the influence of species traits and landscape …ufdcimages.uflib.ufl.edu › uf › e0 › 04 ›...
TRANSCRIPT
THE INFLUENCE OF SPECIES TRAITS AND LANDSCAPE ATTRIBUTES ON THE RESPONSE OF MID- AND LARGE-SIZED NEOTROPICAL MAMMALS TO FOREST
FRAGMENTATION
By
DANIEL HARRY THORNTON
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
2010
1
© 2010 Daniel Harry Thornton
2
To Erin and Sam, Mom and Dad, and everyone else who helped along the way
3
ACKNOWLEDGMENTS
The National Science Foundation, American Society of Mammalogists, Wildlife
Conservation Society-Guatemala Program, University of Florida IGERT Working
Forests in the Tropics, IDEA Wild, and the Department of Wildlife Ecology and
Conservation-University of Florida provided financial support for this research. I thank
the administration of Tikal National Park and the Consejos Nacional de Áreas
Protegidas in Guatemala for supporting this work. I am indebted to Nery Jurado and
Demetrio Cordova for invaluable help in the field. I thank Roan McNab, Rony Garcia,
Jose Moreira, and the staff of WCS-Guatemala for logistical support and advice. I
extend my gratitude to Alejandro Estrada and Tania Urquiza-Haas for access to their
data and information regarding their study areas. Finally, I thank my committee
members Dr. Lyn Branch, Dr. Mel Sunquist, Dr. Michael Binford, Dr. Emilio Bruna, and
Dr. Mary Christman for advice on numerous aspects of my research. Research Protocol
E019 was approved by the UF Institutional Animal Care and Use Committee.
4
TABLE OF CONTENTS page
ACKNOWLEDGMENTS.................................................................................................. 4
LIST OF TABLES............................................................................................................ 7
LIST OF FIGURES.......................................................................................................... 8
ABSTRACT ..................................................................................................................... 9
CHAPTER
1 INTRODUCTION .................................................................................................... 12
2 ANALYZING THE RELATIONSHIP BETWEEN SPECIES TRAITS AND VULNERABILITY TO FRAGMENTATION: PASSIVE SAMPLING EFFECTS AND CROSS-LANDSCAPE COMPARISONS........................................................ 17
Introduction ............................................................................................................. 17 Methods .................................................................................................................. 21
Study Site ......................................................................................................... 21 Species Ecological and Life History Traits........................................................ 23 Mammal Surveys.............................................................................................. 24 Measuring Vulnerability to Fragmentation ........................................................ 26 Influence of Species Traits on Vulnerability...................................................... 29 Cross-Landscape Comparison of Vulnerability................................................. 30
Results.................................................................................................................... 31 Vulnerability to Fragmentation.......................................................................... 32 Influence of Species Traits on Vulnerability...................................................... 32 Cross-Landscape Comparison ......................................................................... 33
Discussion .............................................................................................................. 34 Influence of Species Traits on Vulnerability...................................................... 34 Cross-Landscape Comparison ......................................................................... 38 Species-Level Patterns .................................................................................... 41 Conclusions...................................................................................................... 42
3 EVALUATING THE RELATIVE INFLUENCE OF HABITAT LOSS AND FRAGMENTATION: DO TROPICAL MAMMALS MEET THE TEMPERATE PARADIGM?......................................................................................................... 533
Introduction ........................................................................................................... 533 Methods ................................................................................................................ 555
Study Site ....................................................................................................... 555 Mammal Surveys............................................................................................ 566 Habitat Measurements: Patch-Scale Variables .............................................. 577 Habitat Measurements: Landscape-Scale Variables........................................ 58
5
Additional Predictor Variables .......................................................................... 59 Relative Influence of Habitat Loss and Fragmentation ..................................... 60
Results.................................................................................................................... 63 Discussion .............................................................................................................. 66
4 A REEXAMINATION OF THE INFLUENCE OF LANDSCAPE CONTEXT ON SPECIES PRESENCE AND ABUNDANCE............................................................ 76
Introduction ............................................................................................................. 76 Methods .................................................................................................................. 80
Literature Search.............................................................................................. 80 Collection of Data From Studies....................................................................... 81 Data Analysis ................................................................................................... 83
Results.................................................................................................................... 85 Effects of Predictor Variables ........................................................................... 85 Test of Predictions............................................................................................ 86
Discussion .............................................................................................................. 87
5 CONCLUSION...................................................................................................... 104
General Conclusions ............................................................................................ 104 Conservation Recommendations.......................................................................... 105
APPENDIX
A DETAILS OF COMPILATION OF ECOLOGICAL AND LIFE HISTORY TRAITS . 111
B PARAMETER ESTIMATES FOR OCCUPANCY MODELS.................................. 114
C BEST-FIT MODEL SETS...................................................................................... 115
D AIC WEIGHTS ...................................................................................................... 119
E DATA ON REFERENCES INCLUDED IN REVIEW ............................................. 120
LIST OF REFERENCES ............................................................................................. 123
BIOGRAPHICAL SKETCH.......................................................................................... 147
6
LIST OF TABLES
Table page 2-1 Life history and ecological traits of mammals studied in northern Guatemala. ... 45
2-2 Factor loadings from principle components analysis of ecological and life history traits.. ...................................................................................................... 46
2-3 Sampling effort employed for camera-trapping and visual censusing of mammals in forest patches and intact forest sites. ............................................. 46
2-4 Vulnerability estimates for the 3 landscapes included in the analysis................. 47
2-5 Results from regression analysis using PCA axes as predictor variables and 3 measures of vulnerability to fragmentation as response variables. ................. 48
3-1 Sampling effort employed for camera-trapping and visual censusing of mammals in forest patches................................................................................. 71
3-2 Patch and landscape context metrics for the 50 focal patches in my study area. ................................................................................................................... 72
3-3 Correlations between landscape metrics used to categorize habitat amount and configuration in the landscape surrounding focal patches. .......................... 73
3-4 Parameter estimates for patch-scale variables and habitat loss and fragmentation variables. ..................................................................................... 74
4-1 Example table used for coding species response to landscape context, patch, and within-patch variables. ...................................................................... 97
4-2 Covariates used in the generalized linear mixed models.................................... 98
4-3 Influence of predictor variables on probability of response to landscape context, patch, and within-patch variables. ......................................................... 99
7
LIST OF FIGURES
Figure page 2-1 Location of study area in Guatemala.. ................................................................ 49
2-2 Map showing fragmented and continuous forest study areas............................. 50
2-3 Results of hierarchical partitioning analysis for 3 landscapes............................. 51
2-4 Results of hierarchical partitioning analysis for difference in occupancy between continuous forest and forest patches of the same size. ....................... 52
3-1 Map of a portion of the study area. ..................................................................... 75
4-1 Example of typical focal patch study................................................................. 100
4-2 Characteristics of the studies, and study species, included in the review......... 101
4-3 Comparison of parameter estimates for least squares means of probability of response to landscape context. ........................................................................ 102
4-4 Comparison of parameter estimates for least squares means of probability of response to landscape context and patch variables based on sample size (i.e., # of patches surveyed). ............................................................................ 103
8
Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy
THE INFLUENCE OF SPECIES TRAITS AND LANDSCAPE ATTRIBUTES ON THE RESPONSE OF MID- AND LARGE-SIZED NEOTROPICAL MAMMALS TO FOREST
FRAGMENTATION
By
Daniel Harry Thornton
May 2010
Chair: Mel Sunquist Cochair: Lyn Branch Major: Wildlife Ecology and Conservation
As tropical reserves become smaller and more isolated, the ability of species to
use fragmented landscapes will be a key determinant of species survival. Although a
substantial amount of research has examined species distribution, abundance, and
richness in fragmented landscapes, the response of mid- and large-sized tropical
mammals to fragmentation remains relatively unknown. This gap in knowledge is
significant given that large mammals may be especially vulnerable to fragmentation and
play key roles in the maintenance of tropical forest habitat.
Species use of fragmented habitats is determined by both the ecological and life
history traits of species and the physical attributes of patches and landscapes. Although
several species traits commonly are associated with vulnerability to fragmentation, the
combination of traits that are most highly influential, and the effectiveness of those traits
in predicting vulnerability across distinct landscapes, remains poorly understood.
Considerable uncertainty also exists regarding what characteristics of patches and
landscapes exert the most influence on species. In particular, the relative influence of
habitat loss vs. habitat fragmentation on species is currently an area of intense debate.
9
Consensus from empirical studies in temperate areas is that habitat loss is more
important than fragmentation, but this question has not been investigated for any
tropical and wide-ranging fauna. Based on ecological traits (e.g., a high level of
specialization in tropical faunas), these species may be expected to respond more
strongly to fragmentation than temperate species studied thus far.
I studied use of forest patches by 25 mid- and large-sized neotropical mammals
in northern Guatemala. I determined the relationship between several species traits and
vulnerability to fragmentation and assessed the effectiveness of those traits in predicting
patch occupancy rates of the same mammals on two distinct landscapes in Mexico. I
also determined how patch occupancy patterns of mammals were influenced by habitat
loss and fragmentation in the landscape surrounding patches, as well several patch-
scale variables (i.e., patch size and quality). Additionally, I reviewed 125 multi-scaled
fragmentation studies to further examine the relative influence of patch and landscape-
scale factors on a variety of species.
For mammals in Guatemala, body size, home range size, and hunting
vulnerability were related to occupancy rates, but after controlling for passive sampling
effects only hunting vulnerability strongly influenced sensitivity to fragmentation.
Species that were heavily hunted were less common in forest patches than intact forest
sites. The cross-landscape comparison revealed both similarities and differences in the
species traits that influenced patch occupancy in Guatemala and Mexico. My analysis of
the influence of patch and landscape attributes on occupancy patterns revealed that a
greater number of mammal species responded significantly to habitat fragmentation
than to habitat loss in the landscape, and that this response generally was negative.
10
11
Patch-scale factors also exerted a strong influence on occupancy of mammals. My
review of the literature revealed that species from diverse taxa responded strongly to
within-patch, patch, and landscape-scale variables, but the probability of response to
these factors changed across taxa and according to several methodological variables.
Given the ubiquity of hunting in tropical environments, my findings indicate that
management efforts in fragmented landscapes that do not account for hunting pressure
may be ineffective in conserving tropical mammals. My study also indicates that species
traits may be useful in predicting relative patch occupancy rates and/or vulnerability to
fragmentation across distinct landscapes, but that caution must be used as certain traits
can become more or less influential on different landscapes, even when considering the
same set of species. Finally, my results point to the need for management efforts in
fragmented environments to go beyond habitat preservation and also consider
prevention of habitat fragmentation per se, at least for tropical mammals.
CHAPTER 1 INTRODUCTION
The clearing of tropical forest for food production and urban expansion is
proceeding at a tremendous rate (Asner et al. 2009), resulting in landscapes that
contain small fragments of forest surrounded by a sea of agriculture, cattle pasture, and
urban lands. Until recently, the main focus of conservation effort in the tropics has been
the creation and maintenance of large tracts of uninterrupted forest. Although this
strategy has resulted in reduced forest loss inside protected areas (Nepstad et al. 2006)
and will remain as a key component of effective conservation in the tropics, researchers
increasingly have recognized the potential conservation value of fragmented and
human-dominated landscapes that exist outside of reserves (Daily et al. 2003; Chazdon
et al. 2009). These altered landscapes serve as important refugia for tropical species
caught outside of protected areas (e.g., Sekercioglu et al. 2007) and act as corridors for
species during seasonal, migratory, or dispersal movements (Hansen & DeFries 2007).
Given that fragmented landscapes are becoming more common while at the same time
tropical reserves become smaller and more isolated (deFries et al. 2005), a better
understanding of species response to fragmentation in tropical environments is urgently
needed.
A substantial body of literature exists detailing how loss and fragmentation of
habitat affects the distribution, abundance, richness, and demography of plants and
animals in a variety of ecosystems (see reviews in Harrison & Bruna 1999; Debinski &
Holt 2000; Ewers & Didham 2006; Fleishman & Mac Nally 2007; Laurance 2008).
Factors operating at a variety of scales have been found to strongly influence species
distribution or abundance in fragmented environments, including small-scale variations
12
in patch quality or level of disturbance within patches (e.g., Michalski & Peres 2007;
Schooley & Branch 2009), meso-scale variations in patch size and shape (e.g., Graham
& Blake 2001; Crooks 2002), and large-scale variations in the composition and
configuration of habitat in the landscape or region (e.g., Bakker et al. 2002; Lee et al.
2002). In addition to the physical characteristics of patches and landscapes, species
survival within fragmented landscapes is determined by ecological and life history traits
of species (Henle et al. 2004). Numerous traits have been identified as potentially
influencing species response to fragmentation, including body size, home range size,
natural density, niche breadth, and matrix tolerance, among others (Laurance 1991;
Davies et al. 2000; Henle et al. 2004).
Although habitat loss and fragmentation are well-studied topics in general, the vast
majority of research has been done in temperate ecosystems, with relatively less work
in tropical environments (see Chapter 4). Moreover, tropical studies largely have been
concentrated in a few well-studied areas of Brazil and Mexico, with other regions of the
tropics receiving much less attention. Taxonomic bias in fragmentation studies also is
marked, with birds being the most commonly studied species (e.g., Fahrig 2003).
Only a limited number of studies have examined the response of larger tropical
mammals to habitat fragmentation, and many of these studies were focused exclusively
on primate species (Estrada et al. 1994; Onderdonk & Chapman 2000; Estrada et al.
2002; Michalski & Peres 2005; Anzures-Dadda & Manson 2007; Michalski & Peres
2007; Arroyo-Rodríguez et al. 2008; Urquiza-Haas et al. 2009). This lack of research is
problematic given that mid and large-sized mammals likely cannot maintain viable
populations in a single reserve and move extensively over large areas, necessitating
13
use of fragmented landscapes. Large tropical mammals also are of particular
conservation concern because many are threatened globally or locally, are often
flagship species for conservation programs, and may play key roles in the long-term
maintenance of tropical forest through their roles as top carnivores and seed predators
and dispersers. Despite their conservation significance, we still have a poor
understanding of how these species respond to the changes in patch quality, habitat
area, and configuration that occur with habitat loss and fragmentation, and how species
traits influence interspecies differences in response.
The Maya Biosphere Reserve (MBR) and its surrounding landscape in the
Department of Petén, northern Guatemala, is an important site for addressing the
effects of habitat fragmentation on mammals. The Maya Forest is the largest contiguous
tropical forest in Mesoamerica and maintaining the integrity of this forest is of critical
conservation concern. The human population of the Petén has increased 10-fold over
the last 40 years, and this population explosion along with concurrent agricultural
activity has resulted in deforestation of a large portion of the entire department since
1970 (McNab et al. 2004). The only tropical forest now left within the buffer zone and
surrounding landscape of the MBR exists in scattered and isolated fragments
surrounded by agricultural fields and cattle ranches. The forest within the core area of
the reserve increasingly is threatened by illegal slash-and-burn agriculture and hunting,
with large portions of the western half of the reserve also heavily fragmented (Wildlife
Conservation Society 2007). This region of Guatemala contains a diverse fauna of mid-
and large-sized mammals that includes 5 felid species, 5 ungulates, 2 large rodent
species, 2 primates, and a diversity of smaller carnivores. Several of these species are
14
of global conservation concern (e.g., Baird’s tapir [Tapirus bairdii], jaguar [Panthera
onca] and black howler monkey [Alouatta pigra]), heavily utilized by humans for food
(e.g., paca [Agouti paca], red brocket deer [Mazama americana]) or are poorly known
throughout their range (e.g. margay [Leopardus wiedii], tayra [Eira barbara]).
I examined occupancy patterns of 25 mid- and large-sized mammals in forest
patches and intact forest sites of northern Guatemala to determine how these species
responded to habitat loss and fragmentation. I addressed several questions that have
general significance to fragmentation and landscape ecology theory as well as to
applied conservation management of tropical mammals in fragmented habitats. I also
reviewed 125 fragmentation studies to broaden my analysis and begin a synthesis of
the diverse literature on multi-scaled fragmentation effects.
Chapter 2 is an examination of the relationship between species traits and
vulnerability to fragmentation. Although several ecological and life history traits
commonly are associated with vulnerability to fragmentation, the combination of traits
that are most highly influential for tropical mammals remains poorly understood.
Moreover, the usefulness of species traits to predict response to fragmentation across
diverse landscapes is virtually unstudied. I determined which of 7 species traits most
strongly influence the vulnerability of mammals to habitat loss and fragmentation and
how the results of this analysis were affected by the incorporation of passive sampling
effects. I also examined the influence of species traits on patch occupancy rates of the
same set of mammals on two landscapes in Mexico (using previously published data) to
investigate if similar species traits successfully predict patch occupancy on distinct
landscapes.
15
16
Chapter 3 is an examination of how characteristics of both patches and
landscapes influence patch occupancy patterns of 20 mammal species. Considerable
uncertainty exists regarding which scales of patchiness tend to exert the most influence
on species, and which factors at each scale are most important. One question that has
generated substantial recent debate is whether or not, at the landscape-scale, species
respond more strongly to habitat loss or habitat fragmentation per se (i.e., the breaking
apart of habitat, independent of habitat loss). Empirical evidence from studies
conducted in temperate environments suggests that habitat loss is much more
important, but this question rarely has been investigated in tropical environments. I
tested this hypothesis for tropical mammals by determining if patch occupancy patterns
were influenced more strongly by habitat loss or fragmentation. I also determined the
influence of patch and within-patch factors on species occupancy patterns.
Chapter 4 broadens my analysis and takes advantage of the substantial number of
fragmentation studies that have been published over the last decade to examine in
more detail the relative influence of small and large-scale factors on species distribution
and abundance in patchy landscapes. I summarized the results of 125 studies that have
simultaneously examined species response to within-patch, patch, and landscape-scale
factors in fragmented environments. I tested the influence of taxonomic status and
several methodological variables on the probability that species would respond
significantly to these variables. I conclude my dissertation with a chapter detailing
general conclusions and conservation recommendations.
CHAPTER 2 ANALYZING THE RELATIONSHIP BETWEEN SPECIES TRAITS AND
VULNERABILITY TO FRAGMENTATION: PASSIVE SAMPLING EFFECTS AND CROSS-LANDSCAPE COMPARISONS
Introduction
Preserving biodiversity in the tropics requires integration of conservation efforts
both within and outside of reserves. Protected areas in the tropics only cover 5-10% of
remaining tropical forest (Myers 2002) and are inadequate for the protection of a large
number of species (Rodrigues et al. 2004; Ceballos 2007; Jenkins & Giri 2007).
Moreover, tropical reserves are becoming smaller and more isolated over time because
of forest loss within park borders and in the surrounding landscape (deFries et al. 2005).
Human-modified landscapes outside of tropical reserves therefore will serve an
increasingly important role in preserving species diversity (Chazdon et al. 2009). These
landscapes typically consist of remnant forest patches embedded in a matrix of
agriculture, cattle pasture, and secondary forest regrowth. Although recent studies have
shown that a subset of forest species use agricultural and pastural habitats (Daily et al.
2003; Harvey et al. 2006; Medina et al. 2007; Sekercioglu et al. 2007), remnant forest
patches are probably the most important components to conservation of biodiversity
within human-altered landscapes of tropical forest ecosystems. These patches provide
critical habitat for many forest-dependent tropical species living outside of protected
areas (Turner & Corlett 1996; Laurance & Bierregaard 1997).
Substantial interspecies variation exists in the ability of species to occupy or use
forest fragments (Laurance 1991; Gascon et al. 1999; Laurance 2002; Laurance et al.
2002). Survival of species within forest patches is determined by a combination of patch
and landscape attributes, and the life history or ecological traits of species (Henle et al.
17
2004). An understanding of how species traits influence distribution or abundance in
forest patches is important for identifying generalities in response to habitat loss and
fragmentation (Henle et al. 2004; Ewers & Didham 2006). Such knowledge is beneficial
to predicting and mitigating species loss in human-dominated landscapes (Laurance
1991; Davies et al. 2000).
Several species traits are commonly associated with vulnerability to fragmentation.
However, the combination of traits that are most highly influential, and the degree to
which those traits determine vulnerability on distinct landscapes, remains poorly
understood for most taxa (Henle et al. 2004). Three factors confound analyses of how
species traits influence vulnerability to fragmentation. First, studies do not always
separate “passive sampling effects” from actual effects of habitat loss and fragmentation
(Johnson 2001; Haila 2002). Passive sampling effects are apparent patterns in the
relative distribution or abundance of organisms among habitat fragments that are
merely artifacts of sampling. Such effects occur because less abundant or patchily
distributed species would be expected by chance alone to be present in fewer
fragments than more abundant or evenly distributed species (Bolger et al. 1991;
Johnson 2001; Haila 2002). These species are not necessarily more vulnerable to
habitat loss and fragmentation, but will be categorized that way if passive sampling
effects are not accounted for in the analysis. A second common problem is that issues
of detectability often are not considered in analyses. Species that appear to be
vulnerable could merely be those that are harder to detect (Fleishman & Mac Nally
2007). Inclusion of these elusive species in the analysis without accounting for
detectability will underestimate the actual number of patches occupied by such species
18
(MacKenzie et al. 2002) and make it harder to identify the real relationships between
vulnerability of a species and ecological or life history traits. Finally, the ability to identify
influential ecological and life history traits is hindered by a lack of cross-landscape
comparisons. The opportunity rarely is available to examine how the same set of
species respond to habitat loss and fragmentation on different landscapes. This limits
our ability to identify generalities in how species traits influence response to habitat
patchiness and to evaluate the relative role of context-specific environmental factors vs.
more intrinsic biological factors in controlling response patterns.
I addressed these issues with a study of mammal distribution patterns in a
fragmented tropical landscape in northern Guatemala. Neotropical mammals are an
excellent suite of species for studies of interspecies variability in response to
fragmentation because they are a species-rich group that varies greatly in ecological
and life history traits. Moreover, this suite of mammals includes some of the first species
to disappear from fragmented habitats (Chiarello 1999; Peres 2001). These same
mammals play key roles in maintaining long-term integrity of tropical forests through
their roles as top carnivores and seed predators and dispersers (Terborgh et al. 1999;
Terborgh et al. 2001; Silman et al. 2003; Wright et al. 2007). Thus, the ability of mid-
and large-sized mammals to use, move through, and survive in forest fragments is an
important issue for biodiversity conservation in the tropics. Although studies focused on
understanding differences in relative vulnerability of mammals to habitat loss and
fragmentation remain sparse, previous information from tropical and temperate regions
suggests that several factors may influence vulnerability of mid- and large-sized
mammals, including body size, mobility, matrix tolerance, niche breadth, nearness to
19
range boundary, and vulnerability to hunting pressure (Laurance 1991; Chiarello 1999;
Crooks 2002; Swihart et al. 2003; Michalski & Peres 2007).
I studied use of forest fragments and intact forest sites by 25 mid- and large-
sized mammals to determine how species traits influence sensitivity to habitat loss and
fragmentation. I tested the generality of these results with a comparison of patch
occupancy patterns of all 25 mammal species across three distinct fragmented
landscapes, and addressed several specific questions related to the influence of
species traits on vulnerability:
1) What are the most important species traits influencing vulnerability to habitat
loss and fragmentation of mammals in Guatemala?
2) Does accounting for passive sampling effects alter the relative importance of
species traits in determining vulnerability to habitat loss and fragmentation? I predicted
that traits influencing distribution of species in an unaltered landscape (i.e., body size,
home-range size, reproductive rate, trophic level) would be less important in
determining vulnerability to fragmentation once sampling effects were taken into
account in the analysis.
3) Do the same set of ecological and life history traits explain interspecific
variation in vulnerability to fragmentation on different landscapes? I predicted that the
same species traits would explain vulnerability across landscapes. To address this
question, I re-analyzed data from two studies conducted in nearby Mexican landscapes
with the same species (Estrada et al. 1994, Urquiza-Haas et al. 2009) so that the results
were comparable with my Guatemalan study. To my knowledge, this is the first attempt
20
to look at whether or not similar suites of ecological and life history traits determine
response to fragmentation for the same set of species inhabiting different landscapes.
Methods
Study Site
I conducted this study in a 300,000-ha area in the Petén region of northern
Guatemala (Fig. 2-1). The northernmost part of the study area is an intact landscape of
humid subtropical forest located within Tikal National Park, which itself is situated within
the larger Maya Biosphere Reserve (an UNESCO world heritage site and the largest
contiguous tropical forest in Central America). The southern part of the study area is a
highly fragmented landscape located within the buffer zone of the Maya Biosphere
Reserve and on private lands farther to the south. This area was formerly contiguous
forest, but now consists of a diverse collection of primary rainforest patches embedded
in a matrix of secondary and regenerating forest, cattle pasture and agricultural land
(Fig. 2-2A). Forest destruction and fragmentation began in the late 1970s. Forest
patches are thus no more than 30 years old. Forest cover in this area consists of
subtropical humid rainforest with scattered patches of seasonally inundated bajo forest
and a fringe of savannah forest in the south. Altitudinal variation is minimal (130 to 400
masl). Annual average temperature of this region is 21–24º C, and annual average
precipitation is 1,350 mm, with a marked dry season from December to May when the
average monthly rainfall is only 60 mm.
I selected 50 primary rainforest patches that ranged in size from 2.9 to 445.5 ha
and 12 sites in continuous forest (6 sites each of 20 ha and 100 ha; Fig. 2-2B) as my
study sites. Rainforest patches contained a diverse collection of tree species, but were
often dominated by some combination of Brosimum alicastrum, Manilkara zapota, Ficus
21
sp., Vitex gaumeri, Pouteria sp., Sebastiana longicuspis, Terminalia amazonia, and
Alseis yucatánensis, Bursera simaruba, Spondias mombin, Aspidosperma
megalocarpon, Dendropanax arboreus, Protium copal, Pimenta dioica, and Cedrela
odorata. I did not sample rainforest patches that were severely degraded by logging or
fire, but I sampled a number of lightly degraded sites. Lightly degraded sites were those
that still had a largely intact overstory of trees and where the effects of past fires were
limited to less than 25% of the patch area. I did not sample patches of bajo forest,
savannah forest, or secondary forest. Almost all forest patches in my study were used
to some degree by local inhabitants for collection of non-timber forest products, hunting,
or cattle grazing. Study sites in continuous forest were similar in tree composition and
density to rainforest patches, but were surrounded by uninterrupted forest cover. My
continuous forest sites were subject to low levels of hunting pressure and collection of
non-timber forest products. Based on these characteristics, I made the assumption that
these sites in continuous forest are analogous to pre-fragmentation conditions in the
fragmented part of the study area. Continuous forest sites were situated greater than
500 m, but less than 3 km from the main road in Tikal National Park (Fig. 2-2B). This
distance restriction was established to take advantage of the decreased hunting
pressure on mammals close to the main road and archaeological ruins (which are
patrolled more heavily by park guards), while minimizing disturbance related to traffic
and human visitors to the park. A previous camera-trapping study of mammals in the
park found increased mammal diversity in areas closer to the main road (Kawanishi
1995).
22
Species Ecological and Life History Traits
The mid- and large-sized mammal community in northern Guatemala is quite
diverse, and includes 29 species. However, four of these species are either patchily
distributed throughout the region or have habits that make them very difficult to detect
(grison [Galictis vitatta], northern naked-tailed armadillo [Cabassous centralis], coyote
[Canis latrans], cacomistle [Bassariscus sumichrasti]). After excluding these species, I
could adequately sample 25 species in my study area. Although most species are found
elsewhere in Central and/or South America, the Mexican black howler monkey (Alouatta
pigra) is restricted to Guatemala, Belize, and the Yucatán peninsula of Mexico. Several
species are of high global or local conservation concern, including the tapir (Tapirus
bairdii), jaguar (Panthera onca), spider monkey (Ateles geofroyii), and white-lipped
peccary (Tayassu pecari).
I determined values of 7 life history and ecological traits for each species using
field guides and published literature (Table 2-1). The procedure and literature used to
determine values for each trait category are described in Appendix A. These traits were
chosen because they commonly are hypothesized to influence vulnerability to
fragmentation in mammals based on empirical and theoretical evidence (Laurance
1991; Peres 2001; Henle et al. 2004; Ewers & Didham 2006). Ecological and life history
variables used in my study were correlated. Consequently, I performed a principle
components analysis (PCA) using proc FACTOR in SAS (SAS 2008) to reduce the
number of variables and remove correlations. I log-transformed body mass and home-
range size prior to input in the PCA analysis. Results from the PCA indicate that 83% of
the variation in the seven traits is explained by just three axes. Based on factor
loadings, these three axes each represent a distinct aspect of mammalian biology and
23
ecology (Table 2-2). PCA axis 1 (reproduction/niche specialization axis) represents a
gradient from species with low reproductive rates and specialized diets and habitats, to
species with high reproductive rates and generalized diets and habitats. PCA axis 2
(body size/hunting vulnerability axis) represents a gradient from species with small body
size and low vulnerability to hunting (i.e., are rarely or never hunted) to species with
large body size and high vulnerability to hunting (i.e., heavily hunted/persecuted
species). PCA axis 3 (home range/trophic level axis) represents a gradient from species
with small home ranges and lower trophic levels to species with large home ranges and
higher trophic levels.
Mammal Surveys
I determined mammal presence/absence within forest patches and continuous
sites from January 2006 to August of 2008. I avoided sampling during mid-late wet
season (mid-September to mid-December) because of problems with camera
performance in very wet conditions. I used camera-trapping and visual censuses to
survey for mammal presence within study sites. This combination of techniques gave us
the best chance to detect the presence of elusive arboreal and terrestrial species. I
placed camera-traps in a variety of locations in each study site to maximize the number
of species photographed. These locations included roads, small and large game trails,
water holes, den sites, and other areas containing substantial signs of animal use such
as tracks, digging, or scraping. I placed camera-traps at least 10 m from the edge of
patches, with the sensor approximately 10 - 20 cm off the ground so that smaller
species could not avoid detection by walking under the sensor. I used both passive
(Leaf River™ model C-1BU, Leaf River Outdoor Products, Taylorsville, MS) and active
(Trailmaster® model 1500, Goodson and Associates, Inc., Lenexa, KS) infrared camera
24
traps in approximately equal proportions within each site. I placed more cameras in
larger sites (Table 2-3) and spaced them farther apart in order to cover a larger area. I
included patch size in assessments of detectability to account for potential biases
related to unequal sampling per unit area in the patches.
I deployed camera-traps for a 16-day period in each site. A photograph of a
species at any camera within a site was considered an indication of presence. I
recorded presence/absence for each species within each site after every 4-day interval.
By breaking up the 16-day period into 4-day sessions, I created a series of repeat
detection/non-detection data (i.e., a detection history) for use in modeling detection
probabilities for each species. Because absence of a species could be the result of
either true absence or failure to detect a species, detectability must be included to avoid
underestimating occupancy (MacKenzie et al. 2006).
Camera-trapping was ineffective for sampling arboreal species, as well as two
terrestrial species (white-tailed deer [Odocoileus virginianus] and collared peccaries
[Pecari tajacu]). In order to document presence/absence of these species, I performed
visual censuses of sites in the early morning (between sunrise and 3 hours after
sunrise). Surveys were repeated 5 times within a 2-week period for each site, resulting
in a series of detection/non-detection data for use in modeling detection probabilities. In
order to cover as much of the site as possible and to increase my chance of
encountering species, I did not cut transects for walking within each site. I surveyed
sites by walking along small roads, human foot paths, and game trails, and by walking
through sections without any obvious trails. I walked approximately 1 km per hour and
recorded direct observations of animals, vocalizations, and well-defined tracks as
25
indications of presence within the site. Distance walked varied with patch size (Table 2-
3). For small sites (less than 10 ha), I was able to walk through most or all of the site
during each session. For sites too large to survey completely in one session, I divided
the site into 2-4 sections and randomly chose a section to walk each session. I repeated
this process until I had 5 surveys for the site. During each survey, I chose a random
compass direction to guide my general walking direction within the designated section.
Once an edge was reached (of the section, or of the site itself), I walked a new compass
direction oriented back into the section until the appropriate distance was reached. I
walked greater distances in larger patches, but did not walk the same distance per unit
area in small and large patches (Table 2-3). As with cameras, patch size was included
in assessments of detectability to account for potential biases related to unequal
sampling per unit area in the patches.
For 17 forest patches and 4 continuous sites, I also conducted visual censuses 2
hours prior to sunrise to search for arboreal nocturnal mammals, particularly kinkajous
(Potos flavus). I cut meter-wide transects within multiple sections of each site in order to
move through the fragments during the night. I walked transects at a very slow pace
(0.5km/hour), and used flashlights to search for mammals in the trees. I repeated
nocturnal surveys 5 times, walking a different transect each night. Larger sites received
greater sampling effort (more and longer transects were cut and walked; Table 2-3).
Measuring Vulnerability to Fragmentation
I modeled patch occupancy and detection probabilities for each species using
logistic regression in program PRESENCE (Hines 2006). Because many species in my
study were capable of moving in and out of patches during sampling, the occupancy
estimator is best interpreted as “probability of use” of a patch, rather than probability of
26
occupancy. The detection parameter is best interpreted as probability of being within the
patch and detected during sampling (MacKenzie et al. 2006). For ease of presentation, I
will use traditional occupancy terminology in this paper.
I used three measures of vulnerability: 1) overall proportion of patches occupied
(used) by each species (continuous forest sites not included), 2) difference between
probability of occupancy (use) in continuous vs. fragmented forest sites of 20-ha size,
and 3) difference between probability of occupancy (use) in continuous vs. fragmented
forest sites of 100-ha size. These two later measures account for passive sampling
effects in the analysis by comparing expected patterns of occupancy in continuous
forest sites with observed patterns in forest fragments of similar size. The two sizes (20
ha and 100 ha) were chosen to represent the size range of most patches in my study
area (Table 2-3). Although the ideal way to account for passive sampling effects is to
have pre- and post-fragmentation information on species distribution, such information
rarely is available and my comparative approach is appropriate given similarities
between study sites in intact and fragmented areas.
To determine the overall proportion of patches occupied by each species, I
modeled occupancy and detection probabilities using detection/non-detection data
collected from all 50 forest patches and 12 continuous forest sites for 24 species. For
kinkajous, modeling was limited to the 17 forest patches and 4 continuous forest sites
surveyed during pre-dawn hours. In program PRESENCE, I first modeled detection as a
function of patch size and a categorical variable representing fragmented vs. continuous
sites, keeping occupancy constant. I determined the best fit detection model for each
species using AICc (Burnham & Anderson 2002). Species with less than 7 total
27
detections in all sites combined were considered to have constant detection
probabilities. I then took the best fit or constant detection model, and modeled
probability of occupancy using two covariates: size of the site (i.e., patch size or size of
the site in continuous forest) and fragmented vs. continuous forest status. Based on this
final model, I calculated the overall proportion of patches occupied by summing the
individual occupancy estimates for each patch (excluding occupancy estimates from
continuous sites) and dividing by the total number of patches. This was my first
measure of vulnerability, with higher proportion of patches occupied indicating less
vulnerable species. For my second measure of vulnerability, I used occupancy models
for each species to estimate the probability of occupancy in a 20-ha forest site and a 20-
ha patch, and subtracted the two values. I repeated this process for 100- ha sites, which
was my third measure of vulnerability. High positive values resulting from these two
analyses indicate species that had high occupancy probabilities in continuous forest
sites, but low occupancy probabilities in forest patches of the same size. These species
are considered highly vulnerable to habitat loss and fragmentation. Conversely, high
negative values indicate species that benefit from habitat loss and fragmentation, as
they were more commonly encountered in forest patches than in continuous forest sites
of the same size. For the analysis of vulnerability, I limited the number of covariates
used in modeling detection and occupancy to two because of sparse datasets for
several species (e.g., jaguars, tapir, puma [Puma concolor]). Overall patch occupancy
rates using simple or more complicated models (see Chapter 3) were similar, so I only
present results based on the simple models with limited numbers of covariates.
28
Influence of Species Traits on Vulnerability
I evaluated the influence of ecological and life history traits on vulnerability to
fragmentation with multiple regression and hierarchical partitioning. I used two
approaches to account for correlations between the species traits: 1) I used the
uncorrelated PCA axes of species traits as predictor variables in a multiple regression,
and 2) I used hierarchical partitioning analysis (Chevan & Sutherland 1991) to tease
apart the independent contribution of each individual species trait. For the first
approach, I ran three separate regressions with proc REG in SAS (SAS 2008) using the
three PCA axes as predictor variables and my three measures of vulnerability as
dependent variables. Because the PCA axes are orthogonal, partial R2 values of each
axis can be used to evaluate the relative influence of each axis on the response
variable. For the second approach, I employed hierarchical partitioning analysis with the
original seven species traits as predictor variables using the ‘hierpart’ macro in SAS
(Murray & Conner 2009). Hierarchical partitioning analysis calculates the increase in
model fit associated with each predictor variable by averaging the goodness of fit
increase across the hierarchy of models in which the variable appears (see [Chevan
and Sutherland 1991] for additional explanation). Initial inputs into analyses are the
goodness-of-fit values (e.g., R2) obtained using regression analysis. The hierarchical
partitioning analysis estimates the independent explanatory power of a variable (i.e.,
effect on a response variable attributable solely to that particular predictor) and joint
explanatory power (i.e., effect on a response variable attributable to joint action with
other predictors) of each predictor variable. The independent explanatory power serves
as the appropriate measure of the influence of a predictor variable on a response
(Chevan & Sutherland 1991; Mac Nally 2000). The key advantage of hierarchical
29
partitioning analysis is that it can provide an accurate assessment of the independent
effect of a predictor variable, even in the presence of multicollinearity (Murray & Conner
2009; but see Smith et al. 2009 for a criticism of this approach).
Cross-Landscape Comparison of Vulnerability
I compared results of my analysis to two other studies that involved the same
species (Estrada et al. 1994; Urquiza-Haas et al. 2009). Estrada et al. (1994) studied
patch occupancy of small and large mammals in 35 lowland rainforest fragments of Los
Tuxtlas, Mexico, and Urquiza-Haas et al. (2009) studied patch occupancy of mid- and
large-sized mammals in 147 fragments of tropical dry forest in Yucatán, Mexico. As in
this study, the overall proportion of fragments occupied was determined for each
species, although the methodology of the two studies differed from my own. Estrada et
al. (1994) used live traps and diurnal/noctural visual surveys to determine
presence/absence of mammals within patches, and Urquiza-Haas et al. (2009) used
interviews with landowners to determine patch occupancy. Patch occupancy estimates
were not listed in the original paper for the Los Tuxtlas data, but were obtained from the
author (A. Estrada, personal communication). Results of these two studies were not
corrected for passive sampling effects, and thus are comparable only to my vulnerability
measure of “overall proportion of fragments occupied”.
I limited my cross-landscape comparison to those species shared among the
three landscapes (n = 25 species). Patch occupancy patterns of common and Virginia
opossums (Didelphis marsupialis and Didelphis virginianus) were evaluated at the
genus level in the Yucatán landscape, but were treated as separate species for the Los
Tuxtlas and Guatemalan landscapes. I therefore treated them as separate species for
the Yucatán landscape, and assigned each species the same patch occupancy as was
30
recorded for the two species combined (i.e., 100 %). Whether or not I combine data for
these two species did not alter results of the subsequent analysis. For the Los Tuxtlas
landscape, data were collected on Alouatta palliata and Dasyprocta mexicana instead of
Alouatta pigra and Dasyprocta punctata as in my study and the Yucatán study.
However, because of similarities in morphology and ecology of these species (Reid
1997), I included them in the analysis and treated them as equivalent species.
I tested for general correlations between the rankings of species vulnerability in my
study and the two Mexican studies using Spearman’s correlation coefficients. This
analysis tested whether or not species that tended to be ranked lower in terms of overall
patch occupancy in my study also tended to rank lower in the other two landscapes. I
also performed multiple regression and hierarchical partitioning analysis to determine
which combination of species traits were most important in influencing patch occupancy
on each landscape. I made the assumption that the values used in my study for the
ecological and life history traits of species could be applied to those same species in the
Mexican landscapes. This assumption is likely to hold for some traits such as body size
and trophic level, but may not hold for other traits such as home-range size. However,
given the lack of region specific information for many of these traits and the geographic
proximity of the three study areas, I believe my approach is appropriate as a first
assessment of cross-landscape generalities in species response to fragmentation.
Results
I detected 25 species of mid- and large-sized mammals within my study sites in
12,960 camera-trap nights and 400 kilometers of visual surveys. All species detected in
intact forest sites also were detected in 1 or more fragments, except for white-lipped
peccaries, which only were detected in intact forest. The number of species detected in
31
forest patches varied between 3 and 19, with significantly more species detected in
larger fragments fragments (P < 0.01). I detected between 7 and 16 species in intact
forest sites, with more species detected in the larger 100-ha sites (P < 0.10).
Vulnerability to Fragmentation
Relative vulnerability of species to fragmentation as measured by proportion of
patches occupied varied greatly among species. Occupancy of forest patches in my
study area ranged from 0 (e.g., white-lipped peccaries) to nearly 100% (e.g., kinkajous,
Table 2-4). Best-fit models and parameter estimates for each species are presented in
Appendix B. Inclusion of detectability in the analysis resulted in substantial increases in
overall patch occupancy for species that were difficult to detect (Table 4). For example,
the estimated percentage of patches occupied increased 20.8 % and 19.2 % from the
naive estimate that did not include detectability for the Mexican porcupine (Coendou
mexicanus) and tayra (Eira barbara), respectively.
Relative vulnerability as measured by differences in occupation of continuous and
fragmented forest sites of the same size also varied greatly between species (Table 2-
4). Some species were much more common in continuous forest, such as white-tailed
deer, red brocket deer (Mazama americana), and puma, whereas others were much
more commonly encountered in forest patches, including northern tamandua
(Tamandua mexicana) and northern raccoon (Procyon lotor). Results of the vulnerability
analysis using 20-ha and 100-ha sites were similar for all species.
Influence of Species Traits on Vulnerability
My measure of vulnerability that was uncorrected for sampling effects (overall
proportion of patches occupied) was influenced strongly by PCA axis 2 and 3 (Table 2-
5). This indicates that species that are larger and more heavily hunted, and species that
32
have larger home ranges and higher trophic levels, tended to occupy fewer fragments
than those species that did not have those traits. Parameter estimates and partial R2
values indicate that axis 2 (body size/hunting axis) was the most important predictor.
Overall, the model with all three predictors explained a large amount of variation in
patch occupancy (R2 = 0.58). Results from the hierarchical partitioning analysis
generally confirm the results from the PCA regression analysis, but enabled us to look
at the importance of individual predictors in more detail (Fig. 2-3A). The overall
proportion of fragments occupied was influenced most strongly by body size, home
range, and to a lesser extent by hunting pressure.
Accounting for passive sampling effects altered the relative influence of species
traits on vulnerability to fragmentation. My measure of vulnerability corrected for
sampling effects (the difference in occupancy between fragmented and continuous
forest sites of similar size) was influenced strongly only by axis 2 (Table 2-5). This result
was similar when considering either 20-ha or 100-ha sites. This indicates that species
that are larger and more heavily hunted tended to be much more likely to occupy
continuous forest sites than forest patches of the same size. Overall fit of these models
was slightly lower, with only 41.3 % and 49.2 % of the variation explained by the full
models for 20-ha and 100-ha sites, respectively. Hierarchical partitioning demonstrated
that this measure of vulnerability was driven largely by differences in hunting
vulnerability, which accounted for almost half of the explained variance in both cases
(Fig. 2-4A and B).
Cross-Landscape Comparison
In general, species that had lower levels of patch occupancy in my study also had
lower levels of patch occupancy in the Los Tuxtlas and Yucatán landscapes (r = 0.66
33
and 0.52, respectively). This agreement between studies only holds when comparing
the rankings of species in terms of their relative vulnerability; absolute values of patch
occupancy differed substantially between the studies (Table 2-4). All three PCA axes
had a significant effect on vulnerability to fragmentation for species in Los Tuxtlas (P =
0.022, 0.026, and 0.002, respectively), and these axes explained 54.2 % of the variation
in the response. In Yucatán, the reproduction/niche breadth and body size/hunting axes
were related to vulnerability (P = 0.015 and 0.011, respectively), and all three axes
explained 44.0 % of the variation in proportion of patches occupied. Hierarchical
partitioning generally agreed with the PCA analyses but with some added detail on
individual predictors. For the Los Tuxtlas dataset, home range and reproductive rate
were most influential, accounting for a large amount of the explained variation in
vulnerability (Fig. 2-3B). In Yucatán, body size and habitat breadth had the highest
levels of independent explanatory power (Fig 2-3C).
Discussion
Influence of Species Traits on Vulnerability
After accounting for passive sampling effects, vulnerability to hunting was the
single most important species trait influencing how species responded to fragmentation
in my Guatemalan study site. Species that were more heavily hunted were more
vulnerable to fragmentation. The negative impacts of hunting on densities and/or
abundances of tropical mammals have been well documented in intact forest across the
Americas (Bodmer et al. 1997; Carillo et al. 2000; Hill et al. 2003; Peres & Nascimento
2006). Many of the species included in my study respond negatively to hunting pressure
in continuous forests of Guatemala and Mexico (Baur 1998; Polisar et al. 1998; Naranjo
& Bodmer 2007; Reyna-Hurtado & Tanner 2007). However, the effect of hunting on
34
mammal distribution and abundance has not been documented widely in fragmented
habitats (but see Cullen Jr. et al. 2000), even though the lack of escape habitat and
ease of access of hunters to forest patches makes species especially vulnerable to
hunting within forest remnants (Peres 2001; Parry et al. 2009). I observed direct and
indirect evidence of hunting in 20 of my forest patches during the relatively limited
sampling period, and all fragments in my study were located close to human
communities (< 5 km) and easily accessible via roads/trails. Based on models of the
minimum forest area required to maintain a sustainable harvest, the majority of forest
fragments in tropical environments fall well below the size necessary to support
populations of mid- and large-sized mammals that are experiencing hunting pressure
(Peres 2001). My data provide empirical support for a profound impact of hunting on
tropical vertebrates in fragmented landscapes by showing that more heavily hunted and
persecuted species were most likely to show a large reduction in their occupancy of
forest patches when compared to their normal occupancy patterns in intact forest.
The relative influence of particular species traits changed substantially depending
on whether or not I accounted for sampling effects in my estimate of vulnerability to
fragmentation. As predicted, two traits that are important in determining the
density/distribution of mammals in intact forest (body size and home-range size) were
very important in driving vulnerability to fragmentation when sampling effects were not
removed. Species that have these traits are expected to be present in a smaller
proportion of patches just by virtue of their natural rarity. These traits declined
substantially in importance when I accounted for sampling effects by comparing
occupancy patterns in forest patches with expected occupancy patterns in continuous
35
forest. Thus, body size and home range may not be as important in determining
vulnerability to fragmentation as indicated by an examination of patch occupancy
patterns alone.
Although some studies that correlate species traits with vulnerability to
fragmentation account for sampling effects (e.g., Bolger et al. 1991; Lynam & Billick
1999; Davies et al. 2000; Meyer et al. 2008), a substantial number do not and employ
measures such as overall patch occupancy or raw species-area curves to infer
sensitivity to fragmentation (e.g., Onderdonk & Chapman 2000; Crooks 2002; Virgos et
al. 2002; Viveiros de Castro & Fernandez 2004; Wang et al. 2009). Other authors have
pointed out the pitfalls of using patch occupancy rates or raw species-area curves to
infer vulnerability to fragmentation (Bolger et al. 1991; Johnson 2001; Haila 2002; Meyer
et al. 2008). My results demonstrate that, in some instances, non-removal of sampling
effects could lead to incorrect conclusions regarding the importance of species traits. In
particular, not accounting for passive sampling effects could lead to an increased
emphasis on the importance of traits associated with natural abundance or widespread
distribution such as body size, natural abundance, or potentially niche breadth that may
not be warranted. However, in some systems, these traits exert a heavy influence on
regional extinction proneness (Woodroffe & Ginsberg 1998; Purvis et al. 2000; Kamilar
& Paciulli 2008) and correlate with vulnerability to fragmentation even after the removal
of sampling effects (Davies et al. 2000; Shahabuddin & Ponte 2005). The importance of
these traits cannot be discounted. However, removal of sampling effects will promote a
better understanding of the influence of these types of traits on species’ vulnerability to
fragmentation (sensu Meyer et al. 2008).
36
In addition to biases created by passive sampling effects, vulnerability estimates of
mammals in this study would have been biased had I not accounted for differences in
detectability in the analysis of occupancy and vulnerability patterns. Detectability varies
as a function of species-specific abundance within sites (Royle & Nichols 2003) as well
as species-specific differences in the efficacy of survey techniques. For example, my
visual surveys did an excellent job of detecting monkeys, which are generally fast
moving, noisy, and likely to vocalize, but were much less effective at detecting
porcupines, which move slowly and quietly through the canopy. Although camera-traps
are ideal for collecting data on a large number of species, they do not detect all species
equally well (Barea-Azcón et al. 2007; Tobler et al. 2008). For example, smaller species
are less likely to trigger the cameras (Tobler et al. 2008). Specific characteristics of
camera trap locations, such as being placed on roads, small animal trails, or den sites
also may influence the probability of detection (Trolle & Kery 2005; Larrucea et al. 2008;
Harmsen et al. 2010). Ocelots in my study had much higher encounter rates on roads
than small animal trails, whereas the reverse pattern was true for agoutis (Thornton,
unpublished data). Failure to incorporate species-specific estimates of detectability in
my analysis would have underestimated the proportion of patches occupied for a
number of species that were difficult to detect (e.g., raccoons and tayra). Furthermore,
because detectability was different in intact forest and forest patches for some species
(e.g., species with the fragmented vs. continuous forest covariate included in best-fit
detection models in Appendix B), failure to incorporate dectection probabilities in my
analysis of occupation would have biased my calculation of differences in occupancy
rates of intact and fragmented sites.
37
Cross-Landscape Comparison
Relative rankings of species with respect to patch occupancy were in general
agreement between the three study landscapes, although the correlations were far from
perfect. Overall, species that were ranked lower in terms of patch occupancy on one
landscape tended to be ranked lower on the other landscapes, and vice versa.
However, absolute levels of patch occupancy for the same species differed drastically
among the study sites. In general, species had the lowest levels of patch occupancy in
Los Tuxtlas and highest levels in Yucatán. For example, coatis occupied 20% of the
patches in Los Tuxtlas, 63% of the patches in Guatemala, and 98% of the patches in
Yucatán. Some of this variation could be explained by differences in methodology.
Interviews used in Yucatán may have resulted in higher estimates of patch occupancy
for species because they measured both past and present mammal use of patches.
Mammals were recorded as present if they had been observed within the patch in the
last 5 years (Urquiza-Haas et al. 2009). The generally lower estimates of patch
occupancy in Los Tuxtlas may have been influenced by the use of live traps instead of
camera traps or lower levels of trapping effort. Also, the Los Tuxtlas study was done
before techniques existed for the incorporation of detectability in patch occupancy
estimates. A failure to incorporate dectectability may have biased estimates of
occupancy lower for some species. The three landscapes differed in climate, human
population density, and physical characteristics of patches and the surrounding
landscape and these context-specific landscape characteristics also could drive
differences in absolute levels of patch occupancy among these landscapes.
Similar to the results from Guatemala, species traits strongly influenced variation
in overall patch occupancy patterns of mid- and large-sized mammals in both Mexican
38
landscapes. Averaged across all three landscapes, species traits examined in this study
explained approximately 52% of the variation in overall patch occupancy for this set of
mid- and large-sized mammals. Intrinsic biological traits are thus extremely important in
determining patch occupancy rates for the species considered here.
Even though I limited my comparison to almost the exact same set of species in
each landscape, the relative importance of species traits in determining patch
occupancy patterns differed among the three landscapes. Because this cross-
landscape analysis used a measure of fragmentation sensitivity that did not account for
passive sampling effects, my comparison is perhaps best viewed as an analysis of how
species traits influence patch occupancy, rather than vulnerability to habitat
fragmentation per se. However, my general finding that different traits can emerge as
important on different landscapes for the same set of species is applicable and relevant
to studies of species’ vulnerability to habitat loss and fragmentation.
Occupancy patterns in Yucatán were influenced primarily by body size and habitat
breadth. This agrees largely with the results published in the original study (Urquiza-
Haas et al. 2009), even though the authors included additional species in their analysis
and used a slightly different measure of habitat specificity than I did. Habitat breadth, or
a related characteristic, matrix tolerance, has been found to be an important
determinant of vulnerability to fragmentation in studies on mammals (Laurance 1991)
and other taxa (Gascon et al. 1999; Sekercioglu et al. 2007). Species with a greater
habitat breadth are expected to be less vulnerable to fragmentation because they can
forage more effectively outside of forest patches, use more disturbed habitats, and may
be better able to cross non-forest habitat (Henle et al. 2004). Patch occupancy rates in
39
the Los Tuxtlas landscapes were influenced primarily by home range and reproductive
rate. Home-range size and reproductive rate are correlated with abundance, and thus
may be appear as important primarily because of sampling effects, or because species
with lower abundance are more likely to go extinct in fragments. Home-range size may
be also be influential because species with larger home ranges may be more likely to
come into contact and conflict with humans (Woodroffe & Ginsberg 1998). Reproductive
rate could also exert an influence on species vulnerability to fragmentation through
affects on the ability of animals to rebound from loss of habitat (Henle et al. 2004).
Although difficult to assess based on only three study sites, variation in the
influence of species traits on patch occupancy may be influenced by context-specific
differences between the landscapes. For example, the relatively large influence of
hunting vulnerability on patch occupancy in my Guatemalan landscape compared to the
Mexican landscapes could be related to differences in hunting pressure among the
three areas. Patches in my study area were heavily impacted by hunting pressure
whereas this was not the case in Los Tuxtlas at the time of the study (personal
communication, A. Estrada). Subsistence hunting occurs in Yucatán (personal
communication, T. Urquiza-Haas), but hunting pressure may be less intense than in
northern Guatemala where a rapidly increasing and extremely poor rural population
creates a large demand for wild game. The importance of habitat breadth in the Yucatán
compared to the other landscapes may be related to differences in the type of patches
studied in each area. Patches in the Yucatán were a mixture of primary, secondary, and
disturbed habitats, whereas patches in Guatemala and Los Tuxtlas were largely
undisturbed primary forest sites. Because habitat breadth is important in determining
40
use of disturbed or secondary forest habitats, this trait could be a more important
influence on patch occupancy patterns in Yucatán.
Collectively, results from the cross-landscape comparison indicate both similarities
and differences in how the same species responded to habitat patchiness. The strong
influence of both body size and home-range size in determining patch occupancy on all
three landscapes probably accounts for the correlation between the studies in terms of
the relative rankings of species according to patch occupancy rates. This suggests that
in may be possible to predict with some degree of accuracy relative patch occupancy
patterns of these species on novel landscapes, which is one of the major goals of
analyzing the relationship between species traits and response to landscape change
(Mac Nally & Bennett 1997). However, the influence of other variables, particularly
reproductive rate, habitat breadth, and hunting vulnerability on patch occupancy
patterns changed substantially across landscapes and likely accounts for the less than
perfect nature of the correlations. The search for generalities in species response to
habitat patchiness, and potentially to habitat loss and fragmentation per se, based on
shared ecological and life history characteristics therefore will be made more difficult
because of variability in the importance of traits among landscapes.
Species-Level Patterns
I sampled more fragments than any previous study of tropical mid- and large-sized
mammals in fragmented environments, excluding studies exclusively based on
interviews with landowners (i.e., Michalski & Peres 2005; Urquiza-Haas et al. 2009).
Given this substantial dataset, I conclude that forest fragments in my study area
supported a diverse assemblage of mammals. Larger fragments in my study area
generally supported a greater number of species, but small forest patches were still
41
used by a number of mammal species, some of them heavily forest-dependent (e.g.,
pacas [Agouti paca], agoutis [Dasyprocta punctata], margays [Leopardus wiedii], spider
monkeys, brocket deer). For example, I detected 9 species in a patch of only 7 ha,
including paca, margay, and ocelot (Leopardus pardalis). The suite of species using
fragments in my study area included several wide-ranging and/or forest dependent
species that are often considered most at risk of disappearing from fragmented
environments. However, because forest fragments in my study area were relatively
young, this diversity of mid- and large-sized mammals could be the result of insufficient
time for species to disappear completely from fragments (i,e., an extinction debt, Tilman
et al. 1994).
Although overall mammal diversity in my landscape was high, not all species were
equally adept at using forest patches. Several species appeared to be highly vulnerable
to loss and fragmentation of habitat and had a combination of low patch occupancy and
lower occupancy rates in forest patches than in intact forest sites (e.g., pumas, jaguars,
ocelots, tapir, white-lipped peccaries, spider monkeys). Many of these species are of
the highest conservation priority nationally and/or globally. In contrast, a suite of species
obviously benefited from loss and fragmentation of forest and had high patch occupancy
rates and/or were more commonly encountered in forest patches than continuous
forest. These species included both habitat generalists (e.g., northern raccoon,
jaguarundi [Puma yagouaroundi], hog-nosed skunk) and forest-dependent species (e.g.,
tayra).
Conclusions
My results indicate that vulnerability to hunting drives much of the interspecies
differences in sensitivity to habitat loss and fragmentation in northern Guatemala.
42
Because I was able to incorporate detectability in my analysis, my findings do not reflect
detection differences between species but instead real patterns in vulnerability. Hunting
pressure on mid- and large-sized mammals is common in many fragmented tropical
environment (Peres 2001). Reduction of hunting pressure may have a marked positive
effect on the ability of species to use and persist within fragmented landscapes of the
tropics and thus should be a primary focus of management efforts in human-dominated
environments with high levels of hunting. My work also shows that the way in which
vulnerability to fragmentation is measured, and in particular whether or not passive
sampling effects are accounted for in the analysis, may alter conclusions regarding the
relative influence of species traits on sensitivity to habitat loss and fragmentation.
Finally, my cross-landscape comparison found correlations among mammals on three
distinct landscapes when comparing the relative ranking of species in terms of patch
occupancy, which suggests some degree of similarity in response that could be used to
predict how the same species will react on novel landscapes. However, my comparison
also demonstrates that the relative influence of certain species traits on patch
occupancy patterns (and perhaps to some extent on vulnerability fragmentation)
changes across landscapes, perhaps because of context-specific differences between
landscapes. Moreover, absolute values of patch occupancy were markedly different on
the three landscapes. I found these results even though I was considering almost the
exact same set of species in all three landscapes. These findings therefore suggest
some limitations in the use of species ecological and life history traits to predict variation
in patch occupancy and/or sensitivity to loss and fragmentation of habitat across diverse
43
44
landscapes, at least until we are able to better incorporate extrinsic factors such as
landscape context into the analysis.
Table 2-1. Life history and ecological traits of mammals studied in northern Guatemala. Species Common name Body
mass (kg)
Home range (ha)
Trophic levela
Repro. rateb
Dietary breadthc
Habitat breadthc
Hunting vuln.d
Didelphis marsupialis Common opossum 1.5 12 2 12 7 7 1 Didelphis virginianus Virginia opossum 1.8 12 2 12 7 7 1 Dasypus novemcinctus Nine-banded armadillo 3.0 6 2 4 5 6 3 Tamandua mexicana Northern tamandua 6.2 25 3 1 1 5 1 Alouatta pigra Yucatán black howler 6.9 17 1 0.5 2 2 2 Ateles geoffroyi Central American spider
monkey 7.0 250 1 0.33 1 1 2
Leopardus pardalis Ocelot 9.7 1695 3 0.5 6 3 2 Leopardus wiedii Margay 3.6 1095 3 0.5 6 3 2 Panthera onca Jaguar 65.9 5600 3 1 3 3 3 Puma concolor Puma 45.0 6500 3 1.33 3 4 3 Puma yagouaroundi Jaguarundi 5.2 1065 3 2.5 7 5 2 Urocyon cinereoargenteus Gray Fox 2.7 95 2 4 7 6 1 Conepatus semistriatus Striped hog-nosed skunk 2.5 103 2 4 4 4 1 Eira barbara Tayra 4.5 1375 2 2.5 5 4 1 Nasua narica White-nosed coati 4.6 60 2 3.5 4 3 2 Potos flavus Kinkajou 3.3 23 1 1 4 2 1 Procyon lotor Northern raccoon 5.6 50 2 3.5 8 6 2 Tapirus bairdii Baird’s Tapir 240 125 1 0.5 3 3 3 Pecari tajacu Collared peccary 19.0 249 2 2 6 3 3 Tayassu pecari White-lipped peccary 33.5 2387 2 2 5 1 3 Mazama americana Red brocket deer 22.0 52 1 1 3 2 3 Odocoileus virginianus White-tailed deer 34.0 284 1 1.5 3 5 3 Coendou mexicanus Mexican porcupine 2.0 19 1 1 3 2 1 Agouti paca Paca 8.5 2 1 1.5 4 4 3 Dasyprocta punctata Central American agouti 3.5 2 1 1.5 4 3 3 a Trophic level: 1 = primarily browser/grazer or frugivore, 2 = omnivore, 3 = primarily carnivore/myrmecophage. b Calculated as # of young/year. cValues for dietary and habitat breadth based on number of food or habitat categories used – higher values indicate more generalized diets or habitats. See Appendix A for full description of trait categories. d Hunting vulnerability 1 = rarely/never hunted or killed, 2 = occasionally hunted, 3 = often hunted (e.g., a preferred game species).
45
Table 2-2. Factor loadings from principle components analysis of ecological and life history traits. Factor loadings indicate the correlation coefficients between the original variables and the “new” PCA variables (see text for interpretation of axes). Total explained variance of the three axes = 84%.
Variable PCA Axis 1 PCA Axis 2 PCA Axis 3 Body mass -0.32a 0.84 0.23 Home range -0.21 0.24 0.89 Trophic level 0.29 -0.09 0.87 Reproductive rate 0.82 -0.30 -0.15 Dietary breadth 0.80 -0.12 0.23 Habitat breadth 0.89 -0.15 0.00 Hunting vulnerability -0.13 0.94 -0.05 Eigenvalues 3.07 1.79 0.98 Variation explained (%) 43.82 25.50 14.04
Table 2-3. Sampling effort employed for camera-trapping and visual censusing of mammals in forest patches and intact forest sites. Distances listed were walked 5 times in each fragment.
Patch size # of patches surveyed
# of cameras placed in patch
Distance walked per session for daytime surveys (km)*
Distance walked per session for nighttime surveys (km)*
2.9-10 12 7 0.8 0.5 >10-20 7 10 1 0.8 >20-40 12 14 1.2 1 >40-80 5 17 1.5 1 >80-160 7 20 1.8 1.2 >160-320 4 25 2 1.5 >320 3 28 2 1.5 *Distances listed were walked 5 times in each fragment
46
Table 2-4. Vulnerability estimates for the 3 landscapes included in the analysis.
Species Naïve PFOa
Overall PFOb
Diff. occ. 20 ha
sitesc
Diff. occ. 100 ha sitesc
PFO – Yucatand
PFO – Los Tuxtlase
Didelphis marsupialis
66.0 76.5 -33.5 -33.9 100 54.3
Didelphis virginianus 36.0 31.9 15.3 16.8 100 40.0 Dasypus novemcinctus
70.0 70.4 -0.9 -0.8 100 14.3
Tamandua mexicana
62.0 86.9 -77.4 -77.5 95.9 17.1
Alouatta pigra* 80.0 80.7 23.9 13.0 32.0 45.7 Ateles geoffroyi 38.0 38.3 79.7 63.8 55.1 2.9 Leopardus pardalis 34.0 45.5 24.9 10.5 81.0 2.9 Leopardus wiedii 52.0 57.1 -11.3 -10.6 64.0 5.7 Panthera onca 6.0 9.0 14.5 28.0 55.1 0.0 Puma concolor 6.0 14.7 66.4 84.0 66.7 2.9 Puma yagouaroundi 24.0 36.1 -17.5 -28.1 79.6 14.3 Urocyon cinereoargenteus
42.0 45.7 17.8 17.8 100 0.0
Conepatus semistriatus
56.0 60.2 0.1 0.1 100 8.6
Eira barbara 24.0 43.2 -25.7 -28.3 94.6 14.3 Nasua narica 60.0 62.4 7.3 7.4 98.0 20.0 Potos flavus 88.0 96.9 -25.0 -21.2 93.2 17.1 Procyon lotor 48.0 71.0 -36.0 -38.3 97.3 17.1 Tapirus bairdii 2.0 2.3 26.7 47.9 12.2 0.0 Pecari tajacu 14.0 16.2 50.2 38.6 91.2 14.3 Tayassu pecari 0.0 0.0 0.0 16.7 4.1 0.0 Mazama americana 30.0 31.1 50.2 46.3 84.4 2.9 Odocoileus virginianus
20.0 24.4 74.3 64.2 91.8 0.0
Coendou mexicanus 36.0 56.8 -28.4 -31.3 92.5 22.9 Agouti paca 66.0 66.4 30.3 20.5 98.0 40.0 Dasyprocta punctataf
44.0 44.3 52.5 45.2 98.0 28.6
a Estimated percentage of patches occupied in Guatemala not corrected for detectability (number of fragments where species was detected at least once/total number of fragments). b Estimated percentage of patches occupied in Guatemala corrected for detectability (= sum of individual occupancy probabilities from each patch/total number of patches). c Occupancy probability (expressed as percentage) of continuous forest site – occupancy probability in forest patch of same size. d Percentage of patches occupied in Yucatan based on interviews with landowners (Urquiza-Haas et al. 2009). e Percentage of patches occupied in Los Tuxtlas based on live trapping and visual censuses (Estrada et al. 1994). f For Los Tuxtlas dataset, data were collected on Alouatta palliata and Dasyprocta mexicana.
47
Table 2-5. Results from regression analysis using PCA axes as predictor variables and 3 measures of vulnerability to fragmentation as response variables.
Predictor variables Parameter estimate ± SE
P-value Partial R2
Response variable = Proportion of fragments occupied
Axis 1 (reproduction/specialization) 0.01 ± 0.04 0.76 0.01 Axis 2 (body size/hunting) -0.18 ± 0.04 <0.01 0.45 Axis 3 (home range/trophic level) -0.09 ± 0.04 0.03 0.12
Response variable = Difference in occupancy of 20 ha intact sites and 20 ha patches
Axis 1 (reproduction/specialization) -0.08 ± 0.06 0.24 0.04 Axis 2 (body size/hunting) 0.23 ± 0.06 <0.01 0.34 Axis 3 (home range/trophic level) -0.07 ± 0.06 0.30 0.03
Response variable = Difference in occupancy of 100 ha intact sites and 100 ha forest patches:
Axis 1 (reproduction/specialization) -0.08 ± 0.06 0.22 0.04 Axis 2 (body size/hunting) 0.25 ± 0.06 <0.01 0.44 Axis 3 (home range/trophic level) -0.04 ± 0.06 0.49 0.01
48
Figure 2-1. Location of study area in Guatemala. Inset map shows location of the Maya
Biosphere Reserve in northern Guatemala. Enlarged portion shows location of my study sites in relation to the Maya Biosphere Reserve. Intact forest sites (upper black box) were located in the south-central portion of Tikal National Park, and forest patches (lower black box) were located in the buffer zone of the Maya Biosphere Reserve and private lands farther to the south.
49
igure 2-2. Map showing fragmented and continuous forest study areas. A) Map n
rk
Fshowing a small portion of my fragmented study area, with forest cover iblack, and matrix habitat (ag/pasture and regenerating forest) in white. B) Map showing my intact forest study area, with black boxes indicating the position of 20 and 100 ha study sites in Tikal National Park. The central paroad is also shown in black. These study sites were embedded within completely intact forest cover.
50
Figure 2-3. Results of hierarchical partitioning analysis for 3 landscapes. A) proportion
of patches occupied in Guatemala, B) proportion of patches occupied in Los Tuxtlas, Mexico based on data in Estrada et al. (1994), C) proportion of patches occupied in Yucatán, Mexico based on data in Urquiza-Haas et al. (2009). BS = body size, HR = home range size, TL = tropic level, RR = reproductive rate, DB = dietary breadth, HB = habitat breadth, HV = hunting vulnerability
51
52
Figure 2-4. Results of hierarchical partitioning analysis for difference in occupancy between continuous forest and forest patches of the same size. A) difference in occupancy of intact sites and forest patches of 20-ha size, B) difference in occupancy of intact sites and forest patches of 100-ha size. BS = body size, HR = home range size, TL = tropic level, RR = reproductive rate, DB = dietary breadth, HB = habitat breadth, HV = hunting vulnerability.
CHAPTER 3 EVALUATING THE RELATIVE INFLUENCE OF HABITAT LOSS AND
FRAGMENTATION: DO TROPICAL MAMMALS MEET THE TEMPERATE PARADIGM?
Introduction
Given the rapid landscape transformation occurring in the tropics, knowledge of
efficient actions for reducing biodiversity loss is urgently needed. Although prevention of
habitat loss is an important and obvious way to protect biodiversity, the effect of
reducing habitat fragmentation per se (i.e., breaking apart of habitat independent of
habitat loss) on biodiversity is still unclear. Some theoretical studies predict strong
effects of fragmentation (e.g., With & King 1999), whereas others predict small effects of
fragmentation compared with habitat loss (e.g., Fahrig 1997). Consensus from
numerous empirical studies is that habitat fragmentation has relatively weak effects on
species distribution and abundance compared with habitat loss, and that this effect is at
least as likely to be positive as negative (reviewed in Fahrig 2003; Smith et al. 2009).
Results from these studies are used to suggest that conservationists should be more
concerned with overall habitat loss in the landscape than with how remaining habitat is
arranged (Fahrig 2003; Ritchie et al. 2009; St-Laurent et al. 2009). Consequently,
management plans and design of protected areas should focus on preservation and
restoration of target amounts of habitat, rather than on the details of habitat
arrangement (Fahrig 1997). However, these conclusions are based on empirical studies
of temperate fauna, and this issue rarely has been investigated in tropical systems.
Moreover, most studies have focused on birds, insects, and small mammals, and to my
knowledge this question has not been investigated for any wide-ranging or large-bodied
species such as large mammals. Lack of studies on tropical species and wide-ranging
53
species is significant because two of the major results of habitat fragmentation per se
are an increase in amount of edge and a reduction in mean patch size. Tropical faunas
contain a high percentage of edge-sensitive species (Laurance et al. 2002) and
therefore may be influenced more strongly by habitat fragmentation than temperate
faunas studied thus far (Fahrig 2003). Wide-ranging or large bodied species that require
large areas of habitat for persistence may respond more strongly to reductions in patch
size that occur with habitat fragmentation than less wide-ranging species (Villard et al.
1999).
Often the influence of habitat loss and fragmentation is assessed empirically by
determining presence or abundance of species within focal patches or sites, and then
relating these patterns to variables measured in the surrounding landscape that serve
as correlates of habitat loss (i.e., amount of habitat in the landscape) and habitat
fragmentation (i.e., configuration of habitat in the landscape) (e.g., Drolet et al. 1999;
Klingbeil & Willig 2009; Ritchie et al. 2009). Identifying the proper spatial scale at which
to measure habitat loss and fragmentation is a key challenge for reliable estimates of
the influence of these factors. Different species respond most strongly to landscape-
scale variables measured at different spatial extents and therefore one spatial scale
often is not sufficient in multi-species studies (Holland et al. 2005). Moreover, a single
species may respond most strongly to habitat loss at a different spatial scale than
habitat fragmentation or other landscape-scale variables (Boscolo & Metzger 2009).
Studies that examine a species’ response to a single spatial scale may find a weak
response to habitat loss or fragmentation because the proper scale was not used in the
analysis. A second challenge in determining influence of landscape-scale variables
54
such as habitat loss and fragmentation is control of site or patch-scale variables.
Studies that do not simultaneously evaluate the influence of both small and large scale
factors may attribute a species’ response to either habitat loss or fragmentation in the
landscape when in fact the response is driven by site or patch-scale variables (Koper &
Schmiegelow 2006).
I examined responses of 20 mid- and large-sized neotropical mammals to habitat
loss and fragmentation at multiple spatial extents and assessed their response to patch
size and measures of patch quality in a fragmented landscape of northern Guatemala. I
tested the currently accepted hypothesis developed from temperate studies that habitat
loss is more important than habitat fragmentation in determining distribution patterns of
species. If this hypothesis is true, then patch occupancy patterns of mammals should be
influenced more strongly by habitat loss than habitat fragmentation, after controlling for
the influence of patch-scale variables.
Methods
Study Site
I conducted this study in a 250,000-ha area in the Petén region of northern
Guatemala (Fig. 3-1). The northernmost part of the study area is located within the
buffer zone of the Maya Biosphere Reserve (a UNESCO world heritage site, and the
largest contiguous tropical forest in Central America), and the remainder on private
lands further to the south. This area was formerly contiguous forest, but now consists of
primary rainforest patches embedded in a matrix of regenerating forest, cattle pasture
and agricultural land. Forest cover consists primarily of subtropical humid rainforest.
Forest destruction and fragmentation began in the late 1970s. Forest patches are thus
no more than 30 years old. Annual average temperature of this region is 21–24º C, and
55
annual average precipitation is 1350 mm, with a marked dry season from late
December to May.
My study sites were 50 primary rainforest patches (hereafter focal patches) that
ranged in size from 2.9 to 445.5 ha. I did not sample rainforest patches that were
severely degraded by logging or fire, but I sampled a number of lightly degraded sites.
Almost all forest patches in my study were used to some degree by local inhabitants for
collection of non-timber forest products, hunting, or shade for cattle.
Mammal Surveys
I used camera traps and visual censuses to determine presence/absence of
mammals weighing > 1 kg within focal forest patches from January 2006 to August
2008. I avoided sampling during mid-late wet season (mid-September to mid-
December) because of poor camera-trap performance in wet conditions. I deployed
camera traps for a 16-day period in each focal patch. A photograph of a species at any
camera within a patch was considered an indication of presence. I recorded
presence/absence for each species within each patch after every 4-day interval. By
breaking up the 16-day period into 4-day sessions, I created a series of repeat
detection/non-detection data (i.e., a detection history) for modeling detection
probabilities for each species (MacKenzie et al. 2002).
Camera trapping was ineffective for sampling arboreal species and two terrestrial
species (Odocoileus virginianus and Pecari tajacu). In order to document
presence/absence of these species, I performed visual censuses of focal patches
between sunrise and 3 hours after sunrise. Surveys were repeated 5 times within a 2-
week period for each patch, and presence-absence was assessed after each survey to
create a detection history. I surveyed patches by walking along human foot paths and
56
game trails, and by walking through sections without any obvious trails. I walked
approximately 1 km per hour and recorded direct observations of animals, vocalizations,
and well-defined tracks as indications of presence.
I placed more cameras in larger patches, and walked greater distances during
visual censuses in larger patches (Table 3-1). Because of the wide range in patch sizes,
I could not sample with the same intensity per unit area in large and small patches. I
included patch size in assessments of detectability to account for potential biases in
unequal sampling per unit area.
Habitat Measurements: Patch-Scale Variables
For each focal patch, I calculated focal patch size and patch quality (Table 3-2). I
chose patch quality measures that have been found to affect patch use of mid- and
large-sized mammals in other studies (e.g., basal area of fruiting trees, fire disturbance),
or factors that I thought would affect use of patches in my system (e.g., cattle
disturbance, basal area of small trees/saplings). I determined basal area of trees using
the point-centered quarter method (PCQM) along randomly placed transects within
each focal patch. I spaced sample points further apart, and sampled more points, in
larger patches (Table 3-1). In each quadrat of the PCQM sample, I recorded distance to
the nearest stem in 3 distinct size-classes: 0-10-cm diameter-at-breast-height (dbh),
>10-30-cm dbh, and >30-cm dbh. For the two largest size classes, I identified stems to
species level when possible and to family level otherwise. I determined basal area of
fruiting trees for the two largest size classes combined (>10-cm dbh). Fruiting tree
species were those identified as being important to larger mammals in the area based
on published literature (Ponce-Santizo et al. 2006), local knowledge, and personal
observation. I also determined total basal area of small trees and saplings (stems less
57
than 10-cm dbh). I determined presence/absence of cattle within each focal patch by
looking for signs of cattle presence during other sampling activities. I assessed effects
of past fire along the edge and interior of the patch and created a binary fire variable
(0=no fire effects, 1=fire along edge and/or less than 25% of interior) for the analysis.
Patches with greater than 25% of the interior disturbed by fire were not surveyed.
Habitat Measurements: Landscape-Scale Variables
I created a vegetation map for my study site by performing an unsupervised
classification of 2003 Landsat ETM+ images using ERDAS imagine 9.0. The original
unsupervised classification consisted of 40 classes, which I combined into 4 classes:
water, pasture/agriculture, regenerating forest ≤ 15 years old, primary forest/secondary
forest > 15 years old. Primary forest and older secondary forest were combined into one
class because of my inability to separate these classes with Landsat imagery. Classes
were assigned with the aid of ground-truthed points, examination of .5m resolution 2001
aerial photographs of the study site, as well as my knowledge of the study area. I used
an additional 500 ground-truthed points and 200 random points from aerial photographs
to assess classification accuracy of the landcover classes. Overall classification
accuracy was 80.6 %. Forest was classified with an accuracy of 91.3%, pasture at an
accuracy of 80.5 % and regenerating forest at 70.0 %.
I measured habitat loss and fragmentation within buffers of 6 different sizes (500,
1000, 1500, 2000, 2500, 3000 m) from the edge of each focal patch (Fig. 3-1). Within
each buffer, I used FRAGSTATS software (McGarigal et al. 2002) to calculate
proportion of forest in the landscape to represent habitat loss and 4 other metrics to
represent habitat fragmentation: mean nearest-neighbor distance of forest patches,
density of forest/non-forest edge, mean size of forest patches, and density of forest
58
patches (Table 3-2). Substantial correlations among these 4 fragmentation metrics
suggested that variables were redundant. Moreover, fragmentation metrics were highly
correlated with proportion of forest (Table 3-3). I chose density of patches as my
measure of habitat fragmentation because it was uncorrelated with proportion of forest
within all buffers except the 500-m buffer, where there was a moderate correlation
between the variables (r = 0.49). Proportion of forest and density of patches thus served
as largely independent measures of habitat loss and fragmentation, respectively. A
second analysis in which I statistically removed the correlation between proportion of
forest and fragmentation measures resulted in qualitatively similar results to those
presented here. Density of forest patches was correlated negatively with mean nearest
neighbor distance and mean patch size, and positively with density of forest/non forest
edge. This variable thus represents a gradient from landscapes containing a few, larger,
highly isolated patches and a low total amount of forest edge to landscapes containing
many, smaller, less isolated patches and a high total amount of forest edge. I also
calculated proportion of regenerating forest within each buffer as an additional measure
of landscape context that could affect mammal occupancy by providing low quality
habitat between patches (Table 3-2).
Additional Predictor Variables
I calculated distance from the focal patch to the Maya Biosphere Reserve
boundary as a measure of potential effect of a nearby source. I also calculated distance
from the focal patch to the nearest human community as a measure of hunting
pressure. Distance to nearest village is commonly used as an index of hunting pressure
because hunting tends to be more intense closer to villages in the tropics (Parry et al.
2009). I recorded season of sampling (dry vs. early wet) for each patch. Correlation
59
coefficients between all continuous potential predictor variables included in analyses
(patch, landscape, and additional variables) were small (r < 0.5).
Relative Influence of Habitat Loss and Fragmentation
I used detection/non-detection data for the 50 focal forest patches to examine
patch occupancy and detection probabilities for each species using logistic regression in
program PRESENCE (Hines 2006). Because many species in my study were capable
of moving in and out of patches during sampling, the occupancy estimator is best
interpreted as “probability of use” of a patch, rather than probability of occupancy. The
detection parameter is best interpreted as probability of being within the patch and
detected during sampling (MacKenzie et al. 2006). For ease of presentation, I will use
traditional occupancy terminology in this paper.
I built occupancy models for each species by first determining best-fit detection
models holding occupancy constant. I tested two covariates for detection: patch size
and season of survey. The best fit detection model for each species, determined with
AICc (Burnham & Anderson 2002), then was used in all subsequent analyses of
occupancy. I next determined the spatial scale at which each species responded most
strongly to habitat loss (i.e., proportion of forest in the landscape), habitat fragmentation
(i.e., density of forest patches in the landscape), and proportion of regenerating forest in
the landscape by considering each of these landscape-scale variables independently in
occupancy models. I fit these models for each of the 6 buffer size classes, and
determined, for each variable, the best-fit buffer size class (i.e., the best-fit spatial
scale).
I used best-fit detection models and best-fit spatial scale(s) for each species to
examine relative influence of habitat loss and fragmentation on patch occupancy
60
patterns. I fit occupancy models in program PRESENCE, limiting the number of
occupancy covariates in models to 5 because of sample size constraints. Models
included proportion of forest and density of forest patches in the landscape, and 3
patch-scale variables: patch size, basal area of fruiting trees >10 cm dbh, and basal
area of small trees and saplings (< 10 cm dbh).
I standardized all predictor variables prior to entrance in the analysis, so that
resulting coefficients for predictor variables represent the influence on the response of a
one standard deviation change in the predictor, holding all other predictors in the model
constant. Standardized partial coefficients have been found to be one of the best
methods for estimating the influence of habitat loss and fragmentation variables, even
when the variables are highly correlated, which was not the case in my analysis (Smith
et al. 2009). I explicitly included patch-scale variables in the analyses to be confident
that responses attributed to habitat loss or fragmentation in the landscape were not
driven by response to patch size or quality (Koper & Schmiegelow 2006). The relative
influence of habitat loss and fragmentation was compared by examining values of the
standardized predictors. This analysis also enabled us to assess the relative influence
of patch-scale factors on occupancy patterns. Significance of each of the 5 variables
was assessed using a likelihood ratio test of the full model vs. a reduced model without
the variable. For 3 species, I reduced the number of variables considered in modeling
occupancy to 4 variables (Pecari tajacu, Tamandua mexicana) and 2 variables (Procyon
lotor) because 5-variable occupancy models could not be fit without obtaining
unrealistically large parameter estimates.
61
I calculated autocorrelation in the residuals of my occupancy models using
Moran’s I. To calculate residuals, I subtracted the observed occupancy state from the
probability that a species was present and detected at least once according to the
occupancy models (Moore & Swihart 2005). I converted residuals to Pearson’s
residuals, and used the “autocorr” macro in SAS (SAS 2008) to calculate Moran’s I
values. Because focal patches in my study were spread over such a large study area, I
used relatively large lag distances for the calculation of Moran’s I: 0-4 km, >4-8 km, >8-
12 km, <12-16 km, >16-20 km. A Monte Carlo randomization test with 999 permutations
was used to determine the probability of obtaining an I value as large as the observed
value for each species in each lag category. Significance of Moran’s I values were
tested using Bonferonni corrections, where the ith lag was test at 0.05/I (Fortin & Dale
2005). For species found to have autocorrelated residuals based on Moran’s I values (n
= 3), I calculated an index presented in Moore and Swihart (2005):
autocovi = ij
Jj
iijJj
w
ywi
i
1
1
,
where yj = 1 for all occupied patches in a set Ji that are defined as neighbors of patch i,
and wij = 1/hij, where hij was the distance between patches i and j (see Moore & Swihart
2005 for additional description). I added this index as a covariate for those species with
autocorrelated residuals and re-ran the occupancy model. I could not use a more
traditional autocovariate (e.g., Augustin et al. 1996), because I did not sample all
surrounding patches for presence-absence, nor could I estimate occupancy for
unsampled patches. Although not ideal, the index decreased autocorrelation in
residuals for the three species.
62
I performed an additional analysis within a multi-model testing framework to
develop best-fit models with all 11 habitat variables from my study patches (patch,
landscape, and additional variables). This enabled us to examine if the relative influence
of habitat loss vs. fragmentation qualitatively changed when I considered these
variables in combinations with all predictor variables and to test a larger overall set of
variables that I thought could exert an influence on mammal occupancy patterns. I
developed a priori a candidate list of 56 models to test that were constrained to a
maximum of 6 variables influencing occupancy. These models included all univariate
models and more complex models with patch size, habitat loss and fragmentation in
combination with other variables. I calculated AIC weights for variables to provide a
measure of the relative influence of these variables on patch occupancy (Burnham &
Anderson 2002). I assessed overdispersion in the data for each species with a
goodness-of-fit test implemented in program PRESENCE (MacKenzie & Bailey 2004).
When overdispersion was present, I used QAICc to compare models and calculate AIC
weights (Burnham & Anderson 2002).
Results
I photographed 25 species in over 8800 camera-trap nights. I recorded presence
of 6 species of arboreal mammals within forest fragments in 320 km of visual censusing.
After excluding species with a very small number of detections, I could adequately
analyze 20 species from my study area (Table 3-4). Detectability varied greatly among
species (Table 3-4). Patch size and season of survey had effects on detectability for a
subset of species (Table 3-4).
Substantial interspecies variability occurred in the spatial scale of response to
habitat loss and fragmentation (Appendix C). For example, spider monkeys (Ateles
63
geoffroyi) responded most strongly to habitat fragmentation within a 500-m buffer and
tayra (Eira barbara) responded most within a 3000-m buffer. Although some species
responded most strongly to habitat loss and fragmentation at the same spatial scale
(e.g., agouti [Dasyprocta punctata]), others responded at markedly different scales to
these two variables. (e.g., nine-banded armadillo [Dasypus novemcinctus]).
Based on results from the 5-variable models, habitat fragmentation (i.e., density
of patches) had a significant effect (p < 0.05) on occupancy for 6 species, and a
marginally significant effect (p < 0.10) for 2 other species. In contrast, habitat loss (i.e.,
proportion of forest) had a significant effect on occupancy for only 4 species, and a
marginally significant effect for 2 (Table 3-4). Of those species that responded
significantly to habitat fragmentation, 6 responded negatively and 2 positively to
increasing fragmentation in the landscape surrounding focal patches. Three species
responded negatively and 3 positively to increasing habitat loss in the landscape.
According to standardized effect sizes, 10 of 20 mammals responded more strongly to
habitat fragmentation in the landscape than habitat loss (Table 3-4).
Results from analysis of all 11 variables generally support conclusions from the
5-variable models. According to AIC weights (Appendix D), habitat fragmentation was
one of the top two most influential variables for 10 of 20 species, whereas habitat loss
was one of the top two most influential variables for only 5 of 20 species. Habitat
fragmentation appeared in best-fit models for 7 species: spider monkeys, Virginia
opossums (Didelphis virginianus), tayras, margay (Leopardus wiedii), coatis (Nasua
narica), northern raccoons (Procyon lotor), and gray foxes (Urocyon
cinereoargenteous). Habitat fragmentation also appeared in highly competitive models
64
(< ∆1 AICc unit of best fit model) for agoutis. In contrast, habitat amount appeared in
best-fit models for only tayras, spider monkeys, Mexican anteaters (Tamandua
mexicana) and highly competitive models for coatis (Appendix C).
Patch-scale variables exerted a strong influence on species. For the 18 species
that were modeled with all 5 variables, 13 responded more strongly to at least one
patch-scale variable than either habitat loss or fragmentation based on standardized
effect sizes. This response was related largely to the influence of focal patch size.
Eleven species responded significantly to focal patch size (Table 3-4). Focal patch size
also appeared as one of the top two most influential variables according to AIC weights
for 11 of 20 species (Appendix D), and appeared in best-fit models for 10 of 20 species
(Appendix C). Patch quality measures exerted a more limited influence on species, with
basal area of small tree and saplings having a significant negative influence on 2
species (pacas [Agouti paca], and white-tailed deer) and a marginally negative influence
on one other species (tayras) in the 5-variable models. Basal area of large fruiting trees
within the patch had a significant positive effect on one species (collared peccaries;
Table 3-4).
Other factors not considered in the 5-variable model exerted a strong influence
on patch occupancy patterns of a few species (Appendix C and D). Distance to the
boundary of the Maya Biosphere Reserve appeared in best-fit models and had a
negative influence on 1 species (Mexican porcupine [Coendou mexicanus] and a
positive influence on 2 species (stripped hog-nosed skunk [Conepatus semistriatus],
common opossum [Didelphis marsupialis]). Proportion of regenerating forest in the
landscape surrounding patches appeared in the best-fit model for nine-banded
65
armadillo, and cattle presence appeared in the best fit model for coati. Fire disturbance
and distance to nearest community did not appear in any best-fit models and had low
overall AIC weights for all species.
Discussion
In contrast to the dominant paradigm from temperate studies of a relatively weak
effect of habitat fragmentation, tropical mid- and large-sized mammals in northern
Guatemala responded strongly to habitat fragmentation independent of habitat loss. A
larger number of species responded in a significant manner to my measure of habitat
fragmentation (density of patches) than to habitat loss (proportion of forest). The effect
of fragmentation on species was apparent even after accounting for effects of patch-
scale factors on response patterns. My data therefore do not support the hypothesis
developed from temperate studies that habitat loss is substantially more important than
fragmentation. Importantly, the effect of fragmentation on species was negative for 75%
of the mammals when considering only those species that displayed a significant
response to fragmentation. Thus, landscapes with habitat distributed in a larger number
of smaller patches and higher density of edge tended to be less suitable for mammals
than landscapes with habitat distributed in a few, larger patches and a lower density of
edge. These negative effects were apparent for both forest specialists (e.g., spider
monkeys, margays) and surprisingly for several more generalist species (e.g.,
armadillos and gray foxes). Positive effects of fragmentation on patch occupancy
patterns of two species (tayras, anteaters) could be driven by positive response to
edges or to decreased isolation between patches.
The strong response to fragmentation found in my study may be related to the
fact that I analyzed response of mid- and large-sized mammals, instead of smaller
66
species such as birds that have been the focus of the majority of studies to date. These
mammals, with their relatively large area and energetic requirements, may respond
strongly to changes in patch size that occur with increasing fragmentation. Large area
requirements also may be a reason why certain large-bodied bird species respond more
strongly to fragmentation (Villard et al. 1999). However, of the 5 species in my study
with the largest home ranges, only 1 (margay) responded in a negative manner to
fragmentation, casting some doubt on this explanation. An alternative explanation for
mammals responding strongly to fragmentation is related to the location of my study site
in a tropical ecosystem. Habitat fragmentation may be of greater importance in
determining distribution or abundance of species in tropical than in temperate systems
because tropical species tend to show strong responses to edges (Laurance et al. 2002;
Fahrig 2003). A major result of increasing fragmentation is a higher density of edges.
Edge density may exert effects on species by increased exposure to edge-altered
habitat (Laurance et al. 2002), increased time spent in matrix habitat (Fahrig 2003), or
increased exposure to human persecution (Woodroffe & Ginsberg 1998). Additional
support for this idea comes from two other recent studies conducted in tropical areas
with bats and birds (Develey & Metzger 2006; Klingbeil & Willig 2009; but see Zurita &
Bellocq 2010). In both of these cases, the authors found strong relationships (both
positive and negative) between fragmentation per se and species distribution and/or
abundance for a subset of the species analyzed, even though species were not
particularly large or wide-ranging. Collectively, these finding indicate that prevention or
management of habitat fragmentation per se may be an important component of
strategies to preserve some tropical species. This subject warrants increased attention,
67
with more studies focused on both wide-ranging and more sedentary species in tropical
environments. Increased emphasis on studies of wide-ranging taxa in temperate
regions also would help to untangle whether increased sensitivity of species to habitat
fragmentation found in this study is a product of species characteristics or biome of the
study site.
Species in my study responded to habitat loss and fragmentation measured at a
variety of spatial extents. I echo concerns expressed by other authors (Holland et al.
2005) that investigations using only one landscape size or spatial extent to represent
the “landscape-scale” are problematic. Such studies do not account for the fact that
different species may respond most strongly to habitat variables measured at different
spatial extents, or that a given species may respond to one type of landscape metric at
a different scale than another landscape metric (e.g., Boscolo & Metzger 2009).
The strong positive influence of focal patch size on occupancy patterns of
mammals was likely driven by large area requirements of these species and has been
documented elsewhere (Michalski & Peres 2007). Patch quality also was influential in
determining patch occupancy patterns for a subset of species. Fruiting tree basal area
exerted a positive influence on patch occupancy by collared peccaries, suggesting that
this species was preferentially using patches with higher quality or more abundant fruit
resources. Basal area of small trees and sapling within a patch exerted a negative
influence on 3 species, which was counter to my expectations that species would be
more likely to use patches with more cover. Cattle presence did not exert a strong
influence on occupancy for most species, but when this variable appeared in best-fit
models or model sets for species, cattle always exerted a negative influence on
68
occupation probability. Cattle may reduce available browse and cover or disturb resting
or denning sites and have been found to exert a negative influence on other species
inhabiting fragments (Hannah et al. 2007). The lack of an influence of fire disturbance
within patches on mammal distribution contrasts with another recent study (Michalski &
Peres 2007). Unlike the Michalski and Peres (2007) investigation, I specifically excluded
patches from my study that were severely impacted by fire, which likely explains the
discrepancy between the two studies.
Distance of focal patches from continuous forest was an influential factor for only
a small subset of species, even though the presence of sources often is important in
maintaining tropical mammal populations (Novaro et al. 2000). Given the relatively
recent nature of forest loss and fragmentation in my study area, the lack of influence of
a potential nearby source may be related to insufficient time for species to respond fully
to landscape change (i.e., an extinction debt; Tilman et al. 1994). Positive effects of
increasing distance from continuous forest were apparent for species that are
considered habitat generalists or are highly adaptable to human disturbance such as
stripped hog-nosed skunks. Distance to the nearest community, my measure of hunting
pressure, did not strongly influence patch occupancy patterns of mammals, which is
surprising considering the importance of hunting pressure in determining mammal
distributions in the tropics in general (e.g., Peres & Palacios 2007), as well as overall
vulnerability to fragmentation for mammals in this particular landscape (D.H.T.,
unpublished data). However, distance to nearest community perhaps is too coarse a
measure of hunting pressure and was inadequate for capturing variation in hunting
pressure on my landscape. Moreover, because all fragments in my study area were
69
relatively easy to access through roads or trails, hunting pressure was likely high and
fairly equal in intensity across the entire landscape.
Although conservation strategies will continue to focus on prevention of habitat
loss, my results also highlight the importance of increased attention to prevention of
habitat fragmentation per se. To date, managing landscape pattern to prevent
fragmentation or ameliorate the negative effects of habitat loss has been deemed as an
ineffective approach to biodiversity conservation based on studies conducted in
temperate environments on a few taxa (Fahrig 2003). A greater number of studies with
a broader taxonomic and geographic scope are necessary to inform the debate on the
relative influence of habitat loss and fragmentation and for understanding how much of
the pattern observed in this study is driven by the wide-ranging habitats of mid- and
large-sized mammals versus the tropical environment. Other recent studies on birds
and bats in tropical environments suggest that tropical fauna generally may be more
responsive to fragmentation (Develey & Metzger 2006; Klingbeil & Willig 2009). Given
that deforestation and fragmentation of remaining habitat are significant problems inside
and outside protected areas in the tropics (deFries et al. 2005; Broadbent et al. 2008), a
broader understanding of the effects of fragmentation on species distributions and
population persistent is vitally important for conservation.
70
Table 3-1. Sampling effort employed for camera-trapping and visual censusing of mammals in forest patches.
Patch size (ha) # of patches surveyed
# of cameras within patch
Total distance walkeda
# of vegetation sample sitesb
2.9-10 12 7 4.0 8 >10-20 7 10 5.0 13 >20-40 12 14 5.0 16 >40-80 5 17 7.5 19 >80-160 7 20 9.0 22 >160-320 4 25 10.0 25 >320 3 28 10.0 28 a Total distance walked over 5 visits to the patch. b Sampling sites for point-centered-quarter-method
71
Table 3-2. Patch and landscape context metrics for the 50 focal patches in my study area.
Variable Mean (SD) Description Patch-scale
Focal patch size 68.5 (98.7) Area of focal forest patch (ha) Basal area of large fruiting trees
20.0 (10.1) Basal area per ha of all important fruiting tree species within the patch (stems ≥ 10-cm dbh)
Basal area of small trees/saplings
2.9 (1.2) Basal area per ha of all shrubs/saplings within the patch (stems < 10-cm dbh)
Fire N/A Ranking of fire disturbance on 3-point scale
Cattle N/A Presence-absence of cattle in fragment
Landscape-scale* Proportion of forest 0.21 (0.11) Proportion of primary and secondary
forest (15+ years regrowth) in the landscape
Proportion of regenerating forest
0.32 (0.11) Proportion of regenerating forest (< 15 years regrowth) in the landscape
Density of patches 2.6 (0.94) Density of forest patches in the landscape (#/km2)
Mean patch size 8.9 (6.6) Mean size of all forest patches in the landscape (ha)
Mean nearest neighbor distance
0.15 (0.05) Mean Euclidean nearest neighbor distance of all patches in the landscape (km)
Density of edge 0.32 (0.12) Total density of forest/non-forest edge in the landscape (km/ha)
* Means of landscape-scale variables calculated based on 2000-m buffers from the edge of each sampled patch. Buffers of other sizes resulted in different values for these categories.
72
73
Table 3-3. Correlations between landscape metrics used to categorize habitat amount and configuration in the landscape surrounding focal patches. Values listed are for the 2000-m buffer (different buffers have different values for the correlation coefficients, though the relationships between variables are qualitatively similar for all buffers except the 500 m buffer – see text).
PF* DFP MNN DE MPS PF 1.00 -0.08 -0.61 0.72 0.88 DFP 1.00 -0.42 0.47 -0.37 MNN 1.00 -0.74 -0.35 DE 1.00 0.46 MPS 1.00 * PF = Proportion of forest, DFP = Density of forest patches, MNN = Mean nearest neighbor distance of forest patches, DE = Density of forest/non-forest edge, MPS = Mean forest patch size
Table 3-4. Parameter estimates for patch-scale variables and habitat loss and fragmentation variables.
Speciesa Detection modelb
Mean pc FPSd BALFT BASTS PF DFP
Agouti paca constant 0.73 0.92* 0.00 -0.69** 0.41 -0.37 Alouatta pigra fps+sea 0.65 0.63 -0.07 0.08 0.36 0.31 Ateles geoffroyi constant 0.61 1.19** 0.02 -0.52 1.07** -1.50** Coendou mexicanus constant 0.19 0.29 0.08 0.22 1.11* -0.64 Conepatus semistriatus constant 0.50 0.51 -0.11 -0.43 -0.94** -0.42 Dasyprocta punctata sea 0.87 0.62* 0.22 -0.43 0.08 -0.11 Dasypus novemcinctus fps+sea 0.67 0.76 -0.33 0.64 -1.09** -1.02** Didelphis marsupialis fps+sea 0.53 -0.03 0.72 -1.57 -0.30 0.11 Didelphis virginianus constant 0.33 1.35** 0.16 -0.15 -0.23 -0.69 Eira barbara constant 0.29 0.91 -0.25 -1.05* 0.95 1.01* Leopardus pardalis fps 0.30 3.14** 0.01 -0.26 0.36 0.16 Leopardus wiedii constant 0.42 1.81** 0.05 0.54 0.45 -1.16** Mazama americana constant 0.53 1.15** -0.15 -0.03 0.41 0.37 Nasua narica fps 0.49 1.18** 0.59 0.47 0.75 -1.14** Odocoileus virginianus constant 0.31 1.46** 0.51 -1.88** -0.55 -0.01 Pecari tajacu fps 0.29 1.50** 1.33** -- 0.09 0.24 Procyon lotor constant 0.27 -- -- -- -0.52 -1.35** Puma yagouaroundi constant 0.29 2.81* 1.00 0.39 1.16* -0.84 Tamandua mexicana constant 0.31 -0.88 -1.86 -- -1.56** 3.28* Urocyon cinereoargenteus
sea 0.34 1.32 0.57 0.95 1.10 -1.80** a I did not detect two species in patches (Tayassu pecari, Bassariscus sumichrasti). Seven species with sparse data were excluded from analysis (Panthera onca, Puma concolor, Tapirus bairdii, Spilogale putorius, Galictis vitatta, Philander opossum, Potos flavus). b Lists the best-fit detection model for each species using focal patch size (fps) and wet/dry season (sea) as possible covariates. c Mean detection probabilities per survey for each species. d FPS = Focal patch size, BALFT = Basal area of large fruiting trees, BASTS = Basal area of small trees/saplings, PF = Proportion of forest in the landscape (my measure of habitat loss), DFP = Density of forest patches in the landscape (my measure of habitat fragmentation). * p < .10 based on likelihood ratio tests of full vs. reduced model without each variable. **p < .05 based on likelihood ratio tests of full vs. reduced model without each variable
74
Figure 3-1. Map of a portion of the study area. Rectangle in the inset map in left corner
indicates the location of my overall study site in northern Guatemala. Larger map shows a small portion of the total study area. Arrows indicate the location of two focal forest patches where species presence-absence was measured. Buffers used to calculate landscape context variables are shown around the two focal patches. Buffers were measured from the edge of focal forest patches and did not include area of the focal patch in calculation of landscape metrics. Buffers of 2500 and 3000 m are not shown.
75
CHAPTER 4 A REEXAMINATION OF THE INFLUENCE OF LANDSCAPE CONTEXT ON SPECIES
PRESENCE AND ABUNDANCE
Introduction
Identifying conditions under which large-scale factors strongly influence local
processes is a central focus of landscape ecology (Levin 1992; Turner 2005). In
spatially heterogeneous or “patchy” landscapes, studies often focus on determining the
relative influence of small-scale factors such as within-patch (e.g., vegetation structure)
or patch (e.g., patch size and shape) variables, and large-scale factors such as
landscape context (e.g., amount of habitat in the landscape) on the distribution,
abundance, and richness of organisms. Results of such multi-scale studies have been
used to test theories regarding which scales exert the strongest influence on species
(Cushman & McGarigal 2004a), and because the vast majority of these studies are
conducted in anthropogenically fragmented landscapes, results also may be used to
guide management decisions regarding conservation of species in fragmented
environments (Bakker et al. 2002; Banks et al. 2005; Holland & Bennett 2009). In
particular, whether management efforts should consider the landscape surrounding
patches, or instead focus solely on managing patch-scale factors such as patch size or
quality, is an important decision that is directly informed by these types of studies
(Mazerolle & Villard 1999).
Despite the applied conservation significance, we still have a poor understanding
of whether species respond more strongly to small or large-scale heterogeneity in
habitat (Kotliar & Wiens 1990) and little work has focused on the assessment of trends
in response across taxa or regions. The large number of published studies in patchy
landscapes presents opportunities to synthesize information across studies and arrive
76
at a better understanding of species response to landscape- and patch-scale
characteristics. A review of 61 studies in the late 1990s (Mazerolle & Villard 1999)
detected a significant response to landscape context in 59% of studies examined, and
concluded that local environmental conditions alone were inadequate for predicting
species distributions. This effect was especially pronounced for vertebrate taxa, with
fewer studies on invertebrates finding a response to landscape factors. However,
almost all studies in the review found a response of species to patch-scale factors,
which suggested that the influence of landscape context was complementary to that of
patch characteristics.
Although this review served as an excellent first look at variation in species
response to landscape and patch factors, a large number of new studies have been
conducted over the last ten years that may add additional insights and allow for a
broadening of the scope of the analysis. For example, many methodological factors that
may influence whether or not a species responds to landscape context or patch
variables (such as sample size or type of response variable) were not considered in the
previous review and have received little attention in the literature. Such methodological
considerations may be particularly relevant for studies in patchy landscapes given the
diversity of approaches taken to investigate the relative influence of patch- and
landscape-scale variables. The previous review also did not include any studies on
tropical species, which may react differently than temperate species, nor did the authors
separate the effects of patch variables from within-patch variables, even though these
two scales of heterogeneity may influence species for drastically different reasons.
77
I reviewed 125 focal patch studies published from 1998 to 2009 to determine the
probability of a species’ response to landscape context, patch, and within-patch factors.
Focal patch studies were defined as empirical field-based studies that examined
species response (e.g., presence-absence) within discrete focal patches, and then
related that response to characteristics of the focal patches and the surrounding
landscape via statistical analysis. I also determined if the probability of response to
landscape, patch, and within-patch factors were influenced by taxa, tropical/temperate
affinity, and several methodological variables. I compared the results of my review to
those of Mazerolle and Villard (1999), and tested the following predictions:
1) Invertebrates would have a lower probability of responding to landscape
context than vertebrate taxa. Because of small body size and relatively low vagility,
invertebrates may perceive and interact with their landscape at a much smaller-scale
than other taxa (Mech & Zollner 2002). Thus, large-scale factors will not be as important
in determining their distribution or abundance. This prediction tests the generality of one
of the findings of Mazerolle and Villard (1999).
2) Tropical species would be more likely than temperate species to respond to all
3 categories of variables (landscape, patch, and within-patch factors) because of the
greater specialization of tropical species. Highly specialized species may be particularly
sensitive to habitat fragmentation and/or heterogeneity at a variety of scales (Henle et
al. 2004; Devictor et al. 2008).
3a) Studies that measured landscape context by examining the composition
and/or configuration of the landscape in a buffer around the focal patch would be more
likely to detect an influence of landscape context on species than studies that measured
78
landscape context by only considering simple Euclidean distance or connectivity
measures of isolation of focal patches (Fig. 4-1). Modeling studies have shown that
buffer measures of isolation are more highly correlated with animal movement than
distance-based measures, and thus reflect isolation in a more realistic manner (Bender
et al. 2003; Tischendorf et al. 2003).
3b) I further predicted that studies with buffers around focal patches would be
more likely to detect an influence of landscape context on species if they tested multiple
buffers than if they only used one predetermined buffer to measure landscape context.
Recent studies indicate that even closely related species respond to landscape context
at different spatial extents (e.g., Cooper & Walters 2002; Schmidt & Tscharntke 2005;
Boscolo & Metzger 2009). Thus studies that only consider one spatial extent (i.e., one
buffer) may find little response to landscape context for some species because the
wrong spatial extent was used in analysis.
4) Studies using abundance data would be more likely to find a significant effect
of landscape, patch, and within-patch factors. Abundance data should be a more
sensitive measure of species response than presence-absence (Cushman & McGarigal
2004b).
5) Sample size (number of focal patches surveyed) would exert a positive
influence on the probability of finding a significant response of a species to landscape
context, patch, and within-patch variables because of increases in statistical power
(Quinn & Keough 2002).
79
Methods
Literature Search
I used the ISI Web of Science database to search for articles published between
January 1998 and October 2009 with the search terms “patch” and “landscape”. I used
this range of publication dates to prevent overlap of sources with the Mazerolle and
Villard (1999) review. I confined results to articles in the ecology, conservation, biology,
or forestry literature. My search yielded 3830 articles. I then scanned through abstracts
to further narrow my search to only those articles that were field-based, focal patch
studies. For this review, I only included focal patch studies that measured presence-
absence, abundance, or density of species within patches. Focal patch studies
examining species richness or diversity as the response variable were excluded. I only
considered focal patch studies conducted in primarily terrestrial systems. This included
studies of forest, open (e.g., prairie, meadows), and aquatic (e.g. marshes, wetlands,
ponds) habitat patches embedded in anthropogenically fragmented environments.
These antropogenically fragmented environments had a matrix of agricultural, urban, or
managed forest. Several studies (10) conducted in naturally fragmented landscapes,
but where patches could be clearly defined, also were included. I did not include studies
where the patch was an island, and I excluded studies conducted on plants. I also
excluded studies that used Principle Components Analysis to reduce dimensionality of
predictors when it resulted in variables that were not easily interpretable as solely
belonging to within-patch, patch, or landscape context categories. After excluding
articles based on my criteria, I had a set of 125 focal patch studies on 970 species for
use in the final analysis.
80
Collection of Data From Studies
For each article in the final set for review (Appendix 1), I determined if the species
in the study displayed a significant response to landscape context, patch, and within-
patch variables. Individual species in the studies were used as the unit of analysis. The
number of species investigated in a single study ranged from 1 to 62. Landscape
context variables included habitat composition and configuration measured in buffers
around focal patches, and/or distance-based measures of isolation of the focal patch
such as simple Euclidean distance (e.g., distance to the nearest patch, distance to
nearest occupied patch) or connectivity metrics (e.g., Hanski’s connectivity index) (Fig.
4-1). Patch variables were patch size and measures of patch shape, including
perimeter, perimeter/area ratio, and fractal dimension. Within-patch variables were
measures related to patch quality such as vegetation structure, level of disturbance,
availability of food or water, temperature, pH, etc. Most of the studies included in my
review simultaneously evaluated species response to all three sets of variables,
however, a small number of studies (7) did not evaluate, or held constant statistically,
the influence of patch and within-patch factors. Exclusion of these studies does not alter
the results of subsequent analysis, so they are included.
Because authors used a diversity of statistical approaches to evaluate the
influence of landscape, patch, and within-patch factors on species response, I
developed a set of rules for defining a “significant” response. A particular variable was
defined as having a significant influence on species response if it: 1) had a significant
(p< 0.05) univariate or multivariate influence on the response (I chose only multivariate
analyses when both were given), 2) appeared in a best-fit model based on stepwise
multiple regression analysis, 3) appeared in the best fit model, or a highly competitive
81
model when given (within ∆2 AIC or AICc units of the best fit model), based on
information-theoretical approaches.
I used a binomial response variable to indicate whether species showed a
significant response to landscape context, patch, and within-patch factors, considering
each category of variables separately (Table 4-1). For example, if a particular species in
a study responded to landscape context and patch size, but not measures of within-
patch quality, it would be coded as “1” for both response to landscape context and patch
factors, but a “0” for response to within-patch factors. These data then formed the basis
for three separate generalized linear mixed models that evaluated the influence of
covariates on response to landscape context, patch, and within-patch factors (described
below).
Although I recognize that the vote-counting methodology that I employed has
many limitations as a method for synthesizing the results of multiple studies (Hedges &
Olkin 1980, 1985), the data from the studies in my review were not amendable to a
more rigorous meta-analysis of effect sizes for several reasons. First, studies included
in my review often only reported results of significance tests and did not list effect sizes
for parameters. Also, descriptive statistics for predictor variables were usually not
reported, preventing standardization of effect sizes. Authors used a large diversity of
metrics to represent landscape, patch, and within-patch factors, making it difficult to get
estimates of the same metric for all studies. Furthermore, authors often tested variables
in a multiple regression framework. Because effect sizes of variables in multiple
regression depend on other variables included in the analysis, such data are difficult to
use in meta-analysis.
82
Several other factors potentially may have biased my review. Given that I had to
sort through so many studies containing the keywords “patch” and “landscape”, I may
have overlooked several appropriate focal patch studies. I also may have missed
several focal patch studies that did not contain these keywords. A more significant
potential problem is whether my review is biased toward finding effects of landscape
context on species because only those studies finding an effect were published.
However, several factors may serve to ameliorate this bias. First, the vast majority of
studies simultaneously considered smaller-scale variables (within-patch and patch), and
thus were not specifically focused on just examining response to landscape context.
Moreover, the majority of studies (60%) examined the response of multiple species,
which increased the chance that both significant and non-significant results would be
published. I cannot rule out the possibility that some studies may not have been
published that did not find a significant effect for any within-patch, patch, or landscape
variables for all species examined, although I think this unlikely to account for a large
number of studies.
Data Analysis
My review differs from the Mazerolle and Villard (1999) review in that I used the
species, instead of the study, as the unit of analysis. The advantage of my approach is
that not all species in a single study will respond the same to patch and landscape
factors, and thus we do not lose this information by summarizing data at the study level.
The disadvantage is that because I examined the response of multiple species within
individual studies, correlations in the response of species to landscape, patch, or within-
patch variables may be present (e.g., all the species in a particular study may have
been more or less likely to respond to landscape context or patch or within-patch
83
variables because of the selection of certain metrics, or unaccounted for peculiarities of
sampling, analysis, or study system). To account for these correlations, I used
generalized linear mixed models in Proc GLIMMIX (SAS 2008) to examine patterns in
how species responded to landscape, patch, and within-patch variables. GLMMs
provide a flexible framework for analyzing non-normal data when correlations among
observations are present. GLMMs account for these correlations by incorporating a
random effect variable when analyzing the data, which in my analysis was a variable
denoting the citation from which the data were collected. The response variable in each
GLMM was the presence, or lack thereof, of a significant response to landscape
context, patch, and within-patch factors for each species (Table 4-1).
I considered five predictor variables (i.e., fixed effects) as potentially influencing
the probability that a species would display a significant response to landscape, patch,
and within-patch variables (Table 4-2). To test for effects of these variables on the
probability of response, I ran the GLMMs with all predictor variables simultaneously in
the model. I used a residual pseudo-liklihood as the estimation method and assessed
degrees of freedom using the Kenward-Rogers method (Littell et al. 2006). I used F-
tests to determine the overall significance of each of the predictor variables, and Tukey-
Kramer adjustment for multiple comparisons to test for significant differences between
levels of each predictor variable. I also ran the GLMMs without any covariates to get an
estimate of the overall mean probability of species responding to landscape context,
patch, and within-patch variables.
To test whether or not studies employing multiple buffers had a greater chance of
finding a response to landscape context, I restricted my dataset to focal patch studies
84
that used buffers and re-ran the GLMM with response to landscape context as the
dependent variable. Instead of using the predictor variable “type of study” (Table 4-1), I
used a binomial variable that distinguished between studies employing single or multiple
buffers.
Results
Birds were by far the most common study species in the articles of my review,
accounting for almost 69% of the total number of study species (Fig. 4-2). Although this
bias is not as pronounced when examining the percentage of the total number of
studies that focused on birds, birds still account for the largest number of studies,
followed closely by mammals. Herpetofauna were the least common study species, only
accounting for 6% of the total number of study species. In particular, reptiles were
almost completely unstudied and were the focus of only 3 focal patch studies. Species
from temperate climates also accounted for the vast majority of study species (79%)
and an even greater percentage of the total number of studies were located in
temperate areas (88%).
Based on results of GLMMs without covariates, the mean probability of a species
responding to landscape context was 0.56. The mean probability of species response to
patch and within-patch factors was 0.59 and 0.71, respectively. Probability of species
response to either patch or within patch factors was 0.85.
Effects of Predictor Variables
When considering all predictor variables simultaneously in GLMMs, the probability
of a species responding significantly to landscape context was influenced by taxonomic
category of the species, type of study, location of study, and sample size of the study,
but not by type of response variable. The probability of a species responding
85
significantly to patch variables was only influenced by sample size. The probability of a
species responding significantly to within-patch variables was only influenced by type of
response variable (Table 4-3).
Test of Predictions
Contrary to my first prediction, invertebrates did not have a significantly lower
probability of responding to landscape context than most other vertebrate taxa (Fig. 4-
3A). Tukey adjusted multiple comparison tests indicated that invertebrates, birds, and
herpetofauna did not have significantly different probabilities of responding to landscape
context. However, mammals had a significantly higher probability of response to
landscape context than invertebrates and birds (p < 0.01, p = 0.04, respectively).
Contrary to my second prediction, the probability of tropical species responding to
landscape context was less likely than for temperate species (0.42 and 0.59,
respectively). This difference was marginally significant (p = 0.07).
In agreement with my third prediction, studies that measured landscape context
variables within buffers around the focal patch had a higher probability of detecting an
effect of landscape context than studies that only measured landscape context via
isolation measures. Studies measuring landscape context using both buffer and
isolation measures had a similar probability of finding a response to landscape context
as studies measuring landscape context with buffers only, and both of these types of
studies had a significantly greater chance to detect a response to landscape context
than isolation only studies (Fig. 4-3B). However, the second part of my prediction was
not supported by my analysis. When considering only studies that measured landscape
context in buffers, the use of multiple buffers around the focal patch did not increase the
likelihood of detecting a response to landscape context (p = 0.97).
86
Studies using abundance or density were more likely to detect a response to
within-patch factors than those using presence-absence as a response variable (p =
0.04), which supports my fourth prediction. Probabilities of detecting a response to
within-patch variables were 0.81 and 0.60 for abundance or density and presence-
absence studies, respectively. I did not find an effect of response variable on likelihood
of detecting a response to landscape context or patch factors.
In agreement with my fifth prediction, the probability of detecting species
response to landscape context and patch variables was influenced by sample size.
Studies that surveyed a large number of patches (> 60) were more likely to find a
response to landscape context and patch variables than studies based on smaller
sample sizes (Fig. 4-4). The probability of detecting a response to within-patch variables
was not affected by sample size.
Discussion
Although the relative influence of landscape, patch, and within-patch factors varies
greatly across study sites and between species, my synthesis of focal patch studies
revealed several overarching trends. When considered together, over half of the
species included in this review (56%) were influenced significantly by at least one
measure of landscape context. This agrees closely with the results of the early review
by Mazerolle and Villard (1999), which found that 59% of studies detected a response to
landscape context. Moreover, none of the individual taxa analyzed in my review had a
probability of response to landscape context less than 0.4. In agreement with Mazerolle
and Villard (1999), I conclude that landscape context is an important factor influencing
species distribution and abundance, and that models based on local-scale variables
only will be insufficient for many species in predicting distribution or abundance in
87
patchy landscapes. However, in contrast with the earlier review, I did not find that
invertebrates have a lower probability of response to landscape context than
vertebrates. My findings indicate a more complicated picture, with both invertebrates
and birds having a significantly lower probability of response to landscape context than
mammals, and herpetofauna falling between these two extremes.
The relative high probability of mammals responding to landscape context could
be driven by several factors. Mammals generally may perceive and interact with their
landscape at larger scales than birds or invertebrates, and thus be more responsive to
habitat variables measured at landscape scales. In contrast, birds and invertebrates
may be more strongly influenced by habitat variables in their immediate vicinity than by
diffuse variables acting at coarse scales (Cushman & McGarigal 2004a). Another
possibility for the difference in probability of response of birds and mammals is that the
higher mobility of birds may enable them to move freely among fragments in some
cases, (e.g., Fraser & Stutchbury 2004; e.g., Dale et al. 2009; Churchill & Hannon 2010)
resulting in a low response to isolation, which was the most common measure of
landscape context in my review. However, this clearly does not apply to all birds, as
some species are reluctant to cross small gaps and show altered movement behavior in
fragmented landscapes (e.g., Belisle & Desrochers 2002; Gobeil & Villard 2002;
Robertson & Radford 2009). The higher probability of detecting a response of mammals
to landscape context also may indicate that the coarse-scale habitat maps used to
estimate landscape context variables in focal patch studies (usually vegetation maps
derived from aerial or satellite imagery) were better at depicting habitat for mammals
than for birds or invertebrates, and, consequently, were more accurate representations
88
of landscape context for mammals. The inadequacy of using vegetation types to depict
habitat for butterflies is well known (Dennis 2003; Dennis et al. 2006; Vanreusel & Van
Dyck 2007), and butterflies accounted for 59% of the invertebrate species included in
my review. However, whether or not this same problem would apply to birds or other
invertebrates is unknown.
Tropical species were less likely to respond to landscape context than temperate
species, although this difference was only marginally significant. A recent modeling
study has shown that patch isolation metrics (whether distance or buffer based) tend to
have lower predictive power for specialist than generalist species because specialist
species are influenced more strongly by the matrix composition (Tischendorf et al.
2003). Given that most studies in my review used isolation as the sole measure of
landscape context and did not incorporate matrix quality in estimations of isolation,
studies may not have detected the response of these species to landscape context.
Another possibility is that some other unrecorded methodological variable that has an
effect on probability of response to landscape context may have been covarying with
the location of the study in a tropical or temperate environment. A larger number of focal
patch studies conducted in tropical regions are needed to address this issue and gain a
better understanding of the multi-scaled nature of species response to habitat
fragmentation.
Species in all taxa also responded at high rates to patch and within-patch
variables in my review, which also agrees with the results of the Mazerolle and Villard
(1999) review. However, because Mazerolle and Villard did not separate out patch and
within-patch variables, they could not assess differences in how species responded to
89
these two scales of heterogeneity. My review indicates that although the probability of
species response to landscape context and patch variables (patch size and shape) was
similar, species had an especially high probability of response to within-patch variables.
A high likelihood of response to within-patch variables is notable given the increasing
recognition that fragmentation results in not only increased isolation of patches and a
reduction in patch size, but also changes in habitat quality of patches over time (Holland
& Bennett 2009). Long term patch quality changes in fragmented landscapes, such as
edge-related vegetation changes (Laurance et al. 2002), or degradation due to cattle,
logging, or fire can have a marked effect on species presence or abundance (e.g.,
Hannah et al. 2007; Michalski & Peres 2007). The high frequency of response of
species to within-patch variables that I found in my review also highlights the need to
incorporate patch quality in metapopulation models (e.g, Schooley & Branch 2009).
My review of the literature indicates that several methodological variables had
pronounced effects on the probability of detecting a significant response of species to
landscape, patch, and within-patch factors. This finding suggests that focal patch
studies may fail to find an effect at a particular scale because of the approach that was
used, not necessarily because species were not responding at that scale. Studies that
measured landscape variables within buffers around the focal patch had a higher
probability of detecting a response to landscape context than studies that measured
landscape variables by calculating Euclidean distance or connectivity measures of the
focal patch to surrounding patches. Buffer measures of isolation have been found to be
more highly correlated with animal movement in modeling studies than Euclidean
distance or connectivity measures (Bender et al. 2003; Tischendorf et al. 2003) and
90
thus may be better proxies of overall isolation of the focal patch. If this is the case,
studies employing these buffer measures of isolation would be more likely to find that
species respond significantly to landscape context. Studies that employ buffers also are
more likely to analyze a variety of variables related to habitat composition and
configuration that go beyond measures of isolation (e.g., edge density, matrix quality)
that may increase the chance of detecting a response to one or more landscape-scale
variables.
The use of multiple buffers did not appear to have a marked effect on the
probability of detecting species response to landscape context, which was surprising
considering that many studies have found that spatial scale influences the strength of
response of species to landscape variables (e.g., Boscolo & Metzger 2009; e.g., Cooper
& Walters 2002; Schmidt & Tscharntke 2005). Moreover, a recent simulation study
found that the performance of buffer measures as predictors of connectivity were
sensitive to measures of the buffer radius (Moilanen & Nieminen 2002). Thus, all buffers
are not equivalent and studies that use multiple buffers should have a better chance of
detecting a significant response to landscape context. The lack of effect of multiple
buffers on probability of detecting a response to landscape context may indicate that
authors employing single buffers did a good job of identifying the most influential spatial
scale of influence on their particular species. However, authors are unlikely to know
beforehand what scales may be most important, especially given that even closely
related species can respond most strongly to landscape variables measured at
drastically different scales (Kolozsvary & Swihart 1999; Rizkalla & Swihart 2006;
Kadoya et al. 2008; Klingbeil & Willig 2009). For example, over 50% of single buffer
91
studies of birds in my review used either a 1 or 2 km buffer. However, focal patch
studies employing multiple buffers consistently show that birds respond most strongly to
both larger and smaller buffers than this size (Hinsley et al. 1995; Ribic & Sample 2001;
Sallabanks et al. 2006; Renfrew & Ribic 2008; Yamaura et al. 2008). A more likely
explanation for the lack of relationship between use of multiple buffers and detection of
response to landscape context is the binary nature of my measure of response (i.e.,
significant or non-significant), which does not take into account variation in the strength
of response to landscape context. Measures of landscape context within multiple buffers
often are correlated because larger buffers encompass smaller buffers. A significant
response of a species to landscape context therefore may be found at more than one
scale. However, the overall strength of the response could be quite different (Hedges &
Olkin 1980, 1985).
Studies that measured species response in a large number of focal patches (> 60
patches) were more likely to detect effects of landscape context and patch variables on
species presence or abundance than studies that measured response in a smaller
number of patches. Sample size has been found to influence the detection of changes
in species prevalence or population size (Ward et al. 2008; Nielsen et al. 2009) and the
accuracy of large-scale species distribution modeling (Wisz et al. 2008), but to my
knowledge this is the first indication of the influence of sample size on detection of a
response to fragmentation. Although not a surprising finding, as the power of statistical
tests to find a significant effect of a variable is related to sample size (Quinn & Keough
2002), this is rarely discussed as a possible reason for lack of a response of species to
patch or landscape variables in the literature. This may be a significant oversight, as
92
almost 55% of the studies included in this review examined species response in less
than 60 focal patches, my cut-off for “large” sample size. Although I used an arbitrary
cut-off to determine large vs. medium vs. small sample sizes in my analysis, I do not
think this biased my analysis. I re-ran the GLMMs using sample size as a continuous
variable and found a similar result of increasing probability of response to patch and
landscape variables with increasing sample size. For example, the odds of finding a
significant response to landscape and patch factors increased (holding all other
variables in the model constant) by 2.4% and 3.0%, respectively, for every additional 10
focal patches sampled.
Whether or not species response was measured as presence-absence or
abundance had an effect on the probability of detecting a significant response to within-
patch variables, but not patch or landscape context variables. Studies were more likely
to detect a response of species to within-patch factors when abundance or density were
used as the response variable. My findings support a recent community-level study on
birds that found that abundance data explained more of the total variation in species-
environment relationships than did presence-absence data (Cushman & McGarigal
2004b), particularly when looking at the relationship between species response and
small-scale heterogeneity in vegetation among plots.
Focal patch studies in my review employed a variety of approaches to analyze
data and reach conclusions regarding the importance of landscape, patch, and within-
patch factors. However, I noted several common analytical problems that could easily
be addressed in future studies of this kind. Only 20 of the 125 studies in my review
tested for spatial autocorrelation in their datasets (i.e., the residuals of regression
93
models). Autocorrelation is problematic for classic statistical test such as regression that
rely on independently distributed errors (Legendre 1993) and may lead to erroneous
conclusions regarding the significance of covariates in studies of species-environment
relationships (Lichstein et al. 2002; Christman 2008). A large separation distance
between focal patches may have led authors to conclude that there was no need to
estimate spatial autocorrelation and may explain the low number of studies testing for
autocorrelation. I cannot address this issue because most studies did not provide
information on separation distances of patches. However, even study sites that are far
apart (e.g., focal patches with non-overlapping buffers) may still show autocorrelation
(Lichstein et al. 2002; Schooley 2006).
Forty of the 125 studies in my review did not measure correlations between
predictor variables (landscape context, patch, and within-patch factors), although the
problem of multicollinearity for regression models in ecology and conservation biology
has been acknowledged for quite some time (Mac Nally 2000). Correlations between
predictors in multi-scaled studies of species-environment relationships are likely
commonplace and can lead to incorrect conclusion regarding the importance of
predictors (Koper & Schmiegelow 2006).
Finally, only 34% of studies that employed a buffer approach to measuring
landscape context variables used multiple buffers. Although my own analysis showed
that use of single vs. multiple buffers does not have a major impact on the probability of
finding a significant response to landscape context, use of single vs. multiple buffers
may have a major effect on the strength of the response to landscape context (i.e.,
overall effect size). I thus list this as a potential deficiency in the analysis of focal patch
94
studies, with the caveat that more work must be done to determine the utility of multiple
vs. single buffers in analyzing the effect of landscape context on species.
My review is limited by the use of a vote-counting procedure to synthesize
results, which cannot account for differences in power between studies to find
significant effects and ignores information on effect sizes. The fact that sample size had
a significant effect on probability of detecting response to landscape and patch factors
indicates the importance of synthesizing information across studies in a more rigorous
manner. Meta-analysis of effect sizes would provide a better way to combine
information from multiple studies on the relative influence of landscape, patch, and
within-patch factors (Osenberg et al. 1999a). If data could be published in a form more
amenable to meta-analysis, this would greatly aid future reviews. For example, focal
patch studies could be commonly published with standardized effect sizes, instead of
just significance tests of no effect, or with raw data available in appendices for a re-
analysis of datasets in a form more amenable to meta-analytical approaches (Osenberg
et al. 1999b). Additionally, more work could be done to narrow down the diverse set of
landscape metrics (e.g., Cushman et al. 2008) used to just a few that are of primary
relevance and that could be applied repeatedly across studies and ecosystems. My
review suggests issues that would be interesting to address in a meta-analytic
framework, such as a comparison of effect sizes of certain landscape variables when
assessed in single or multiple buffers or a comparison of effect sizes of patch isolation
when isolation is measured in different ways.
The literature on species response to patchiness and fragmentation already is so
large and diverse that substantial opportunities exist to synthesize information in order
95
96
to address particular questions that cannot be resolved with a single study and to look
for general trends in species response across ecosystems (Lindenmayer & Fischer
2006). My review revealed that the probability of species responding to landscape
context, as well as patch and within-patch factors, was influenced by a variety of
methodological aspects of the studies. Such study design issues are rarely discussed
by authors as reasons why a particular study did not find an effect, but should be given
more consideration because of their relatively large influence. My study also confirmed
the general importance of the multi-scaled approach in focal patch studies, as species
responded at high rates to heterogeneity measured at the landscape, patch, and within-
patch scales. Landscape context was generally important in determining distribution or
abundance of species within focal patches, and conservation efforts in patchy
landscape will need to consider characteristics of the surrounding landscape, or risk
missing important influences for approximately half the species in a given system.
Landscape context appears to be especially important for mammals, and whether this is
driven by differences in scales of perception or movement behavior of taxa, or the
accuracy of habitat maps (or other methodological differences) in studies of mammals
vs. other taxa remains unknown. My review also indicates that more multi-scaled
studies of species response to fragmentation or spatial heterogeneity should be done
that focus on herpetofauna, as well as tropical species of all taxa, given the under
representation of these groups in my review.
Table 4-1. Example table used for coding species response to landscape context, patch, and within-patch variables. These data form the basis of three separate general linear mixed models.
Citationa Species Common name Significant response to landscape contextb
Significant response to patch variables
Significant response to within-patch variables
1 Andropadus latirostris
Yellow-whiskered Greenbul
1 0 1
1 Andropadus virens
Little Greenbul 0 0 1
1 Aviceda cuculoides
African Cuckoo-Hawk 1 0 1
1 Baeopogon clamans
Sjostedt's Greenbul 1 1 1
2 Mniotilta varia Black-and-white Warbler 0 0 0 2 Momotus
momota Blue-crowned Motmot 0 1 0
2 Myiozetetes similis
Social Flycatcher 1 1 0
3 Climacteris affinis
White-browed Treecreeper
0 1 1
4 Parus major Great Tit 1 1 1 4 Poecile
montana Willow Tit 1 0 1
4 Sitta europaea Eurasian Nuthatch 0 1 1 a Citation identifies the reference used to obtain the data for each species. This variable was used as the random effect in generalized linear mixed models to account for correlations present in the dataset because I recorded the response of multiple species from single studies (e.g., citations 1, 2, and 4 in the table).b A “1” indicates that a species was found to respond significantly to landscape context (or patch/within patch factors), whereas a “0” indicates no significant response found.
97
Table 4-2. Covariates used in the generalized linear mixed models. Covariates Categories Description
1) Birds (B) Class Aves 2) Mammals (M) Class Mammalia 3) Herpetofauna (H) Includes Class Amphibia and
Reptilia
Taxonomic category of study species
4) Invertebrates (I) Includes Class Insecta, Arachnida, and Gastropoda
Type of study 1) Focal patch buffer study (FPBS)
Landscape context variables were measured in buffers around the focal patch (distances of buffers varied between studies) and included composition and configuration metrics
2) Focal patch buffer and isolation study (FPBIS)
Landscape context variables were measured in buffers around the focal patch, and also included isolation metrics from the focal patch (Euclidean distances or connectivity measures)
3) Focal patch isolation study (FPIS)
Landscape context variables were only isolation metrics from the focal patch (Euclidean distances or connectivity measures)
Tropical/temperate affinity
1) Tropical (TROP) Studies conducted in tropical regions
2) Temperate (TEMP) Studies conducted in temperate regions
Sample size 1) Small (SM) Studies that examined 30 or less patches
2) Medium (MD) Studies that examined between 31 and 60 patches
3) Large (LG) Studies that examined greater than 60 patches
Response variable* 1) Presence-Absence (P) Species presence-absence was assessed within patches
2) Abundance/Density (A)
Species abundance or density was assessed within patches
*For studies that measured and analyzed both abundance/density and presence-absence of the same species within patches, I randomly chose one of the response variables to consider in the review for each species and excluded the other
98
Table 4-3. Influence of predictor variables on probability of response to landscape context, patch, and within-patch variables. Values listed based on three separate generalized linear mixed models.
Variable
Landscape context F-value P-value
Patch F-value P-value
Within-patch F-value P-value
Taxonomic category of study species
4.39 >0.01 0.52 0.67 1.04 0.38
Type of study 5.58 >0.01 2.18 0.12 0.06 0.94 Tropical/temperate affinity 3.62 0.06 0.02 0.90 1.90 0.17 Sample size 3.53 0.04 3.61 0.03 1.13 0.33 Response variable 0.51 0.48 0.20 0.65 4.41 0.04
99
Figure 4-1. Example of typical focal patch study. Forest patches are in gray, and matrix habitat (which may or may not be homogenous) in white. Animal presence-absence, abundance or density is recorded within the focal patch, along with within-patch variables such as vegetation structure or level of disturbance. Size and/or shape of the focal patch is measured, as well as landscape context variables. Landscape context variables may be based on isolation measures such as distance to the nearest patch (A in figure), distance to nearest occupied patch (B in figure), distance to nearest source, or connectivity measures (e.g., Hanski’s connectivity index). Landscape context variables also may be based on composition and/or configuration of habitat in the landscape surrounding the focal patch measured within buffers of various distances from the focal patch (C and D in figure). Studies may just use one buffer to determine composition/configuration, or may use multiple buffers at different distances and chose the buffer that provides the most explanatory power. Statistical test are used to determine which factors (within-patch, patch, and landscape context variables) exert a significant effect on species presence or abundance.
100
Figure 4-2. Characteristics of the studies, and study species, included in the review. Data are organized according to categories of predictor variables tested in the generalized linear mixed models. Because single studies sometimes included multiple species, the proportions of species and studies represented by each category differ somewhat. Abbreviations of predictor variable categories are given in Table 1.
101
Figure 4-3. Comparison of parameter estimates for least squares means of probability
of response to landscape context. A) Taxonomic category as the predictor variable B) Type of study as predictor variable. The y-axis provides the least squares mean estimate of the response on the logit scale. Error bars represent 95 % confidence intervals around the mean. Numbers in parentheses have been converted from the logit scale to probabilities and indicate the mean probability that a species in each category would respond significantly to landscape context. Means with different letters above the error bars are significantly different.
102
103
Figure 4-4. Comparison of parameter estimates for least squares means of probability
of response to landscape context and patch variables based on sample size (i.e., # of patches surveyed). A) Landscape context as the response B) Patch variables as the response. The y-axis provides the least squares mean estimate of the response on the logit scale. Error bars represent 95 % confidence intervals around the mean. Numbers in parentheses have been converted from the logit scale to probabilities and indicate the mean probability that a species in each category would respond significantly to landscape context. Means with different letters above the error bars are significantly different.
CHAPTER 5 CONCLUSION
General Conclusions
My study demonstrates that species traits were important determinants of mammal
response to fragmentation. Body size, home range size, and hunting pressure were
related to occupancy rates, but after controlling for passive sampling effects only
hunting pressure strongly influenced vulnerability to fragmentation. Heavily hunted
species were much less common in forest patches than in continuous forest sites.
Given the ubiquity of hunting in tropical environments, my work suggests that
management efforts in fragmented landscapes that do not account for hunting pressure
may be ineffective in conserving tropical mammals. My work also shows that the way in
which vulnerability to fragmentation is measured, and in particular whether or not
passive sampling effects are accounted for in the analysis, may alter conclusions
regarding the relative influence of species traits on sensitivity to fragmentation. My
cross-landscape analysis for sites in Mexico and Guatemala indicates that species traits
may be useful in predicting relative patch occupancy rates and/or vulnerability to
fragmentation across distinct landscapes, but that caution must be used as certain traits
can become more or less influential on different landscapes, even when considering the
same set of species.
Characteristics of patches and landscapes also heavily influenced mammal
response to fragmentation. Mammal use of forest fragments was heavily influenced by
small-scale variables such as patch size and quality, as well as large-scale variables
such as loss and fragmentation of habitat in the surrounding landscape. My finding that
habitat fragmentation in the landscape exerted an equal or greater influence on
104
mammal occupancy patterns as habitat loss runs counter to the current paradigm
developed from temperate studies. My results point to the need to go beyond habitat
preservation and also consider prevention of habitat fragmentation per se, at least for
tropical mammals.
When broadening my analysis and considering a larger collection of
fragmentation literature, my study confirms the general importance of the multi-scaled
approach in focal patch studies. Species from several different taxa responded at high
rates to variables measured at the landscape, patch, and within-patch scales.
Landscape context was generally important in determining distribution or abundance of
species within focal patches, and conservation efforts in patchy landscape will need to
consider characteristics of the surrounding landscape, or risk missing important
influences for approximately half the species in a given system. My data also suggest
that methodological variables, such as sample size or type of landscape metric, could
heavily influence the findings of any particular study. Such variables must be considered
more frequently in the literature as potential confounding factors for why certain species
or groups of species do or do not exhibit responses to within-patch, patch and
landscape factors.
Conservation Recommendations
Given that hunted mammals are highly vulnerable to fragmentation in northern
Guatemala, an increased emphasis by government agencies on reduction of hunting
and enforcement of existing hunting laws in fragmented landscapes of Guatemala
would be highly beneficial to the continued persistence of a number of mammal species
in forest fragments. Hunting for personal consumption or for bush meat markets in
northern Guatemala is quite common (Baur 1998; Polisar et al. 1998; McNab 1999).
105
However, restriction of hunting through enforcement, sustainability education, or
provision of alternative protein has not been particularly successful in northern
Guatemala, even in well-protected intact forest of the Maya Biosphere Reserve (R.
McNab, personal communication). Reduction of hunting would likely be even more
difficult in privately owned, fragmented landscapes. An alternative approach would be to
encourage individual landholders that own forest patches to try and limit hunting on their
property, as long as the risks and costs associated with active patrolling of forest
fragments is not too high. Several landholders in my study site already have begun to
do this as they try to develop successful ecotourism operations.
My finding of the strong negative effect of habitat fragmentation per se on
mammal use of forest patches suggests that in additional to the preservation of as much
habitat as possible, management efforts should encourage the retention of habitat in
fewer, larger patches of forest habitat, rather than in a greater number of smaller
patches. In privately held lands in Guatemala, this could be achieved, in part, through
programs such as PINFOR (Programa de Incentivos Forestales), which pays private
landowners for the preservation and maintenance of primary rainforest patches or for
reforestation efforts. Through this program or similar government incentives,
landowners (solely, or in cooperation with surrounding landholders) could be
encouraged to maintain patches of forest habitat, or to reforest, in a manner that
maintains forest cohesion across the entire landscape. However, only a small number of
landholders that maintain primary forest patches on their property take advantage of
PINFOR (e.g., only 9 of 50 landholders in my study were enrolled in PINFOR at the time
of my study), which may limit the effectiveness of this program in addressing
106
conservation issues on private lands at the landscape-scale, at least until it is more
widely adopted.
Several patch quality factors influenced mammal use of forest patches, and their
negative influence potentially could be limited through better management of forest
patches. Cattle presence in patches exerted a negative influence on a subset of
mammals in our study. Preventing cattle from seeking shelter within forest fragments
through fencing or other means would be an easy way to increase the value of forest
patches for mammals. This approach may have the added benefit to landholders of
reducing conflicts with large carnivores (de Azevedo & Murray 2007), which still occurs
frequently in northern Guatemala. Low basal area of fruiting trees was detrimental for
one mammal in our study, as well as a large ground dwelling bird (Crax rubra,
unpublished data). Selective logging that frequently occurs in forest patches in
Guatemala therefore may be detrimental to some species and could be reduced
through programs such as PINFOR that pay landholders to maintain forest patches in
their current state. Although we did not find an influence of limited fire disturbance within
patches on mammal occupancy patterns, we could not address the detrimental
influence of more widespread fire within patches, as we did not sample any patches that
had been burned over more than 25% of their area. Anecdotal evidence from our study
area suggests that patches with extensive fire damage have completely altered
vegetation structure. Moreover, these patches have a large standing crop of dead trees,
and therefore will likely disappear over time. Fire frequently enters into forest patches
during the dry season, and efforts to reduce impacts of fire on forest patches, such as
107
fire breaks, may be one of the most important steps to the long-term maintenance of
forest cover in the fragmented zones of northern Guatemala.
Although detailed accounts of how each individual species responded to
fragmentation is not discussed in this dissertation, several species-level patterns are of
particular note from an applied conservation perspective. Margays were detected in a
majority of the patches studied, including several small (e.g., 5 ha) and isolated
patches, and had similar occupancy rates in large intact and fragmented forest sites.
Although the arboreal margay is commonly referred to as being highly vulnerable to
habitat loss and fragmentation, little information exists on its ecology in intact or
fragmented habitats (Sunquist & Sunquist 2002). Limited data suggest that it is absent
from highly altered habitats such as shade coffee, pasture, or disturbed fragments as
well as heavily settled areas (Estrada et al. 1994, Daily et al 2003, Urquiza-Haas et al
2009). My data suggest the margay is quite capable of persisting in forest patches of
northern Guatemala (at least in the short-term), and hence may be more adaptable to
disturbance than previously thought. Tayras were similarly resilient to fragmentation,
even though this species also is thought to be heavily forest dependent and highly
arboreal.
Jaguars only occupied 3 patches, all of which were located close to the reserve
boundary. This is true even though I sampled several large patches located far away
from the Maya Biosphere Reserve. Two other studies of relatively large, isolated forest
fragments in other parts of northern Guatemala also have failed to detect the presence
of jaguars (López et al. 2008), which indicates that jaguars may have difficulty moving
or dispersing across fragmented landscapes. This is of concern given the emphasis
108
being placed on connecting jaguar populations throughout their range (Salom-Pérez et
al. 2007; Rabinowitz & Zeller 2010). Many of the proposed jaguar corridors in
Guatemala, Mexico and Honduras cross highly fragmented landscapes ( Rabinowitz &
Zeller 2010) that may be very difficult for jaguars to utilize. Increased attention to jaguar
ecology in fragmented habitats may help to achieve a better understanding of the
usefulness of fragmented landscapes to serve as corridors for this wide-ranging
carnivore.
Two species of conservation concern, white-lipped peccaries and Baird’s tapir,
were extremely uncommon in my study area. White-lipped peccaries were the only
species completely absent from forest patches in my landscape. Even though this
species also was rare in intact forest, and thus did not appear to be overly vulnerable to
fragmentation according to my analysis, the scale of my study was likely inadequate to
assess the response of this species to fragmentation. The vulnerability of this species to
hunting (Peres 1996; Cullen Jr. et al. 2000), and its unique ecology (living in very large
social groups and ranging over large areas) suggest that it will not be able to persist
within, or move through fragmented landscapes. Baird’s tapirs were only present within
one forest patch, and were much more common in intact forest.
Current management efforts by NGOs and government agencies within the intact
forest of the Maya Biosphere Reserve that emphasize work on jaguar, white-lipped
peccaries, and tapir, are well-focused given that these species struggle greatly to use
fragmented landscapes outside of the reserve and are species of local and regional
conservation concern. I would suggest that other species, such as puma, white-tailed
deer, red brocket deer, and spider monkeys should perhaps receive more attention at
109
110
the local level, as these species also are highly vulnerable to fragmentation in northern
Guatemala. These species are important for other reasons, such as their roles as top
carnivores (puma), prey species for large carnivores and people (white-tailed deer and
red brocket deer), seed dispersers (spider monkeys), or as ecotourism attractions for
private enterprises located outside the reserve.
APPENDIX A DETAILS OF COMPILATION OF ECOLOGICAL AND LIFE HISTORY TRAITS
Body mass: We determined body mass as the average of male and female body
mass. When only a range was given, we took the midpoint of the range. Body mass
values were taken from Reid (1997) for all non-felid mammals. Values for the 5 cat
species were taken from Sunquist and Sunquist (2002).
Home range size: We determined home range size from field guides (Walker
1999, Sunquist and Sunquist 2002) and recently published telemetry studies (Sanchez-
Rojas et al. 1997; Saenz & Vaughan 1998; Beck-King & von Helversen 1999; Carrillo et
al. 2000; Foerster & Vaughan 2002; Maffei & Taber 2003; Michalski & Peres 2007;
Aliaga-Rossel et al. 2008; Dillon & Kelly 2008). We took the average of male and female
ranges and the average of wet and dry ranges when listed separately. When multiple
home range sizes were given for a species, we preferentially used ranges from adjacent
forest in Belize and Mexico or tropical forest environments in general when available.
For species that only had home range estimates from North America (Urocyon
cinereoargenteus, Dasypus novemcinctus), we used data from the most southernly
range. For species that did not have a home range listed (Conepatus semistriatus,
Coendou mexicana), we took home range estimates from closely related species.
Trophic level: We categorized trophic levels as follows based on dietary habits
listed in Reid (1997) and Sunquist and Sunquist (2002): 1 = primarily browser/grazer or
frugivore, 2 = omnivore, 3 = primarily carnivore/myrmecophage.
Reproductive rate: We calculated reproductive rate as the number of young
produced per year. This was determined by multiplying the average litter size by
average number of litters per year. Data were taken from field guides (Emmons & Feer
111
1997; Reid 1997; Nowak 1999; Sunquist & Sunquist 2002). When data from seasonal
and non-seasonal environments were listed, we used data from seasonal areas
because of the marked dry season in Guatemala. When not specified, we assumed one
litter per year. When multiple litter sizes were listed, we only used data from wild
populations and averaged listed values. When data were not available for a species
(e.g., Alouatta pigra), we used data from closely related species. For those species with
data from many regions, we used information preferentially from southern/tropical
climates.
Dietary breadth: Dietary breadth was calculated as the number of categories of
different prey types eaten by each species. When we could determine relative
contributions of prey items, we only counted items that made up at least 5% of the diet.
Data were taken from Reid (1997), Emmons and Feer (1997) and Sunquist and
Sunquist (2002). Dietary categories included in the analysis were: grass, browse,
crustacean/fish, insects/arthropods, hard mast, soft mast, small mammals, large
mammals, reptiles/amphibians, birds, carrion, and domestic crops.
Habitat breadth: We calculated habitat breadth based on information on general
habitat preferences given in Reid (1997), Emmons and Feer (1997), and Sunquist and
Sunquist (2002). For additional information on use of secondary and open habitats, we
also consulted additional literature (Reyna-Hurtado & Tanner 2005; Michalski et al.
2006; Parry et al. 2007). Habitat categories were limited to those commonly
encountered in northern Gautemala: Tropical/sub-tropical forest,
secondary/regenerating forest, marsh/wet savanna, dry savanna/grassland, pasture,
cropland, and urban environments.
112
113
Hunting vulnerability: Hunting vulnerability was categorized as follows: 1 =
rarely/never hunted or killed, 2 = occasionally hunted or killed, but not a preferred game
species or actively persecuted species, 3 = often hunted or killed – a preferred game
species (e.g., Mazama americana) or actively persecuted species (e.g., Panthera onca)
species. Rankings of the species were determined independently by 2 biologists from
the area with a long history of work in the region (Garcia-Anleu, personal
communication, Radachowsky, personal communication) and one former hunter
(Cordova, personal communication). The median value from the 3 rankings was used
as the estimate of hunting vulnerability for this study. Anecdotal data from 2 years in the
field by the PI support this general ranking of species. Furthermore, game species listed
as highly vulnerable in our study were shown to be the most highly preferred game
species in two studies of hunting off-takes of local communities within the Maya
Biosphere Reserve (Baur 1998; Polisar et al. 1998). Although hunting vulnerability is not
an intrinsic biological characteristic of a species, but instead a product of region-specific
hunter preference, we feel its inclusion as a “species trait” is appropriate. The likelihood
of a species being hunted or persecuted is affected by several intrinsic characteristics,
such as body size, mobility, escape characteristics, trophic level, and palatability.
APPENDIX B PARAMETER ESTIMATES FOR OCCUPANCY MODELS
These models were used to calculate the proportion of fragments occupied and
the difference in occupancy between intact and fragmented forest sites. SZ = size of the
site or patch, FR = forest patch vs intact forest status (0 = forest patch, 1 = intact forest
site), p = probability of detection, ψ = probability of occupation
Species Model Agouti paca p(0.96)ψ(0.79+0.74*SZ+1.82*FR) Alouatta pigraa p(0.78+0.84*SZ)ψ(1.59+0.83*SZ) Ateles geoffroyia p(0.43)ψ(-0.90+.90*SZ) Coendou mexicana p(-1.54)ψ(0.33+0.66*SZ-1.30*FR) Conepatus semistriatus p(-0.41-1.10*FR)ψ(0.44+.42*SZ+0.00*FR) Dasyprocta punctata p(1.0)ψ(-0.23+0.47*SZ+2.77*FR) Dasypus novemcinctus p(1.13+0.42*SZ-0.77*FR)ψ(0.89+0.38*SZ-0.04*FR) Didelphis marsupialis p(0.37+0.34*SZ)ψ(1.18-0.06*SZ-1.47*FR) Didelphis virginianus p(-0.33) ψ(-0.79+0.47*SZ+0.69*FR) Eira barbara p(-1.53+0.61*SZ) ψ(-0.28+0.26*SZ-1.40*FR) Leopardus pardalis p(-0.89+0.17*SZ) ψ(0.50+2.95*SZ+1.08*FR) Leopardus wiedii p(-0.31) ψ(0.46+1.09*SZ-0.47*FR) Mazama americana p(0.25) ψ(-0.83+1.22*SZ+2.27*FR) Nasua narica p(0.09+0.68*SZ) ψ(-0.13+0.84*SZ+0.31*FR) Odocoileus virginianus p(-0.87-0.01*SZ+0.07*FR) ψ(-1.11+0.95*SZ+3.92*FR) Panthera onca p(-1.22) ψ(-2.95+1.07*SZ+1.94*FR) Pecari tajacu p(-0.76) ψ(-0.19+0.92*SZ+2.33*FR) Potos flavus p(-0.51-0.86*FR) ψ(3.51+0.34*SZ-2.40*FR) Procyon lotor p(-1.08-1.71*FR) ψ(0.97-0.78*SZ-1.62*FR) Puma concolor p(-1.58+0.64*FR) ψ(-3.00+2.90*SZ+5.25*FR) Puma yagouaroundi p(-1.13) ψ(-0.57+0.99*SZ-1.40*FR) Tamandua mexicana p(-0.90+0.40*SZ) ψ(2.06-0.57*SZ-4.17*FR) Tapirus bairdii p(-0.62) ψ(-4.89+1.07*SZ+4.47*FR) Tayassu pecarib Urocyon cinereoargenteus p(-0.13) ψ(-0.17+0.00*SZ+0.72*FR) a A. pigra and A. geoffroyi had 100% occupancy in intact forest, and therefore the FR variable could not be included in the occupancy models without obtaining unrealistically large parameter and standard error estimates. We fit models without this variable, and assumed 100% occupancy in intact forest sites for the purposes of calculating estimates of vulnerability. b Occupancy models for T. pecari could not be calculated because it was not detected in any patches, and only detected in 1 intact 100 ha forest site. We used a value of 0.0 for proportion of patches occupied, 0.0 for difference between occupancy in intact sites and forest patches of 20 ha, and 0.17 (=1 occupied site/6 total 100 ha sites) for the difference between occupancy in intact and forest sites of 100 ha.
114
APPENDIX C BEST-FT MODEL SETS
Models shown are all models that fall within ∆ 2 AICc or ∆ 2 QAICc units of the
best-fit model. AIC weights are rescaled based only on models that fall within ∆2 AICc
units of the best-fit model. # Par = # of parameters in the model, FPS = size of the
sampled patch, BALFT = basal area of large fruiting trees, BASTS = basal area of small
trees and saplings, CAT = presence of cattle within patch (1/0 variable), FIRE =
evidence of fire within patch (1/0 variable), PF = proportion of forest in the landscape
(our measure of habitat loss), DFP = density of forest patches in the landscape (our
measure of habitat fragmentation), PREG = proportion of regenerating forest in the
landscape, DISTC = distance to nearest community, DISTS = distance to source
(boundary of the Maya Biosphere Reserve), SEA = season of occupancy survey (1/0
variable, wet season = 1), AUTO = autocovariate index (see text for explanation). Buffer
size lists the buffer size (in meters) at which each species responded most strongly to
proportion of forest, density of patches, and proportion of regenerating forest in the
landscape surrounding focal patches (see text for explanation). These were the scales
used in the occupancy modeling for each species.
Model #Par ∆ AICc/ QAICc weight
Buffer size PF
Buffer size DFP
Buffer size
PREG Agouti paca 3000 500 1500
p(.)ψ(-BASTS) 3 0.00 0.18 p(.)ψ(-CAT) 3 0.50 0.14 p(.)ψ(FPS) 3 0.55 0.13 p(.)ψ(.) 2 0.65 0.13 p(.)ψ(FPS+SEA) 4 1.26 0.09 p(.)ψ(SEA) 3 1.34 0.09 p(.)ψ(PREG) 3 1.35 0.09 p(.)ψ(FPS+PREG) 4 1.38 0.09 p(.)ψ(PF) 3 1.90 0.07
115
Model #Par ∆ AICc/ QAICc weight
Buffer size PF
Buffer size DFP
Buffer size
PREG Alouatta pigra 2000 500 500
p(fps+sea)ψ(.) 4 0.00 0.25 p(fps+sea)ψ(PF) 5 1.09 0.15 p(fps+sea)ψ(-CAT) 5 1.10 0.15 p(fps+sea)ψ(FPS) 5 1.44 0.12 p(fps+sea)ψ(DFP) 5 1.56 0.12 p(fps+sea)ψ(DISTS) 5 1.58 0.11
Ateles geoffroyi 500 500 500 p(.)ψ(FPS+PF-DFP+AUTO) 6 0.00 0.54 p(.)ψ(FPS+PF-PREG-DFP+AUTO) 7 1.48 0.26
p(.)ψ(FPS-DFP) 5 1.90 0.21 Coendou mexicanus 3000 2000 500
p(.)ψ(FPS -DISTS) 4 0.00 1.00 Conepatus semistriatus 3000 500 3000
p(.)ψ(DISTS) 3 0.00 0.17 p(.)ψ(FPS+DISTS-DISTC) 5 0.38 0.14 p(.)ψ(FPS+DISTS) 4 0.75 0.12 p(.)ψ(-PF) 3 1.04 0.10 p(.)ψ(FPS -PF) 4 1.05 0.10 p(.)ψ(-DFP+DISTS-DISTC) 4 1.13 0.10 p(.)ψ(FPS-PF-DFP) 5 1.21 0.10
Dasyprocta punctata 500 500 1500 p(sea)ψ(FPS) 4 0.00 0.22 p(sea)ψ(FPS-DFP) 5 0.48 0.18 p(sea)ψ(.) 3 0.87 0.15 p(sea)ψ(-BASTS) 4 1.19 0.12 p(sea)ψ(-CAT) 4 1.30 0.12 p(sea)ψ(-DISTC) 4 1.37 0.11 p(sea)ψ(-DFP) 4 1.60 0.10
Dasypus novemcinctus 500 2000 2000 p(fps+sea)ψ(-PREG+SEA) 6 0.00 1.00
Didelphis marsupialis 1000 1000 500 p(fps-sea)ψ(BALFT-BASTS+DISTS)
7 0.00 0.51
p(fps-sea)ψ(DISTC+DISTS-SEA)
7 1.27 0.27
p(fps-sea)ψ(DISTS) 5 1.68 0.22 Didelphis virginianus 3000 1000 1000
p(.)ψ(FPS-DFP) 4 0.00 0.54 p(.)ψ(DISTS) 3 1.66 0.23 p(.)ψ(FPS) 3 1.70 0.23
116
Model #Par ∆ AICc/ QAICc weight
Buffer size PF
Buffer size DFP
Buffer size
PREG Eira barbara 500 3000 3000
p(.)ψ(PF+DFP+SEA) 5 0.00 1.00 Leopardus pardalis 2500 2500 1500
p(fps)ψ(FPS) 4 0.00 0.38 p(fps)ψ(FPS+PREG) 5 0.19 0.34 p(fps)ψ(FPS-DISTS) 5 1.94 0.14 p(fps)ψ(FPS+PF) 5 1.95 0.14
Leopardus wiedii 2000 2000 3000 p(.)ψ(FPS-DFP) 4 0.00 0.70 p(.)ψ(FPS) 3 1.65 0.30
Mazama americana 500 1500 500 p(.)ψ(FPS) 3 0.00 0.29 p(.)ψ(FPS+PREG) 4 0.63 0.21 p(.)ψ(FPS+DFP) 4 1.34 0.15 p(.)ψ(FPS+PF) 4 1.44 0.14 p(.)ψ(FPS+DISTS) 4 1.87 0.11 p(.)ψ(FPS-CAT+FIRE) 5 1.95 0.11
Nasua narica 2500 500 1000 p(fps)ψ(-CAT-DFP+AUTO) 6 0.00 0.23 p(fps)ψ(-CAT-DFP+PF+AUTO) 7 0.03 0.23
p(fps)ψ(-CAT+AUTO) 5 0.03 0.23 p(fps)ψ(-CAT-DFP+FIRE+AUTO) 7 0.56 0.17
p(fps)ψ(-CAT+FIRE+AUTO) 6 0.93 0.14 Odocoileus virginianus 500 1000 500
p(.)ψ(FPS+BALFT-BASTS+AUTO) 6 0.00 0.71
p(.)ψ(-BASTS+AUTO) 4 1.82 0.29 Pecari tajacu 500 1500 500
p(.)ψ(FPS+BALFT-BASTS) 5 0.00 1.00 Procyon lotor 500 2500 500
p(.)ψ(-DFP) 3 0.00 0.39 p(.)ψ(-DFP-CAT+FIRE) 5 0.86 0.25 p(.)ψ(-PF-PREG-DFP) 5 1.24 0.21 p(.)ψ(-PF-DFP-CAT+FIRE) 6 1.87 0.15
Puma yagouaroundi 3000 2500 1500 p(.)ψ(FPS+SEA) 4 0.00 0.22 p(.)ψ(FPS) 3 0.21 0.20 p(.)ψ(SEA) 3 1.06 0.13 p(.)ψ(FPS-DISTS) 4 1.11 0.13 p(.)ψ(FPS+PREG) 4 1.17 0.12 p(.)ψ(.) 2 1.53 0.10
117
118
Model #Par ∆ AICc/ QAICc weight
Buffer size PF
Buffer size DFP
Buffer size
PREG p(.)ψ(FPS+PF) 4 1.56 0.10
Tamandua mexicana 1500 1500 3000 p(.)ψ(-PF-PREG+DFP) 5 0.00 0.40 p(.)ψ(CAT+FIRE+DISTS) 5 1.29 0.21 p(.)ψ(-PF+DFP) 4 1.36 0.20 p(.)ψ(DFP-SEA) 4 1.61 0.18
Urocyon cinereoargenteus 1500 500 500 p(sea)ψ(-DFP) 3 0.00 0.45 p(sea)ψ(-DFP+SEA) 5 1.65 0.20 p(sea)ψ(.) 3 1.75 0.19 p(sea)ψ(SEA) 4 1.99 0.17
APPENDIX D AIC WEIGHTS
AIC weights calculated from 56 patch occupancy models fit using combinations of 11 variables. AIC weights for each
variable were calculated by summing the weights of the first ten models in which the variable appears.
Species FPS* PF DFP PREG BALFT BASTS CAT FIRE DISTS DISTC SEAA. paca 0.31 0.13 0.13 0.16 0.13 0.19 0.15 0.11 0.11 0.05 0.21A. pigra 0.18 0.18 0.14 0.12 0.06 0.06 0.12 0.9 0.15 0.10 0.15A. geoffroyi 0.87 0.68 0.86 0.28 0.06 0.07 0.03 0.03 0.03 0.01 0.06C. mexicanus 0.60 0.04 0.03 0.02 0.04 0.04 0.02 0.02 0.93 0.28 0.12C.semistriatus 0.30 0.38 0.35 0.20 0.08 0.08 0.11 0.11 0.50 0.28 0.13D. punctata 0.38 0.15 0.24 0.09 0.08 0.12 0.11 0.09 0.15 0.13 0.12D. novemcinctus 0.00 0.23 0.23 0.99 0.00 0.00 0.00 0.00 0.02 0.00 0.99D. marsupialis 0.14 0.04 0.04 0.10 0.40 0.44 0.03 0.03 0.73 0.31 0.25D. virginianus 0.47 0.18 0.28 0.12 0.05 0.06 0.05 0.05 0.23 0.12 0.11E. barbara 0.35 0.79 0.68 0.25 0.03 0.03 0.08 0.08 0.07 0.04 0.39L. pardalis 0.73 0.20 0.13 0.22 0.05 0.05 0.08 0.07 0.15 0.06 0.10L. wiedii 0.90 0.28 0.53 0.11 0.03 0.03 0.06 0.06 0.08 0.03 0.07M. americana 0.88 0.23 0.23 0.19 0.04 0.04 0.09 0.09 0.12 0.04 0.07N. narica 0.09 0.25 0.44 0.15 0.02 0.02 0.64 0.39 0.07 0.04 0.05O. virginianus 0.67 0.09 0.08 0.06 0.44 0.56 0.13 0.07 0.11 0.08 0.11P. tajacu 0.75 0.13 0.13 0.12 0.67 0.66 0.05 0.05 0.14 0.04 0.23P. lotor 0.10 0.36 0.69 0.30 0.05 0.07 0.20 0.21 0.12 0.07 0.20P. yagouaroundi 0.51 0.14 0.12 0.16 0.06 0.06 0.08 0.07 0.17 0.10 0.28T. mexicana 0.10 0.46 0.50 0.29 0.07 0.06 0.13 0.13 0.37 0.22 0.22U. cinereoargenteus 0.14 0.14 0.40 0.08 0.08 0.09 0.14 0.14 0.08 0.05 0.23*FPS = size of sampled patch, PF = proportion of forest in the landscape (i.e., habitat loss), DFP = density of forest patches in the landscape (i.e., habitat fragmentation), PREG = proportion of regenerating forest in landscape, BALFT = basal area of large fruiting trees, BASTS = basal area of small trees and saplings, CAT = cattle presence, FIRE = fire, DISTS = distance to source, DISTC = distance to nearest community, SEA = season of survey (wet vs. dry).
119
APPENDIX E DATA ON REFERENCES INCLUDED IN REVIEW
Citation Taxonomic class studied
Number of species studied
Anderson et al. (2007) Mammalia 1 Anzures-Dadda & Manson (2007) Mammalia 1 Arroyo-Rodríguez et al. (2008) Mammalia 1 Askins et al. (2007) Aves 7 Attum et al. (2008) Reptilia 4 Austen et al. (2001) Aves 48 Bakermans & Rodewald (2006) Aves 1 Bakker et al. (2002) Aves 10 Banks et al. (2005) Mammalia 1 Barbaro et al. (2005) Aves, Insecta, Arachnida 34, 7,8 Bauerfeind et al. (2009) Insecta 1 Biedermann (2004) Insecta 2 Bonte et al. (2003) Arachnida 1 Bradford et al. (2003) Amphibia 1 Brotons & Herrando (2001) Aves 26 Brouwers & Newton (2009) Insecta 1 Brown (2007) Aves 41 Cagle (2008) Reptilia 7 Calme & Desrochers (2000) Aves 18 Cassel-Lundhagen et al. (2008) Insecta 1 Castellón & Sieving (2006) Aves 1 Chandler et al. (2009) Aves 9 Cox et al. (2003) Mammalia 3 Cozzi et al. (2008) Insecta 3 Cristóbal-Azkarate et al. (2005) Mammalia 1 Crooks (2002) Mammalia 11 Crozier & Niemi (2003) Aves 7 Denoël & Lehmann (2006) Amphibia 1 Díaz et al. (1998) Aves 30 Elzinga et al. (2007) Insecta 4 Estades & Temple (1999) Aves 21 Estrada et al. (2002) Mammalia 1 Eyre (2006) Mammalia 1 Fahrig & Jonsen (1998) Insecta 6 Fedriani et al. (2002) Mammalia 1 Fernández-Juricic (2004) Aves 8 Ficetola & Bernardi (2004) Amphibia 5 Fitzgibbon et al. (2007) Mammalia 1 Fleishman et al. (2002) Insecta 1 Forys & Humphrey (1999) Mammalia 1 González-Varo et al. (2008) Aves 1
120
Citation Taxonomic class studied
Number of species studied
Graham & Blake (2001) Aves 36 Gray et al. (2004) Amphibia 4 Grundel & Pavlovic (2007) Insecta 1 Guadagnin & Maltchik (2007) Aves 37 Guerry & Hunter (2002) Amphibia 9 Habel et al. (2007) Insecta 1 Hanser & Huntly (2006) Mammalia 1 Harper et al. (2008) Mammalia 2 Haynes et al. (2007) Insecta 1 Heisswolf et al. (2009) Insecta 2 Hinsley et al. (1995) Aves 22 Hokit et al. (1999) Reptilia 2 Holland & Bennett (2009) Mammalia 5 Hurme et al. (2008) Mammalia 1 Incoll et al. (2001) Mammalia 2 James et al. (2003) Insecta 1 Jansson & Angelstam (1999) Aves 1 Jaquiéry et al. (2008) Mammalia 1 Johnson & Collinge (2004) Mammalia 1 Kappes et al. (2009) Gastropoda 17 Knapp et al. (2003) Amphibia 1 Kolozsvary & Swihart (1999) Amphibia 9 Koper & Schmiegelow (2006) Aves 10 Krauss et al. (2005) Insecta 1 Kurosawa & Askins (2003) Aves 25 Lawes et al. (2006) Aves 1 Lawes et al. (2000) Mammalia 3 Lee et al. (2002) Aves 3 Lomolino & Perault (2001) Mammalia 9 MacDonald & Rushton (2003) Aves 28 Maclean et al. (2006) Aves 6 Maes & Bonte (2006) Insecta, Arachnida 3,2 Magle & Crooks (2009) Mammalia 1 Manu et al. (2007) Aves 62 Mapelli & Kittlein (2009) Mammalia 1 Matter et al. (2009) Insecta 1 McAlpine et al. (2006) Mammalia 1 Michalski & Peres (2005) Mammalia 18 Miller & Cale (2000) Aves 8 Moore & Swihart (2005) Mammalia 5 Mortelliti & Boitani (2007) Mammalia 2 Mortelliti & Boitani (2008) Mammalia 2 Naugle et al. (1999) Aves 3
121
122
Citation Taxonomic class studied
Number of species studied
Nunes & Galetti (2007) Aves 1 Nupp & Swihart (2000) Mammalia 7 Onderdonk & Chapman (2000) Mammalia 3 Otto et al. (2007) Amphibia 1 Perkins et al. (2003) Aves 30 Piha et al. (2007) Amphibia 3 Pita et al. (2007) Mammalia 1 Rabasa et al. (2008) Insecta 1 Radford & Bennett (2004) Aves 1 Radford & Bennett (2006) Aves 1 Renfrew & Ribic (2008) Aves 3 Ribic & Sample (2001) Aves 4 Rickman & Connor (2003) Insecta 8 Riffell et al. (2003) Aves 15 Rizkalla & Swihart (2006) Reptilia 5 Rukke (2000) Insecta 5 Rukke & Midtgaard (1998) Insecta 1 Sallabanks et al. (2006) Aves 24 Santos et al. (2002) Aves 37 Scharine et al. (2009) Mammalia 1 Schooley & Branch (2009) Mammalia 1 Silva et al. (2005) Mammalia 5 Sun et al. (2003) Aves 1 Swihart et al. (2007) Mammalia 2 Thomas et al. (2001) Insecta 3 Thornton (unpublished data) Mammalia 20 Uezu et al. (2005) Aves 5 Urquiza-Haas et al. (2009) Aves, Mammalia 4,9 Vergara & Armesto (2009) Aves 20 Vergara & Marquet (2007) Aves 1 Virgos (2001) Mammalia 1 Virgos & Garcia (2002) Mammalia 2 Virgos et al. (2002) Mammalia 2 Walker et al. (2003) Mammalia 1 WallisDeVries (2004) Insecta 1 Weyrauch & Grubb (2004) Amphibia 9 Wilson et al. (2002) Insecta 2 Wilson et al. (2009) Aves 3 Winter et al. (2006) Aves 3 Yamaura et al. (2008) Aves 8 Zabel & Tscharntke (1998) Insecta 24
LIST OF REFERENCES
Aliaga-Rossel, E., R. W. Kays, and J. M. V. Fragoso. 2008. Home-range use by the Central America agouti (Dasyprocta punctata) on Barro Colorado Island, Panama. Journal of Tropical Ecology 24:367-374.
Anderson, J., G. Cowlishaw, and J. M. Rowcliffe. 2007. Effects of forest fragmentation
on the abundance of Colobus angolensis palliatus in Kenya's Coastal Forests. International Journal of Primatology 28:637-655.
Anzures-Dadda, A., and R. H. Manson. 2007. Patch- and landscape-scale effects on
howler monkey distribution and abundance in rainforest fragments. Animal Conservation 10:69-76.
Arroyo-Rodríguez, V., S. Mandujano, and J. Benítez-Malvido. 2008. Landscape
attributes affecting patch occupancy by howler monkeys (Alouatta palliata mexicana) at Los Tuxtlas, Mexico. American Journal of Primatology 70:69-77.
Askins, R. A., B. Zuckerberg, and L. Novak. 2007. Do the size and landscape context of
forest openings influence the abundance and breeding success of shrubland songbirds in southern New England? Forest Ecology and Management 250:137-147.
Asner, G. P., T. K. Rudel, T. M. Aide, R. DeFries, and R. Emerson. 2009. A
contemporary assessment of change in humid tropical forests. Conservation Biology 23:1386-1395.
Attum, O., Y. M. Lee, J. H. Roe, and B. A. Kingsbury. 2008. Wetland complexes and
upland-wetland linkages: landscape effects on the distribution of rare and common wetland reptiles. Journal of Zoology 275:245-251.
Augustin, N. H., M. A. Mugglestone, and S. T. Buckland. 1996. An autologistic model for
the spatial distribution of wildlife. Journal of Applied Ecology 33:339-347. Austen, M. J., C. M. Francis, D. M. Burke, and M. S. W. Bradstreet. 2001. Landscape
context and fragmentation effects on forest birds in southern Ontario. Condor 103:701-714.
Bakermans, M. H., and A. D. Rodewald. 2006. Scale-dependent habitat use of acadian
flycatcher (Empidonax virescens) in central Ohio. Auk 123:368-382. Bakker, K. K., D. E. Naugle, and K. F. Higgins. 2002. Incorporating landscape attributes
into models for migratory grassland bird conservation. Conservation Biology 16:1638-1646.
123
Banks, S. C., G. R. Finlayson, S. J. Lawson, D. B. Lindenmayer, D. Paetkau, S. J. Ward, and A. C. Taylor. 2005. The effects of habitat fragmentation due to forestry plantation establishment on the demography and genetic variation of a marsupial carnivore, Antechinus agilis. Biological Conservation 122:581-597.
Barbaro, L., L. Pontcharraud, and F. Vetillard. 2005. Comparative responses of bird,
carabid, and spider assemblages to stand and landscape diversity in maritime pine plantation forests. Ecoscience 12:110-121.
Barea-Azcón, J. M., E. Virgós, E. Ballesteros-Duperón, M. Moleón, and M. Chirosa.
2007. Surveying carnivores at large spatial scales: a comparison of four broadly-applied methods. Biological Conservation 16:1213-1230.
Bauerfeind, S. S., A. Theisen, and K. Fischer. 2009. Patch occupancy in the
endangered butterfly Lycaena helle in a fragmented landscape: effects of habitat quality, patch size and isolation. Journal of Insect Conservation 13:271-277.
Baur, E. H. 1998. Informe final: estudio de la cacería de subsistencia en la concesión
forestal de Carmelita, San Andres, Petén. Propeten/Conservation International, Flores, Guatemala.
Beck-King, H., and O. von Helversen. 1999. Home range, population density, and food
resources of Agouti paca (Rodentia: Agoutidae) in Costa Rica: a study using alternative methods. Biotropica 31:675-685.
Belisle, M., and A. Desrochers. 2002. Gap-crossing decisions by forest birds: an
empirical basis for parameterizing spatially-explicit, individual-based models. Landscape Ecology 17:219-231.
Bender, D., L. Tischendorf, and L. Fahrig. 2003. Using patch isolation metrics to predict
animal movement in binary landscapes. Landscape Ecology 18:17-39. Biedermann, R. 2004. Patch occupancy of two hemipterans sharing a common host
plant. Journal of Biogeography 31:1179-1184. Bodmer, R. E., J. F. Eisenberg, and K. H. Redford. 1997. Hunting and the likelihood of
extinction of Amazonian mammals. Conservation Biology 11:460-466. Bolger, D. T., A. C. Alberts, and M. E. Soule. 1991. Occurrence patterns of bird species
in habitat fragments: sampling, extinction, and nested species subsets. American Naturalist 137:155-166.
Bonte, D., L. Lens, J.-P. Maelfait, M. Hoffmann, and E. Kuijken. 2003. Patch quality and
connectivity influence spatial dynamics in a dune wolfspider. Oecologia 135:227-233.
124
Boscolo, D., and J. P. Metzger. 2009. Is bird incidence in Atlantic forest fragments influenced by landscape patterns at multiple scales? Landscape Ecology 24:907-918.
Bradford, D. F., A. C. Neale, M. S. Nash, D. W. Sada, and J. R. Jaeger. 2003. Habitat
patch occupancy by toads (Bufo punctatus) in a naturally fragmented desert landscape. Ecology 84:1012-1023.
Broadbent, E. N., G. P. Asner, M. Keller, D. E. Knapp, P. J. C. Oliveira, and J. N. Silva.
2008. Forest fragmentation and edge effects from deforestation and selective logging in the Brazilian Amazon. Biological Conservation 141:1745-1757.
Brotons, L., and S. Herrando. 2001. Factors affecting bird communities in fragments of
secondary pine forests in the north-western Mediterranean basin. Acta Oecologia 22:21-31.
Brouwers, N. C., and A. C. Newton. 2009. The influence of habitat availability and
landscape structure on the distribution of wood cricket (Nemobius sylvestris) on the Isle of Wight, UK. Landscape Ecology 24:199-212.
Brown, W. P. 2007. Body mass, habitat generality, and avian community composition in
forest remnants. Journal of Biogeography 34:2168-2181. Burnham, K., and D. Anderson 2002. Model selection and multimodel inference.
Springer-Verlag, New York. Cagle, N. L. 2008. Snake species distributions and temperate grasslands: A case study
from the American tallgrass prairie. Biological Conservation 141:744-755. Calme, S., and A. Desrochers. 2000. Biogeographic aspects of the distribution of bird
species breeding in Québec's peatlands. Journal of Biogeography 27:725-732. Carillo, E., G. Wong, and A. D. Cuaron. 2000. Monitoring mammal populations in Costa
Rican protected areas under different hunting restrictions. Conservation Biology 14:1580-1591.
Cassel-Lundhagen, A., P. Sjögren-Gulve, and S.-A. Berglind. 2008. Effects of patch
characteristics and isolation on relative abundance of the scarce heath butterfly Coenonympha hero (Nymphalidae). Journal of Insect Conservation 12:477-482.
Castellón, T. D., and K. E. Sieving. 2006. Landscape history, fragmentation, and patch
occupancy: models for a forest bird with limited dispersal. Ecological Applications 16:2223-2234.
Ceballos, G. 2007. Conservation priorities for mammals in megadiverse Mexico: the
efficiency of reserve networks. Ecological Applications 17:569-578.
125
Chandler, R. B., D. I. King, and S. DeSteffano. 2009. Scrub-shrub bird habitat associations at multiple spatial scales in beaver meadows in Massachusetts. Auk 126:186-197.
Chazdon, R. L., C. A. Harvey, O. Komar, D. M. Griffith, B. G. Ferguson, M. Martinez-
Ramos, H. Morales, R. Nigh, L. Soto-Pinto, M. V. Breugel, and S. M. Philpott. 2009. Beyond reserves: a research agenda for conserving biodiversity in human-modified tropical landscapes. Biotropica 41:142-153.
Chevan, A., and M. Sutherland. 1991. Hierarchical partitioning. The American
Statistician 45:90-96. Chiarello, A. G. 1999. Effects of fragmentation of the Atlantic forest on mammal
communitites in south-eastern Brazil. Biological Conservation 89:71-82. Christman, M. C. 2008. Statistical modeling of observation data with spatial
dependencies. Journal of Wildlife Management 72:22-33. Churchill, J. L., and S. J. Hannon. 2010. Off-territory movement of male American
Redstarts (Setophaga ruticilla) in a fragmented agricultural landscape is related to song rate, mating status and access to females. Journal of Ornithology 151:33-44.
Cooper, C. B., and J. R. Walters. 2002. Independent effects of woodland loss and
fragmentation on Brown Treecreeper distribution. Biological Conservation 105:1-10.
Cox, M. P., C. R. Dickman, and J. Hunter. 2003. Effects of rainforest fragmentation on
non-fllying mammals of the Eastern Dorrigo Plateau, Australia. Biological Conservation 115:175-189.
Cozzi, G., C. B. Muller, and J. Krauss. 2008. How do local habitat management and
landscape structure at different spatial scales affect fritillary butterfly distribution on fragmented wetlands? Landscape Ecology 23:269-283.
Cristóbal-Azkarate, J., J. J. Veá, N. Asensio, and E. Rodriguez-Luna. 2005.
Biogeographical and floristic predictors of the presence and abundance of mantled howlers (Alouatta palliata mexicana) in rainforest fragments at Los Tuxtlas, Mexico. American Journal of Primatology 67:209-222.
Crooks, K. R. 2002. Relative sensitivities of mammalian carnivores to habitat
fragmentation. Conservation Biology 16:488-502. Crozier, G. E., and G. J. Niemi. 2003. Using patch and landscape variables to model
bird abundance in a naturally heterogeneous landscape. Canadian Journal of Zoology 81:441-452.
126
Cullen Jr., L., R. E. Bodmer, and C. V. Padua. 2000. Effects of hunting in habitat fragments of the Atlantic forests, Brazil. Biological Conservation 95:49-56.
Cushman, S. A., and K. McGarigal. 2004a. Hierarchical analysis of forest bird species-
environment relationships in the Oregon coast range. Ecological Applications 14:1090-1105.
Cushman, S. A., and K. McGarigal. 2004b. Patterns in the species-environment
relationship depend on both scale and choice of response variables. Oikos 105:117-124.
Cushman, S. A., K. McGarigal, and M. C. Neel. 2008. Parsimony in landscape metrics:
strength, universality, and consistency. Ecological Indicators 8:691-703. Daily, G. C., G. Ceballos, J. Pacheco, G. Suzan, and A. Sanchez-Azofeifa. 2003.
Countryside biogeography of neotropical mammals: conservation opportunities in agricultural landscapes of Costa Rica. Conservation Biology 17:1814-1826.
Dale, S., O. Steifetten, T. S. Osiejuk, K. Losak, and J. P. Cygan. 2009. How do birds
search for breeding areas at the landscape level? Ecography 29:886-898. Davies, K. F., C. R. Margules, and J. F. Lawrence. 2000. Which traits of species predict
population declines in experimental forest fragments? Ecology 81:1450-1461. de Azevedo, F. C. C., and D. L. Murray. 2007. Evaluation of potential factors
predisposing livestock to predation by jaguars. Journal of Wildlife Management 71:2379-2386.
Debinski, D. M., and R. D. Holt. 2000. A survey and overview of habitat fragmentation
experiments. Conservation Biology 14:342-355. deFries, R., A. Hansen, A. C. Newton, and M. C. Hansen. 2005. Increasing isolation of
protected areas in tropical forests over the past twenty years. Ecological Applications 15:19-26.
Dennis, R. L. H. 2003. Towards a functional resource-based concept for habitat: a
butterfly biology viewpoint. Oikos 102:417-426. Dennis, R. L. H., T. G. Shreeve, and H. Van Dyck. 2006. Habitat and resources: the
need for a resource-based definition to conserve butterflies. Biodiversity and Conservation 15:1943-1966.
Denoël, M., and A. Lehmann. 2006. Multi-scale effect of landscape processes and
habitat quality on newt abundance: Implications for conservation. Biological Conservation 130:495-504.
127
Develey, P. F., and J. P. Metzger. 2006. Emerging threats to birds in Brazilian Atlantic Forests: the roles of forest loss and configuration in a severely fragmented ecosystem. Pages 269-290 in W. F. Laurance, and C. A. Peres, editors. Emerging threats to tropical forests. University of Chicago Press, Chicago.
Devictor, V., R. Julliard, and F. Jiguet. 2008. Distribution of specialist and generalist
species along spatial gradients of habitat disturbance and fragmentation. Oikos 117:507-514.
Díaz, M., R. Carbonell, T. Santos, and J. L. Tllería. 1998. Breeding bird communities in
pine plantations of the Spanish plateaux: biogeography, landscape and vegetation effects. Journal of Applied Ecology 35:562-574.
Dillon, A., and M. J. Kelly. 2008. Ocelot home range, overlap and density: comparing
radio telemetry with camera trapping. Journal of Zoology 275:391-398. Drolet, B., A. Desrochers, and M. Fortin. 1999. Effects of landscape structure on nesting
songbird distribution in a harvested boreal forest. Condor 101:699-704. Elzinga, J. A., S. van Nouhuys, D.-J. van Leeuwen, and A. Biere. 2007. Distribution and
colonisation ability of three parasitoids and theiry herbivorous host in a fragmented landscape. Basic and Applied Ecology 8:75-88.
Emmons, L. H., and F. Feer 1997. Neotropical rainforest mammals: a field guide.
University of Chicago Press, Chicago. Estades, C., and S. A. Temple. 1999. Deciduous-forest bird communities in a
fragmented landscape dominated by exotic pine plantations. Ecological Applications 9:573-585.
Estrada, A., R. Coates-Estrada, and D. J. Meritt. 1994. Non flying mammals and
landscape changes in the tropical rain forest region of Los Tuxtlas, Mexico. Ecography 17:229-241.
Estrada, A., A. Mendoza, L. Castellanos, R. Pacheco, S. van Belle, Y. García, and D.
Muñoz. 2002. Population of the black howler monkey (Alouatta pigra) in a fragmented landscape in Palenque, Chiapas, Mexico. American Journal of Primatology 58:45-55.
Ewers, R. M., and R. K. Didham. 2006. Confounding factors in the detection of species
responses to habitat fragmentation. Biological Reviews 81:117-142. Eyre, T. J. 2006. Regional habitat selection of large gliding possums at forest stand and
landscape scales in southern Queensland, Australia I. Greater glider (Petuaroides volans). Forest Ecology and Management 235:270-282.
128
Fahrig, L. 1997. Relative effects of habitat loss and fragmentation on population extinction. Journal of Wildlife Management 61:603-610.
Fahrig, L. 2003. Effects of habitat fragmentation on biodiversity. Annual Review of
Ecology Evolution and Systematics 34:487-515. Fahrig, L., and I. Jonsen. 1998. Effect of habitat patch characteristics on abundance
and diversity of insects in an agricultural landscape. Ecosystems 1:197-205. Fedriani, J. M., M. Delibes, P. Ferreras, and J. Roman. 2002. Local and landscape
habitat determinants of water vole distribution in a patchy Mediterranean environment. Ecoscience 9:12-19.
Fernández-Juricic, E. 2004. Spatial and temporal analysis of the distribution of forest
specialists in an urban-fragmented landscape (Madrid, Spain): Implications for local and regional bird conservation. Landscape and Urban Planning 69:17-32.
Ficetola, G. F., and F. D. Bernardi. 2004. Amphibians in a human-dominated landscape:
the community structure is related to habitat features and isolation. Biological Conservation 119:219-230.
Fitzgibbon, S. I., D. A. Putland, and A. W. Goldizen. 2007. The importance of functional
connectivity in the conservation of a ground-dwelling mammal in an urban Australian landscape. Landscape Ecology 22:1513-1525.
Fleishman, E., and R. Mac Nally. 2007. Measuring the response of animals to
contemporary drivers of fragmentation. Canadian Journal of Zoology 85:1080-1090.
Fleishman, E., C. Ray, P. Sjögren-gulve, and C. L. Boggs. 2002. Assessing the roles of
patch quality, area, and isolation in predicting metapopulation dynamics. Conservation Biology 16:706-716.
Foerster, C. R., and C. Vaughan. 2002. Home range, habitat use, and activity of Baird's
Tapir in Costa Rica. Biotropica 34:423-437. Fortin, M.-J., and M. Dale 2005. Spatial analysis: a guide for ecologists. Cambridge
University Press, Cambridge. Forys, E., and S. R. Humphrey. 1999. The importance of patch attributes and context to
the management and recovery of an endangered lagomorph. Landscape Ecology 14:177-185.
Fraser, G. S., and B. J. M. Stutchbury. 2004. Area-sensitive forest birds move
extensively among forest patches. Biological Conservation 118:377-387.
129
Gascon, C., T. E. Lovejoy, R. O. Bierregaard Jr., J. R. Malcolm, P. C. Stouffer, H. L. Vasconcelos, W. F. Laurance, B. Zimmerman, M. Tocher, and S. Borges. 1999. Matrix habitat and species richness in tropical forest remnants. Biological Conservation 91:223-229.
Gobeil, J. F., and M. A. Villard. 2002. Permeability of three boreal forest landscape
types to bird movements as determined from experimental translocations. Oikos 98:447-458.
González-Varo, J. P., J. V. López-Bao, and J. Guitián. 2008. Presence and abundance
of the Eurasian nuthatch Sitta europaea in relation to the size, isolation and the intensity of management of chestnut woodlands in the NW Iberian Peninsula. Landscape Ecology 23:79-89.
Graham, C. H., and J. G. Blake. 2001. Influence of patch- and landscape-level factors
on bird assemblages in a fragmented tropical landscape. Ecological Applications 11:1709-1721.
Gray, M. J., L. M. Smith, and R. I. Leyva. 2004. Influence of agricultural landscape
structure on a Southern High Plains, USA, amphibian assemblage. Landscape Ecology 19:719-729.
Grundel, R., and N. B. Pavlovic. 2007. Resource availability, matrix quality,
microclimate, and spatial pattern as predictors of patch use by the Karner blue butterfly. Biological Conservation 135:135-144.
Guadagnin, D. L., and L. Maltchik. 2007. Habitat and landscape factors associated with
neotropical waterbird occurrence and richness in wetland fragments. Biodiversity and Conservation 16:1231-1244.
Guerry, A. D., and M. L. Hunter. 2002. Amphibian distributions in a landscape of forests
and agriculture: an examination of landscape composition and configuration. Conservation Biology 16:745-754.
Habel, J. C., T. Schmitt, W. Härdtle, M. Lütkepohl, and T. Assmann. 2007. Dynamics in
a butterfly-plant-ant system: influence of habitat characteristics on turnover rates of the endangered lycaenid Maculinea alcon. Ecological Entomology 32:536-543.
Haila, Y. 2002. A conceptual genealogy of fragmentation research: from island
biogeography to landscape ecology. Ecological Applications 12:321-334. Hannah, D., J. C. Z. Woinarski, C. P. Catterall, J. C. Mccosker, N. Y. Thurgate, and R.
J. Fensham. 2007. Impacts of clearing, fragmentation and disturbance on the bird fauna of Eucalypt savanna woodlands in central Queensland, Australia. Austral Ecology 32:261-276.
130
Hansen, A. J., and R. DeFries. 2007. Ecological mechanisms linking protected areas to
surrounding lands. Ecological Applications 17:974-988. Hanser, S. E., and N. J. Huntly. 2006. The biogeography of small mammals of
fragmented sagebrush-steppe landscapes. Journal of Mammalogy 87:1165-1174.
Harmsen, B. J., R. J. Foster, S. Silver, L. Ostro, and C. P. Doncaster. 2010. Differential
use of trails by forest mammals and the implications for camera-trap studies: a case study from Belize. Biotropica 42:126-133.
Harper, M. J., M. A. McCarthy, and R. Van Der Ree. 2008. Resources at the landscape
scale influence possum abundance. Austral Ecology 33:243-252. Harrison, S., and E. Bruna. 1999. Habitat fragmentation and large-scale conservation:
what do we know for sure? Ecography 22:225-232. Harvey, C. A., A. Medina, D. M. Sánchez, S. Vilchez, B. Hernández, J. C. Saenz, J. M.
Maes, F. Casanoves, and F. L. Sinclair. 2006. Patterns of animal diversity in different forms of tree cover in agricultural landscapes. Ecological Applications 16:1986-1999.
Haynes, K. J., T. Diekötter, and T. O. Crist. 2007. Resource complementation and the
response of an insect herbivore to habitat area and fragmentation. Oecologia 153:511-520.
Hedges, L. V., and I. Olkin. 1980. Vote-counting methods in research synthesis.
Psychological Bulletin 88:359-369. Hedges, L. V., and I. Olkin 1985. Statistical methods for meta-analysis. Academic
Press, New York. Heisswolf, A., S. Reichmann, H. J. Poethke, B. Schröder, and E. Obermaier. 2009.
Habitat quality matters for the distribution of an endangered leaf beetle and its egg parasitoid in a fragmented landscape. Journal of Insect Conservation 13:165-175.
Henle, K., K. F. Davies, M. Kleyer, C. Margules, and J. Settele. 2004. Predictors of
species sensitivity to fragmentation. Biodiversity and Conservation 13:207-251. Hill, K., G. McMillan, and R. Farina. 2003. Hunting-related changes in game encounter
rates from 1994 to 2001 in the Mbaracayu Reserve, Paraguay. Conservation Biology 17:1312-1323.
131
Hines, J. E. 2006. PRESENCE2- Software to estimate patch occupancy and related parameters. USGS-PWRC.
http://www.mbr-pwrc.usgs.gov/software/presence.html. Hinsley, S. A., P. E. Bellamy, I. Newton, and T. H. Sparks. 1995. Habitat and landscape
factors influencing the presence of individual breeding bird species in woodland fragments. Journal of Avian Biology 26:94-104.
Hokit, D. G., B. M. Stith, and L. C. Branch. 1999. Effects of landscape structure in florida
scrub: a population perspective. Ecological Applications 9:124-134. Holland, G. J., and A. F. Bennett. 2009. Differing responses to landscape change:
implications for small mammal assemblages in forest fragments. Biodiversity and Conservation 18:2997-3016.
Holland, J. D., L. Fahrig, and N. Cappuccino. 2005. Body size affects the spatial scale
of habitat-beetle interactions. Oikos 110:101-108. Hurme, E., M. Mönkkönen, P. Reunanen, A. Nikula, and V. Nivala. 2008. Temporal
patch occupancy dynamics of the Siberian flying squirrel in a boreal forest landscape. Ecography 31:469-476.
Incoll, R. D., R. H. Loyn, S. J. Ward, R. B. Cunningham, and C. F. Donnelly. 2001. The
occurrence of gliding possums in old-growth forest patches of mountain ash (Eucalyptus regnans) in the Central Highlands of Victoria. Biological Conservation 98:77-88.
James, M., F. Gilbert, and S. Zalat. 2003. Thyme and isolation for the sinai baton blue
butterfly (Pseudophilotes sinaicus). Oecologia 134:445-453. Jansson, G., and P. Angelstam. 1999. Threshold levels of habitat composition for the
presence of the long-talied tit (Aegithalos caudatus) in a boreal landscape. Landscape Ecology 14:283-290.
Jaquiéry, J., J. Guélat, T. Broquet, L. Berset-Brändli, E. Pellegrini, R. Moresi, A. H.
Hirzel, and N. Perrin. 2008. Habitat-quality effects on metapopulation dynamics in greater white-toothed shrews, Crocidura russula. Ecology 89:2777-2785.
Jenkins, C. N., and C. Giri. 2007. Protection of mammal diversity in Central America.
Conservation Biology 22:1037-1044. Johnson, D. H. 2001. Habitat fragmentation effects on birds in grasslands and wetlands:
a critique of our knowledge. Great Plains Research 11:211-231. Johnson, W. C., and S. K. Collinge. 2004. Landscape effects on black-tailed prairie dog
colonies. Biological Conservation 115:487-497.
132
Kadoya, T., S. Suda, Y. Tsubaki, and I. Washitani. 2008. The sensitivity of dragonflies to landscape structure differs between life-history groups. Landscape Ecology 23:149-158.
Kamilar, J. M., and L. M. Paciulli. 2008. Examining the extinction risk of specialized
folivores: a comparative study of colobine monkeys. American Journal of Primatology 70:816-827.
Kappes, H., K. Jordaens, F. Hendrickx, J.-P. Maelfait, L. Lens, and T. Backeljau. 2009.
Response of snails and slugs to fragmentation of lowland forest in NW Germany. Landscape Ecology 24:685-697.
Kawanishi, K. 1995. Camera monitoring of human impacts on rain forest wildlife in Tikal
National Park, Guatemala. M.S. Thesis. Frostburg State University, Frostburg, Maryland.
Klingbeil, B. T., and M. R. Willig. 2009. Guild-specific responses of bats to landscape
composition and configuration in fragmented Amazonian rainforest. Journal of Applied Ecology 46:203-213.
Knapp, R. A., K. R. Matthews, H. K. Preisler, and R. Jellison. 2003. Developing
probabilistic models to predict amphibian site occupancy in a patchy landscape. Ecological Applications 13:1069-1082.
Kolozsvary, M. B., and R. K. Swihart. 1999. Habitat fragmentation and the distribution of
amphibians: patch and landscape correlates in farmland. Canadian Journal of Zoology 77:1288-1299.
Koper, N., and F. K. A. Schmiegelow. 2006. A multi-scaled analysis of avian response
to habitat amount and fragmentation in the Canadian dry mixed-grass prairie. Landscape Ecology 21:1045-1059.
Kotliar, N. B., and J. A. Wiens. 1990. Multiple scales of patchiness and patch structure:
a hierarchical framework for the study of heterogenetiy. Oikos 59:253-260. Krauss, J., I. Steffan-Dewenter, C. B. Müller, and T. Tscharntke. 2005. Relative
importance of resource quantity, isolation and habitat quality for landscape distribution of a monophagous butterfly. Ecography 28:465-474.
Kurosawa, R., and R. A. Askins. 2003. Effects of habitat fragmentation on birds in
deciduous forests in Japan. Conservation Biology 17:695-707. Larrucea, E. S., P. F. Brussard, M. M. Jaeger, and R. H. Barrett. 2008. Cameras,
coyotes, and the assumption of equal detectability. Journal of Wildlife Management 71:1682-1689.
133
Laurance, W. F. 1991. Ecological correlates of extinction proneness in Australian tropical rain forest mammals. Conservation Biology 5:79-89.
Laurance, W. F. 2002. Hyperdynamism in fragmented habitats. Journal of Vegetation
Science 13:595-602. Laurance, W. F. 2008. Theory meet reality: How habitat fragmentation research has
transcended island biogeographic theory. Biological Conservation 141:1731-1744.
Laurance, W. F., and R. O. Bierregaard 1997. Tropical forest remnants: ecology,
management, and conservation of fragmented communities. University of Chicago Press, Chicago.
Laurance, W. F., T. E. Lovejoy, H. L. Vasconcelos, E. M. Bruna, R. K. Didham, P. C.
Stouffer, C. A. Gascon, R. O. J. Bierregaard, S. G. Laurance, and E. Sampaio. 2002. Ecosystem decay of Amazonian forest fragments: a 22-year investigation. Conservation Biology 16:605-618.
Lawes, M. J., S. Fly, and S. E. Piper. 2006. Gamebird vulnerability to forest
fragmentation: patch occupancy of the crested guineafowl (Guttera edouardi) in Afromontane forests. Animal Conservation 9:67-74.
Lawes, M. J., P. E. Mealin, and S. E. Piper. 2000. Patch occupancy and potential
metapopulation dynamics of three forest mammals in fragmented afromontane forest in South Africa. Conservation Biology 14:1088-1098.
Lee, M., L. Fahrig, K. Freemark, and D. J. Currie. 2002. Importance of patch scale vs
landscape scale on selected forest birds. Oikos 96:110-118. Legendre, P. 1993. Spatial autocorrelation: trouble or new paradigm? Ecology 74:1659-
1673. Levin, S. A. 1992. The problem of pattern and scale in ecology. Ecology 73:1943-1967. Lichstein, J. W., T. R. Simons, S. A. Shriner, and K. E. Franzreb. 2002. Spatial
autocorrelation and autoregressive models in ecology. Ecological Monographs 72:445-463.
Lindenmayer, D. B., and J. Fischer 2006. Habitat fragmentation and landscape change.
Island Press, Washington. Littell, R. C., G. A. Milliken, W. W. Stroup, R. D. Wolfinger, and O. Schabenberber 2006.
SAS for Mixed Models. SAS Institute, Cary, NC.
134
Lomolino, M. V., and D. R. Perault. 2001. Island biogeography and landscape ecology of mammals inhabiting fragmented, temperate rain forests. Global Ecology and Biogeography 10:113-132.
López, G., R. B. McNab, D. Thornton, R. García, J. Moreira, V. Méndez, M. Córdova, A.
Xol, G. Ponce, M. Mérida, and G. Ruano. 2008. Validación de la presencia de Jaguares en dos fragmentos importantes de su corredor en Guatemala: Xutilhá y Sierra Santa Cruz. Informe interno. Sociedad para la Conservación de la Vida Silvestre (WCS-Guatemala), Programa para la Conservación del Jaguar. 17 pp.
Lynam, A. J., and I. Billick. 1999. Differential responses of small mammals to
fragmentation in a Thailand tropical forest. Biological Conservation 91:191-200. Mac Nally, R. 2000. Regression and model-building in conservation biology,
biogeography and ecology: The distinction between - and reconciliation of - "predictive" and explanatory models. Biodiversity and Conservation 9:655-671.
Mac Nally, R., and A. F. Bennett. 1997. Species-specific predictions of the impact of
habitat fragmentation: local extinction of birds in the box-ironbark forest of central Victoria, Australia. Biological Conservation 82:147-155.
MacDonald, D. W., and S. Rushton. 2003. Modelling space use and dispersal of
mammals in real landscapes: a tool for conservation. Journal of Biogeography 30:607-620.
MacKenzie, D. I., and L. L. Bailey. 2004. Assessing the fit of site-occupancy models.
Journal of Agricultural, Biological, and Environmental Statistics 9:300-318. MacKenzie, D. I., J. D. Nichols, G. B. Lachman, S. Droege, J. A. Royle, and C. A.
Landtimm. 2002. Estimating site occupancy rates when detection probabilities are less than one. Ecology 83:2248-2255.
MacKenzie, D. I., J. D. Nichols, A. J. Royle, K. H. Pollock, L. L. Bailey, and J. E. Hines
2006. Occupancy estimation and modeling: inferring patterns and dynamics of species occurrence. Elsevier, Boston.
Maclean, I. M. D., M. Hassall, R. R. Boar, and I. R. Lake. 2006. Effects of disturbance
and habitat loss on papyrus-dwelling passerines. Biological Conservation 131:349-358.
Maes, D., and D. Bonte. 2006. Using distribution patterns of five threatened
invertebrates in a highly fragmented dune landscape to develop a multispecies conservation approach. Biological Conservation 133:490-499.
Maffei, L., and A. B. Taber. 2003. Home range of a Mazama americana (Cervidae) in a
dry forest of Bolivia. Ecologia en Bolivia 38:179-180.
135
Magle, S. B., and K. R. Crooks. 2009. Investigating the distribution of prairie dogs in an urban landscape. Animal Conservation 12:192-203.
Manu, S., W. Peach, and W. Cressell. 2007. The effects of edge, fragment size and
degree of isolation on avian species richness in highly fragmented forest in West Africa. Ibis 149:287-297.
Mapelli, F. J., and M. J. Kittlein. 2009. Influence of patch and landscape characteristics
on the distribution of the subterranean rodent Ctenomys porteousi. Landscape Ecology 24:723-733.
Matter, S. F., M. Ezzeddine, E. Duermit, J. Mashburn, R. Hamilton, T. Lucas, and J.
Roland. 2009. Interactions between habitat quality and connectivity affect immigration but not abundance or population growth of the butterfly, Parnassius smintheus. Oikos 118:1461-1470.
Mazerolle, M. J., and M. Villard. 1999. Patch characteristics and landscape context as
predictors of species presence and abundance: a review. Ecoscience 6:117-124. McAlpine, C. A., J. R. Rhodes, J. G. Callaghan, M. E. Bowen, D. Lunney, D. L. Mitchell,
D. V. Pullar, and H. P. Possingham. 2006. The importance of forest area and configuration relative to local habitat factors for conserving forest mammals: A case study of koalas in Queensland, Australia. Biological Conservation 132:153-165.
McGarigal, K., S. A. Cushman, M. C. Neel, and E. Ene. 2002. FRAGSTATS: Spatial
Pattern Analysis Program for Categorical Maps. Computer software program produced by the authors at the University of Massachusetts, Amherst. Available at the following web site: www.umass.edu/landeco/research/fragstats/fragstats.html.
McNab, R. B. 1999. Comparative impacts of chicle and xate harvests on wildlife of the
Maya Biosphere Reserve, Guatemala. M.A. Thesis. University of Florida, Gainesville.
McNab, R. B., N. S. Monroy, V. H. Ramos, J. López, and T. Dubón. 2004. The current
distribution of the ocellated turkey (Meleagris ocellata) in Guatemala: preliminary report. WCS Guatemala/IRG-FIPA/CONAP.
Mech, S. G., and P. A. Zollner. 2002. Using body size to predict perceptual range. Oikos
98:47-52. Medina, A., C. A. Harvey, D. Sanchez, S. Vilchez, and B. Hernández. 2007. Bat
diversity and movement in a Neotropical agricultural landscape. Biotropica 39:120-128.
136
Meyer, C. F. J., J. Frund, W. P. Lizano, and E. K. V. Kalko. 2008. Ecological correlates of vulnerability to fragmentation in Neotropical bats. Journal of Applied Ecology 45:381-391.
Michalski, F., P. G. Crawshaw, T. G. d. Oliveira, and M. E. Fabian. 2006. Notes on
home range and habitat use of three small carnivore species in a disturbed vegetation mosaic of southeastern Brazil. Mammalia 70:52-57.
Michalski, F., and C. A. Peres. 2005. Anthropogenic determinants of primate and
carnivore local extinctions in a fragmented forest landscape of southern Amazonia. Biological Conservation 124:383-396.
Michalski, F., and C. A. Peres. 2007. Disturbance-mediated mammal persistence and
abundance-area relationships in Amazonian forest fragments. Conservation Biology 21:1626-1640.
Miller, J. R., and P. Cale. 2000. Behavioral mechanisms and habitat use by birds in a
fragmented agricultural landscape. Ecological Applications 10:1732-1748. Moilanen, A., and M. Nieminen. 2002. Simple connectivity measures in spatial ecology.
Ecology 83:1131-1145. Moore, J. E., and R. K. Swihart. 2005. Modeling patch occupancy by forest rodents:
incorporating dectectability and spatial autocorrelation with hierarchically structured data. Journal of Wildlife Management 69:933-949.
Mortelliti, A., and L. Boitani. 2007. Estimating species' absence, colonization and local
extinction in patchy landscapes: an application of occupancy models with rodents. Journal of Zoology 273:244-248.
Mortelliti, A., and L. Boitani. 2008. Interaction of food resources and landscape structure
in determining the probability of patch use by carnivores in fragmented landscapes. Landscape Ecology 23:285-298.
Murray, K., and M. M. Conner. 2009. Methods to quantify variable importance:
implications for the analysis of noisy ecological data. Ecology 90:348-355. Myers, N. 2002. Biodiversity and biodepletion: the need for a paradigm shift. Pages 46-
60 in T. O'Riordan, and S. Stoll-Kleemann, editors. Biodiversity, sustainability, and human communities: protecting beyond the protected. Cambridge University Press, Cambridge.
Naranjo, E. J., and R. E. Bodmer. 2007. Source-sink systems and conservation of hunted ungulates in the Lacandon Forest, Mexico. Biological Conservation 138:412-420.
137
Naugle, D. E., K. F. Higgins, S. M. Nusser, and W. C. Johnson. 1999. Scale-dependent habitat use in three species of prairie wetland birds. Landscape Ecology 14:267-276.
Nepstad, D., S. Schwartzman, B. Bamberger, M. Santilli, D. Ray, P. Schlesinger, P.
Lefebvre, A. Alencar, E. Prinz, G. Fiske, and A. Rolla. 2006. Inhibition of Amazon deforestation and fire by parks and indigenous lands. Conservation Biology 20:65-73.
Nielsen, S. E., D. L. Haughland, E. Bayne, and J. Schieck. 2009. Capacity of large-
scale, long-term biodiversity monitoring programmes to detect trends in species prevalence. Biodiversity and Conservation 18:2961-2978.
Novaro, A. J., K. H. Redford, and R. E. Bodmer. 2000. Effects of hunting in source-sink
systems in the Neotropics. Conservation Biology 14:713-721. Nowak, R. M. 1999. Walker's mammals of the world. Johns Hopkins University Press,
Baltimore. Nunes, M. F. C., and M. Galetti. 2007. Use of forest fragments by blue-winged macaws
(Primolius maracana) within a fragmented landscape. Biodiversity and Conservation 16:953-967.
Nupp, T. E., and R. K. Swihart. 2000. Landscape-level correlates of small-mammal
assemblages in forest fragments of farmland. Journal of Mammalogy 81:512-526.
Onderdonk, D. A., and C. A. Chapman. 2000. Coping with forest fragmentation: the
primates of Kibale National Park, Uganda. International Journal of Primatology 21:587-611.
Osenberg, C. W., O. Sarnelle, S. D. Cooper, and R. D. Holt. 1999a. Resolving
ecological questions through meta-analysis: goals, metrics, and models. Ecology 80:1105-1117.
Osenberg, C. W., O. Sarnelle, and D. E. Goldberg. 1999b. Meta-analysis in ecology:
concepts, statistics, and applications. Ecology 80:1103-1104. Otto, C. R. V., D. C. Forester, and J. W. Snodgrass. 2007. Influences of wetland and
landscape characteristics on the distribution of carpenter frogs. Wetlands 27:261-269.
Parry, L., J. Barlow, and C. A. Peres. 2007. Large-vertebrate assemblages of primary and secondary forests in the Brazilian Amazon. Journal of Tropical Ecology 23:653-662.
138
Parry, L., J. Barlow, and C. A. Peres. 2009. Allocation of hunting effort by Amazonian smallholders: implications for conserving wildlife in mixed-use landscapes. Biological Conservation 142:1777-1786.
Peres, C. A. 1996. Population status of white-lipped Tayassu pecari and collared
peccaries T. tajacu in hunted and unhunted Amazonian forests. Biological Conservation 77:115-123.
Peres, C. A. 2001. Synergistic effects of subsistence hunting and habitat fragmentation
on Amazonian forest vertebrates. Conservation Biology 15:1490-1505. Peres, C. A., and H. S. Nascimento. 2006. Impact of game hunting by the Kayapo of
south-eastern Amazonia: implications for wildlife conservation in tropical forest indigenous reserves. Biodiversity and Conservation 15:2627-2653.
Peres, C. A., and E. Palacios. 2007. Basin-wide effects of game harvest on vertebrate
population densities in Amazonian forests: implications for animal-mediated seed dispersal. Biotropica 39:304-315.
Perkins, M. W., R. J. Johnson, and E. E. Blankenship. 2003. Response of riparian
avifauna to percentage and pattern of woody cover in an agricultural landscape. Wildlife Society Bulletin 31:642-660.
Piha, H., M. Luoto, and J. Merila. 2007. Amphibian occurrence is influenced by current
and historic landscape characteristics. Ecological Applications 17:2298-2309. Pita, R., P. Beja, and A. Mira. 2007. Spatial population structure of the Cabrera vole in
Mediterranean farmland: the relative role of patch and matric effects. Biological Conservation 134:383-392.
Polisar, J., R. B. McNab, H. Quigley, M. J. Gonzalez, and M. Cabrera. 1998. Preliminary
assessment of the effects of subsistence hunting in the Mayan Biosphere Reserve. Part 1: Progress report - Games populations in Tikal National Park and Uaxactún. Wildlife Conservation Society, Petén, Flores. Guatemala.
Ponce-Santizo, G., E. Andresen, and E. Cano. 2006. Dispersión primaria de semillas
por primates y dispersión secundaria por escarabajos coprófagos en Tikal, Guatemala. Biotropica 38:390-397.
Purvis, A., J. L. Gittleman, G. Cowlishaw, and G. M. Mace. 2000. Predicting extinction
risk in declining speciess. Proceeding of the Royal Society of London - Series B Biological Sciences 267:1947-1952.
Quinn, G. P., and M. J. Keough 2002. Experimental design and data analysis for
biologists. Cambridge University Press, Cambridge .
139
Rabasa, S. G., D. Gutiérrez, and A. Escudero. 2008. Relative importance of host plant patch geometry and habitat quality on the patterns of occupancy, extinction and density of the monophagous butterfly Iolana iolas. Oecologia 156:491-503.
Rabinowitz, A. and K.A. Zeller. 2010. A range-wide model of landscape connectivity and
conservation for the jaguar, Panthera onca. Biological Conservation 143:939-945.
Radford, J. Q., and A. F. Bennett. 2004. Thresholds in landscape parameters:
occurrence of the white-browed treecreeper Climacteris affinis in Victoria, Australia. Biological Conservation 117:375-391.
Radford, J. Q., and A. F. Bennett. 2006. Factors affecting patch occupancy by the
White-browed Treecreeper Climacteris affinis in an agricultural landscape in north-west Victoria, Australia. Pacific Conservation Biology 13:195-206.
Reid, F. A. 1997. A field guide to the mammals of Central America and Southeast
Mexico. Oxford University Press, New York. Renfrew, R. B., and C. A. Ribic. 2008. Multi-scale models of grassland passerine
abundance in a fragmented system in Wisconsin. Landscape Ecology 23:181-193.
Reyna-Hurtado, R., and G. W. Tanner. 2005. Habitat preferences of ungulates in
hunted and nonhunted areas in the Calakmul Forest, Campeche, Mexico. Biotropica 37:676-685.
Reyna-Hurtado, R., and G. W. Tanner. 2007. Ungulate relative abundance in hunted
and non-hunted sites in Calakmul Forest (southern Mexico). Biodiversity and Conservation 16:743-756.
Ribic, C. A., and D. W. Sample. 2001. Associations of grassland birds with landscape
factors in southern Wisconsin. American Midland Naturalist 146:105-121. Rickman, J. K., and E. F. Connor. 2003. The effect of urbanization on the quality of
remnant habitats for leaf-mining Lepidoptera on Quercus agrifolia. Ecography 26:777-787.
Riffell, S. K., B. E. Keas, and T. M. Burton. 2003. Birds in North American Great Lakes
coastal wet meadows: is landscape context important? Landscape Ecology 18:95-111.
Ritchie, L., M. Betts, G. Forbes, and K. Vernes. 2009. Effects of landscape composition
and configuration on northern flying squirrels in a forest mosaic. Forest Ecology and Management 257:1920-1929.
140
Rizkalla, C. E., and R. K. Swihart. 2006. Community structure and differential responses of aquatic turtles to agriculturally induced habitat fragmentation. Landscape Ecology 21:1361-1375.
Robertson, O. J., and J. Q. Radford. 2009. Gap-crossing decisions of forest birds in a
fragmented landscape. Austral Ecology 34:435-446. Rodrigues, A. S. L., S. J. Andelman, M. I. Bakarr, L. Boitani, T. M. Brooks, R. M.
Cowling, L. D. C. Fishpool, G. A. B. d. Fonseca, K. J. Gaston, M. Hoffmann, J. S. Long, P. A. Marquet, J. D. Pilgrim, R. L. Pressey, J. Schipper, W. Sechrest, S. N. Stuart, L. G. Underhill, R. W. Waller, M. E. J. Watts, and X. Yan. 2004. Effectiveness of the global protected area network in representing species diversity. Nature 428:640-643.
Royle, J. A., and J. D. Nichols. 2003. Estimating abundance from repeated presence-
absence data or point counts. Ecology 84:777-790. Rukke, B. A. 2000. Effects of habitat fragmentation: increased isolation and reduced
habitat size reduces the incidence of dead wood fungi beetles in a fragmented forest landscape. Ecography 23:492-502.
Rukke, B. A., and F. Midtgaard. 1998. The importance of scale and spatial variables for
the fungivorous beetle Bolitophagus reticulatus (Coleoptera, Tenebrionidae) in a fragmented forest landscape. Ecography 21:561-572.
Saenz, J. C., and C. Vaughan. 1998. Home range and habitat use by two groups of
Odocoileus virginianus (Artiodactyla: Cervidae) relocated in a tropical environment. Revista de Biologica Tropical 46:1185-1197.
Sallabanks, R., R. A. Riggs, L. E. Cobb, and S. W. Dodson. 2006. Bird-habitat
relationships in grand fir forests of the Blue Mountains, Oregon. Forest Science 52:489-502.
Salom-Pérez, R., E. Carrillo, J. C. Sáenz, and J. M. Mora. 2007. Critical condition of the
jaguar Panthera onca population in Corcovado National Park, Costa Rica. Oryx 41:51-56.
Sanchez-Rojas, G., S. Gallina, and S. Mandujano. 1997. Area of activity and habitat
usage of two white-tailed deer (Odocoileus virginianus) in a tropical forest of the Jalisco coast, Mexico. Acta Zoologica Mexicana Nueve Serie 72:39-54.
Santos, T., J. L. Telleria, and R. Carbonell. 2002. Bird conservation in fragmented
Mediterraean forests of Spain: effects of geographical location, habitat and landscape degradation. Biological Conservation 105:113-125.
SAS. 2008. SAS for Windows, Version 9.2. SAS Institute, Cary, North Carolina, USA.
141
Scharine, P. D., C. K. Nielson, E. M. Schauber, and L. Rubert. 2009. Swamp rabbits in floodplain ecosystems: influence of landscape- and stand-level habitat on relative abundance. Wetlands 29:615-623.
Schmidt, M. H., and T. Tscharntke. 2005. Landscape context of sheetweb spider
(Araneae: Linyphiidae) abundance in cereal fields. Journal of Biogeography 32:467-473.
Schooley, R. L. 2006. Spatial heterogeneity and characteristic scales of species-habitat
relationships. Bioscience 56:533-537. Schooley, R. L., and L. C. Branch. 2009. Enhancing the area-isolation paradigm: habitat
heterogeneity and metapopulation dynamics of a rare wetland mammal. Ecological Applications 19:1708-1722.
Sekercioglu, C. H., S. R. Loarie, F. O. Brenes, P. R. Ehrlich, and G. C. Daily. 2007.
Persistence of forest birds in the Costa Rican agricultural countryside. Conservation Biology 21:482-494.
Shahabuddin, G., and C. A. Ponte. 2005. Frugivorous butterfly species in tropical forest
fragments: correlates of vulnerability to extinction. Biodiversity and Conservation 14:1137-1152.
Silman, M. R., J. W. Terborgh, and R. A. Kiltie. 2003. Population regulation of a
dominant rain forest tree by a major seed predator. Ecology 84:431-438. Silva, M., L. Hartling, and S. B. Opps. 2005. Small mammals in agricultural landscape of
Prince Edward Island (Canada): Effects of habitat characteristics at three different spatial scales. Biological Conservation 126:556-568.
Smith, A. C., N. Koper, C. M. Francis, and L. Fahrig. 2009. Confronting collinearity:
comparing methods for disentangling the effects of habitat loss and fragmentation. Landscape Ecology 24:1271-1285.
St-Laurent, M.-H., C. Dussault, J. Ferron, and R. Gagnon. 2009. Dissecting habitat loss
and fragmentation effects following logging in boreal forest: conservation perspectives from landscape simulations. Biological Conservation 142:2240-2249.
Sun, Y.-H., Z.-J. Piao, and J. E. Swenson. 2003. Occurrence of hazel grouse Bonasa
bonasia in a heavily human-impacted landscape near the Changbai Mountains, northeastern China. Wildlife Biology 9:371-375.
Sunquist, M., and F. Sunquist 2002. Wild cats of the world. University of Chicago Press,
Chicago.
142
Swihart, R. K., T. M. Gehring, M. B. Kolozsvary, and T. E. Nupp. 2003. Responses of 'resistant' vertebrates to habitat loss and fragmentation: the importance of niche breadth and range boundaries. Diversity and Distributions 9:1-18.
Swihart, R. K., J. R. Goheen, S. A. Schnelker, and C. E. Rizkalla. 2007. Testing the
generality of patch and landscape-level predictors of tree squirrel occurrence at a regional scale. Journal of Mammalogy 88:564-572.
Terborgh, J., J. A. Estes, P. Paquet, K. Ralls, D. Boyd-Heger, B. J. Miller, and R. F.
Noss. 1999. The role of top carnivores in regulating terrestrial ecosystems. Pages 39-64 in M. E. Soule, and J. Terborgh, editors. Continental conservation: scientific foundations of regional reserve networks. Island Press, Washington.
Terborgh, J., L. Lopez, V. P. Nunez, M. Rao, G. Shahabuddin, G. Orihuela, M. Riveros,
R. Ascanio, G. H. Adler, T. D. Lambert, and L. Balbas. 2001. Ecological meltdown in predator-free forest fragments. Science 294:1923-1926.
Thomas, J. A., N. A. D. Bourn, R. T. Clarke, K. E. Stewart, D. J. Simcox, G. S.
Pearman, R. Curtis, and B. Goodger. 2001. The quality and isolation of habitat patches both determine where butterflies persist in fragmented landscapes. Proceedings of the Royal Society: Biological Sciences 268:1791-1796.
Tilman, D., R. May, C. Lehman, and M. Nowak. 1994. Habitat destruction and the
extinction debt. Nature 371:65-66. Tischendorf, L., D. Bender, and L. Fahrig. 2003. Evaluation of patch isolation metrics in
mosaic landscapes for specialist vs. generalist dispersers. Landscape Ecology 18:41-50.
Tobler, M. W., S. E. Carillo-Percastegui, R. L. Pitman, R. Mares, and G. Powell. 2008.
An evaluation of camera traps for inventorying large- and medium-sized terrestrial rainforest mammals. Animal Conservation 11:169-178.
Trolle, M., and M. Kery. 2005. Camera-trap study of ocelot and other secretive
mammals in the northern Pantanal. Mammalia 69:409-416. Turner, I. M., and R. T. Corlett. 1996. The conservation value of small, isolated
fragments of lowland tropical rain forest. Trends in Ecology and Evolution 11:330-333.
Turner, M. G. 2005. Landscape ecology in North America: past, present, and future.
Ecology 86:1967-1974. Uezu, A., J. P. Metzger, and J. M. E. Vielliard. 2005. Effects of structural and functional
connectivity and patch size on the abundance of seven Atlantic Forest bird species. Biological Conservation 123:507-519.
143
Urquiza-Haas, T., C. A. Peres, and P. M. Dolman. 2009. Regional scale effects of
human density and forest disturbance on large-bodied vertebrates throughout the Yucatan Peninsula, Mexico. Biological Conservation 142:134-148.
Vanreusel, W., and H. Van Dyck. 2007. When functional habitat does not match
vegetation types: a resource-based approach to map butterfly habitat. Biological Conservation 135:202-211.
Vergara, P. M., and J. J. Armesto. 2009. Responses of Chilean forest birds to
anthropogenic habitat fragmentation across spatial scales. Landscape Ecology 24:25-38.
Vergara, P. M., and P. A. Marquet. 2007. On the seasonal effect of landscape structure
on a bird species: the thorn-tailed rayadito in a relict forest in northern Chile. Landscape Ecology 22:1059-1071.
Villard, M. A., M. K. Trzcinski, and G. Merriam. 1999. Fragmentation effects on forest
birds: relative influence of woodland cover and configuration on landscape occupancy. Conservation Biology 13:774-783.
Virgos, E. 2001. Role of isolation and habitat quality in shaping species abundance: a
test with badgers (Meles meles L.) in a gradient of forest fragmentation. Journal of Biogeography 28:381-389.
Virgos, E., and F. J. Garcia. 2002. Patch occupancy by stone martens Martes foina in
fragmented landscapes of central Spain: the role of fragment size, isolation and habitat structure. Acta Oecologica 23:231-237.
Virgos, E., J. L. Telleria, and T. Santos. 2002. A comparison of the response to forest
fragmentation by medium-sized Iberian carnivores in central Spain. Biodiversity and Conservation 11:1063-1079.
Viveiros de Castro, E. V. B., and F. A. S. Fernandez. 2004. Determinants of differential
extinction vulnerabilities of small mammals in Atlantic forest fragments in Brazil. Biological Conservation 119:73-80.
Walker, R. S., A. Novaro, and L. C. Branch. 2003. Effects of patch attributes, barriers,
and distance between patches on the distribution of a rock-dwelling rodent (Lagidium viscacia). Landscape Ecology 18:185-192.
WallisDeVries, M. F. 2004. A quantitative conservation approach for the endangered
butterfly Maculinea alcon. Conservation Biology 18:489-499.
144
Wang, Y., J. Zhang, K. J. Feeley, P. Jiang, and P. Ding. 2009. Life-history traits associated with fragmentation vulnerability of lizards in the Thousand Island Lake, China. Animal Conservation 12:329-337.
Ward, D., D. Pepler, and R. Botha. 2008. The influence of sample size on the
determination of population trends in the vulnerable Lesser Kestrel Falco naumanni overwintering in South Africa. Ostrich 79:199-204.
Weyrauch, S. L., and T. C. J. Grubb. 2004. Patch and landscape characteristics
associated with the distribution of woodland amphibians in an agricultural fragmented landscape: an information-theoretic approach. Biological Conservation 115:443-450.
Wildlife Conservation Society. 2007. Estimación de la deforestación en la Reserva de la
Biosfera Maya, periodo 2006-2007. Biodiversity Conservation at the Landscape Scale. USAID-EGAT GCPII-Maya Biosphere Reserve Landscape-Guatemala. Monitoring ecological integrity of the MBR, Petén, Guatemala.
Wilson, R. T., S. Ellis, J. S. Baker, M. E. Lineham, R. W. Whitehead, and C. D. Thomas.
2002. Large-scale patterns of distribution and persistence at the range margins of a butterfly. Ecology 83:3357-3368.
Wilson, T. L., E. J. Johnson, and J. A. Bissonette. 2009. Relative importance of habitat
area and isolation for bird occurrence patterns in a naturally patchy landscape. Landscape Ecology 24:351-360.
Winter, M., D. H. Johnson, J. A. Shaffer, T. M. Donovan, and W. D. Svedarsky. 2006.
Patch size and landscape effects on density and nesting success of grassland birds. Journal of Wildlife Management 70:158-172.
Wisz, M. S., R. J. Hijmans, J. Li, A. T. Peterson, C. H. Graham, and A. Guisan. 2008.
Effects of sample size on the performance of species distribution models. Diversity and Distributions 14:763-773.
With, K. A., and A. W. King. 1999. Extinction thresholds for species in fractal
landscapes. Conservation Biology 13:314-326. Woodroffe, R., and J. R. Ginsberg. 1998. Edge effects and the extinction of populations
inside protected areas. Science 280:2126-2128. Wright, S. J., K. E. Stoner, N. Beckman, R. T. Corlett, R. Dirzo, H. C. Muller-Landau, G.
Nunez-Iturri, C. A. Peres, and B. C. Wang. 2007. The plight of large animals in tropical forests and the consequences for plant regeneration. Biotropica 39:289-291.
145
146
Yamaura, Y., K. Katoh, and T. Takahashi. 2008. Effects of stand, landscape, and spatial variables on bird communities in larch plantations and deciduous forest in central Japan. Canadian Journal of Forest Resources 38:1223-1243.
Zabel, J., and T. Tscharntke. 1998. Does fragmentation of Urtica habitats affect
phytophagous and predatory insects differentially? Oecologia 116:419-425. Zurita, G. A., and M. I. Bellocq. 2010. Spatial patterns of bird community similarity: bird
responses to landscape composition and configuration in the Atlantic forest. Landscape Ecology 25:147-158.
BIOGRAPHICAL SKETCH
Daniel Thornton received his bacelor’s degree in geology from Carleton College
(Northfield, MN) in 1998. He enrolled in the graduate program at the University of
Florida in 2000. In May 2003, he received his master’s degree in wildlife ecology and
conservation. In August 2003, he continued his graduate program at the University of
Florida. He received his Ph.D. in wildlife ecology and conservation in May of 2010.
147