
0.00
0.00
61
965.57
ф(1) p(time) f(0)
1224.20
194.95
0.00
0.00
60
986.52
aф, site fidelity; p, recapture probability; f, immigration onto the study area. Parameters with “(1)” and “(0)” indicate the
parameter is fixed. “.” indicates a constant value.
bLocation is a group, where animals have either a mean-capture distance greater than zero or zero from the grid edge. Lure is a
time effect between the non-lure and lure sampling periods
56
cTime as recapture probability variation by each sampling night (24-hour period from 12:00 PM to 11:59 AM)
57
Table 3. Top abundance and density estimation models (model likelihood >0.125) incorporating detection probability variation of the Malagasy
civet (Fossa fossana), sampled with and without lure from 26 camera stations at the Sahamalaotra trail system within the rainforests of
Ranomafana National Park, Madagascar from 9 June-8 August, 2008.
Analysis
Dataset
Model Selectiona
AICc
ΔAICc
wi
Model
Likelihood
No.
Parameters
Deviance
Superpopulation ( s)
Complete
b+h+sex+distedge+lure
2855.40
0.00
0.93
1.00
11
2831.30
Non-Lure
b+h+distedge
619.76
0.00
0.67
1.00
5
609.68
Non-Lure
b+h+distedge+sex
621.39
1.62
0.30
0.44
6
609.26
Lure
b+h+distedge+sex
528.84
0.00
0.59
1.00
6
516.67
Lure
h+distedge+sex
530.02
1.18
0.33
0.55
5
519.90
Core-only
Abundance ( )
Complete
b+h+sex+Lure
1197.54
0.00
0.98
1.00
8
1183.40
Non-Lure
b+h+sex
325.70
0.00
0.64
1.00
5
315.48
58
Non-Lure
b+h
327.26
1.56
0.29
0.46
6
317.04
Lure
b+h
274.32
0.00
0.47
1.00
4
266.11
Lure
h
274.49
0.17
0.43
0.98
3
268.37
Maximum-
Likelihood Spatially-
Explicit Capture-
Recapture
Density(
Complete
Negexp g0(sex+lure) σ(h)
2479.68
0.00
0.99
1.00
6
2426.87
Non Lure
Negexp g0(h) σ(h)
1165.86
0.00
0.64
1.00
5
1116.42
Non Lure
Negexp g0(.) σ(h)
1166.98
1.12
0.36
0.571
4
1121.16
Lure
Negexp g0(sex) σ(sex)
1668.47
0.00
0.68
1.00
4
1622.65
Lure
Negexp g0(h) σ(h)
1669.93
1.46
0.33
0.48
5
1620.49
ab, behavior effect; h, heterogeneity; sex, males vs. females; distedge, mean capture distance from grid edge; lure, a time effect between
the non-lure and lure sampling period; „.‟, fixed parameter; Negexp, negative-exponential distance function; g0, detection probability at a
home range center; σ, spatial scalar detection probability away from a home range center
59
Table 4. Population estimates, the coefficient of variation (CV), and trap success of the
Malagasy civet (Fossa fossana), sampled with and without lure from 26 camera stations at
the Sahamalaotra trail system within the rainforests of Ranomafana National Park,
Madagascar from 9 June-8 August, 2008.
Population Analysisa
Sampling Period
Abundance ± SE
CV
Trap success
(Capture Events /
100 Trap Nights)b
Superpopulation ( s)
Complete
22.74 ± 1.02
0.04
32.53
No Lure
25.08 ± 3.79
0.15
23.89
Lure
21.51 ± 2.21
0.10
45.86
Core-only (
Complete
8.07 ± 0.28
0.04
22.37
No Lure
8.12 ± 0.41
0.05
14.85
Lure
8.00 ± 0.02
0.002
33.99
aSuperpopulation is the population attributed to the sampling area plus an unknown amount
of the surrounding area; Core-only is the population of the individuals that were on-average
detected within the interior of the sampling area and assumed to be attributed to only the grid
bCapture events are all photographs within a 0.5 hour period; trap nights are the number of
complete 24-hour periods during which at least one camera was functioning at a sampling
station.
60
Table 5. Four density estimates and associated coefficient of variation (CV) of the Malagasy civet (Fossa fossana), sampled with and without
lure from 26 camera stations at the Sahamalaotra trail system within the rainforests of Ranomafana National Park, Madagascar from 9 June-8
August, 2008.
Density
Estimatora
Buffer Typeb
Sampling
Periodc
Buffer
Value (m)
ESA (km2)a
Density
(Individuals km-2 ± SE)
CV
Significance
(α = 0.05)d
s / ESA
½MMDM*
No Lure
356
8.09
3.10 ± 0.47
0.15
A
Lure
347
7.91
2.72 ± 0.28
0.10
A
½MMDM-Core*
No Lure
588
11.05
2.27 ± 0.33
0.14
B
Lure
591
11.47
1.88 ± 0.19
0.10
B
MMDM*
No Lure
712
13.07
1.92 ± 0.29
0.15
C
Lure
694
12.94
1.66 ± 0.17
0.10
C
MMDM-Core*
No Lure
1175
17.73
1.41 ± 0.21
0.15
D
Lure
1181
17.77
1.21 ± 0.12
0.01
D
/ A
-
No Lure
-
6.53
1.24 ± 0.06
0.05
E
Lure
-
6.53
1.23 ± 0.003
0.002
E
ML-SECR
-
No Lure
-
-
1.57 ± 0.35
0.22
F
61
Lure
-
-
1.27 ± 0.29
0.23
F
B-SECR
-
No Lure
-
-
1.22 ± 0.17
0.14
G
Lure
-
-
1.22 ± 0.19
0.15
G
as, superpopulation; , Core-only abundance; ESA, effective sampling area; A, sampling area; ML-SECR, maximum-likelihood
spatially-explicit capture-recapture (SECR) model; B-SECR, Bayesian SECR model
b½MMDM* and MMDM* are the expected half and full mean maximum distance moved. MMDM-Core* values are based on a subset of
Core animals
cSampling nights: 36 non-lure, 25 lure
dSignificance between the non-lure and lure datasets within each method, where different capital letters are significant, as determined
when 95% confidence intervals of two means overlap no more than half the average margin of error (Cumming and Finch, 2005)
62
Figure 1. We placed 26 camera stations over a 6.53 km2 area along the Sahamalaotra trail
system within the rainforests of Ranomafana National Park, Fianarantsoa province in
southeastern Madagascar from 9 June–8 August, 2008.
63
Figure 2. Density and 95% confidence limits using four methods of estimation on the complete dataset for the Malagasy civet
(Fossa fossana) with statistical significance (95% confidence interval of two means overlap less than half the average margin
of error) indicated with different capital letters (α = 0.05). Sampling was carried out using 26 camera stations at the
Sahamalaotra trail system within the rainforests of Ranomafana National Park, Madagascar from 9 June-8 August, 2008.
½MMDM*
½MMDM-Core*
MMDM*
MMDM-Core*
Core-only
ML-SECR
B-SECR
SECR
0.00
0.50
1.00
1.50
2.00
2.50
3.00
3.50
Density (civet km-2)
HHH
I
I
J
J
J
J
s/ESA
64
Chapter 3 - The impact of forest logging and fragmentation on carnivore species
composition, density, and occupancy in Madagascar’s southeastern rainforests
Abstract:
The endemic carnivores of Madagascar‟s southeastern rainforest face ongoing threats of
forest loss and fragmentation, yet we know little about their ecology or how they respond to such
disturbances. We evaluated carnivore species composition, density of two endemic carnivores
(Fossa Cryptoprocta ferox and Malagasy civet Fossa fossana), and carnivore occupancy in
continuous and fragmented rainforests in southeastern Madagascar to gain insight into the
responses of these carnivores to forest loss, fragmentation, and degradation. We used
photographic-sampling to survey carnivores at two contiguous (Primary and Selectively-logged)
and two fragmented rainforests (Fragments <2.5 km and >15 km from intact primary forest).
Carnivore composition varied among rainforests, with a higher number of native and lower
number of exotic carnivores at the contiguous than fragmented rainforests. We found F. fossana
absent from both fragmented rainforests and at a lower density ± SE in the Selectively-logged
(1.38 ± 0.22 individual/km2) than at the Primary rainforest (3.19 ± 0.55 individual/km2). C.
ferox was detected briefly in the Fragments <2.5 km from intact rainforest and had similar
densities between the Primary and Selectively-logged forests (0.12 ± 0.05 and 0.09 ± 0.04
adults/km2, respectively), but was absent at the Fragments >15 km rainforest. We identified only
two protected areas that could maintain >300 adult C. ferox in all of Madagascar. Broad-striped
mongoose (Galidictis fasciata) occupancy in the fragmented rainforests was positively related to
fragment size, while the ring-tailed mongoose (Galidia elegans elegans) was negatively
associated with increasing exotic-wild cat (Felis catus and Felis silvestris, grouped) trap success
65
within and outside rainforest fragments. Our findings indicate that degraded rainforest fragments
are of limited conservation value in supporting an intact endemic Malagasy carnivore community
compared to primary and selectively-logged contiguous rainforest. However, degraded
fragments may be of significant value in maintaining connectedness of carnivore populations
across the landscape.
1. Introduction:
Mammalian carnivores are generally sensitive to habitat loss, fragmentation, and
degradation (Crooks 2002; Sergio et al. 2008) due to their low densities, high area and energy
requirements, and persecution by humans (Cardillo et al. 2004; Noss et al. 1996). The decline or
extirpation of a large- or meso-carnivore can alter ecosystem dynamics and result in trophic
cascades, which could lead to further species extirpations (Berger et al. 2001; Ripple and Beschta
2006; Roemer et al. 2009). Carnivores can also be useful indicators of habitat disturbance (Soulé
and Terborgh 1999) and biodiversity (Sergio et al. 2006), making them effective focal species for
conservation planning (Carroll et al. 2001; Thorne et al. 2006). As such, understanding
carnivore population ecology in a region can be critical in planning for the best actions to protect
biodiversity and ecosystem integrity (Noss et al. 1996; Sergio et al. 2008; Terborgh et al. 1999).
The island of Madagascar is renowned for its wealth of endemic flora and fauna (Vences
et al. 2009), but also for its forest loss, fragmentation, and degradation, making it a global
biodiversity conservation priority. Only an estimated 16% of primary forest cover remains in
Madagascar, with much of the remaining forest isolated in fragments with little area > 1 km from
a forest edge (Harper et al. 2007). While studies of numerous Malagasy taxa have directly
benefited conservation planning, the dearth of studies on Madagascar‟s carnivore species has
66
excluded them from consideration (Kremen et al. 2008). Knowledge is lacking on basic ecology
and human impacts on the population ecology and habitat use of all Malagasy carnivores (Irwin
et al. 2010), especially in the eastern rainforests where only short-term or rapid assessment
studies have been conducted (Dollar 1999; Gerber et al. 2010; Goodman and Pidgeon 1999).
Some insight can be gained from longer studies of conspecifics in the western dry forests (Dollar
et al. 2007; Hawkins and Racey 2005), but due to increased primary production and variability of
resources in rainforest, there is likely to be significant variation between forest types. Given the
importance of carnivores to ecosystem dynamics and as focal species for conservation, it is
critical to better understand Malagasy carnivore ecology.
Of the nine extant endemic carnivore species within the endemic family Eupleridae
(Goodman and Helgen 2010; Yoder et al. 2003), five are known to occupy the southeastern
rainforests (broad striped mongoose Galidictis fasciata, fossa Cryptoprocta ferox, Malagasy
civet Fossa fossana, ring-tailed mongoose Galidia elegans elegans, small-toothed civet Eupleres
goudotii; Gerber et al. 2010). All species are thought to be declining, except G. e. elegans which
has an unknown population trend (IUCN 2010). Of particular concern is the population ecology
of C. ferox and G. e. elegans as both are known to prey on lemurs (Goodman 2003c), many of
which are also threatened (IUCN 2010). In the dry forests, lemurs can comprise greater than
50% of C. ferox’s diet. While there is likely significant geographic variation in C. ferox’s diet
(Hawkins and Racey 2008), their impact can be considerable with extirpations of lemurs by C.
ferox reported from contiguous and fragmented rainforests (Irwin et al. 2009).
Our objectives were to 1) use non-invasive photographic-sampling to quantify carnivore
species composition, density, and occupancy across a continuum of disturbed rainforests to gain
insight of the impacts of forest logging and fragmentation on Malagasy carnivores, 2) investigate
67
the influence of exotic carnivore species, human activity, landscape characteristics, and micro-
habitat on the occupancy of Malagasy rainforest carnivores, and 3) estimate population sizes of
C. ferox and F. fossana in Ranomafana National Park (RNP), and of C. ferox island-wide. These
data will allow the incorporation of carnivore abundance, distribution, and habitat use into
conservation planning in the rainforests of Madagascar, which is focused on the creation and
restoration of forest corridors linking fragmented and continuous and protected and un-protected
forest areas (Kremen et al. 2008; Norris 2006).
2. Materials and Methods:
2.1. Study Area
We carried out this research in four survey grids placed along Madagascar‟s eastern
rainforest escarpment (Fig.1). We sampled the contiguous rainforests at the Valohoaka-
Vatoranana (Primary) and Sahamalaotra (Selectively-logged) trail-systems located in the
boundaries of RNP. We also sampled two fragmented rainforests at Mahatsinjo,Tsinjoarivo
(Fragments <2.5 km) and Ialatsara Forest Station (Fragments >15 km). In this paper, we
consider forest fragmentation to include the cumulative or synergistic effects of forest loss,
fragmentation, and degradation as such disturbances are often concurrent (Laurance and
Cochrane 2001).
RNP is recognized for its exceptional floristic and faunal diversity within Madagascar
and among rainforests worldwide (Wright and Andriamihaja 2003). RNP encompasses 399 km2
which experienced spatially-patchy logging prior to its establishment in 1991; it currently
protects 329 km2 of mid-altitude rainforest (Conservation International 2010). Our Primary
rainforest grid was located within RNP and consisted of a mixture of unlogged forest with little
68
anthropogenic disturbance and an area of forest with minimal selective-logging (Wright 1997).
Our Selectively-logged rainforest grid located within RNP experienced pre-1991 moderate
logging pressure (JC. Razafimahaimodison, Pers. Comm.). Northwest of RNP, we sampled a
protected fragmented rainforest separated by >15 km from the main tract of contiguous primary
rainforest by a human-dominated landscape (Fig. 1). The Fragments >15 km grid encompassed
four fragments ranging in size from 2-5 ha, five fragments 10-64 ha, and one fragment which
was 240 ha, as determined by GPS-based mapping in November, 2008. These rainforest
fragments were surrounded by a matrix of shrub, open-burned, exotic pine and eucalyptus, and
minimal agriculture. Local people did not live within or between the fragments, but actively
burned the matrix and used the area for travel and forest products. Lastly, we sampled a fourth
grid 150 km north of RNP, located within a network of unprotected hilltop rainforest fragments
that from its closest edge was <2.5 km from the large tract of contiguous primary rainforest
(Fig.1). The Fragments <2.5 km grid was comprised of 19 fragments less than 7 ha, eight
fragments of 10-40 ha and one fragment of 192 ha, as determined by combining GPS mapping in
December, 2009 with IKONOS 2001 satellite imagery. Fifty-three household units with 1-6
houses per unit occupied the matrix surrounding the fragments, such that the matrix was
dominated by agriculture and naturally regenerating shrub. Local people from these households
used the fragments readily for a diversity of forest products (Irwin 2006). At both fragmented
grids, forest edges were hard transitions to open habitat, caused by human activities, such as
burning, grazing, and farming.
Surveys were conducted over a period of two years from May-December, 2008 (Primary,
Selectively-logged, and Fragments >15 km grids) and October-December, 2009 (Fragments <2.5
km grid). The two contiguous grids within RNP were sampled in the cold-dry season (April-
69
October) and the two fragmented grids in the warm-dry season (November-February, Tecot
2008). A five year average (2005-2009) min-max daily temperature was 13-20 ˚C at RNP during
the cold-dry season. The temperature during the warm-dry season sampling at the Fragments
>15 km and Fragments <2.5 km grids were 15-24˚C and 12-26˚C, respectively. The five year
average (2005-2009) daily rainfall ± SD was 6.2 ± 11.4 mm at RNP during the cold-dry season.
The average rainfall ± SD during the warm-dry season sampling at the Fragments >15 km and
Fragments <2.5 km grids were 5.1 ± 8.4 mm and 7.4 ± 14.0 mm, respectively.
2.2. Rainforest Habitat Sampling
To evaluate impacts of habitat on carnivore species composition and population ecology,
we measured landscape-, grid-, and camera-level (see sec 2.3) habitat features. Camera-level
vegetation was sampled along four transects radiating in each cardinal direction from camera
sampling stations. Along each transect, we established sampling points at 25, 50, and 75 m from
the camera station. These transect points along with a point at the camera station made a total of
13 vegetation sampling points per camera station. At each of the 13 points, we recorded the
diameter at breast height (D130) and the distance of the nearest tree (D130 ≥ 10 cm) in each quarter
surrounding the center point. Also, at each point we recorded presence or absence of vegetation
cover every 2 m for a total of 10 m in each cardinal direction for twenty point-intercepts, thus a
total of 260 point-intercepts per camera station, following Davis et al. (In Press) methodology.
Point intercept vegetation cover measurements included percent cover for down and dead trees
(≥ 15cm D130), herbaceous ground plants (0-0.5 m), woody-shrubs (0.5-5 m, ≤ 10 cm D130), low
canopy trees (0-5 m, ≥ 10 cm D130), medium canopy trees (5-15 m, ≥ 10 cm D130), and high
canopy trees (≥ 15 m, ≥ 10 cm D130). We used the nearest tree distances and D130 to estimate tree
70
density and basal area using an unbiased point-centered-quarter estimator (Pollard 1971). To test
for grid-level differences in vegetation characteristics, we calculated the mean and standard error
for each variable and used Welch‟s unequal variance ANOVA to test for differences among
rainforest sites (Zar, 1998). We normalized data using either a log or square-root transformation.
When the global test indicated a difference among grids, we used the Dunnett-Tukey-Kramer
pairwise multiple comparison test adjusting for unequal variances and sample size (Dunnett,
1980).
We used ArcGIS 9.2 (ESRI, Redlands, CA) to measure landscape- and camera-level
variables for each rainforest grid. We obtained data layers from available resources
(Conservation International 2010; Kremen et al. 2008) and through collaborations with
Madagascar National Parks, The Institute for the Conservation of Tropical Environments,
Sadabe, and on the ground GPS mapping. We quantified fragment area, shape, and nearest
neighbor metrics using FragStats (McGarigal et al. 2002). Additionally, we calculated nearest
distances from camera stations to the matrix and villages.
2.3. Photographic-Sampling
We used a random starting point to establish a systematic grid of passive infrared camera
stations along established trail-systems to photographically sample carnivores. Following
recommendations for survey design (Karanth and Nichols 2002; White et al. 1982) and using the
results from a preliminary study (Gerber et al. 2010), we deployed a minimum of 26 camera
stations for at least 52 days at each rainforest grid (Table 1). Among all grids, the camera station
density ± SD was 4.43 ± 0.54 stations km-2 with an average spacing ± SD among adjacent
stations of 555 ± 100 m. Camera stations consisted of two independently-operating passive
71
infrared cameras, mounted on opposite sides of a trail. This allowed a photographic-capture of
both flanks of every animal, thus improving individual identification in recaptures. We attached
cameras 20 cm above the ground and set them to be active for 24 hours/day. Camera types
included Deercam DC300 (DeerCam, Park Falls, USA) and Reconyx PC85 (Reconyx, Inc.
Holmen, Wisconsin). We often used chicken meat as a scent lure to effectively sample these
rare/elusive species and modeled effects on detection probability in occupancy and density
analyses. Lure was deployed for the entire sampling period at the Primary and Fragments >15
km grids, but was only deployed for the second half of sampling at the Selectively-logged and
Fragments <2.5 km grids. Previous work has found no effect of lure on the activity patterns of
native and exotic carnivores at contiguous and fragmented rainforests (Gerber et al., in prep,
Chapter 4), as well as no effect on F. fossana geographic closure, abundance and density
estimation, or maximum movement distances (Gerber et al., in review). We checked camera
stations every three to five days to ensure continued operation, replacing lure, batteries, film, and
memory cards when necessary.
2.3.1. Carnivore Species Composition
Among the four rainforest grids, we compared both native and exotic carnivore species
composition. While a positive detection leaves no doubt a species occurs within the study area,
non-detection does not ensure a true absence. We used a binomial model using capture
frequency and sampling effort to evaluate the number of trap nights needed to obtain a 95%
probability of a single detection of native and exotic carnivores at each rainforest grid (Tobler et
al. 2008). A trap night is defined as a 24-hour period during which at least one of the two
cameras at a station was functioning.
72
2.3.2. Carnivore Density
We estimated carnivore density for C. ferox and F. fossana using capture-recapture
analyses, as these species are individually-identifiable (Gerber et al. 2010). To minimize bias,
two researchers independently agreed on the individual-identity of capture events to construct the
necessary capture-histories for analyses. A capture event was defined as all photographs of the
same individual within a 0.5 hour period. To evaluate the geographic closure assumption for F.
fossana, we estimated the study grid population growth rate (λ; Gerber et al., in review) using the
Pradel model (Pradel 1996) implemented in program MARK (V. 5.1; White 2008). If closure is
not violated, λ will equal one. Closure was evaluated for C. ferox using a hypothesis test that
assumes only detection probability heterogeneity (Otis et al. 1978), as this dataset was too small
for the Pradel model. To estimate density, we employed a maximum-likelihood spatially-explicit
capture-recapture model (ML-SECR), implemented in program DENSITY (V. 4.4; Efford et al.
2009). We used the ML-SECR model rather than traditional ad hoc density estimators as it is
better designed to estimate density and its variance when the sampling area is ill-defined due to
geographic closure violations by temporary emigration from the study grid (Obbard et al. 2010;
Gerber et al., in review). We applied a habitat-mask to remove villages, roads, and agricultural
land (non-habitat) from the area estimation. We compared the fit of three detection functions
(half-normal, hazard-rate, and negative-exponential) to model detection probability variation
away from an animal‟s home range center. We used a conditional-likelihood function to model
both g0 (detection process when a single trap is located at the center of an animal‟s home range)
and σ (spatial scale detection process away from the center of the home range) using a priori
biologically plausible models: including time variation (time), the effect of lure (lure) as a
73
difference between the non-lure and lure sampling periods, behavior effect (b), individual
heterogeneity (h; Pledger‟s 2-point mixture), and males versus females (sex). Model fit was
evaluated using Akaike‟s Information Criterion with a small sample size bias correction (AICc).
We model-averaged parameter estimates to incorporate model selection uncertainty (Burnham
and Anderson 2002). We defined a significant statistical difference between estimates when the
95% confidence intervals of two means overlap no more than half the average margin of error.
This is the equivalent to a conservative hypothesis test at α = 0.05 (Cumming and Finch 2005).
2.3.3. Carnivore Occupancy
We estimated the 1) average occupancy of our study grids for native and exotic
carnivores, and 2) probability of occupancy, which we interpret as “use” of a 75 m radius
surrounding the camera stations. We assumed closure while sampling and applied a single-
season occupancy model, available in program PRESENCE (MacKenzie et al. 2005). We
analyzed each species by grouping data from the two contiguous rainforest grids (Primary and
Selectively-logged) and grouping separately the two fragmented rainforest grids (Fragments <2.5
km and >15 km), as each group was sampled within the same season and vegetation structure
was more similar within a group. This allowed us to test differences between similar forests,
include variables appropriate for both grouped forest types, and potentially share information
across grids to more robustly estimate parameters.
We included covariates for both the probability of occupancy and detection to evaluate
biologically-driven hypotheses. Covariates included camera-level vegetation and canopy cover
variables (see sec 2.2), as well as the effect of lure as a survey covariate (lure). Landscape
variables included rainforest fragment size (Area), nearest distance from camera stations to the
74
matrix (Distmatrix), camera station distance to the nearest village (DistVillage), Euclidian
distance from each fragment to nearest neighboring fragment (ENN), camera station forest cover
type as either rainforest or matrix (ForestType), a difference between grouped rainforest grids
(Grid), the fragment shape as a shape index (Shape), and trail type as either maintained or
secondary trail (Trail). Additionally, we evaluated the trap success of exotic carnivores and local
people at camera stations as a measure of disturbance that may influence both native and exotic
carnivore species. Trap success was calculated as the frequency of capture events by species at a
station per 100 trap nights. Covariates were standardized (subtracting the mean and dividing by
the standard deviation) to prevent numerical constraints with parameter estimation. To avoid
multicollinearity, we examined correlations among variables and did not include variables with r
> 0.6 in the same model (Graham 2003). We used a two-step iterative modeling process by first
including covariates that influence detection probability and secondly those covariates
influencing occupancy using the best model for detection probability (Bailey et al. 2004). Our
most parameterized (global) models were constrained to six parameters per response variable,
based on sample size (Burnham and Anderson 2002). We evaluated global model fit by
assessing the degree of overdispersion (ĉ) using a goodness-of-fit test with 10,000 bootstraps and
adjusted ĉ if chi-square tests indicate lack of fit (P < 0.05; MacKenzie and Bailey 2004). We
evaluated competing models using AICc as described in sec 2.3.2.
2.3.4. RNP and Madagascar Carnivore Population Estimates
We extrapolated C. ferox and F. fossana densities to estimate potential carnivore
populations. We used ArcGIS to quantify available habitat from the most recent primary forest
cover data (2005; Conservation International 2010) and to quantify the extent of current
75
protected areas (Kremen et al. 2008). Protected areas used here include a mixture of national
parks, special reserves, and other designations with varying levels of protection status, as defined
by Kremen et al. (2008). To investigate F. fossana and C. ferox abundance at RNP, we
extrapolated this study‟s density estimates to available habitat defined as RNP primary forest
coverage.
In addition, we estimated the potential island-wide population size of C. ferox, the
number of distinct populations, and the connectedness of those populations by extrapolating this
study‟s rainforest, adult-only, density estimates and an adult-only, dry forest estimate (0.18 km-2;
Hawkins and Racey, 2005). We assumed C. ferox were forest dependent (Hawkins and Racey
2005) and classified Madagascar‟s forests into zones of eastern rainforest, western dry forest,
and southern spiny forest (Harper et al. 2007). We excluded the spiny forests from population
estimates as there are no C. ferox density estimates from that region. In the rainforest and dry
forest, we first identified forest fragments that could contain a single adult C. ferox (based on
density estimates) and estimated population size of each fragment. Second, we removed
fragments <25 ha to eliminate highly fragmented areas. Finally, we spatially joined all
remaining fragments ≤4.9 km from each other as a single population of a forest complex, as
movements of C. ferox in the human-dominated landscape are likely constrained to this distance
(this study; Kotschwar 2010).
3. Results:
3.1. Vegetation Structure of Rainforest Grids
Grid-level vegetation structure differed greatly among all forests, but was most similar
between fragmented forests (Table 2). We found the highest tree density and basal area in the
Primary forest, followed by the Selectively-logged, and then the fragmented forests. There was
76
no difference in tree density and basal area between the two fragmented forests. High canopy
cover was greatest at the Primary forest, while the Selectively-logged forest had the highest mid-
canopy cover. Low canopy cover was generally low at all forests, except for the Fragments <2.5
km, which had ten times the low coverage compared to the Primary forest. We observed >78%
shrub canopy cover in all forests, while dead and down cover was found to be highest at the
Primary and Selectively-logged forests, which is consistent with our observations of ongoing
dead wood extraction at the fragmented forests.
3.2. Carnivore Species Composition
We detected all known endemic southeastern rainforest carnivores in the Primary and
Selectively-logged contiguous rainforests. We found reduced numbers of endemic carnivore
species in the fragmented rainforests, with G. fasciata and G. e. elegans detected at both
fragmented forests, F. fossana and E. goudotii absent from both fragmented forests, and C. ferox
only detected at the Fragments <2.5 km. We observed exotic carnivores in all four rainforests.
Only the domestic dog (Canis familiaris) was detected in the Primary and Selectively-logged
forests, while three exotic species (C. familiaris, exotic-wild cat Felis catus and Felis silvestris,
grouping the domestic and introduced African wildcat, F. catus/silvestris, as they are difficult to
distinguish from each other), and the exotic small Indian Civet Viverricula indica) were found in
both fragmented forests.
Among all rainforests, 423 trap nights or less were needed to achieve a 95% probability
of a single detection of most carnivores. The two exceptions were E. goudotii at the Primary
forest (1081 trap nights needed) and C. ferox at the Fragments <2.5 km (1375 trap nights
needed). Assuming capture frequencies were similar among forests where species were detected
77
and not detected, our sampling effort at each forest was above the necessary trap nights to
achieve a 95% probability of detecting all carnivores in each forest, except for C. ferox at the
Fragments >15 km (Table 1). However, even at that site we had very close to the needed number
of trap nights to detect C. ferox.
3.3. Carnivore Density
Two researchers independently agreed on the individual-identity of >96% of F. fossana
and 88% of C. ferox capture events. We excluded two juvenile C. ferox each from the Primary
and Selectively-logged forests due to lack of adequate recaptures; density estimates are thus for
adults only. We detected a single C. ferox at the Fragmented <2.5 km grid, but could not
conduct a density analysis with a single individual. The sex ratios of detected C. ferox and F.
fossana were equal, except at the Primary forest where one more female F. fossana was detected
(Table 1).
Closure was not rejected for F. fossana as λ was not different than 1.0 at the Primary
(1.00 ± 0.004) or Selectively-logged contiguous rainforests (1.00 ± 0.008). Similarly, closure
was not rejected for C. ferox at the Primary (Z = 1.281, P = 0.899) or Selectively-logged forests
(Z = 1.362, P = 0.913).
We found the detection probabilities of F. fossana and C. ferox were affected by both sex
and h (Table 3). We identified a single top model for F. fossana density estimation at both the
Primary and Selectively-logged forests which included sex affecting detection at the home range
center (go; males higher than females) and h among individuals in the spatial scalar over which
detection declines away from the home range center (σ). Our use of lure at the Selectively-
logged forest increased the detection probability of F. fossana at the home range center (go), but
78
not the spatial scalar (σ). We found less definitive evidence of effects on go and σ for C. ferox,
but the inclusion of sex and h did improve fit over the null model.
F. fossana density was highest in the Primary forest, while in the Selectively-logged
forest it was less than half of the Primary forest estimate (Table 4). C. ferox density was an
order of magnitude lower than F. fossana in the Primary and Selectively-logged rainforests, but
not significantly different between these forests (α = 0.05). In the Fragmented <2.5 km forest,
we concluded only that C. ferox density may be greater than zero as one individual was detected
at two camera stations in the largest rainforest fragment (192 ha) in that grid, or zero if the
animal was a transient.
3.4. Carnivore Occupancy
Our fully parameterized models fit the data (P > 0.05), thus we maintain an
overdispersion factor of 1.0. The detection probabilities and occupancies for native and exotic
carnivores were influenced by both landscape- and camera-level variables (Table 5). In the
contiguous rainforest grids, we found C. ferox and G. fasciata detection probability increased the
further a camera station was from the matrix (Distmatrix; Table 5 and Table 6). C. ferox was
also detected less often on smaller secondary trails compared to maintained trails (Trail). G.
fasciata detections were negatively influenced by increasing trap success of C. familiaris (Dog).
We found F. fossana detection probability increased with the use of lure (Lure). C. familiaris
detections were positively associated with local people trap success (Locals). In the fragmented
rainforests, our results indicate that detection probability for native carnivores varied by grid
(Grid), with higher detection at the protected Fragments >15 km (Table 5 and Table 6). Exotic
carnivore detection probability within the fragmented rainforests varied significantly, with higher
79
detections within the matrix compared to rainforest fragments (ForestType), increasing
detections with local people trap success (Locals), and increasing detections with the use of lure
(Lure; Table 5 and Table 6).
In the contiguous rainforests, G. fasciata occupancy was higher on average in the
Primary forest due to larger basal area compared to the Selectively-logged forest (Table 7, Fig.
2). In comparison, our data showed that G. fasciata occupancy in the fragmented rainforests was
determined by fragment size (Table 5); fragments greater than 50 ha had a probability of
occupancy >95% (Table 7, Fig. 2). G. fasciata was never detected outside the rainforest
fragments. In the contiguous rainforests, our data reveal no clear habitat association for C. ferox
or differences in occupancy in the Primary or Selectively-logged forests (Table 5 and Table 6).
We could not model occupancy for either F. fossana or G. e. elegans due to their high naïve
occupancy in the contiguous rainforests (Table 8). However, in the fragmented forests we found
G. e. elegans had a higher occupancy within the fragments compared to the matrix (ForestType)
and was negatively affected by increasing F. catus/silvestris trap success (Table 7 and Fig. 3). C.
familiaris occupancy in the contiguous rainforests was negatively affected by increasing distance
to the closest village from camera stations (DistVillage; Table 5 and Table 7). On average, C.
familiaris occupancy was lower at the contiguous rainforest than the fragmented rainforest and
was highest in the Fragments <2.5 km (Table 8). V. indica and F. catus/silvestris occupancy was
not different between fragmented forests. While both these exotic carnivores used the rainforest
fragments and matrix, V. indica had a significantly higher occupancy within the matrix
(ForestType; Table 7 and Table 8). We could not evaluate E. goudotii occupancy in the
contiguous rainforest or C. ferox at the Fragmented <2.5 km forest due to poor detection
probability (Table 8).
80
3.5. RNP and Madagascar Carnivore Populations
RNP may protect a total population of 516 to 1,193 individual F. fossana, but only 39
adult C. ferox. We determined from our average adult C. ferox rainforest density that ≥ 9.5 km2
of primary rainforest was necessary for a single adult, while ≥ 5.5 km2 of dry forest was
necessary according to estimates from the west (Hawkins and Racey 2005). Across Madagascar,
we estimated a total population of 8,626 adult C. ferox with 4,476 rainforest animals belonging
to 32 populations and 4,150 dry forest animals belonging to 38 populations. Ninety-five percent
of the total rainforest population was contained in two forest complexes (Fig. 4), separated
immediately north of Lac Alaotra Biological Reserve (UTM: 900815E, 8074030S). The
northeast and central-east rainforest complexes could support 2,638 and 1,627 C. ferox adults,
respectively. Dry forests were more fragmented with 95% of the total population divided into
nine separate populations. Two dry forest complexes could support >300 adult C. ferox, a
central-west and north-west population, estimated at 2,877 and 456, respectively. Considering
the protected forest cover only, we estimated an island-wide protected C. ferox population of
only 2,635 adults (780 Dry forest, 1,855 Rainforest). The protected areas of Masaola-Makira
rainforest and Zahamena-Mantadia-Vohidrazana rainforest complex were the only areas
identified as potentially holding >300 adult C. ferox. At Masaola-Makira, Madagascar‟s largest
protected area, we estimated 5,445 km2 of primary forest, thus a potential protected population of
572 adult C. ferox.
4. Discussion:
81
Carnivores are globally threatened by habitat loss, fragmentation, and degradation
(Karanth and Chellam 2009). Malagasy carnivores are no exception. Our data suggest that
within Madagascar‟s southeastern rainforests, forest loss/fragmentation/degradation, selective-
logging and exotic carnivore presence have significant negative implications for the conservation
of native carnivores. Congruent with other carnivore community studies (Crooks 2002;
Michalski and Peres 2005), rainforest fragmentation has altered the species composition of
Malagasy carnivores, such that contiguous rainforests hold a higher number of native and lower
number of exotic carnivore species compared to fragmented rainforests. Fragmentation
sensitivity appears to differ by species as observed in other studies (Crooks 2002); the larger
body-sized Euplerinae species (C. ferox, E. goudotii, F, fossana) were absent or nearly so from
the fragmented rainforests, suggesting greater sensitivity compared to the smaller-bodied
Galidiinae species (G. e. elegans, G. fasciata) which occupied fragmented forests. The single C.
ferox detected within the largest fragment (192 ha) in the Fragments <2.5 km site may have been
a transient animal; the low number of recaptures suggests this animal did not occupy the area, at
least in the warm-dry season we sampled. Although we were 56 trap nights (2.15 actual nights
with the 26 camera stations we deployed) short of a 95% probability of a single detection of C.
ferox in the Fragments >15 km (assuming the same capture frequency at the Fragments <2.5
km), we are confident this animal was absent, as a local ecological knowledge study conducted
in the area found no local people to have observed C. ferox in recent years (2004-2009;
Kotschwar 2010).
Selective-logging is known to have complex and often species-specific effects on
mammals (Bicknell and Peres 2010; Meijaard et al. 2005; Paviolo et al. 2009). Among Asian
civets, logging can have both positive and negative consequences within this diverse group, with
82
logging increasing the distribution of some civets, while negatively affecting the abundance of
others (Meijaard and Sheil 2008). Despite our Selectively-logged forest having reduced basal
area and tree density and an altered canopy cover compared to the Primary forest, we found no
difference in the carnivore community composition between these areas. Either selective-
logging prior to 1991 had not altered the carnivore composition, or contiguous unlogged forest
faciliated recolonization in the last two decades.
Similar to findings on tigers (Panthera tigris; Rayan and Mohamad 2009), our
structurally-altered Selectively-logged rainforest had the same density of C. ferox as compared to
the Primary forest. We hypothesize that C. ferox’s flexible diet, which may explain why this
species occupies diverse forests across Madagascar, may also allow C. ferox to tolerate a certain
level of forest disturbance (Robert et al. 2003). In agreement with previous findings that C. ferox
rainforest density is lower than dry forest density (Gerber et al. 2010), our average adult-only
rainforest density (0.105 km-2) was also lower than the adult-only dry forest density (0.18 km-2;
Hawinks and Racey, 2005). This lower rainforest density may reflect a combination of more
patchily distributed resources common in rainforests (Turner 1996), lower primate prey biomass
compared to Madagascar‟s dry forests (Ganzhorn et al. 2003) or lower prey accessibility related
to the more complex habitat structure (Balme et al. 2007).
F. fossana appear particularly sensitive to forest disturbance. While we found F. fossana
absent from both fragmented rainforests <2.5 km and >15 km from intact primary forest, they are
also unlikely to occupy Madagascar‟s open human-dominated landscape (Kotschwar 2010) and
have been described as intolerant to degraded forests (Kerridge et al. 2003). Some evidence
suggests F. fossana are able to use fragments <2.5 km from intact forest (Dehgan 2003). Dehgan
83
(2003) visually observed F. fossana at three fragments 93, 745 and 1,368 m (200, 2.5, and 30 ha,
respectively) from the intact rainforest, but not at his farthest fragment at 2,438 m (31 ha).
These losses of F. fossana with forest fragmentation and reduced densities with selective
logging are similar to studies of rainforest civets in Borneo and Malaysia (Colón 2002; Heydon
and Bulloh 1996). Carnivore density is generally positively correlated with prey biomass
(Carbone and Gittleman 2002), thus observed decreases in carnivore density with selective-
logging may be explained by reduced prey species richness and/or abundance. Forest logging in
Madagascar‟s rainforests has been shown to reduce amphibian abundance and diversity (Vallan
2002; Vallan et al. 2004), simplify aquatic macroinvertebrate communities (Benstead and Pringle
2004), and affect native small-mammals (Lehtonen et al. 2001), all of which may impact F.
fossana populations. Selective-logging is also thought to have a disproportionate effect on
terrestrial species with narrow diets that are primarily frugivorous, carnivorous, or insectivorous
(Meijaard et al. 2008). Although F. fossana diet has generally been described as omnivorous,
there is some evidence that this species may exhibit seasonal diet specialization on insectivores
in the warm-wet season and vertebrates in the cold-dry season (Goodman et al. 2003). This
potential seasonal specialization may make them more vulnerable to selective-logging-induced
alterations of prey communities.
The two native carnivores found to occupy fragmented rainforest were the Galidiinae
mongooses, G. e. elegans and G. fasciata. While G. e. elegans has been known to use rainforest
fragments (Goodman 2003a), G. fasciata’s presence is more surprising, given their local rarity in
the primary rainforest (Goodman 2003b). Despite both species presence in fragments, they were
both still constrained by disturbances, as G. fasciata was constrained to the larger fragments and
G. e. elegans constrained by increasing levels of activity by the exotic F. catus/silvestris. F.
84
catus/silvestris’ broad distribution across Madagascar, their negative impact on G. e. elegans,
and their documented predation upon diurnal lemurs (Brockman et al. 2008) make them a
considerable conservation threat to endemic wildlife.
We found at least one exotic carnivore in each rainforest site, with C. familiaris being the
most ubiquitous. In the contiguous rainforest, C. familiaris’ use of the forests was negatively
related to the distance to the closest village, such that beyond 4 km from a village C. familiaris is
unlikely to use intact forest. Considering the many villages surrounding RNP, we predicted that
15% and 65% of RNP has greater than 50% and 5% probability of use by C. familiaris,
respectively. This is a great concern as C. familiaris are expected to affect medium and small
carnivores through interference competition and are known to serve as disease vectors (Vanak
and Gompper 2009). In the dry forests, C. familiaris and F. catus/silvestris have been implicated
in the transmission of a number of diseases to C. ferox (Dollar 2006). Even when forests are
protected from direct forest loss, disease transmission from domestic or exotic animals remains a
critical threat to native wildlife, with significant negative effects on the survival and persistence
of many species (Deem et al. 2001).
The exotic V. indica is thought to occur widely throughout Madagascar, mostly within
disturbed habitats (Dollar 2006). While there is evidence this species occasionally uses the
edges of intact rainforest (Gerber et al. 2010), as well as occurs outside the forest, their absence
from the contiguous rainforests suggests they cannot permanently occupy intact rainforests. In
contrast, in the fragmented forests, V. indica was found using the matrix more than the rainforest,
which is consistent with habitat use in their native range (Chen et al. 2009). However, any use of
the fragments poses the possibility of competition with the native mongooses that also occupy
these fragments.
85
5. Conservation Implications:
Rainforest species are generally considered sensitive to fragmentation effects (Turner
1996), thus it may be expected that Madagascar‟s increasingly fragmented forests (Harper et al.
2007) pose a threat to the conservation of rainforest carnivores. With species persistence in a
fragmented landscape often due to the ability to tolerate matrix conditions (Turner 1996), the
Malagasy rainforest carnivores‟ intolerance to the open human-dominated landscape (This study;
Kotschwar 2010) necessitates protection and management of forests for native carnivore
persistence. Given the documented negative effects of fragmentation in particular on F. fossana,
its populations throughout Madagascar are likely to vary greatly in size and be highly isolated.
Additionally, human hunting of F. fossana and C. ferox may significantly reduce numbers of
these species outside protected areas, as both are known to be consumed (Golden 2009; Kerridge
et al. 2003).
Although our island-wide C. ferox population extrapolation is imperfect, we found it a
useful exercise to enhance understanding of the potential population sizes and isolation of this
threatened carnivore and known lemur predator. There is a great need in the rainforest and dry
forest to increase the sizes of protected areas to maintain even moderate protected populations of
C. ferox. This is especially true for dry forests, as the largest protected dry forest area was
Kirindy Metea National Park, with 839 km2 of primary forest, thus a potential population of only
151 adult C. ferox. To protect C. ferox populations, expansion of protected areas may not need
to be in the form of national parks and thus exclusion of local people, but minimally a protection
of C. ferox from human predation and maintenance of forest structure. In addition to expanding
protected areas, establishing forested corridors among fragments and protected areas could
86
encourage and maintain movements of potentially isolated carnivore populations. Two areas in
the eastern rainforest which should be a high priority for forest restoration include 1) the gap
between the two eastern rainforest complexes, which were only linked by a series of small (<25
ha) fragments, and 2) the area 16 km north of RNP (UTM: 751529E, 7689031S) which had a gap
of approximately 4.5 km between forest fragments. Corridors could benefit carnivore
movement, as well as increase forest connectivity as a means to ameliorate the effects of
vegetation shifts predicted by climate change (Hannah et al. 2008). However, re-establishment
of carnivores in fragments, especially C. ferox, may create a risk to lemur population persistence
(Irwin et al. 2009; Kotschwar 2010).
Our findings indicate that degraded rainforest fragments are of limited conservation value
in supporting an intact endemic Malagasy carnivore community compared to primary and
selectively-logged contiguous rainforest. However, degraded fragments may be of significant
value in maintaining connectedness of carnivore populations across the landscape. We
recommend protecting intact primary rainforest to conserve Madagascar's endemic carnivore
populations, protecting fragments to maintain connectedness of the forest landscape, and
continuing to study how local people and carnivores coexist in forests that provide forest
products and are of carnivore conservation value.
In order to significantly improve our understanding of C. ferox and its populations, future
research should investigate the 1) minimum fragment size required for occupancy, 2) movement
dynamics within and between contiguous and fragmented areas, and 3) density variation within
and between all forest types, especially in the spiny forests where no quantitative data exists.
The spiny forests cover 20,267 km2; if C. ferox occupied this unique habitat at even a moderate
87
density, it would significantly increase our population estimates and would be of great
conservation value to this species‟ long-term survival.
Acknowledgments:
Funding provided by Virginia Tech, National Geographic Society Committee on
Research and Exploration, Sigma Xi Virginia Tech Master‟s Degree and Grants-in-Aid Awards,
National Science Foundation Graduate Research Fellowship Program, and Burd Sheldon
McGinnes Graduate Fellowship. We thank the Government of Madagascar, Madagascar
National Parks, the Direction des Eaux et Forêts, and CAFF/CORE for permission to conduct
this research. We were greatly assisted by ICTE/MICET, Centre ValBio, Sadabe, Mitchell
Irwin, and many remarkable research assistants. Design and analyses were greatly improved by
the advice of D. Stauffer, M. Kelly, J. Cohen, and J. Ivan.
88
References:
Bailey, L.L., Simons, T.R., Pollock, K.H., 2004. Estimating site occupancy and species detection
probability parameters for terrestrial salamanders. Ecological Applications 14, 692-702.
Balme, G., Hunter, L., Slotow, R., 2007. Feeding habitat selection by hunting leopards Panthera
pardus in a woodland savanna: prey catchability versus abundance. Animal Behaviour
74, 589-598.
Benstead, J.P., Pringle, C.M., 2004. Deforestation alters the resource base and biomass of
endemic stream insects in eastern Madagascar. Freshwater Biology 49, 490-501.
Berger, J., Stacey, P.B., Bellis, L., Johnson, M.P., 2001. A Mammalian Predator-Prey Imbalance:
Grizzly Bear and Wolf Extinction Affect Avian Neotropical Migrants. Ecological
Applications 11, 947-960.
Bicknell, J., Peres, C.A., 2010. Vertebrate population responses to reduced-impact logging in a
neotropical forest. Forest Ecology and Management 259, 2267-2275.
Brockman, D., Godfrey, L., Dollar, L., Ratsirarson, J., 2008. Evidence of Invasive Felis silvestris
Predation on Propithecus verreauxi at Beza Mahafaly Special Reserve, Madagascar.
International Journal of Primatology 29, 135-152.
Burnham, K.P., Anderson, D.R., 2002. Model selection and multimodel inference: a practical
information-theoretic approach. Springer-Verlag, New York,.
Carbone, C., Gittleman, J., 2002. A common rule for scaling. Science 295, 2273-2276.
Cardillo, M., Purvis, A., Sechrest, W., Gittleman, J.L., Bielby, J., Mace, G.M., 2004. Human
Population Density and Extinction Risk in the World's Carnivores. PLoS Biology 2,
e197.
89
Carroll, C., Noss, R.F., Paquet, P.C., 2001. Carnivores as focal species for conservation planning
in the Rocky Mountain region Ecological Applications 11, 961-980.
Chen, M.-T., Tewes, M.E., Pei, K.J., Grassman, L.I., 2009. Activity patterns and habitat use of
sympatric small carnivores in southern Taiwan. Mammalia 73, 20-26.
Colón, C.P., 2002. Ranging behaviour and activity of the Malay civet (Viverra tangalunga) in a
logged and an unlogged forest in Danum Valley, East Malaysia. Journal of Zoology 257,
473-485.
Conservation International, 2010. Center for Applied Biodiversity Science:
https://learning.conservation.org/spatial_monitoring/Forest/Pages/default.aspx. Data
Accessed Last: May 10, 2010.
Crooks, K.R., 2002. Relative sensitivities of mammalian carnivores to habitat fragmentation.
Conservation Biology 16, 488-502.
Cumming, G., Finch, S., 2005. Inference by eye: Confidence intervals and how to read pictures
of data. American Psychologist 60, 170-180.
Davis, M.L., Kelly, M.J., Stauffer, D.F., Carnivore co-existence and habitat use in the Mountain
Pine Ridge Forest Reserve, Belize. Animal Conservation, In press.
Deem, S.L., Karesh, W.B., Weisman, W., 2001. Putting theory into practice: wildlife health in
conservation. Conservation Biology 15, 1224-1233.
Dehgan, A., 2003. The Behavior of Extincion: Predicting the incidence and local extinction of
lemurs in fragmented habitats of southeastern Madagascar, In Biological Sciences.
University of Chicago, Chicago.
90
Dollar, L., 1999. Preliminary report on the status, activity Cycle, and ranging of Cryptoprocta
ferox in the Malagasy rainforest, with Implications for conservation. Small Carnivore
Conservation 20.
Dollar, L., 2006. Morphometrics, diet, and conservation of Cryptoprocta ferox. Duke University.
Dollar, L., Ganzhorn, U., Goodman, S.M., 2007. Primates and Other Prey in the Seasonally
Variable Diet of Cryptoprocta ferox in the Dry Deciduous Forest of Western
Madagascar, In Primate Anti-Predator Strategies. eds S.L. Gursky, K.A.I. Nekaris, pp.
63-76. University of Chicago Press, Chicago.
Dunnett, C.W., 1980. Pairwise Multiple Comparisons in the Unequal Variance Case. Journal of
the American Statistical Association 75, 796-800.
Efford, M.G., Dawson, D.K., Borchers, D.L., 2009. Population density estimated from locations
of individuals on a passive detector array. Ecology 90, 2676-2682.
Ganzhorn, J.U., Klaus, S., Ortmann, S., Schmid, J., 2003. Adaptations to seasonality: some
primate and non-primate examples, In Primate Life Histories and Socioecology. eds P.M.
Kappeler, M.E. Pereira, pp. 132–148. University of Chicago Press, University, Chicago,
USA.
Gerber, B., Karpanty, S.M., Crawford, C., Kotschwar, M., Randrianantenaina, J., 2010. An
assessment of carnivore relative abundance and density in the eastern rainforests of
Madagascar using remotely-triggered camera traps. Oryx 44, 219-222.
Golden, C.D., 2009. Bushmeat hunting and use in the Makira Forest, north-eastern Madagascar:
a conservation and livelihoods issue. Oryx 43, 386-392.
91
Goodman, S.M., 2003a. Carnivora: Galidia elegans, Ring-tailed Mongoose, Vontsira Mena, In
The Natural History of Madagascar. eds S.M. Goodman, J.P. Benstead, pp. 1351-1354.
The University of Chicago Press, Chicago.
Goodman, S.M., 2003b. Galidictis, Broad-striped Mongoose, Vontsira Fotsy, In Natural History
of Madagascar. eds S.M. Goodman, J.P. Benstead, pp. 1354-1357. The University of
Chicago Press, Chicago.
Goodman, S.M., 2003c. Predation on Lemurs, In The Natural History of Madagascar. eds G.S.
M., B. J.P., pp. 1221–1227. The University of Chicago Press, Chicago.
Goodman, S.M., Helgen, K.M., 2010. Species limits and distribution of the Malagasy carnivoran
genus Eupleres (Family Eupleridae). Mammalia 74, 177-185.
Goodman, S.M., Kerridge, F.J., Ralisoamalala, R.C., 2003. A note on the diet of Fossa fossana
(Carnivora) in the central eastern humid forests of Madagascar. Mammalia 67, 595-597.
Goodman, S.M., Pidgeon, M., 1999. Carnivora of the Reserve Naturelle Integrale d'Andohahela,
Madagascar. Fieldiana: Zoology, 259-268.
Graham, M.H., 2003. Confronting multicollinearity in ecological multiple regression. Ecology
84, 2809–2815.
Hannah, L., Dave, R., Lowry, P.P., Andelman, S., Andrianarisata, M., Andriamaro, L., Cameron,
A., Hijmans, R., Kremen, C., MacKinnon, J., Randrianasolo, H.H., Andriambololonera,
S., Razafimpahanana, A., Randriamahazo, H., Randrianarisoa, J., Razafinjatovo, P.,
Raxworthy, C., Schatz, G.E., Tadross, M., Wilmé, L., 2008. Climate change adaptation
for conservation in Madagascar. Biology Letters 4, 590-594.
92
Harper, G.J., Steininger, M.K., Tucker, C.J., Juhn, D., Hawkins, F., 2007. Fifty years of
deforestation and forest fragmentation in Madagascar. Environmental Conservation 34,
325-333.
Hawkins, C.E., Racey, P.A., 2005. Low population density of a tropical forest carnivore,
Cryptoprocta ferox: implications for protected area management. Oryx 39, 35-43.
Hawkins, C.E., Racey, P.A., 2008. Food habits of an endangered carnivore, Cryptoprocta ferox,
in the dry deciduous forests of western Madagascar. Journal of Mammalogy 89, 64–74.
Heydon, M.J., Bulloh, P., 1996. The impact of selective logging on sympatric civet species in
Borneo. Oryx 30, 31-36.
Irwin, M., 2006. Ecological Impacts of Forest Fragmentation on Diademed Sifakas (Propithecus
diadema) at Tsinjoarivo, eastern Madagascar: Implications for Conservation in
Fragmented Landscapes, In Anthropology. Stonybrook University, Stonybrook, NY.
Irwin, M.T., Raharison, J.-L., Wright, P.C., 2009. Spatial and temporal variability in predation
on rainforest primates: do forest fragmentation and predation act synergistically? Animal
Conservation 12, 220-230.
Irwin, M.T., Wright, P.C., Birkinshaw, C., Fisher, B.L., Gardner, C.J., Glos, J., Goodman, S.M.,
Loiselle, P., Rabeson, P., Raharison, J.-L., Raherilalao, M.J., Rakotondravony, D.,
Raselimanana, A., Ratsimbazafy, J., Sparks, J.S., Wilmé, L., Ganzhorn, J.U., 2010.
Patterns of species change in anthropogenically disturbed forests of Madagascar.
Biological Conservation 143, 2351-2362.
IUCN, 2010. IUCN Red List of Threatened Species. Accessed 22 April, 2010.
Karanth, K.U., Chellam, R., 2009. Carnivore conservation at the crossroads. Oryx 43, 1-2.
93
Karanth, K.U., Nichols, J.D., 2002. Monitoring Tigers and Their Prey: A Manual for
Researchers, Managers and Conservationists in Tropical Asia. Centre for Wildlife
Studies, Banglamore.
Kerridge, F.J., Ralisoamalala, R.C., Goodman, S.M., Pasnick, S.D., 2003. Fossa fossana,
Malagasy Striped Civet, Fanaloka, In Natural History of Madagascar. eds S.M.
Goodman, J.P. Benstead, pp. 1363-1365. The University of Chicago Press, Chicago.
Kotschwar, M., 2010. Variation in predator communities and anti-predator behaviors of lemurs
in southeastern Madagascar. Virginia Polytechnic Institute and State University,
Blacksburg.
Kremen, C., Cameron, A., Moilanen, A., Phillips, S.J., Thomas, C.D., Beentje, H., Dransfield, J.,
Fisher, B.L., Glaw, F., Good, T.C., Harper, G.J., Hijmans, R.J., Lees, D.C., Jr., E.L.,
Nussbaum, R.A., Raxworthy, C.J., Razafimpahanana, A., Schatz, G.E., Vences, M.,
Vieites, D.R., Wright, P.C., Zjhra, M.L., 2008. Aligning conservation priorities across
taxa in Madagascar with high-resolution planning tools Science 320, 222-225.
Laurance, W.F., Cochrane, M.A., 2001. Special Section: Synergistic effects in fragmented
landscapes. Conservation Biology 15, 1488-1489.
Lehtonen, J.T., Mustonen, O., Ramiarinjanahary, H., Niemela, J., Rita, H., 2001. Habitat use by
endemic and introduced rodents along a gradient of forest disturbance in Madagascar.
Biodiversity Conservation 10, 1185-1202.
MacKenzie, D., Bailey, L.L., 2004. Assessing the fit of site-occupancy models. Journal of
Agricultural, Biological, and Environmental Statistics 9, 300-318.
94
MacKenzie, D.I., Nichols, J.D., Sutton, N., Kawanishi, K., Bailey, L.L., 2005. Improving
inferences in population studies of rare species that are detected imperfectly. Ecology 86,
1101-1113.
McGarigal, K., Cushman, S.A., Neel, M.C., Ene, E., 2002. FRAGSTATS: Spatial Pattern
Analysis Program for Categorical Maps, p. Computer software program produced by the
authors at the University of Massachusetts, Amherst.
Meijaard, E., Sheil, D., 2008. The persistence and conservation of Borneo's mammals in lowland
rain forests managed for timber: observations, overviews and opportunities. Ecological
Research 23.
Meijaard, E., Sheil, D., Marshall, A.J., Nasi, R., 2008. Phylogenetic age is positively correlated
with sensitivity to timber harvest in Bornean mammals. Biotropica 40.
Meijaard, E., Sheil, D., Nasi, R., Augeri, D., Rosenbaum, B., Iskandar, D., Setyawati, T.,
Lammertink, A., Rachmatika, I., Wong, A., Soehartono, T., Stanley, S., O'Brien, T.,
2005. Life after logging: reconciling wildlife conservation and production forestry in
Indonesian Borneo. CIFOR, Bogor, Indonesia.
Michalski, F., Peres, C.A., 2005. Anthropogenic determinants of primate and carnivore local
extinctions in a fragmented forest landscape of southern Amazonia. Biological
Conservation 124, 383-396.
Norris, S., 2006. Madagascar Defiant. Bioscience 56, 960-965.
Noss, R.F., Quigley, H.B., Hornocker, M.G., Merrill, T., Paquet, P.C., 1996. Conservation
biology and carnivore conservation in the Rocky Mountains. Conservation Biology 10,
949-963.
95
Obbard, M.E., Howe, E.J., Kyle, C.J., 2010. Empirical comparison of density estimators for large
carnivores. Journal of Applied Ecology 47, 76-84.
Otis, D.L., Burnham, K.P., White, G.C., Anderson, D.R., 1978. Statistical inference from capture
data on closed animal populations. Wildlife Monograph 62, 3-135.
Paviolo, A.n., Di Blanco, Y.E., De Angelo, C.D., Di Bitetti, M.S., 2009. Protection Affects the
Abundance and Activity Patterns of Pumas in the Atlantic Forest. Journal of Mammalogy
90, 926-934.
Pollard, J.H., 1971. On distance estimators of density in randomly distributed forests. Biometrics
27, 991-1002.
Pradel, R., 1996. Utilization of capture-mark-recapture for the study of recruitment and
population growth rate. Biometrics 52, 703-709
Rayan, D.M., Mohamad, S.W., 2009. The importance of selectively logged forests for tiger
Panthera tigris conservation: a population density estimate in Peninsular Malaysia. Oryx
43, 48-51.
Ripple, W.J., Beschta, R.L., 2006. Linking a cougar decline, trophic cascade, and catastrophic
regime shift in Zion National Park. Biological Conservation 133, 397-408.
Robert, K.S., Thomas, M.G., Mary Beth, K., Thomas, E.N., 2003. Responses of 'resistant'
vertebrates to habitat loss and fragmentation: the importance of niche breadth and range
boundaries. Diversity and Distributions 9, 1-18.
Roemer, G.W., Gompper, M.E., Valkengurgh, B.V., 2009. The Ecological Role of the
Mammalian Mesocarnivore. Bioscience 59, 165-173.
96
Sergio, F., Caro, T., Brown, D., Clucas, B., Hunter, J., Ketchum, J., McHugh, K., Hiraldo, F.,
2008. Top Predators as Conservation Tools: Ecological Rationale, Assumptions, and
Efficacy. Annual Review of Ecology, Evolution, and Systematics 39, 1-19.
Sergio, F., Newton, I., Marchesi, L., Pedrini, P., 2006. Ecologically justified charisma:
preservation of top predators delivers biodiversity conservation. Journal of Applied
Ecology 43, 1049-1055.
Soulé, M.E., Terborgh, J., 1999. Continental conservation: scientific foundations of regional
reserve networks. Island Press, Washington D.C., USA.
Tecot, S., 2008. Seasonality and predictability: The hormonal and behavioral responses of the
Red-bellied Lemur, Eulemur rubriventer, in southeastern Madagascar. The University of
Texas, Austin.
Terborgh, J., Estes, J., Paquet, P., Ralls, K., Boyd-Heger, D., Miller, B., Noss, R., 1999. The role
of top carnivores in regulating terrestrial ecosystems, In Continental Conservation:
Scientific Foundations of Regional Reserve Networks. eds M.E. Soulé, J. Terborgh.
Island Press, Washington.
Thorne, J.H., Cameron, D., Quinn, J.F., 2006. A conservation design for the central coast of
California and the evaluation of mountain lion as an umbrella species. Natural Areas
Journal 26, 137-148.
Tobler, M.W., Carrillo-Percastegui, S.E., Pitman, R.L., Mares, R., Powell, G., 2008. An
evaluation of camera traps for inventorying large- and medium-sized terrestrial rainforest
mammals. Animal Conservation 11, 169-178.
Turner, I.M., 1996. Species Loss in Fragments of Tropical Rain Forest: A Review of the
Evidence. Journal of Applied Ecology 33, 200-209.
97
Vallan, D., 2002. Effects of anthropogenic environmental changes on amphibian diversity in the
rain forests of eastern Madagascar. Journal of Tropical Ecology 18, 725-742.
Vallan, D., Andreone, F., Raherisoa, V.H., Dolch, R., 2004. Does selective wood exploitation
affect amphibian diversity? The case of An‟Ala, a tropical rainforest in eastern
Madagascar. Oryx 38, 410-417.
Vanak, A.T., Gompper, M.E., 2009. Dogs Canis familiaris as carnivores: their role and function
in intraguild competition. Mammal Review 39, 265-283.
Vences, M., Wollenberg, K.C., Vieites, D.R., Lees, D.C., 2009. Madagascar as a model region of
species diversification. 24, 456-465.
White, G.C., 2008. Closed population estimation models and their extensions in Program MARK
Environmental and Ecological Statistics 15, 89-99.
White, G.C., Anderson, D.R., Burnham, K.P., Otis, D.L., 1982. Capture–recapture and removal
methods for sampling closed populations. Los Alamos National Laboratory, Los Alamos.
Wright, P.C., 1997. The Future of Biodiversity in Madagascar, In Natural Change and Human
Impact in Madagascar. eds S.M. Goodman, B.D. Patterson, pp. 381-405. Smithsonian
Institution Scholarly Press, Washington D.C.
Wright, P.C., Andriamihaja, B.A., 2003. The conservation value of long-term research: A case
study from Parc National de Ranomafana, In Natural History of Madagascar. eds S.
Goodman, J. Benstead. University of Chicago Press, Chicago.
Yoder, A.D., Burns, M.M., Zehr, S., Delefosse, T., Veron, G., Goodman, S.M., Flynn, J.J., 2003.
Single origin of Malagasy Carnivora from an African ancestor. Nature 421, 734-737.
Zar, J.H., 1998. Biostatistical Analysis. Prentice Hall, Upper Saddle River.
98
Table 1. Photographic-sampling grids at four rainforest sites in Madagascar and summary capture/recapture data of the individually-
identifiable Cryptoprocta ferox and Fossa fossana. Sampling occurred from May-December, 2008 and October-December, 2009.
Rainforest Grid
Mean Camera Spacing
(m ± SD)
(No. Camera Stations)
Season
Sampling
Dates
Total Trap
Nightsa
C. ferox Adults:
male-female
(total recaptures)b
F. fossana Individuals:
male-female
(total recaptures)b
Primary
513 ± 93
(27)
Cold-Dry
08/12/08 –
10/09/08
1446
3-3, (75)
16-17, (644)
Selectively-Logged
567 ± 94
(26)
Cold-Dry
06/09/08 –
08/09/08
1417
2-2, (37)
11-11, (428)
Fragmented <2.5 km
558 ± 128
(31)
Warm-Dry
10/15/09 –
12/07/09
1379
1-0, (1)
0-0, (0)
Fragmented >15 km
584 ± 86
(27)
Warm-Dry
10/20/08 –
12/11/08
1323
0-0, (0)
0-0, (0)
aTrap nights are total sampling effort where at least one of two remote cameras were functioning at a station per 24-hour period
bTotal recaptures are for the spatially-explicit capture-recapture model, where individuals can be recaught at multiple stations on a
single sampling occasion
99
Table 2. Grid-level vegetation structure ± SE at four rainforest sites with increasing levels of disturbance in southeastern Madagascar.
Sampling occurred from May-December, 2008 and October-December, 2009. Different letters associated with vegetation variables
across sites indicate sites differ significantly in that variable (experiment-wise α = 0.05).
Vegetation Structure
Primary
(N= 27)
Selectively-logged
(N=26)
Fragmented <2.5 kma
(N=27)
Fragmented >15 kma
(N=13)
Tree Density (≥ 10 cm D130 stems/ha)
1612 ± 27 A
622 ± 17 B
337 ± 11 C
500 ± 11 BC
Basal Area (m2/ha, stems ≥ 10 cm D130)
102 ± 13 A
24 ± 2 B
8 ± 4 C
13 ± 3 C
Down/Dead Cover % (≥ 15cm D130)
5 ± 2 A
4 ± 0.5 A
2 ± 0.3 B
0.6 ± 0.6 C
Ground Cover % ( 0-0.5 m)
70 ± 4 A
91 ± 1 B
57 ± 4 A
91 ± 3 B
Shrub Cover % (0.5-5 m)
87 ± 4 AB
94 ± 2 A
78 ± 4 B
85 ± 3 AB
Low-Tree Canopy Cover % (0-5 m)
5 ± 2 A
11 ± 0.9 B
45 ± 4 C
16 ± 2 B
Mid-Tree Canopy Cover % (5-15 m)
58 ± 5 A
86 ± 2 B
27 ± 4 C
31 ± 4 C
High-Tree Canopy Cover % (≥ 15 m)
66 ± 4 A
38 ± 4 B
6 ± 2 C
4 ± 2 C
a Rainforest habitat only was quantified in these two fragmented forest sites which are respectively <2.5 and > 15 km from the
contiguous rainforest
100
Table 3. Model selection (model likelihood >0.125) using a maximum-likelihood spatially-explicit capture-recapture model to
estimate density of the fossa (Cryptoprocta ferox) and Malagasy civet (Fossa fossana) in a Primary and Selectively-logged contiguous
rainforest grid within Madagascar‟s eastern rainforests. Sampling occurred from May-December, 2008 and October-December, 2009.
Species
Rainforest Grid
Modelsa
AICc
ΔAICc
wi
Model
Likelihood
k
Deviance
Cryptoprocta
Primary
HalfNormal g0(sex) σ(.)
865.61
0.00
0.446
1.00
3
839.27
ferox
HalfNormal g0(h) σ(.)
865.66
0.05
0.435
0.98
3
839.32
HalfNormal g0(.) σ(.)
868.42
2.81
0.109
0.25
2
842.08
Selectively-Logged
NegExp g0(h) σ(h)
438.14
0.00
0.713
1.00
4
465.41
NegExp g0(Sex) σ(Sex)
442.02
3.88
0.102
0.14
4
469.29
Fossa fossana
Primary
Hazard Rate g0(sex) σ(h)
3858.47
0.00
0.99
1.00
6
3782.57
Selectively-Logged
NegExp g0(sex+lure) σ(h)
2479.68
0.00
0.99
1.00
6
2426.87
101
aDistance functions: Hazard Rate, Half-Normal, Negative-Exponential; g0 is the detection probability when a single trap is located at
the center of an animal‟s home range; σ is the spatial scale detection probability away from the center of the home range; Sex is males
versus females; Lure is chicken meat used as a scent-lure and unavailable for consumption by carnivores; h is individual heterogeneity
102
Table 4. Density estimates of the fossa (Cryptoprocta ferox) and Malagasy civet (Fossa fossana) in four rainforest areas within
Madagascar with statistical significance (95% confidence interval of two means overlap less than half the average margin of error)
indicated with different capital letters (α = 0.05). Sampling occurred from May-December, 2008 and October-December, 2009.
Rainforest Grid
Cryptoprocta ferox
Density ± SE
(adults/km2)
Fossa fossana
Density ± SE
(individuals/km2)
Primary
0.12 ± 0.05 A
3.19 ± 0.55 C
Selectively-Logged
0.09 ± 0.04 A
1.38 ± 0.22 D
Fragmented <2.5 km
≥0a
0 B
Fragmented >15 km
0 B
0 B
aOne animal was detected, which precluded estimation
103
Table 5. Model selection (model likelihood >0.125) of occupancy and detection probability for native and exotic rainforest carnivores
in southeastern Madagascar at two contiguous (Primary and Selective-logged) and two fragmented (< 2.5 and > 15 km from
contiguous forest) sampling grids. Sampling occurred from May-December, 2008 and October-December, 2009.
Rainforest
Grid
Species
Modelsa
AICc
ΔAICc
wi
Model
Likelihood
k
-2*log
likelihood
Contiguous
Canis familiaris
ψ(DistVillage) p(Locals)
235.92
0.00
0.61
1.00
4
227.92
ψ(Grid + DistVillage),p(Locals)
237.65
1.73
0.26
0.42
5
227.65
Cryptoprocta ferox
ψ(Grid+Locals) p(Trail + DistMatrix)
993.86
0.00
0.41
1.00
6
981.86
ψ(Locals) p(Trail + DistMatrix)
995.92
2.06
0.15
0.36
5
985.92
ψ(.) p(Trail + DistMatrix)
997.21
3.35
0.08
0.19
4
989.21
ψ(Grid) p(Trail + DistMatrix)
998.02
4.16
0.05
0.13
5
988.02
Fossa fossana
ψ(.) p(Lure)
3434.08
0.00
0.99
1.00
3
3428.08
Galidictis fasciata
ψ(Grid + BasalArea) p(DistMatrix +
Dogs)
528.36
0.00
0.92
1.00
6
515.21
104
Fragmented
Canis familiaris
ψ(Grid) p(Locals + ForestType + Lure)
1278.88
0.00
0.84
1.00
6
1266.88
ψ(.) p(Locals + ForestType + Lure)
1282.34
3.46
0.15
0.18
5
1270.34
Felis catus/silvestris
ψ(DD) p(Dog)
373.05
0.00
0.66
1.00
4
364.22
ψ(Grid+DD) p(Dog)
375.29
2.24
0.22
0.33
5
364.01
Galidia elegans
elegans
ψ(ForestType) + Cat p(Grid)
973.43
0.00
0.85
1.00
5
962.25
Galidictis fasciata
ψ(Area) p(Grid)
456.53
0.00
0.99
1.00
4
447.76
Viverricula indica
ψ(ForestType + Locals) p(ForestType)
541.85
0.00
0.40
1.00
5
531.85
ψ(ForestType) p(ForestType)
542.93
1.08
0.23
0.58
4
534.93
ψ(ForestType + Cat) p(ForestType)
544.33
2.48
0.12
0.29
5
534.33
ψ(ForestType + Dog) p(ForestType)
544.72
2.87
0.10
0.24
5
534.72
aψ : occupancy, p: detection probability, Area: rainforest fragment area, BasalArea: basal area, Cat: Felis catus/silvestris trap success,
DD: down and dead cover, DistMatrix: closest distance to non-rainforest, DistVillage: distance to closest village, Dog: Canis
familiaris trap success, ForestType: difference at fragmented grids between rainforest habitat and matrix, Grid: difference between
105
contiguous rainforest grids or fragmented rainforest grids, Locals: local human trap success, Lure: effect of using lure versus not using
at only Selectively-logged and Fragments <2.5 km grids, Trail: maintained trails versus small trails
106
Table 6. Occupancy analyses of detection probability (p) regression coefficients (β (SE)) for the top models of endemic and exotic
carnivores at two fragmented and two contiguous rainforest grids within Madagascar‟s eastern forests. Sampling occurred from May-
December, 2008 and October-December, 2009.
Rainforest
Grids
Parametera
Canis
familiaris
Cryptoprocta
feroxb
Felis
catus/silvestrisc
Fossa
fossanac
Galidia
elegans
elegansd
Galidictis
fasciata
Viverricula
indicac
Contiguous
Intercept
-3.90 (0.31)
-2.57 (0.10)
-
1.44 (0.09)
-
-4.02 (0.24)
-
DistMatrix
-
0.43 (0.10)
-
-
-
1.19 (0.23)
-
Dog
-
-
-
-
-
-1.16 (0.38)
-
Locals
0.60 (0.18)
-
-
-
-
-
-
Lure
-
-
-
0.90 (0.10)
-
-
-
Trail
-
-1.62 (0.30)
-
-
-
-
-
Fragmented
Intercept
-3.07 (0.18)
-
-3.63 (0.19)
-
-1.77 (0.09)
-2.67 (0.15)
-3.66 (0.31)
Dog
-
-
0.81 (2.44)
-
-
-
-
ForestType
0.95 (0.16)
-
-
-
-
-
0.96 (0.36)
Grid
-
-
-
-
-1.90 (0.27)
-1.91 (0.43)
-
107
Locals
0.43 (0.05)
-
-
-
-
-
-
Lure
0.68 (0.20)
-
-
-
-
-
-
aDash(-) indicates the parameter is not applicable and bold indicates the estimate is statistically different than zero. DistMatrix: closest
distance to non-rainforest, Dog: Canis familiaris trap success, ForestType: difference at fragmented grids between rainforest habitat
(1) and matrix (0), Grid: difference between fragmented rainforest grids (Fragments >15 km (0) and Fragments <2.5 km (1)) or
contiguous rainforest grids (Priamry (0) and Selectively-Logged (1)), Locals: local human trap success, Lure: effect of using lure
versus not using at only Selectively-logged and Fragments <2.5 km grids, Trail: maintained (0) trails versus small secondary trails (1)
bC. ferox was detected at one of the fragmented grids, but poor recaptures precluded modeling
cSpecies was not detected at either the fragmented or the contiguous rainforests
dG. e. elegans was detected at both contiguous forests, but naïve occupancy equals one at those grids, thus precluded modeling
108
Table 7. Occupancy analyses (ψ) regression coefficients (β (SE)) for the top models of each carnivore species at two fragmented and
contiguous rainforest grids within Madagascar‟s eastern forests. Sampling occurred from May-December, 2008 and October-
December, 2009.
Rainforest
Grids
Parametera
Canis
familiaris
Cryptoprocta
feroxb
Felis
catus/silvestrisc
Fossa
fossanac
Galidia
elegans
elegansd
Galidictis
fasciata
Viverricula
indicac
Contiguous
Intercept
-0.91 (0.51)
1.85 (1.43)
-
3.95 (1.10)
-
-0.17 (0.42)
-
BasalArea
-
-
-
-
-
0.99 (0.38)
-
DistVillage
-1.21 (0.54)
-
-
-
-
-
-
Grid
-
2.33 (1.27)
-
-
-
-0.10 (0.45)
-
Locals
-
3.15 (2.62)
-
-
-
-
-
Fragments
Intercept
0.188 (0.40)
-
1.30 (1.17)
-
1.19 (0.72)
6.34 (3.13)
0.28 (1.42)
Area
-
-
-
-
-
12.05 (4.74)
-
Cat
-
-
-
-
-1.84 (0.80)
-
-
DD
-
-
4.17 (2.17)
-
-
-
-
ForestType
-
-
-
-
-2.71 (1.09)
-
2.63 (1.06)
109
Grid
1.18 (0.44)
-
-
-
-
-
-
Locals
-
-
-
-
-
-
5.86 (8.37)
aDash (-) indicates the parameter is not applicable and bold indicates the estimate is statistically different than zero. Area: rainforest
fragment area, BasalArea: basal area, Cat: Felis catus/silvestris trap success, DD: down and dead cover, DistVillage: distance to
closest village, ForestType: difference at fragmented grids between rainforest habitat (1) and matrix (0), Grid: difference between
fragmented rainforest grids (Fragments >15 km (0) and Fragments <2.5 km (1)) or contiguous rainforest grids (Priamry (0) and
Selectively-Logged (1)), Locals: local human trap success
bC. ferox was detected at the fragmented sites, but with no recaptures at only two sites precluded occupancy estimation
cSpecies was not detected at either the contiguous or fragmented sites
d G. e. elegans was detected at both contiguous forests, but naïve occupancy equals one at those grids, thus precluded modeling
110
Table 8. Model-averaged occupancy (SE) of native and exotic carnivores in two contiguous and two fragmented rainforest sites within
Madagascar‟s eastern forests. Sampling occurred from May-December, 2008 and October-December, 2009.
Contiguous Rainforest Grids
Fragmented Rainforest Grids
Primary
Selectively-
Logged
<2.5 km from Primary
>15 km from Primary
Rainforest
Matrix
Rainforest
Matrix
Canis familiaris
0.27 (0.08)
0.39 (0.10)
0.87 (0.07)
0.87 (0.07)
0.67 (0.06)
0.67 (0.06)
Cryptoprocta ferox
0.93 (0.05)
0.72 (0.08)
0.069a
0.00
0.00
0.00
Eupleres goudotii
0.07a
0.31a
0.00
0.00
0.00
0.00
Fossa fossana
0.98 (0.02)
0.98 (0.02)
0.00
0.00
0.00
0.00
Galidia elegans elegans
1.00a
1.00a
0.77 (0.10)
0.19 (0.12)
0.77 (0.10)
0.19 (0.12)
Galidictis fasciata
0.57 (0.13)
0.23 (0.10)
0.67 (0.09)
0.04 (0.04)
0.67 (0.09)
0.04 (0.04)
Felis catus/silvestris
0.00
0.00
0.64 (0.12)
0.64 (0.12)
0.41 (0.09)
0.41 (0.09)
Viverricula indica
0.00
0.00
0.57 (0.08)
0.94 (0.04)
0.57 (0.08)
0.94 (0.04)
aNaïve occupancy, due to limited dataset
bNaïve occupancy equals one at those grids, thus precluded modeling
111
Figure 1. Carnivores were sampled at four photographic-sampling sites within Madagascar‟s
eastern rainforests from May-December, 2008 and October-December, 2009. Top insert map
shows political provinces and location on the island of Madagascar, while the bottom insert map
shows the camera station sampling layout among rainforest fragments at Mahatsinjo, Tsinjoarivo
as an example. Rainforest extent from Conservation International, 2010.
112
Figure 2. Broad-striped mongoose (Galidictis fasciata) occupancy variation by basal area in
contiguous rainforests within Ranomafana National Park and occupancy variation by fragment
size in fragmented rainforests (<2.5 km and >15 km from intact contiguous rainforest), sampled
within Madagascar‟s eastern forests from May-December, 2008 and October-December, 2009.
0
50
100
150
200
250
0
50
100
150
200
250
00.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Rainforest Patch Size (ha)
Basal Area (m2 / ha)
Probability of Occupancy
Basal Area (Primary)
Basal Area (Selectively-Logged)
Rainforest Fragment Size (Fragments)
113
Figure 3. Ring-tailed mongoose (Galidia elegans elegans) occupancy variation within rainforest
fragments and surrounding matrix (open human-dominated habitat) and the effect of exotic-wild
cats (Felis catus/silvestris). Sampling occurred within Madagascar‟s eastern forests from May-
December, 2008 and October-December, 2009.
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
012345678910
Probability of Occupancy
Wild Cat Trap Success (Capture Events / 100 Trap Nights)
Rainforest Fragments
Matrix
114
Figure 4. Madagascar‟s primary forest cover (2005) and estimated fossa (Cryptoprocta ferox)
populations >300 adults. Ninety-five percent of the total rainforest population is contained in
two forest complexes, while 80% of the total dry forest population is contained in two forest
complexes. Forest fragments were joined into a single population when forested areas large
enough to hold ≥ 1 fossa were <4.9 km from each other.
115
Chapter 4 - Temporal activity patterns of Malagasy rainforest carnivores
Abstract:
Understanding the temporal activity patterns of animals can provide meaningful insight
into behavioral responses to habitat disturbance and mechanisms of co-existence among
sympatric species. We used photographic-sampling to quantify the temporal activity patterns of
Madagascar‟s native carnivores (Cryptoprocta ferox, Fossa fossana, Galidia elegans elegans,
Galidictis fasciata, Eupleres goudotii), exotic carnivores (Viverricula indica, Canis familiaris,
Felis catus, Felis silvestris) and local people within the eastern rainforests. We sampled
carnivores in two contiguous (Primary and Selectively-logged) and two fragmented rainforests
(Fragments <2.5 km and >15 km from intact primary rainforest). To avoid issues of non-
independence of consecutive photographs, we defined an activity-sample as the median time of
all photographs of the same species or individual taken within a 0.5 hour period and applied a
kernel density estimator to quantify activity patterns. We compared the overlap in the daily
activity patterns among all species and assessed how each individual species used dawn, day,
dusk, and night time periods with respect to their availability in the diel cycle. Malagasy
carnivores had diverse activity patterns (diurnal, nocturnal, crepuscular) with varied temporal
activity overlap (5.8 to 88.8%) among species. Species with greater temporal overlap differed
from each other in other aspects of their ecology, such as diet, in accordance with predictions
from niche theory. Rainforest site and season had little effect on the diurnal G. elegans elegans
or the nocturnal G. fasciata, which were detected at all sites. C. ferox in the contiguous
rainforest selected the crepuscular hours greater than would be expected by the availability of
this time period in the diel cycle, but overall the activity pattern of this species could be
described as cathemeral. As a major lemur predator, understanding C. ferox activity patterns will
116
help to better understand lemur predation risk and the evolution of anti-predator behaviors,
specifically cathemerality of many lemur species at risk of predation by C. ferox. The temporal
overlap of exotic carnivores with native carnivores is of significant conservation concern due to
the potential for both direct resource competition and disease transmission.
Introduction:
The distribution of a species‟ activity through time is an important niche-dimension that
has profound implications for its ecology and evolution (Kronfeld-Schor and Dayan 2003;
Pianka 1973). How species use time is inextricably linked to morphological and physiological
adaptations (Aschoff et al. 1982; Hayward and Slotow 2009). For example, most carnivores
have vision that is highly adaptable for activity throughout the diel-cycle (Kavanau and Ramos
1975), so it is not surprising that many carnivores also exhibit a high degree of plasticity in
temporal activity patterns (Gittleman 1986; Zielinski 2007). Carnivore temporal activity has
been shown to be influenced by prey availability (Zielinski et al. 1983), daily or seasonal
temperature variation (Lourens and Nel 1990; Zub et al. 2009), human activity (Beckmann and
Berger 2003; Griffiths and Schaik 1993; Kolowski et al. 2007), interspecific competition
(Hayward and Slotow 2009; Hunter and Caro 2008; Romero-Muñoz et al. 2010), and intra-guild
predation (Palomares and Caro 1999).
While the activity patterns of many carnivores appear temporally flexible, we know little
about how carnivores may alter their activity patterns in response to new disturbances and
ecological changes, such as those caused by forest loss and fragmentation (Beckmann and Berger
2003). If an animal‟s temporal activity is at all constrained by phylogeny (Roll and Kronfeld-
Schor 2006), it may be unable to adapt in ecological time to a changing environment. Impacts of
117
habitat loss are often evaluated with regard to altering species richness and abundance.
However, to fully understand ecological processes at the population level, we also need
knowledge of the impact on individual‟s behavior (i.e. activity patterns) and ecological
relationships (Norris et al. 2010; Presley et al. 2009; Sutherland and Dolman 1994).
The cue to which many animals maintain their activity patterns is the day/night cycle
(Kronfeld-Schor and Dayan 2003), which is of course unaltered by habitat change. Day-length
(or night-length) can act as an ecological constraint that limits an animal‟s time to perform its
essential activities (Hill et al. 2003). However, not all mammals exhibit simple unimodal
activity, as either diurnal or nocturnal. More complex or multimodal patterns, such as
crepuscular activity (bimodal peaks in activity at dawn and dusk), are common among carnivores
(Gittleman 1986). Cathemerality, which is defined by an evenness of activity throughout the
diel-cycle, or when significant activity occurs during both the day and night (Tattersall 2006), is
also common across mammalian taxa (Curtis et al. 2006; Halle and Stenseth 2000; Schaik and
Griffiths 1996). Crepuscular and cathemeral activity patterns may allow individuals to adjust to
habitat alterations or new disturbance events (i.e. human activity) compared to strict
diurnal/nocturnal patterns which constrain an individual to activity just during those limited
hours (Hill et al. 2003).
Like many carnivores, the activity patterns of Madagascar‟s rainforest carnivores are still
poorly understood. Of the nine extant endemic carnivore species within the endemic family
Eupleridae (Goodman and Helgen 2010; Yoder et al. 2003), five are known to occupy the
southeastern rainforests (broad striped mongoose Galidictis fasciata, fossa Cryptoprocta ferox,
Malagasy civet Fossa fossana, ring-tailed mongoose Galidia elegans elegans, small-toothed
civet Eupleres goudotii; Gerber et al. 2010). Malagasy carnivore activity patterns have only
118
generally been described (Albignac 1972; Dollar 1999; Hawkins 1998). As these Malagasy
rainforest carnivores co-occur at the local scale (i.e. at a sampling station; B. Gerber,
unpublished data), temporal activity separation may be an important factor in their co-existence.
Additionally, C. ferox and G. e. elegans are known lemur predators, which suggests their activity
patterns may have direct implications for lemur predation risk and the evolution of anti-predator
behaviors, including lemur activity patterns (Colquhoun 2006; Karpanty and Wright 2007).
Our objectives were to sample Madagascar‟s endemic rainforest carnivores, exotic
carnivores (domestic dog, Canis familiaris; exotic-wild cat, Felis catus and Felis silvestris,
grouping the domestic and introduced African wildcat, F. catus/silvestris, as they are difficult to
distinguish from each other; small-Indian civet Viverricula indica) and local people using non-
invasive remote photographic-sampling to 1) quantify temporal activity patterns, 2) compare
temporal activity overlap among native and exotic carnivores and local people across a
continuum of increasingly disturbed rainforests, and 3) evaluate the selection or avoidance of the
dawn, day, dusk, and night time-periods through the diel-cycle.
Materials and Methods:
Study Areas:
We worked in the southeastern rainforests of Madagascar at four study sites (Fig. 1). We
sampled the contiguous rainforests at the Valohoaka-Vatoranana (Primary) and Sahamalaotra
trail-systems (Selectively-logged) located in the boundaries of Ranomafana National Park
(RNP). We also sampled two fragmented rainforests at Mahatsinjo,Tsinjoarivo (Fragments <2.5
km from intact rainforest) and Ialatsara Forest Station (Fragments >15 km from intact rainforest).
119
Our Primary rainforest site had little anthropogenic disturbance (Wright 1997), while our
Selectively-logged rainforest experienced moderate logging pressure before RNP was
established in 1991 (JC. Razafimahaimodison, Pers. Comm.). Our Fragments <2.5 km site was
located 150 km north of RNP within a network of unprotected hilltop rainforest fragments <2.5
km from the large tract of contiguous primary rainforest. Fragments were surrounded by a
human-dominated landscape of agriculture and naturally regenerating shrub. Our Fragments >15
km site was northwest of RNP, located in a protected fragmented rainforest >15 km from the
main tract of contiguous primary rainforest. Fragments were surrounded by shrub, exotic pine
and eucalyptus, open areas recently burned, and minimal agriculture. In both fragmented sites,
forest edges were hard transitions to open habitat, caused by human activities such as burning,
grazing, and farming. Local people used both fragmented study areas for travel and forest
products, but only at the Fragments <2.5 km site did people live between the fragments.
Sampling took place over a period of two years from May-December, 2008 (Primary,
Selectively-logged, and Fragments >15 km) and October-December, 2009 (Fragments <2.5 km).
The two contiguous rainforests at RNP were sampled in the cold-dry season (April-October) and
the two fragmented grids in the warm-dry season (November-February, Tecot 2008). A five year
average (2005-2009) min-max daily temperature was 13-20 ˚C at RNP during the cold-dry
season. The temperature during the warm-dry season sampling at the Fragments >15 km and
Fragments <2.5 km sites were 15-24˚C and 12-26˚C, respectively. The five year average (2005-
2009) daily rainfall ± SD was 6.2 ± 11.4 mm at RNP during the cold-dry season. The average
rainfall ± SD during the warm-dry season sampling at the Fragments >15 km and Fragments
<2.5 km sites were 5.1 ± 8.4 mm and 7.4 ± 14.0 mm, respectively.
120
Sampling:
We used a random starting point to establish a systematic grid of passive-infrared camera
stations along established trail-systems to photographically sample carnivores at each rainforest
site. We deployed 26 to 31 camera stations for at least 52 days in each site. The average spacing
± SD among adjacent stations was 555 ± 100 m. Camera stations consisted of two
independently-operating passive infrared cameras, mounted on opposite sides of a trail. This
allowed a photographic-capture of both flanks of every animal, thus improving individual
identification in recaptures for C. ferox and F. fossana, which are individually-identifiable
(Gerber et al. 2010). We attached cameras 20 cm above the ground and set them to be active for
24 hours/day. Camera types included Deercam DC300‟s (DeerCam, Park Falls, USA) which
were to set to a 1-min delay between consecutive photographs and Reconyx PC85‟s (Reconyx,
Inc. Holmen, Wisconsin), which were set with no time delay. Each photograph recorded the date
and time of the sample. We used chicken meat as a scent-lure to effectively sample these
rare/elusive species at all sites. However, in the Selectively-logged and Fragments <2.5 km sites,
we sampled for approximately half the time with and half without scent-lure to evaluate the
effect on carnivore activity patterns. We checked camera stations every three to five days to
ensure continued operation, replacing batteries, scent-lure, film, and memory cards when
necessary.
Data Analyses:
Temporal Activity Pattern Distribution and Overlap
We defined an activity-sample as the median time of all photographs of the same species
or individual (C. ferox and F. fossana) detected at a camera station within a 0.5 hour period, thus
121
avoiding non-independence of consecutive photographs (O'Brien et al. 2003; Ridout and Linkie
2009). To assess whether this effectively reduced the temporal dependency of activity samples
within our datasets, we calculated the median time between consecutive samples of individuals
within a rainforest site, or consecutive samples of individually-unidentifiable species at each
camera station. To evaluate the effect of scent-lure on carnivore activity, we tested if species
activity distributions (over total diel-cycle) from data collected with and without lure were
different using the non-parametric circular Mardia-Watson-Wheeler statistical test (MWW;
Batschelet 1981). Analyses using this test were restricted to datasets with ≥ 10 activity samples
(Fisher 1993). We considered P < 0.1 as significant.
We then tested for individual, sex, and rainforest site effects on activity patterns. For the
individually-identifiable C. ferox and F. fossana, we first tested for differences among
individual‟s activity distributions within a site using MWW. If no differences were observed
among individuals, we pooled datasets at each site and tested for differences among sites and
sexes. For all other carnivores and local people, we were unable to test for individual variation
and thus recognize the pooling of activity samples at each site is possibly pseudoreplication,
depending on if there was individual variation in activity patterns (Aebischer et al. 1993).
However, we still used these pooled data to test for differences of species‟ activity distributions
among sites using MWW.
For all species, we pooled datasets when activity distributions were not different across
forest sites or forest sites and sex. We then used Rao‟s test of uniformity on these pooled/un-
pooled species‟ activity distributions to evaluate whether species exhibit a uniform or cathemeral
activity pattern throughout the diel cycle using the R package Circular (R Development Core
Team 2010). The Rao‟s test was preferred over Rayleigh‟s or Kuiper's V Test as it is more
122
powerful when the data is multimodal (Mardia and Jupp 2000), as is common with carnivores
(Gittleman 1986). If distributions were non-uniform, we calculated the directional median and
standard error of activity in the diel cycle.
To quantify the overall activity pattern of each species and overlap among sympatric
species, we utilized a kernel density analysis (Ridout and Linkie 2009). Following Ridout and
Linkie (2009) we attempted to minimize bias with small or large sample sizes by employing two
kernel estimators; we used their equation 3.1 (page 325) with a smoothing parameter of 1.25
when sample sizes were ≤ 50 and their equation 3.3 (page 325) with a smoothing parameter of
1.00 when sample sizes were ≥ 50.
Relative Preference and Selection/Avoidance of Time Periods in the Diel-Cycle
To determine the relative preference of, and selection and/or avoidance of, different
periods of the diel-cycle by carnivores and local people, we categorized the diel-cycle based on
sunrise/sunset times at each rainforest site into dawn, day, dusk, and night. We defined the
crepuscular hours of dawn and dusk as ± 1 hour before and after sunrise and sunset. For C. ferox
and F. fossana, from which we could identify the sexes, we used a generalized multinomial
model with a logit link (PROC LOGISTIC; SAS Institute Inc.) to test whether the proportion of
use among dawn, day, dusk and night differed by site and sex. For other species for which we
could not identify the sexes, we tested whether the proportion of use among dawn, day, dusk, and
night differed by site using a likelihood-ratio chi-square test in a contingency table (Zar 1998).
If we found differences, we used partial chi-square cell values to interpret the contribution of
individual cells to overall significance. When we found no differences among forest sites or
forest sites and sex, we pooled datasets.
123
We used the subsequent datasets to examine whether Malagasy carnivores and local
people select or avoid dawn, day, dusk and night, given their availability. For C. ferox and F.
fossana, we were able to use the individual as the sampling unit and employed compositional
analyses (Aebischer et al. 1993). Although we could not identify individuals of G. e. elegans,
we still used compositional analyses using camera stations as the sampling unit. Camera stations
are likely to capture some variation among individuals, and serve as a proxy for individual,
within the forest sites as home ranges for this species are likely small compared to camera station
spacing (Dunham 1998). Additionally, sample sizes of G. e. elegans activity samples were
adequate at camera stations which made this feasible. We used the kernel density analysis to
derive the proportions of use among temporal classes from individuals and camera stations with
≥ 10 activity samples from the kernel probability distributions. We were unable to use the
camera station as a sampling unit for other individually-unidentifiable species due to too few
activity samples per camera station.
Compositional analysis calculates the difference in log-ratios of proportions of used and
available temporal classes for each individual. We used a randomization procedure with 1000
iterations to overcome issues when log-ratio differences were not multivariate normal and
calculated Wilks‟ lambda (λ), a multivariate analog to the t-test (Mardia et al. 1979), to test if
temporal activity classes were used randomly. If temporal classes were used non-randomly (P
<0.1), this implied that the species selected or avoided certain temporal classes and we then
performed a series of pair-wise comparisons to determine relative preference (Aebischer et al.
1993).
For all other species (C. familiaris, E. goudotii, F. catus/silvestris, G. fasciata, V. indica,
local people), we determined selection/avoidance of temporal classes using Bailey‟s
124
simultaneous confidence intervals constructed around the proportional use of temporal classes
(Cherry 1996). As before, we are unable to account for individual variation and recognize this is
possibly pseudoreplication (Aebischer et al. 1993). We used the kernel density analysis to derive
the proportions of use among temporal classes from datasets with ≥10 activity samples from the
kernel probability distributions. A Bonferroni correction was used to control for the non-
independence of multi-comparisons, thus we increased the Type II error to fix the experiment-
wise error rate at P <0.1 (Byers et al. 1984; Cherry 1996). When the available temporal class
was below the lower confidence limit interval, the time class was selected. When the available
proportion was higher than the upper confidence limit interval, the time class was avoided.
Otherwise, the time class was used in proportion to its availability. When there were no
observations in a given temporal class and the availability was large, we assumed the species
avoided this time period. These analyses cannot account for relative preference, thus there are
no rankings of temporal classes for the individually-unidentifiable species.
Results:
We collected activity data on all five native southeastern rainforest carnivores, three
exotic carnivores, and local people at our four study sites (Table 1). Species presence and
number of activity samples at the forest sites varied, and the limited datasets (≤10 activity
samples) of C. ferox at Fragments <2.5 km, and C. familiaris and E. goudotii at the Primary
rainforests, precluded certain species comparisons across all sites.
We identified fifteen C. ferox individuals in total, but only three individuals each at the
Primary and Selectively-logged forests had ≥ 10 observations per individual for analyses. We
identified fifty-five F. fossana individuals total at the Primary and Selectively-logged sites, of
125
which thirty-one could be used for analyses. F. fossana were absent from the two fragmented
forest sites. The median time between consecutive photographs of C. ferox and F. fossana
individuals at a forest site was greater than 17.5 hours. The median time between consecutive
photographs of individually-unidentifiable species at a camera station was greater than 19 hours,
except at the Primary site where G. e elegans was observed a median time of every seven hours.
We found no differences between activity distributions of native or exotic carnivores when
sampled with or without scent-lure at either the Selectively-logged or Fragments <2.5 km
rainforests where we conducted this comparison; at the other two sites lure was used for the
entire sampling period (Table 2).
For C. ferox and F. fossana, we found no differences among individual animal‟s activity
distributions within sites (C. ferox individuals: Primary, W = 6.90, P = 0.14, Selectively-logged,
W = 5.46, P = 0.24; F. fossana individuals: Primary, W = 39.86, P = 0.11, Selectively-logged, W
= 17.97, P = 0.59) and thus pooled individuals at each site. Comparing the activity distributions
of each species across sites, we found significant differences for C. familiaris, F. fossana, G. e.
elegans, V. indica and local people (Table 1). For example, the median time of C. familiaris’
diel-activity was similar at the fragmented sites, but occurred much earlier in the day at the
Selectively-logged site (Table 3). While temporal activity patterns for F. fossana, G. e. elegans,
V. indica, and local people were different among some of the sites, the actual median time of use
was not greatly different. Of all species, only F. catus/silvestris demonstrated a uniform or
cathemeral activity pattern throughout the diel cycle (Table 3). We found species-specific and
site-specific temporal activity patterns and thus overlap varied by species and sites compared
(Table 4, Figure 2). Temporal activity patterns were remarkably similar between the Primary
and Selectively-logged sites, e.g., between C. ferox and F. fossana there was only a difference of
126
0.6% in their overlap. Across all four forest sites, the temporal activity overlap of G. e. elegans
and local people remained high at 75-88%.
We found that activity during the dawn, day, dusk, and night time periods for G. e.
elegans, V. indica, and local people differed significantly across sites (Table 5). For G. e.
elegans, we found less than expected use of dawn and greater use of dusk at the Selectively-
logged site (Fig. 3) compared to the other sites; activity in this dusk time period accounted for
70% of the total chi-square value (Dawn χ2 = 6.19, Dusk χ2 = 18.36, and Total χ2 = 34.94).
There were no differences in activity use of the different time periods across the other sites, thus
we pooled the data at the Primary, Fragments <2.5 and >15 km sites for G. e. elegans (χ2 = 5.55,
df = 6, P = 0.67). For local people, we found less than expected use of dawn at the Primary site
and greater use of dawn at the Fragments <2.5 km site compared to the other sites, which
accounted for 49% of the total chi-square value (Primary χ2 of Dawn = 4.12, Fragments <2.5 km
χ2 = 3.69, and Total χ2 =15.87). There were no differences in activity use of temporal classes
across the other sites, thus we pooled the Selectively-logged and Fragments <2.5 km sites (χ2 =
0.34, df = 2, P = 0.844) for local people. All other species showed no differences in use of
different time periods across sites, thus we pooled their datasets across all forest sites in which
they were detected.
C. ferox, F. fossana, and G. e. elegans exhibited non-random use of dawn, day, dusk and
night (Table 6). C. ferox preferred the crepuscular hours, with no significant difference between
activity at dawn and dusk. C. ferox were active during the day and night, but daytime hours were
used less than the other time classes (Table 6, Figure 3). We found F. fossana to be
predominantly nocturnal, but they also used the crepuscular hours. At all sites, G. e. elegans was
highly diurnal. We found G. fasciata to be highly nocturnal, with use of the crepuscular hours in
127
proportion to their availability and no activity during the day (Figure 4). As for the exotic V.
indica, we found them to use the night and dusk significantly more than available, while
avoiding the daytime.
Discussion:
Understanding species‟ temporal activity patterns and how those patterns may vary
across seasons, habitats, or other changing ecological conditions provides important insight into
mechanisms of species coexistence and details of ecological relationships between those
coexisting species. How animals use time can also be important in understanding and predicting
how species can persist within a changing landscape (Kronfeld-Schor and Dayan 2003; Norris et
al. 2010). In particular, changing landscapes due to habitat loss and fragmentation are often
accompanied by changing patterns of human use or altered abundances of other species that may
impact activity patterns of all species in that system. This may be especially important in
understanding predator persistence, as their activity patterns can be strongly influenced by
human activity (Griffiths and Schaik 1993) and relationships among sympatric competitors
(Schoener 1974).
We quantified the temporal activity patterns of the Malagasy rainforest carnivores across
a gradient of disturbed rainforests to gain insight into the temporal structure of carnivore
communities and how a changing landscape may influence their persistence. Our photographic-
sampling grids provided an efficient means to sample the entire carnivore community activity in
contrast to radiotelemetry which is often limited to few individuals of a single or few species.
Despite predictions that lure or bait could influence species temporal activity (Schlexer 2008),
we found that scent-lure at camera stations had no effect on the activity patterns of native or
128
exotic carnivores at the contiguous and fragmented rainforests. To account for with the non-
independence of consecutive photographs, our 30-min interval adequately removed temporal
dependence between consecutive photographs and thus reduced any systematic bias within our
analyses.
Our detailed evaluation of Malagasy rainforest carnivore activity was generally consistent
with natural history accounts, but also provided new insights. For example, we provide the first
systematic, detailed evaluation of C. ferox’s temporal activity patterns. This is of particular
importance to discussions concerning lemur behavior as one explanation for cathemerality in
some lemur species (e.g. Eulemur fulvus rufus, red-fronted brown lemur) is that it developed as
an anti-predator strategy, such that lemurs are temporally cryptic and thus unavailable to the
assumed „cathemeral‟ C. ferox (Colquhoun 2006). While C. ferox has previously been described
as nocturnal, crepuscular, and cathemeral in different studies (Albignac 1972; Albignac 1973;
Dollar 1999), we found them to prefer crepuscular activity (using both dawn and dusk greater
than would be expected given their availability in the diel cycle) and secondly to select night
over the daytime. However, it is notable that 23.6% of all C. ferox activity occurred during the
daytime. Whether C. ferox activity can be described as cathemeral is expressly contingent on the
definition of cathemerality (Tattersal 2006). If cathemerality is defined as uniform activity
throughout the diel-cycle, C. ferox cannot be characterized as cathemeral from our findings.
However, if cathemerality is significant activity during the day and night, then our findings
would classify C. ferox as cathemeral under this less strict definition (Tattersall 2006).
To effectively evaluate whether lemur cathemerality may be an anti-predator tactic
against the crepuscular or cathemeral C. ferox, it is necessary to understand how C. ferox
predation rates on lemurs vary throughout the entire diel-cycle, which is currently unknown. To
129
hypothesize that temporal crypticity is an effective means to reduce predation risk assumes the
predator synchronizes their activity pattern to match their prey, which is not always the case.
Predator temporal activity will often generally correspond to the time period when prey are most
vulnerable, which is a function of detectability and catchability (Zielinski 2007). Predators that
forage primarily by sight and sound use prey activity as a foraging cue, such that we would
expect predators to track their prey‟s temporal activity to maximize predation potential.
However, if a predator primarily locates and can access inactive (cryptic or resting) prey, for
example by olfaction, we would expect the predator‟s activity to be asynchronous with that of
the prey. C. ferox’s unique morphology precludes any meaningful prediction as to which of the
preceding scenarios is more likely as they possess strong olfactory, visual, and auditory acuity
with a large rhinarium similar to Viverrids, a tapetum lucidum common to many carnivores, and
large rounded ears which resemble species of Felidae (Kohncke and Leonhardt 1986).
Interestingly, cathemeral lemurs often exhibit a peak activity at the crepuscular hours
(Donati and Borgognini-Tarli 2006), which if C. ferox are cueing in on prey activity would
suggest a high predation risk for cathemeral lemurs given our findings of crepuscular activity
patterns by C. ferox. Predation risk is no doubt a strong selective force on prey behavior,
however, the relationship between lemur predation risk and C. ferox and lemur temporal activity
patterns is likely more complex than simple temporal activity avoidance through crypticity.
More extensive diet studies of C. ferox are needed to understand if cathemeral lemurs are more
or less at risk of predation by this carnivore than would be expected if diet selection were
random. We also hypothesize that if lemurs are intending to be temporally cryptic as to reduce
predation risk by C. ferox, a random activity pattern should be preferred.
130
In addition to providing new data for debates concerning carnivore-lemur interactions,
our quantification of native Malagasy carnivore activity patterns showed a lack of any major
effects of habitat and season on temporal activity patterns. Among the Malagasy carnivores,
there was a wide diversity in the diel-activity patterns (diurnal, nocturnal, and crepuscular) and
selection of dawn, day, dusk and night time periods. Temporal overlap varied considerably
between species, ranging from 5.8 to 88.8%.
Niche theory provides a meaningful framework to understand these varying degrees of
temporal overlap, as we expect a high degree of overlap in one niche component should be
associated with a low degree of overlap in one or more niche dimensions (Schoener 1974). As
similarly-sized carnivores often have high dietary overlap, it is informative to explore patterns of
temporal niche overlap and separation by comparing similar body-sized carnivores (Donadio and
Buskirk 2006; Woodward and Hildrew 2002). For example, the activity of the two small
mongoose species, G.e. elegans and G. fasciata overlapped only 7.5 to 16.1% across all
rainforests, while the two medium-sized civets F. fossana and E. goudotii activity overlapped
88.6%. In addition to body-size, natural history observation suggest the mongoose species share
a similar diet (Goodman and Benstead 2003), which may explain why these two co-occurring
rainforest carnivores have such divergent temporal activity patterns. In contrast, the two civets
have high temporal activity pattern overlap, but may be segregated by habitat and diet
specializations; E. goudottii appears to be strongly associated with wetland habitats and has a
unique dietary specialization on earthworms and insects (Albignac 1973, 1974) while F. fossana
may be more of a generalist in both habitat use and diet (Kerridge et al. 2003). The larger body
size of C. ferox may preclude it from foraging competition with the other sympatric carnivores,
131
thus explaining the high overlap in temporal activity patterns (30 to 88.8%) with the other
carnivores.
Reducing temporal activity overlap, and overlap in other niche dimensions, among
carnivores can be important in reducing competition and especially intraguild predation, which
in some populations can account for 89% of mortality and actually suppress the victim
population (Donadio and Buskirk 2006; Kamler et al. 2003; Palomares and Caro 1999). There is
currently no evidence to suggest whether Malagasy rainforest carnivores exhibit intraguild
predation. However, intraguild predation is common among other carnivore communities, with
predation highest among vertebrate predators that share some dietary overlap, and when the
larger animal is 2 to 5.4 times greater in body size than the smaller animal (Donadio and Buskirk
2006). Among Malagasy rainforest carnivores, we would thus predict a high potential for
antagonistic behavior and possibly killing between C. ferox F. fossana and F. fossana G.
fasciata. Intraguild predation of G. fasciata by F. fossana is one explanation for G. fasciata’s
general rarity in contiguous rainforest where F. fossana is ubiquitous, while at fragmented
rainforests F. fossana is absent and G. fasciata is observed frequently (this thesis, Chapter 3;
Goodman and Benstead 2003).
Malagasy carnivores showed no marked changes in their temporal activity that could be
attributed to fragmentation effects, selective logging, season, or human and exotic carnivore
activity. This is consistent with findings from other studies which have shown subtle changes in
activity pattern due to human activity, resource availability, or habitat disturbance (Ngoprasert et
al. 2007; Norris et al. 2010; Presley et al. 2009; Zielinski 1988); marked changes, such as an
animal shifting its activity from primarily diurnal to nocturnal appears rare (Griffiths and Schaik
1993; Kitchen et al. 2000). Of the native Malagasy carnivores, G. e. elegans and F. fossana both
132
showed only minor differences in their diel activity patterns across sites, maintaining their
overall respective diurnal (median range, 10:39-12:21) and nocturnal activity (median range,
23:29-23:37) patterns regardless of rainforest site. Among all species, the activity patterns of C.
familiaris and local people varied the most across sites. Local people consistently maintained a
diurnal activity pattern, but demonstrated varied patterns of activity within the day, especially in
how they selected or avoided the crepuscular hours. While this variation is likely a result of the
different distances local people lived relative to the rainforest sites, their changing activity
patterns may affect the behavior of carnivore species differently, especially for the diurnal G. e.
elegans or the crepuscular C. ferox.
Temporal activity overlap between species does not have to be large to incur costs of
competition, especially interactions between native and exotic carnivores, which have no shared
evolutionary history. For example, at the fragmented rainforests, we previously found a
increasing F. catus/silvestris trap success correlated with declining occupancy of G. e. elegans
within rainforest fragments and the surrounding matrix (this thesis, Chapter 3), despite their
moderate activity overlap (37-53%). The direct interaction between C. ferox and C. familiaris in
terms of activity overlap was less than 50%, however, this is likely enough potential contact for
disease transmission (Whiteman et al. 2007). Lastly, the overlap in activity patterns between the
introduced V. indica and the native G. fasciata increased from 56% at the Fragments <2.5 km
site to 86% at the Fragments >15 km site suggesting that the effects of fragmentation on activity
pattern overlap may vary depending on distance to contiguous forest and other undescribed
factors (i.e. number of households and people). More study is needed to determine if this
increased temporal overlap may adversely affect the ability of G. fasciata to persist in the
landscape for the long-term.
133
Our ability to explore changes in Malagasy carnivores‟ temporal activity due to habitat
alteration and the activity of local people and exotic carnivores was limited by the absence of F.
fossana, E. goudotii, and C. ferox from our fragmented forest study areas. The absence of these
carnivores begs the questions as to what degree temporal activity overlap with exotic carnivores
and local people, as well as limited plasticity in activity patterns, prohibits these carnivores from
occupying fragmented rainforests. To direct effective conservation action and encourage
carnivore persistence in a disturbed landscape, we need disentangle the dietary-spatial-temporal
constraints that are limiting carnivore populations within Madagascar‟s human-dominated
landscape.
Acknowledgments:
Funding provided by Virginia Tech, National Geographic Society Committee on
Research and Exploration, Sigma Xi Virginia Tech Master‟s Degree and Grants-in-Aid Awards,
National Science Foundation Graduate Research Fellowship Program, and Burd Sheldon
McGinnes Graduate Fellowship. We thank the Government of Madagascar, Madagascar
National Parks, the Direction des Eaux et Forêts, and CAFF/CORE for permission to conduct
this research. We were greatly assisted by ICTE/MICET, Centre ValBio, Sadabe, Mitchell
Irwin, and many remarkable research assistants. Design and analyses were greatly improved by
the advice of D. Stauffer, M. Kelly, J. Cohen, and J. Ivan.
134
References:
Aebischer, N.J., Robertson, P.A., Kenward, R.E., 1993. Compositional Analysis of Habitat Use
From Animal Radio-Tracking Data. Ecology 74, 1313-1325.
Albignac, R., 1972. The Carnivora of Madagascar, In Biogeography and ecology in Madagascar.
eds R. Battistini, G. Richard-Vindard. Dr. W. Junk B.V. Publishers, The Hague.
Albignac, R., 1973. Mammifêres Carnivores. Faune de Madagascar 36. ORSTOM/CNRS, Paris.
Albignac, R., 1974. Observations Eco-ethologiques sur le Genre Eupleres, viverride de
Madagascar. Terre Vie 28, 321-351.
Aschoff, J., Daan, S., Groos, G., 1982. Vertebrate circadian systems. Springer, Berlin.
Batschelet, E., 1981. Circular Statistics in Biology. Academic Press, New York.
Beckmann, J.P., Berger, J., 2003. Rapid ecological and behavioural changes in carnivores: the
responses of black bears (Ursus americanus) to altered food. Journal of Zoology 261,
207-212.
Byers, C., Steinhorst, R., Krausman, P., 1984. Clarification of a technique for analysis of
utilization-availability data. Journal of Wildlife Management 48, 1050-1053.
Cherry, S., 1996. A Comparison of Confidence Interval Methods for Habitat Use-Availability
Studies. Journal of Wildlife Management 60, 653-658
Colquhoun, I.C., 2006. Predation and cathemerality - Comparing the impact of predators on the
activity patterns of lemurids and ceboids. Folia primatologica, 143-165.
Curtis, D., Donati, G., Rasmussen, M., 2006. Cathemerality. Folia primatologica 77, 1-194.
Dollar, L., 1999. Preliminary report on the status, activity Cycle, and ranging of Cryptoprocta
ferox in the Malagasy rainforest, with Implications for conservation. Small Carnivore
Conservation 20.
135
Donadio, E., Buskirk, Steven W., 2006. Diet, Morphology, and Interspecific Killing in
Carnivora. The American Naturalist 167, 524-536.
Donati, G., Borgognini-Tarli, S.M., 2006. From darkness to daylight: cathemeral activity in
primates. Journal of Anthropological Sciences 84, 7-32.
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.
Fisher, N.I., 1993. Statistical analysis of circular data. Cambridge University Press, Cambridge.
Gerber, B., Karpanty, S.M., Crawford, C., Kotschwar, M., Randrianantenaina, J., 2010. An
assessment of carnivore relative abundance and density in the eastern rainforests of
Madagascar using remotely-triggered camera traps. Oryx 44, 219-222.
Gittleman, J.L., 1986. Carnivore brain size, behavioral ecology, and phylogeny. Journal of
Mammalogy 67, 23-36.
Goodman, S.M., Benstead, J.P., 2003. Natural History of Madagascar. The Univeristy of
Chicago Press, Chicago.
Goodman, S.M., Helgen, K.M., 2010. Species limits and distribution of the Malagasy carnivoran
genus Eupleres (Family Eupleridae). Mammalia, DOI 10.1515/MAMM.2010.1018.
Griffiths, M., Schaik, C.P.v., 1993. The impact of human traffic on the abundance and activity
periods of Sumatran rain forest wildlife. Conservation Biology 7, 623-626.
Halle, S., Stenseth, N.C., 2000. Activity patterns in small mammals. An ecological approach.
Springer, New York.
Hawkins, C.E., 1998. Behaviour and ecology of the fossa, Cryptoprocta ferox (Carnivora,
Viverridae) in a dry deciduous forest, western Madagascar. University of Aberdeen,
Aberdeen.
136
Hayward, M.W., Slotow, R., 2009. Temporal Partitioning of Activity in Large African
Carnivores: Tests of Multiple Hypotheses. South African Journal of Wildlife Research
39, 109-125.
Hill, R.A., Barrett, L., Gaynor, D., Weingrill, T., Dixon, P., Payne, H., Henzi, S.P., 2003. Day
length, latitude and behavioural (in)flexibility in baboons (Papio cynocephalus ursinus).
Behavioral Ecology and Sociobiology 53, 278-286.
Hunter, J., Caro, T., 2008. Interspecific competition and predation in American carnivore
families. Ethology Ecology & Evolution 20, 295 - 324.
Kamler, J.F., Ballard, W.B., Gilliland, R.L., II, P.R.L., Mote, K., 2003. Impacts of Coyotes on
Swift Foxes in Northwestern Texas. Journal of Wildlife Management 67, 317-323.
Karpanty, S., Wright, P.C., 2007. Predation on lemurs in the rainforest of madagascar by
multiplenpredator species: Observations and experiments, In Primate Anti-Predator
Strategies. eds S.L. Gursky, K.A.I. Nekaris, pp. 77-99. University of Chicago Press,
Chicago.
Kavanau, J.L., Ramos, J., 1975. Influences of Light on Activity and Phasing of Carnivores. The
American Naturalist 109, 391-418.
Kerridge, F.J., Ralisoamalala, R.C., Goodman, S.M., Pasnick, S.D., 2003. Fossa fossana,
Malagasy Striped Civet, Fanaloka, In Natural History of Madagascar. eds S.M.
Goodman, J.P. Benstead, pp. 1363-1365. The University of Chicago Press, Chicago.
Kitchen, A.M., Gese, E.M., Schauster, E.R., 2000. Changes in coyote activity patterns due to
reduced exposure to human persecution. Canadian Journal of Zoology 78, 853-857.
Kohncke, M., Leonhardt, K., 1986. Cryptoprocta ferox. Mammalian Species 254, 1-5.
137
Kolowski, J.M., Katan, D., Theis, K.R., Holekamp, K.E., 2007. Daily patterns of activity in the
spotted hyena. Journal of Mammalogy 88, 1017-1028.
Kronfeld-Schor, N., Dayan, T., 2003. Partitioning of time as an ecological resource. Annual
Review of Ecology, Evolution, and Systematics 34, 153-181.
Lourens, S., Nel, J., 1990. Winter activity of bat-eared foxes Otocyon megalotis in the Cape
West Coast. South African Journal of Zoology 25, 124-132.
Mardia, K.V., Jupp, P.E., 2000. Statistics of directional data, 2nd Edition edn. John Wiley &
Sons, Chicester.
Mardia, K.V., Kent, J.T., Bibby, J.M., 1979. Multivariate Analysis. Academic Press., New York.
Ngoprasert, D., Lynam, A.J., Gale, G.A., 2007. Human disturbance affects habitat use and
behaviour of Asiatic leopard Panthera pardus in Kaeng Krachan National Park, Thailand.
Oryx 41, 343-351.
Norris, D., Michalski, F., Peres, C.A., 2010. Habitat patch size modulates terrestrial mammal
activity patterns in Amazonian forest fragments. Journal of Mammalogy 91, 551-560.
O'Brien, T., Kinnaird, M., Wibisono, H., 2003. Crouching tigers, hidden prey: Sumatran tiger
and prey populations in a tropical forest landscape. Animal Conservation 6, 131-139.
Palomares, F., Caro, T.M., 1999. Interspecific Killing Among Mammalian Carnivores. The
American Naturalist 153, 492-508.
Pianka, E.R., 1973. The Structure of Lizard Communities. Annual Review of Ecology and
Systematics 4, 53-74.
Presley, S.J., Willig, M.R., Castro-Arellano, I., Weaver, S.C., 2009. Effects of habitat conversion
on temporal activity patterns of Phyllostomid bats in lowland Amazonian rain forest.
Journal of Mammalogy 90, 210-221.
138
R Development Core Team, 2010. R: A Language and Environment for Statistical Computing. R
Foundation for Statistical Computing, Vienna.
Ridout, M.S., Linkie, M., 2009. Estimating overlap of daily activity patterns from camera trap
data. Journal of Agricultural, Biological, and Environmental Statistics 14, 322-337.
Roll, U., Kronfeld-Schor, T.D.N., 2006. On the role of phylogeny in determining activity
patterns of rodents. Evolutionary Ecology 20, 479-490.
Romero-Muñoz, A., Maffei, L., Cuéllar, E., Noss, A.J., 2010. Temporal separation between
jaguar and puma in the dry forests of southern Bolivia. Journal of Tropical Ecology 26,
303-311.
SAS, Institute Inc. SAS 9.2 Cary, NC.
Schaik, C.P.v., Griffiths, M., 1996. Activity Periods of Indonesian Rain Forest Mammals.
Biotropica 28, 1996.
Schlexer, F.V., 2008. Attracting Animals to Detection Devices, In Noninvasive Survey Methods
for Carnivores. eds R.A. Long, P. MacKay, W.J. Zielinski, J.C. Ray, pp. 263-326. Island
Press, Washington.
Schoener, T.W., 1974. Resource partitioning in ecological communities. Science 185, 27-39.
Sutherland, W.J., Dolman, P.M., 1994. Combining Behaviour and Population Dynamics with
Applications for Predicting Consequences of Habitat Loss. Proceedings: Biological
Sciences 255, 133-138.
Tattersall, I., 2006. The Concept of Cathemerality: History and Definition. Folia primatologica
77, 7-14.
Tecot, S., 2008. Seasonality and predictability: The hormonal and behavioral responses of the
139
Red-bellied Lemur, Eulemur rubriventer, in southeastern Madagascar. The University of
Texas, Austin.
Whiteman, C.W., Matushima, E.R., Cavalcanti Confalonieri, U.E., Palha, M.d.D.C., da Silva,
A.d.S.L., Monteiro, V.C., 2007. Human and domestic animal populations as a potential
threat to wild carnivore conservation in a fragmented landscape from the Eastern
Brazilian Amazon. Biological Conservation 138, 290-296.
Woodward, G., Hildrew, A.G., 2002. Body-size determinants of niche overlap and intraguild
predation within a complex food web. Journal of Animal Ecology 71, 1063-1074.
Wright, P.C., 1997. The Future of Biodiversity in Madagascar, In Natural Change and Human
Impact in Madagascar. eds S.M. Goodman, B.D. Patterson, pp. 381-405. Smithsonian
Institution Scholarly Press, Washington D.C.
Yoder, A.D., Burns, M.M., Zehr, S., Delefosse, T., Veron, G., Goodman, S.M., Flynn, J.J., 2003.
Single origin of Malagasy Carnivora from an African ancestor. Nature 421, 734-737.
Zar, J.H., 1998. Biostatistical Analysis. Prentice Hall, Upper Saddle River.
Zielinski, W.J., 1988. The influence of daily variation in foraging cost on the activity of small
carnivores. Animal Behaviour 36, 239-249.
Zielinski, W.J., 2007. Weasels and Martens - Carnivores in Northern Latitudes, In Activity
patterns in small mammals: an ecological approach. eds S. Halle, N.C. Stenseth, N.C.
Stenseth, pp. 191-217. Springer-Verlag, New York.
Zielinski, W.J., Spencer, W.D., Barrett, R.H., 1983. Relationship between Food Habits and
Activity Patterns of Pine Martens. Journal of Mammalogy 64, 387-396
Zub, K., Szafrańska, P.A., Konarzewski, M., Redman, P., Speakman, J.R., 2009. Trade-offs
140
between activity and thermoregulation in a small carnivore, the least weasel Mustela
nivalis. Proceedings of the Royal Society B: Biological Sciences 276, 1921-1927.
141
Table 1. The number of photographic-activity samples for Malagasy carnivores, exotic carnivores, and local people from four
rainforest sites, sampled in two seasons within Madagascar‟s eastern rainforests. Also, a non-parametric circular test of differences
among activity distributions (continuous time) across all sites and a pair-wise analysis when the overall test indicated a significant
difference; sites were not included when sample sizes were < 10.
Species a
Primary /
Selectively-
Logged /
Fragments
<2.5 km /
Fragments
>15 km /
Datasetb
Global
Mardia-Watson-Wheeler Test
Cold-Dry
Cold-Dry
Warm-Dry
Warm-Dry
W
P
Canis familiaris
7
17 A
169 B
97 B
Site
18.98
< 0.001
Cryptoprocta ferox
91
51
2
0
Sex / Site
1.84 / 0.69
0.40 / 0.71
Eupleres goudotii
2
16
0
0
Site
NA
NA
Fossa fossana
780 A
460 B
0
0
Sex / Site
3.64 / 27.76
0.80 / 0.002
Felis catus/silvestris
0
0
23
16
Site
4.53
0.10
Galidia elegans elegans
1554 AB
427 A
28 AC
185 BC
Site
10.03
0.04
Galidictis fasciata
52
10
10
27
Site
1.23
0.98
Viverricula indica
0
0
22 A
49 B
Site
7.64
0.02
Local People
75 AB
149 A
434 C
58 BC
Site
52.42
< 0.001
142
aAn activity sample was the median time of all photographs of the same species or individual (C. ferox and F. fossana) detected at a
camera station within a 0.5 hour period. The same letter within a species indicates the activity distributions are not statistically
different, experiment-wise P <0.1.
bSpecies activity distributions were tested for differences across sites or sites and sexes. C. ferox and F. fossana were the only species
for which sexes could be identified
143
Table 2. Comparison of activity distributions of Malagasy and exotic carnivores before and after
the use of scent-lure to detect carnivores at photographic sampling stations. Sampling occurred
within Madagascar‟s eastern rainforest from May-August, 2008 and October-December, 2009 at
a Selectively-logged and fragmented rainforest <2.5 km from intact primary rainforest,
respectively.
Species
Rainforest Site
Mardia-Watson-
Wheeler Test
Temporal Overlap
(% ± SE)a
Canis familiaris
Fragments <2.5 km
W = 2.68, P = 0.26
82.47 ± 0.06
Cryptoprocta ferox
Selectively-Logged
W = 1.52, P = 0.47
83.43 ± 0.09
Fossa fossana
Selectively-Logged
W = 0.38, P = 0.83
95.51 ± 0.02
Felis catus/silvestris
Fragments <2.5 km
W = 1.35, P = 0.51
72.08 ± 0.11
Galidia elegans elegans
Selectively-Logged
W = 3.63, P = 0.16
81.75 ± 0.06
Galidia elegans elegans
Fragments <2.5 km
W = 1.54, P = 0.46
84.62 ± 0.12
Viverricula indica
Fragments <2.5 km
W = 3.34, P = 0.19
71.72 ± 0.15
a Temporal overlap of activity data collected with and without using lure by applying a kernel
density analyses
144
Table 3. Test of uniform activity distributions of Malagasy carnivores, exotic carnivores, and
local people throughout the diel-cycle. Photographic-activity samples were collected at four
rainforest sites within Madagascar‟s eastern rainforests, sampled from May-December, 2008 and
October-December, 2009; data were pooled if activity distributions among sites were not
statistically different.
Species
Rainforest Sitesa
Rao‟s Test of Uniformity
Median Direction of
Activity ± SEb
Canis familiaris
SL
U = 227.221, P < 0.01
08:30 ± 00:48
Fragments <2.5 km
U = 153.974, P < 0.01
16:18 ± 01:01
Fragments >15 km
U = 144.026, P < 0.10
18:37 ± 02:09
Cryptoprocta ferox
Primary + SL
U = 145.250, P < 0.05
23:34 ± 00:51
Eupleres goudotii
Primary + SL
U = 175.750, P < 0.01
01:41 ± 01:04
Felis catus/silvestris
Fragments <2.5 + >15 km
U = 148.750, P > 0.10
NA
Fossa fossana
Primary
U = 202.418, P < 0.01
23:37 ± 00:07
SL
U = 203.856, P < 0.01
23:29 ± 00:11
Galidia elegans
Primary
U = 208.263, P < 0.01
11:19 ± 00:05
elegans
SL
U = 214.189, P < 0.01
12:21 ± 00:09
Fragments <2.5 km
U = 202.929, P < 0.01
10:47 ± 00:35
Fragments >15 km
U = 200.568, P < 0.01
10:39 ± 00:17
Galidictis fasciata
Primary + SL + Fragments
<2.5 + >15 km
U = 210.091, P < 0.01
23:31 ± 00:18
Viverricula indica
Fragments <2.5 km
U = 197.795, P < 0.01
01:40 ± 00:46
Fragments >15 km
U = 206.515, P < 0.01
23:18 ± 00:30
145
Local people
Primary
U = 221.700, P < 0.01
13:06 ± 00:22
SL
U = 213.678, P < 0.01
11:40 ± 00:13
Fragments <2.5 km
U = 194.220, P < 0.01
11:17 ± 00:12
Fragments >15 km
U = 194.031, P < 0.01
10:03 ± 00:32
aPrimary: little anthropogenic disturbance, SL: selectively-logged prior to 1991, Fragments <2.5
km and Fragments >15 km: fragmented rainforest and their distance from intact primary
rainforest
bThe median time, or central location, in the 24-hr period of all observations.
146
Table 4. Percentage of temporal overlap (% ± SE) across four rainforest sites within Madagascar‟s eastern rainforests by species pairs
using kernel density analyses. Photographic activity-sampling occurred from May-December, 2008 and October-December, 2009. A)
Primary (top triangle) and Selectively-logged (bottom triangle) rainforest sites, B) Fragments <2.5 km from intact forest (top triangle)
and Fragments >15 km from intact forest (bottom triangle).
A)
Species temporal overlap at the Primary rainforest site
Species temporal overlap at the Selectively-
logged rainforest site t
C.ferox
X
71.17 ±
3.97
32.67 ±
4.06
64.84 ±
5.04
LSS
NA
LSS
NA
28.72 ± 4.01
F. fossana
70.62 ±
5.31
X
5.80 ±
0.55
88.82 ±
3.97
LSS
NA
LSS
NA
5.39 ± 1.93
G. e. elegans
30.07 ±
5.05
8.21 ±
0.85
X
7.47 ±
0.63
LSS
NA
LSS
NA
74.86 ± 4.72
G. fasciata
63.21 ±
2.24
81.95
±15.92
13.71 ±
3.07
X
LSS
NA
LSS
NA
5.83 ± 1.58
E. goudotii
77.03 ±
10.21
88.61 ±
9.72
22.00 ±
5.62
78.34 ±
15.24
X
LSS
LSS
LSS
LSS
V. indica
NA
NA
NA
NA
NA
X
NA
NA
NA
C. familiaris
43.18 ±
7.14
20.20 ±
5.01
49.84 ±
7.55
18.73 ±
7.25
25.25 ±
8.01
NA
X
NA
LSS
F.
catus/silvestris
NA
NA
NA
NA
NA
NA
NA
X
NA
Local People
30.00 ±
5.00
8.77 ±
1.34
88.17 ±
3.07
4.19 ±
1.16
18.78 ±
4.07
NA
52.21 ± 8.46
NA
X
aLSS: low sample size <10 activity samples, NA: species did not co-occur
147
B)
Species temporal overlap at the Fragments <2.5 km rainforest site
Cryptoprocta
ferox
Fossa
fossana
Galidia
elegans
elegans
Galidictis
fasciata
Eupleres
goudotii
Viverricula
indica
Canis
familiaris
Felis catus/
silvestris
Local
People
Species temporal overlap at the Fragments
>15 km rainforest site
C.ferox
X
NA
NA
NA
NA
NA
NA
NA
NA
F. fossana
NA
X
NA
NA
NA
NA
NA
NA
NA
G. e. elegans
NA
NA
X
13.72 ±
3.59
NA
10.81 ±
2.75
54.49 ±
5.44
52.79 ± 9.7
77.78 ±
6.77
G. fasciata
NA
NA
16.07 ±
3.2
X
NA
56.00 ±
14.67
37.80 ±
8.79
55.33 ± 9.52
5.19 ±
1.11
E. goudotii
NA
NA
NA
NA
X
NA
NA
NA
NA
V. indica
NA
NA
13.12 ±
1.82
85.70 ±
7.95
NA
X
36.76 ±
7.30
45.12 ± 9.93
8.86 ±
1.56
C. familiaris
NA
NA
55.14 ±
4.18
51.79 ±
5.69
NA
54.28 ±
5.21
X
75.38 ± 7.31
60.22 ±
2.87
F.catus/
silvestris
NA
NA
37.19 ±
7.25
60.00 ±
11.37
NA
71.12 ±
11.06
70.88 ±
6.66
X
52.51 ±
8.06
Local People
NA
NA
83.69 ±
6.16
13.63 ±
3.37
NA
13.27 ±
3.24
50.36 ±
5.22
36.45 ± 7.95
X
aLSS: low sample size <10 activity samples, NA: species did not co-occur
148
Table 5. Chi-square test results for differences between activity periods (Dawn, Day, Dusk,
Night) of Malagasy carnivores and local people across four rainforest sites and sexes.
Photographic activity-sampling occurred within Madagascar‟s eastern rainforests from May-
December, 2008 and October-December, 2009.
Species
Datasetsa
X2
df
Pb
Canis familiaris
Site (SL, Frag <2.5 km, Frag >15 km)
27.32
6
<0.001
Cryptoprocta ferox
Site (Primary and SL) x Sex (Male-Female)
2.54
3
0.47
Eupleres goudotii
Site (Primary, SL)
NA
NA
NA
Felis catus/silvestris
Site (Frag <2.5 km, Frag >15 km)
1.70
3
0.42
Fossa fossana
Site (Primary and SL) x Sex (Male-Female)
0.39
3
0.94
Galidia elegans elegans
Site (Primary, SL, Frag <2.5 km, Frag >15 km)
35.20
9
<0.001
Galidictis fasciata
Site (Primary, SL, Frag <2.5 km, Frag >15 km)c
5.42
6
0.49
Local People
Site (Primary, SL, Frag <2.5 km, Frag >15 km)
26.47
9
0.02
Viverricula indica
Site (Frag <2.5 km, Frag >15 km)c
0.08
3
0.96
aPrimary: little anthropogenic disturbance, SL: selectively-logged prior to 1991, Frag <2.5 km
and Frag >15 km: fragmented rainforest and their distance from intact primary rainforest. C.
ferox and F. fossana were the only species the sexes were identifiable, thus activity differences
were evaluated across sites and sexes
cRemoved day category due to no observations
149
Table 6. Compositional analyses of three Malagasy carnivores sampled across four rainforest sites, sample size, test of random use of
dawn, day, night, and dusk, and the ranking of selection in order of relative preference. Photographic activity-sampling occurred
within Madagascar‟s eastern rainforests from May-December, 2008 and October-December, 2009.
Speciesa
Nb
Wilk's λ
Randomized P
Ranking of Temporal Activity Classesc
Cryptoprocta ferox
6
0.0032
0.0314
Dusk>Dawn>>>Night>>>Day
Fossa fossana
31
0.0422
0.0001
Night>>>Dusk>Dawn>>>Day
Galidia e. elegans (SL)
17
0.0194
0.0010
Day>>>Dusk>Dawn>>>Night
Galidia e. elegans (Primary + Frag <2.5 + >15 km)
34
0.0428
0.0010
Day>>>Dawn>>>Dusk>>>Night
a Galidia elegans elegans Primary: little anthropogenic disturbance, SL: selectively-logged prior to 1991, Frag <2.5 km and Frag >15
km: fragmented rainforest sites and their distance from intact primary rainforest
bNumber of individuals, except G. e. elegans where camera stations were used as the sampling unit
cRelative preference of time classes, ranked from the most preferred to the least, > indicates a non-significant difference between two
classes while >>> indicates a significant difference (P < 0.1)
150
Figure 1. Carnivores were sampled at four photographic-sampling sites within Madagascar‟s
eastern rainforests from May-December, 2008 and October-December, 2009. Top insert map
shows political provinces and location on the island of Madagascar, while the bottom insert map
shows the camera station sampling layout among rainforest fragments at Mahatsinjo, Tsinjoarivo
as an example. Rainforest extent from Conservation International, 2010.
151
152
153
Figure 2. Temporal activity of Malagasy rainforest carnivores, exotic carnivores, and local people. Species activity samples were
collected at four rainforest sites within Madagascar‟s eastern forests; data were pooled across sites when activity distributions were not
different. Solid ( ) lines indicate the use of all data from all relevant sites for each species, circles (○) indicate data from the Primary
site, dotted (…) lines indicate data from the Selectively-logged site, dashed (---) lines indicate data from Fragments <2.5 km site, and
plus signs (+) indicate data from the Fragments >15 km site. Tick marks on the x-axis represent all activity samples for each species.
Y-axis is not consistent among all graphs.
154
Figure 3. Proportional activity of Malagasy carnivores used for compositional analyses and average availability of the temporal
classes. Species activity samples were collected at four rainforest sites within Madagascar‟s eastern forests, which were pooled when
the proportional use of activity time classes across sites were not different. Photographic activity-sampling occurred from May-
December, 2008 and October-December, 200
0%
20%
40%
60%
80%
100%
C. ferox
F. fossana
G. e. elegans
G. e. elegans
Proportion of Temporal Use/Available
Day
Dawn
Dusk
Night
(All Data) (All Data) (Selectively-logged) (Primary + Fragments
<2.5 and >15 km)
Average Availability
155
a) Canis familiaris
(Fragments <2.5 km)
156
157
Figure 4. Selection and/or avoidance of dawn, day, dusk, and night by Galidictis fasciata, exotic carnivores, and local people. Species
activity samples were sampled at four rainforest sites across Madagascar, which were pooled when the activity proportions of time
classes were not different across sites. The bars indicate the available time for each time class, while the circular-points indicate the
proportion of activity-use and confidence interval in each time class; dark grey bars indicate the time class was selected (lower
confidence limit of activity-use is above availability) or avoided (upper confidence limit of activity-use is below availability), while
white bars indicate use was in proportion to availability (confidence limits include availability). When there were no observations in a
temporal class and the availability was large, we assumed the species avoided this time period.
158
Appendix A: Native/Exotic Carnivore Species Accounts for Southeastern Madagascar
Native Carnivores
Fossa (Cryptoprocta ferox):
Figure 1. Rebioma (http://www.rebioma.net) fossa (Cryptoprocta ferox) extent of occurrence and
area of occupancy.
159
C. ferox is Madagascar‟s largest extant, endemic terrestrial species. In one study at
Kirindy forest (dry forest), C. ferox averaged a total length of 1.4 m and adult males weighed 7.4
± 1.2 kg while adult females weighed 6.1 ± 0.6 kg (Hawkins and Racey 2005). In separate study
from the dry forests of Ankarafantsika and Kirindy Mitea National Parks, adult males weighed
7.44 ± 1.97 kg and adult females 6.38 ± 1.05 kg (Dollar 2006). A single adult male captured in
the eastern rainforest weighed 8.1 kg (Dollar 1999). C. ferox have semi-retractable claws,
reversible ankles, and a tail length equal to body length. Additionally, C. ferox use both
plantigrade and digitigrade locomotion (Laborde 1986; Laborde 1986; Laborde 1987). Juvenile
female C. ferox between the age of one and two have been found to have temporary
masculinization, where they develop an enlarged, spinescent clitoris, supported by an os
clitoridis, and a pigmented secretion on the ventral fur that in adults only occurs with males.
This may serve to reduce harassment by male or female conspecifics (Hawkins et al. 2002).
C. ferox’s current geographic extent covers most forest types on Madagascar, including
areas above the tree line (Goodman et al. 1997; Hawkins 2003). In the western dry forests, C.
ferox has a mean homerange of 14.8 km2 and a density of 0.24 individuals/km2 and 0.18
adults/km2 (Hawkins and Racey 2005). In comparison, the pilot study to this thesis estimated C.
ferox rainforest at 0.17 ± 0.06 individuals/km2 (Gerber et al. 2010). The temporal activity
pattern of C. ferox has been generally characterized in natural history accounts as crepuscular
and nocturnal (Albignac 1972; Albignac 1975). Two males tracked via radiotelemetry in the
eastern rainforest showed limited activity during the daylight hours, high activity at night, and a
peak at dawn and pre-dusk (Dollar 1999). Temporal activity analyses of C. ferox in the dry
forests using radiotelemetry showed no seasonal differences in activity patterns and some
activity during the daylight hours, but most activity was concentrated at night (Hawkins 1998).
160
Whether or not the diet of C. ferox varies across forest types is still unknown. However,
they have been recorded to prey on diverse species, including rodents, lipotyphlans, lemurs,
birds, amphibians, reptiles, and insects (Dollar et al. 2007; Hawkins and Racey 2008). In the dry
forests, greater than 50% of C. ferox’s diet can be made up of lemurs (Hawkins and Racey 2008).
One study in the montane forests at Montagne d‟Ambre National Park found that C. ferox diet
was made up of 40% lemur biomass, however, when considering numbers of individuals, rodents
made up the majority of consumed prey items (Rasolonandrasana 1994). Lemur species known
to be preyed upon by C. ferox include a large range of small (Microcebus sp.) to large bodied
lemurs (Propithecus sp.; Goodman et al. 1997; Goodman 2003c; Karpanty and Wright 2007).
While C. ferox predation pressure on lemurs is likely to vary across forest types (Goodman
2003c), the impact can be considerable, including documented cases in which fossa predation
resulted in the breakup of social groups or extirpation of lemur species from both contiguous and
fragmented forests (Irwin et al. 2009). In one study, the frequency of prey consumed by C. ferox
was correlated with the relative abundance of prey, suggesting C. ferox are opportunistic
predators (Hawkins and Racey 2008). This flexible diet is likely what allows C. ferox to have
such a broad distribution among vegetation types in Madagascar (Goodman 2003).
161
Malagasy Small-toothed Civet (Eupleres goudotii):
Figure 2. Rebioma (http://www.rebioma.net) Small-toothed civet (Eupleres goudotii) extent of
occurrence and area of occupancy.
162
Currently, there are two species recognized as belonging to the genus Eupleres: E.
goudotii and E. major (Goodman and Helgen 2010), which were previously identified as
subspecies of E. goudotii (Albignac 1973). E.goudotii occupies the entire length of the eastern
rainforest, including occupation of a mixture of spiny and gallery forest in the southeast
(Goodman and Helgen 2010). In contrast, E. major is found in western Madagascar in the
northwest Sambirano region and as far south as Baie de Baly National Park (Hawkins 1994;
Goodman and Helgen 2010). E. goudotii is Madagascar‟s second largest carnivore, ranging in
total body length from 455-650 mm and weighing 1.5-4.6 kg; the subspecies E. major is
generally larger (Albignac 1973; Hawkins 1994). Prior to the cold-dry season (June-August)
during which food is less plentiful, E. goudotii are known to store additional fat, especially in the
tail. This extra weight has been thought to be used during the cold months for hibernation
(Albignac 1973), however, this species has been observed to be active during this period (Dollar
1999). This species morphology is unique to Malagasy carnivores, having an elongated snout
with small, fine, conical shaped dentition (Albignac 1972; Dollar 1999).
Almost no quantitative information is available regarding the ecology of E. goudotii.
Their diet is poorly known, but they are believed to consume large quantities of earthworms,
insects and some lizards, rodents, and birds. E. goudotii has been observed using its large non-
retractable claws on their forepaws to excavate insects as well as dig burrows used for shelter
(Albignac 1972; Albignac 1984). Their activity pattern is assumed to be nocturnal and/or
crepuscular based on observations of captive individuals (Albignac 1972; Albignac 1974),
however, they have been observed in the wild during daylight hours (Dollar 1999).
163
Malagasy Civet (Fossa fossana):
Figure 3. Rebioma (http://www.rebioma.net) Malagasy civet (Fossa fossana) extent of
occurrence and area of occupancy.
F. fossana is Madagascar‟s third largest endemic carnivore, averaging a total length of
913 mm and weighing between 1.3-2.1 kg. There is no sexual dimorphism between sexes,
164
possibly due to their suspected pattern of monogamous pairing. Similar to E. goudotii, F.
fossana is also known to store fat in its tail prior to the cold-dry months (Kerridge, Ralisoamalala
et al. 2003). F. fossana are found in lowland and montane eastern rainforest and in the northwest
Sambriano region. F. fossana are believed not to occur in secondary or disturbed habitats
(Kerridge, Ralisoamalala et al. 2003), however, no data prior to this current study existed to
support this conclusion. Currently, the only density estimate is from the pilot study to this thesis,
which estimated F. fossana density in the rainforest at 2.23 ± 0.36 individuals/km2 (Gerber et al.
2010). Kerridge et al. (2003) used trapping and subsequent radiotelemetry data and estimated
their homerange as between 0.073 and 0.552 km2. Although no density estimate was calculated
in that study, Kerridge et al. (2003) remarked that 22 individuals were trapped in 2 km2 of
rainforest habitat (Kerridge et al. 2003).
F. fossana is considered nocturnal, however, no actual systematically-collected data to
support this conclusion are available. From spool and line tracking and trapping efforts, F.
fossana are thought to prefer wet areas, including streams and marshes (Kerridge et al. 2003).
Presumed to be strictly terrestrial based on morphology, F. fossana are known to exhibit arboreal
activity, however, it is unknown how often they exhibit this behavior (Albignac 1971; Goodman,
Kerridge et al. 2003). F. fossana diet is diverse, including rodents, lipotyphlans, crustaceans,
snakes, frogs, lizards, and many insect taxa (Rand 1935; Albignac 1971; Albignac 1973;
Kerridge, Ralisoamalala et al. 2003). A single diet analysis from the eastern rainforest
(Ranomafana National Park and Vevembe) indicated a high consumption of Arthropoda (60%),
but also found evidence of numerous vertebrate prey (33%), including small-mammals, reptiles,
and amphibians (Goodman et al. 2003). Lemur depredation has not been officially reported for
F. fossana, but given its ability to climb trees, there is a potential for it to occur.
165
Broad-striped Mongoose (Galidictis fasciata):
Figure 4. Broad-striped mongoose (Galidictis fasciata) extent of occurrence based on forest
cover (2005).
166
There are two species recognized as Galidictis: G. fasciata occurs broadly throughout the
eastern rainforests and G. grandidieri occurs in the southern spiny forests. G. fasciata averages
857 ± 50.4 mm in total length and weighs 640 ± 105.4 g. In contrast, G. grandidieri average
total length is 393.3 ± 11.9 mm and weighs 1207.5 ± 195.2 g. Both species are thought to be
nocturnal, but no systematically collected temporal activity data are known. G. fasciata is
thought to be broadly distributed in low densities from lowland forests up to 1500 m elevation.
There is currently no information on Galdictis sp. diet, but is believed to eat rodents, reptiles, and
small amphibians (Goodman 2003).
167
Ring-tailed Mongoose (Galidia elegans):
Figure 5. Ring-tailed mongoose (Galidia elegans elegans) extent of occurrence based on forest
cover (2005).
G. elegans is the most conspicuous of the Malagasy carnivores due to its diurnal activity
pattern and boldness for stealing chickens from villages and raiding research camps (Goodman
168
2003a). There are currently three recognized subspecies of G. elegans- G.e. elegans which
occupy the entirety of the eastern rainforests, G. e. dambrensis which are isolated in the
Sambirano region, and G. e. occidentalis which occur in the northern-western dry forests
(Bennett et al. 2009). Within each region, G. elegans is thought to be fairly ubiquitous from sea
level to tree line (Albignac 1984; Goodman and Pidgeon 1999; Goodman 2003). There have
been occasional reports of G. elegans in the southern spiny forest (Goodman and Pidgeon 1999),
including fossil remains (Muldoon et al. 2009), suggesting occupation, however, there is no
verification of current and prolonged occurrence of the spiny forest.
G. e. elegans total length ranges from 560 to 670 mm and weighs from 760 to 1085 g
(Dunham 1998). G. elegans morphology includes non-retractable claws which allow arboreal
activity and membranes between the bases of all toes which assists in semi-aquatic activity while
hunting. There are no absolute density estimates of G. elegans, but observations indicate they
occur in high densities with home ranges of 20-25 ha (Goodman 2003). Dunham (1998)
captured 12 individuals with 1,040 trap nights and estimated a density index of 37
individuals/km2. Radiotelemetry locations indicated that 11 animals shared approximately 20 ha.
Unlike other Malagasy carnivores, G. elegans is known to live in family groups that often
consist of a single adult male and female, their most recent offspring, and occasionally an older
juvenile offspring. G. elegans’ diet has not been well studied, but is known to include numerous
insect taxa, rodents, lipotyphlans, crayfish, lizards, snakes, and amphibians. In addition, G.
elegans is a known lemur predator, having been observed predating upon Cheirogaleus major
and Microcebus rufus (Goodman 2003c).
169
Exotic Carnivores:
Domestic Dog (Canis familiaris):
Figure 6. Domestic Dog (Canis familiaris)
C. familiaris are common throughout Madagascar‟s villages and are viewed by local
people as assisting in pest-control and security (Dollar 2006). It is common to find C. familiaris
moving outside villages through native vegetation, including intact rainforest (Gerber et al.
2010), which may have significant consequences to native wildlife. C. familiaris are expected to
affect medium and small carnivores through interference competition, as well as serve as disease
vectors (Vanak and Gompper 2009). At Ankarafantsika C. familiaris has been implicated in
transmitting several infectious diseases, including Canine Distemper and Canine Parvovirus, to
C. ferox (Dollar 2006). In addition, there is increasing evidence that C. familiaris may be an
important predator of large diurnal lemurs. They have been observed harassing, attacking, and
attempting to kill Propithecus verreauxi at Beza Mahafaly Special Reserve. P. verreauxi respond
to C. familiaris with typical ground-predator anti-predator behaviors, such as moving up in the
canopy, visually scanning down, and alarm calling (Broackman et al. 2008).
170
The roles of C. familiaris as a predator of native wildlife, potential competitor with
Malagasy carnivores, and disease vector are still poorly understood. The ubiquity of C.
familiaris throughout Madagascar‟s villages as well as their activity within and around natural
areas demands increasing research attention and creative solutions to ameliorate potentially
harmful impacts of these domestic animals.
171
Wild/Feral Cat (Felis sp.):
Figure 7. Wild/Feral Cat (Felis sp.)
Based on general pelage patterns, Madagascar may contain two exotic species of the
genus Felis: Felis catus and Felis silvestris. F. catus is known as the domesticated housecat,
while F. silvestris is a wild cat that may have been introduced from mainland Africa. Genetic
analyses of domesticated, feral, and wild cats are needed to better understand the phylogeny of
this/these exotic carnivore(s). Felis sp. occurs throughout Madagascar within and near villages,
as well as in native- and non-native vegetated habitats away from villages. Based on
morphological and behavioral characteristics, local people and research often do make a
distinction between wild cats (F. silvestris?) which occur outside of villages and domestic or
feral (F. catus?) that may be more closely associated with villages (Dollar 2006; Brockman et al.
2008; Kotschwar, 2010). The wildcat may be a significant lemur predator, having been
implicated in the predation of Propithecus verreaxi at Beza Mahafaly Special Reserve. P.
verreaxi respond to F. silvestris typical ground-predator anti-predator behaviors, including
moving up in the canopy and visually scanning downwards (Brockman et al. 2008).
A) Felis catus
B) Felis silvestris
172
There is currently no known data on the potential competition between F. silvestris and
Madagascar‟s native carnivores. The use of natural areas by wildcats may also increase the risk
of disease transmission to native carnivores. At Ankarafantsika. F. silvestris has been implicated
in transmitting Feline Calicivirus to C. ferox (Dollar 2006). F. silvestris’ potential impacts on
both native carnivores and lemurs demands more attention, including conservation action that
could limit their populations, especially within native habitats.
173
Small Indian Civet (Vivirricula indica):
Figure 8. Small Indian Civet (Vivirricula indica)
V. indica is an introduced carnivore to Madagascar, native to south Asia, where it
occupies a mosaic of secondary and plantation forests with considerable edge habitat (Chen et al.
2009). This species was likely brought to Madagascar for the use of their scent glands to make
perfumes, but have since established throughout the island. V. indica are most noticed in
disturbed areas adjacent to villages, but have also been detected in degraded and fragmented
forests, both in the eastern rainforest and western dry forests (Dollar 2006). There is some
indication V. indica can use intact rainforest, as they have been detected on the edges of
Ranomafana National Park (Gerber et al. 2010).
There is currently no known data on the potential competition between V. indica and
Madagascar‟s native carnivores. V. indica is presumed to be able to predate upon small lemurs
(Brockman et al. 2008), however, they are generally known to be scavengers that are not able to
climb trees (Ewer 1973), such their impact on Madagascar‟s lemurs in terms of direct predation
are likely negligible.
174
References:
Albignac, 1975. Breeding the fossa (Cryptoprocta ferox) at Montpellier Zoo. International Zoo
Yearbook 15, 147-150.
Albignac, R., 1971. Notes ethologiques sur quelques carnivores Malagaches : le Fossa fossa
(Schreber). Revue d'Ecologie 24, 383-394.
Albignac, R., 1972. The Carnivora of Madagascar, In Biogeography and ecology in Madagascar.
eds R. Battistini, G. Richard-Vindard. Dr. W. Junk B.V. Publishers, The Hague.
Albignac, R., 1973. Mammifêres Carnivores. Faune de Madagascar 36. ORSTOM/CNRS, Paris.
Albignac, R., 1974. Observations Eco-ethologiques sur le Genre Eupleres, viverride de
Madagascar. Terre Vie 28, 321-351.
Albignac, R., 1984. The Carnivores, In Madagascar. eds A. Jolly, P. Oberle, R. Albignac.
Pergamon Press, Oxford.
Bennett, C.E., Pastorini, J., Dollar, L., Hahn, W.J., 2009. Phylogeography of the Malagasy ring-
tailed mongoose, Galidia elegans, from mtDNA sequence analysis. Mitochondrial DNA
20, 7-14.
Brockman, D., Godfrey, L., Dollar, L., Ratsirarson, J., 2008. Evidence of Invasive Felis silvestris
Predation on Propithecus verreauxi at Beza Mahafaly Special Reserve, Madagascar.
International Journal of Primatology 29, 135-152.
Chen, M.-T., Tewes, M.E., Pei, K.J., Grassman, L.I., 2009. Activity patterns and habitat use of
sympatric small carnivores in southern Taiwan. Mammalia 73, 20-26.
Dollar, L., 1999a. Notice of Eupleres goudoutii in the rainforest of southeastern Madagascar.
Small Carnivore Conservation 20, 30-31.
175
Dollar, L., 1999b. Preliminary report on the status, activity Cycle, and ranging of Cryptoprocta
ferox in the Malagasy rainforest, with Implications for conservation. Small Carnivore
Conservation 20.
Dollar, L., 2006. Morphometrics, diet, and conservation of Cryptoprocta ferox. Duke University.
Dollar, L., Ganzhorn, U., Goodman, S.M., 2007. Primates and Other Prey in the Seasonally
Variable Diet of Cryptoprocta ferox in the Dry Deciduous Forest of Western
Madagascar, In Primate Anti-Predator Strategies. eds S.L. Gursky, K.A.I. Nekaris, pp.
63-76. University of Chicago Press, Chicago.
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.
Ewer, R.F., 1973. The Carnivores. Cornell University Press, Ithaca, NY.
Gerber, B., Karpanty, S.M., Crawford, C., Kotschwar, M., Randrianantenaina, J., 2010. An
assessment of carnivore relative abundance and density in the eastern rainforests of
Madagascar using remotely-triggered camera traps. Oryx 44, 219-222.
Goodman, S.M., 2003a. Carnivora: Galidia elegans, Ring-tailed Mongoose, Vontsira Mena, In
The Natural History of Madagascar. eds S.M. Goodman, J.P. Benstead, pp. 1351-1354.
The University of Chicago Press, Chicago.
Goodman, S.M., 2003b. Galidictis, Broad-striped Mongoose, Vontsira Fotsy, In Natural History
of Madagascar. eds S.M. Goodman, J.P. Benstead, pp. 1354-1357. The University of
Chicago Press, Chicago.
Goodman, S.M., 2003c. Predation on Lemurs, In The Natural History of Madagascar. eds G.S.
M., B. J.P., pp. 1221–1227. The University of Chicago Press, Chicago.
176
Goodman, S.M., Helgen, K.M., 2010. Species limits and distribution of the Malagasy carnivoran
genus Eupleres (Family Eupleridae). Mammalia 74, 177-185.
Goodman, S.M., Kerridge, F.J., Ralisoamalala, R.C., 2003. A note on the diet of Fossa fossana
(Carnivora) in the central eastern humid forests of Madagascar. Mammalia 67, 595-597.
Goodman, S.M., Langrand, O., Rasolonandrasana, B.P.N., 1997. The food habits of
Cryptoprocta ferox in the high mountain zone of the Andringitra Massif, Madagascar
(Carnivora, Viverridae). Mammalia: journal de morphologie, biologie, systematique des
mammiferes 61, 185-192.
Goodman, S.M., Pidgeon, M., 1999. Carnivora of the Reserve Naturelle Integrale d'Andohahela,
Madagascar. Fieldiana: Zoology, 259-268.
Hawkins, C.E., 1998. Behaviour and ecology of the fossa, Cryptoprocta ferox (Carnivora,
Viverridae) in a dry deciduous forest, western Madagascar. University of Aberdeen,
Aberdeen.
Hawkins, C.E., 2003. Cryptoprocta ferox, fossa, fosa. In S.M. Goodman & J.P. Benstead (Eds.)
The Natural History of Madagascar, In The Natural History of Madagascar. eds G.S. M.,
B. J.P., pp. 1360–1363. The University of Chicago Press, Chicago.
Hawkins, C.E., Dallas, J.F., Fowler, P.A., Woodroffe, R., Racey, P.A., 2002. Transient
masculinization in the Fossa, Cryptoprocta ferox (Carnivora, Viverridae). Biology of
Reproduction 66, 619-615.
Hawkins, C.E., Racey, P.A., 2005. Low population density of a tropical forest carnivore,
Cryptoprocta ferox: implications for protected area management. Oryx 39, 35-43.
Hawkins, C.E., Racey, P.A., 2008. Food habits of an endangered carnivore, Cryptoprocta ferox,
in the dry deciduous forests of western Madagascar. Journal of Mammalogy 89, 64–74.
177
Hawkins, F., 1994. Eupleres goudotii in west Malagasy deciduous forest. Small Carnivore
Conservation 11.
Irwin, M.T., Raharison, J.-L., Wright, P.C., 2009. Spatial and temporal variability in predation
on rainforest primates: do forest fragmentation and predation act synergistically? Animal
Conservation 12, 220-230.
Karpanty, S., Wright, P.C., 2007. Predation on lemurs in the rainforest of madagascar by
multiplenpredator species: Observations and experiments, In Primate Anti-Predator
Strategies. eds S.L. Gursky, K.A.I. Nekaris, pp. 77-99. University of Chicago Press,
Chicago.
Kerridge, F.J., Ralisoamalala, R.C., Goodman, S.M., Pasnick, S.D., 2003. Fossa fossana,
Malagasy Striped Civet, Fanaloka, In Natural History of Madagascar. eds S.M.
Goodman, J.P. Benstead, pp. 1363-1365. The University of Chicago Press, Chicago.
Kotschwar, M., 2010. Variation in predator communities and anti-predator behaviors of lemurs
in southeastern Madagascar. Virginia Polytechnic Institute and State University,
Blacksburg.
Laborde, C., 1986a. Adaptive characteristics of Crytpoprocta ferox limb bones to arboreal
locomotion as compared to other viverrid carnivores. Annales des Sciences Naturelles.
Zoologie et Biologie Animale 8, 25-39.
Laborde, C., 1986b. Description de la locomotion arboricole de Cryptoprocta ferox (Carnivore
Viverride malgache). Mammalia: journal de morphologie, biologie, systematique des
mammiferes 50, 369-378.
178
Laborde, C., 1987. Adaptive characteristics of Cryptoprocta ferox limb boes to arboreal
locomotion as compared to other Vicerrid carnivores. Annales des Sciences Naturelles.
Zoologie et Biologie Animale 8, 25-39.
Muldoon, K.M., de Blieux, D.D., Simons, E.L., Chatrath, P.S., 2009. The Subfossil Occurrence
and Paleoecological Significance of Small Mammals at Ankilitelo Cave, Southwestern
Madagascar. Journal of Mammalogy 90, 1111-1131.
Rand, A.L., 1935. On the habits of some Madagascar mammals. Journal of Mammalogy 16, 89-
104.
Rasolonandrasana, B.P.N., 1994. Contribution a' l'etude de l'alimentation de Cryptoprocta ferox
Bennett (1833) dans son milieu naturel. University of Antananarivo, Antananarivo,
Madagascar, Antananarivo.
Vanak, A.T., Gompper, M.E., 2009. Dogs Canis familiaris as carnivores: their role and function
in intraguild competition. Mammal Review 39, 265-283.
179
Appendix B: Relevant literature from the ISI Web of Science used to investigate common
mark-recapture analyses.
Balme, G.A., Hunter, L.T.B., Slotow, R., 2009a. Evaluating methods for counting cryptic
carnivores. Journal of Wildlife Management 73, 433-441.
Balme, G.A., Slotow, R., Hunter, L.T.B., 2009b. Impact of conservation interventions on the
dynamics and persistence of a persecuted leopard (Panthera pardus) population.
Biological Conservation 142, 2681-2690.
Di Bitetti, M.S., Paviolo, A., De Angelo, C., 2006. Density, habitat use and activity patterns of
ocelots (Leopardus pardalis) in the Atlantic Forest of Misiones, Argentina. Journal of
Zoology 270, 153-163.
Di Bitetti, M.S., Paviolo, A., De Angelo, C.D., Di Blanco, Y.E., 2008. Local and continental
correlates of the abundance of a neotropical cat, the ocelot (Leopardus pardalis). Journal
of Tropical Ecology 24, 189-200.
Dillon, A., Kelly, M.J., 2007. Ocelot Leopardus pardalis in Belize: The impact of trap spacing
and distance moved on density estimates. Oryx 41, 469-477.
Haines, A.M., Janecka, J.E., Tewes, M.E., Grassman Jr, L.I., Morton, P., 2006. The importance
of private lands for ocelot Leopardus pardalis conservation in the United States. Oryx 40,
90-94.
Harihar, A., Pandav, B., Goyal, S., 2009. Subsampling photographic capture-recapture data of
tigers (Panthera tigris) to minimize closure violation and improve estimate precision: a
case study. Population Ecology 51, 471-479.
180
Heilbrun, R.D., Silvy, N.J., Peterson, M.J., Tewes, M.E., 2006. Estimating Bobcat abundance
using automatically triggered cameras. Wildlife Society Bulletin 34, 69-73.
Jackson, R.M., Roe, J.D., Wangchuk, R., Hunter, D.O., 2006. Estimating snow leopard
population abundance using photography and capture-recapture techniques. Wildlife
Society Bulletin 34, 772-781.
Johnson, A., Vongkhamheng, C., Hedemark, M., Saithongdam, T., 2006. Effects of
humancarnivore conflict on tiger (Panthera tigris) and prey populations in Lao PDR.
Animal Conservation 9, 421-430.
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., Nichols, J.D., 1998. Estimation of tiger densities in India using photographic
captures and recaptures. Ecology 79, 2852-2862 | 2852.
Karanth, K.U., Nichols, J.D., Kumar, N.S., Hines, J.E., 2006. Assessing tiger population
dynamics using photographic capture-recapture sampling. Ecology 87, 2925-2937.
Karanth, U., Nichols, J., Kumar, S., 2004. Photographic Sampling of Elusive Mammals in
Tropical Forest, In Sampling rare or elusive species. ed. W. Thompson, pp. 229-247.
Island Press, Washington.
Linkie, M., Haidir, I.A., Nugroho, A., Dinata, Y., 2008. Conserving tigers Panthera tigris in
selectively logged Sumatran forests. Biological Conservation 141, 2410-2415.
Lynam, A.J., Rabinowitz, A., Myint, T., Maung, M., Latt, K.T., Po, S.H.T., 2009. Estimating
abundance with sparse data: tigers in northern Myanmar. Population Ecology 51, 115-
121.
181
Maffei, L., Cue‟llar, E., Noss, A., 2004. One thousand jaguars (Panthera onca) in Bolivia‟s
Chaco? Camera trapping in the Kaa-Iya National Pa. Journal of Zoology 262, 295-304.
Maffei, L., Noss, A.J., 2007. How small is too small? Camera trap survey areas and density
estimates for ocelots in the Bolivian Chaco. Biotropica 40, 71–75.
Marnewick, K., Funston, P.J., Karanth, K.U., 2009. Evaluating camera trapping as a method for
estimating cheetah abundance in ranching areas. South African Journal of Wildlife
Research 38, 59-65.
McCarthy, K.P., Fuller, T.K., Ming, M., McCarthy, T.M., Waits, L., Jumabaev, K., 2009.
Assessing Estimators of Snow Leopard Abundance. Journal of Wildlife Management 72,
1826-1833.
Paviolo, A.n., Di Blanco, Y.E., De Angelo, C.D., Di Bitetti, M.S., 2009. Protection Affects the
Abundance and Activity Patterns of Pumas in the Atlantic Forest. Journal of Mammalogy
90, 926-934.
Rayan, D.M., Mohamad, S.W., 2009. The importance of selectively logged forests for tiger
Panthera tigris conservation: a population density estimate in Peninsular Malaysia. Oryx
43, 48-51.
Rios-Uzeda, B., mez, H.G., Wallace, R.B., 2007. A preliminary density estimate for Andean bear
using camera-trapping methods. Ursus 18, 124-128.
Salom-Pérez, R., Carrillo, E., Sáenz, J.C., Mora, J.M., 2007. Critical condition of the jaguar
Panthera onca population in Corcovado National Park, Costa Rica. Oryx 41, 51-56.
Sarmento, P., Cruz, J., Eira, C., Fonseca, C., 2009. Evaluation of Camera Trapping for
Estimating Red Fox Abundance. Journal of Wildlife Management 73, 1207-1212.
182
Silver, S.C., Ostro, L.E.T., Marsh, L.K., Maffei, L., Noss, A.J., Kelly, M.J., Wallace, R.B.,
Gomez, H., Ayala, G., 2004. The use of camera traps for estimating jaguar Panthera
onca abundance and density using capture/recapture analysis. Oryx 38, 148-154.
Simcharoen, S., Pattanavibool, A., Karanth, K.U., Nichols, J.D., Kumar, N.S., 2007. How many
tigers Panthera tigris are there in Huai Kha Khaeng Wildlife Sanctuary, Thailand? An
estimate using photographic capture-recapture sampling. Oryx 41, 447-453.
Trolle, M., Kery, M., 2003. Estimation of ocelot density in the Pantanal using capture-recapture
analysis of camera-trapping data. Journal of Mammalogy 84, 607-614 | 607.
Trolle, M., Noss, A.J., Lima, E.D.S., Dalponte, J.C., 2007. Camera-trap studies of maned wolf
density in the Cerrado and the Pantanal of Brazil. Biodiversity and Conservation 16,
1197–1204.
Wang, S.W., Macdonald, D.W., 2009. The use of camera traps for estimating tiger and leopard
populations in the high altitude mountains of Bhutan. Biological Conservation 142, 606-
613.
