|UFDC Home||myUFDC Home | Help|
This item has the following downloads:
THE EFFECTS OF LOGGING ON PRIMATE-HABITAT INTERACTIONS:
A CASE STUDY OF REDTAIL MONKEYS (Cercopithecus ascanius)
IN KIBALE NATIONAL PARK, UGANDA
CLAUDIA MARGRET STICKLER
A THESIS PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF SCIENCE
UNIVERSITY OF FLORIDA
Claudia M. Stickler
To the memory of my grandmother, who had talent and intelligence in abundance, but
lacked the freedom and opportunity to follow all of her dreams, and to my parents,
siblings and other family members, who have consistently served as both inspiration
I extend my deep thanks to my advisor, Colin Chapman, for seeing me
through this sometimes difficult process, devoting many hours to helping me in
innumerable ways in the field and on-campus. I thank him for providing the generosity
and support I needed to accomplish this project. The other members of my committee
were no less important. Michael Binford took me under his wing and spent countless
hours guiding me towards appropriate methods, critical resources, and important
perspectives, expressed confidence in me, and provided encouragement at crucial
moments. Sue Boinski first took a chance and hired me to work as part of her field crew
in Suriname and encouraged and facilitated my application to the University of Florida.
Since then, she has been a constant source of professional and personal support. Graeme
Cumming introduced me to the importance of a spatial approach and helped me to
proceed through every step of my thesis, always reminding me of the essential elements
of my research, demanding great conscientiousness, and encouraging a sense of pride and
confidence in my work.
Other faculty members at the University of Florida offered critical support
throughout this stage of my education. Jane Southworth initially helped me to understand
the applications of a spatial perspective on social and ecological questions. Since then,
she has not only helped to guide and support my research, but has also been an
unwavering source of encouragement and support. I want to extend my thanks to
Marianne Schmink for helping me to maintain my roots in the practical and social aspects
of conservation and development. Steven Humphrey helped to create a flexible and rich
environment for taking on the difficult task of gaining an interdisciplinary education.
Thanks are due also to Karen Kainer, Jon Dain, Robert Buschbacher and Dan Zarin for
advice and encouragement on how to bridge the gap between basic and applied science
and between the natural and social sciences.
In the field, I was assisted by a team of incredibly skilled and knowledgeable
people. First and foremost, I thank Clovice Adyeeri Kaganzi for his tireless dedication to
following the redtails in all kinds of weather and vegetation and to mapping every last
tree in the forest. I attribute the quality of this work in large part to the careful and
thoughtful documentation of his observations and to his vast store of knowledge of
Kibale. Richard Kaserengenyu was a cheerful and conscientious observer, who took
every change of plans in stride and took great pride in his work. James Apuuli Magaro
and Chris Kaija were invaluable in accurately inventorying each tree in the selected study
sites, and entertained and educated me in innumerable ways. Lawrence Tusiime provided
his impressive skills as a botanist, and proved a steadfast partner in sampling habitat
characteristics for many months. Tinkasiimire Astone enthusiastically took on each new
job of seeking out and planting seeds, and monitoring seedlings. He taught me a great
deal about the history of Kibale and its surroundings. I extend my deep thanks to Patrick
Omeija and Jackson Efitre for the all the support they provided. Last but not least, Alice
Akiki provided companionship and dedicated care.
My friends and colleagues at the University of Florida have been invaluable.
Beyond providing friendship, they are also great sources of knowledge and experience.
Tracy Van Holt, Rodrigo Vergara, Connie Clark and John Poulsen took enormous
amounts of time to help me with my work and through the most difficult times. Meredith
Evans, Anna Prizzia, Lin Cassidy, Wendy-Lin Bartels, Franklin Paniagua, Alfredo Rios,
Tom Gillespie and Cedric Worman formed an extraordinary and patient support team,
without whom I would not have arrived at this point with any humor or sanity.
Finally, I thank my family for setting high standards for what individuals can
achieve, no matter what their vocation. They provide me with inspiration and a sense of
responsibility to the broader world. My parents afforded me great opportunities and
freedom to follow my dreams, even if those took me far away. My twin, Alexander, and
my brother Peter and my sister Carin have cheered me on and always created a place in
their homes for me.
My research and education were supported through a National Science
Foundation Graduate Research Fellowship. The Office of the President, Uganda, the
Uganda National Council for Science and Technology, and the Uganda Wildlife
Authority granted me permission to work in Uganda. Gilbert Isabirye-Basuta and John
Kasenene graciously allowed me to conduct research at the Makerere University
Biological Field Station.
TABLE OF CONTENTS
A C K N O W L E D G M E N T S ................................................................................................. iv
LIST OF TABLES ..................................................... ix
L IST O F FIG U RE S .............. ......................... ........................... ....................... .. .. .... .x
1 GEN ER AL IN TR OD U CTION ...................................... ...................... ...............1...
2 COPING WITH LOGGING: BEHAVIORAL MECHANISMS OF REDTAIL
MONKEYS (Cercopithecus ascanius) IN KIBALE NATIONAL PARK,
W E STERN U G A N D A ...................... ...............................................................7......
Introduction ................................................................................ ..........................7
M eth o d s ...................................................................................................... .......... 9
S tu d y A re a ..................................................................................................... 9
B ehavioral O observations ...................................... ....................... ............... 10
R source and H habitat A ssessm ent.................................................. ............... 12
A n a ly se s .............................................................................................................. 1 3
R results ............................................................................ .................... 16
A activity B budget .......................................................................................... 16
H om e R ange Size and U se ..................................... ..................... ............... 17
D aily R an g in g ...................................................................................................... 18
Overall D iet ...................................... ..................... ... .... ............... 18
D ietary Selection and D iversity..................................................... 19
H ab itat U se .............. ........................................................................................... 2 0
M ixed Species A associations ........................................................... ................ 2 1
D isc u ssio n ................................................................................................................ .. 2 2
R source A availability .. .. ...... ............ .............................................. 23
C o n c lu sio n ............................................................................................................... .. 2 5
3 THE EFFECTS OF PATCH STRUCTURE ON REDTAIL MONKEY
(Cercopithecus ascanius) HABITAT USE IN UNLOGGED AND HEAVILY
LOGGED AREAS OF KIBALE NATIONAL PARK, UGANDA ...........................46
In tro d u ctio n ................................................................................................................ 4 6
M e th o d s ...................................................................................................................... 4 9
Study A rea ...........................................................................................................49
B ehavioral and H habitat A ttributes.................................................... ................ 50
A naly ses .............. ..................................................................... ......5 1
R e su lts.. . ........ ........................................................... ......................................5 5
H habitat U se ........................................................................................................55
Spatial Structure of Habitat Attributes ..............................................................56
P ath A analysis ................................................................................................. 57
D iscussion........................................... ..................... 60
Factors Explaining Habitat Use Patterns .......................................... ............... 61
C conclusion .............. ........................................................................ . ..... 63
4 SUM M ARY AND CONCLUSION S .................................. .............. ................ 79
L IST O F R E FE R E N C E S ........................................................... ............... ................ 85
BIOGRAPH ICAL SKETCH ................... ............................................................... 94
LIST OF TABLES
2-1 Test of independence for group activity samples pooled by 60-min or 15-min
sam ple for redtail groups ......................................... ........................ ................ 28
2-2 Monthly and mean home range size or redtail groups in logged and unlogged forest
in Kibale National Park, Uganda between August and December 2002...............29
2-3 Time spent eating different types of foods by redtail groups...............................30
2-4 Dietary diversity and overlap across months for redtail groups in logged and
unlogged forest of Kibale National Park, Uganda. .............................................32
2-5 Characteristics of the tree community in home range areas of focal redtail groups.33
3-1 Habitat attributes examined for two redtail monkey home ranges in unlogged and
heavily logged forest ......................... ........................ ................................... 68
3-2 The six components produced in Principal Components Analysis describing the
major variation in habitat structure in areas of unlogged and heavily logged forest.69
3-3 Standardized direct, indirect and total effect coefficients for path analysis models
(Model 1) examining habitat variables on probability of redtail occurrence...........70
3-4 Effect coefficients for path analysis models examining the effect of unripe fruit
trees on feeding site selection by redtails ........................................... ................ 71
LIST OF FIGURES
Figure. 2-1. The Kanyawara study site is located in Kibale National Park, western
Uganda. Maps are adapted from MUBFS 2003.................................................34
Figure 2-2. Percentage of time engaged in different activities by redtails in unlogged
and heavily logged forest ......................................... ........................ ................ 35
Figure 2-3. Monthly total grid cells entered and cumulative grid cells entered by
redtails in unlogged and heavily logged areas..................................... ................ 36
Figure 2-4. Pattern of home range use by redtail groups in heavily logged and
u n lo g g ed fo rest......................................................................................................... 3 7
Figure 2-5. Daily distance traveled by redtail monkeys in logged and unlogged areas. 38
Figure 2-6. Average percent of monthly feeding observations on different items by
redtail monkeys in unlogged and heavily logged forest. ...................................39
Figure 2-7. Strauss selectivity index (measures selectivity vs. dominance) for top
80% of food species selected by redtails............................................. ................ 40
Figure 2-8. Height in canopy of redtails in unlogged and heavily logged forest...............41
Figure 2-9. Distribution of number of canopy layers observed in
hom e ranges of redtails. .............. .............. ............................................ 42
Figure 2-10. Understory vegetation associated with redtail location in unlogged and
heavily logged forest.. ............. ................ .............................................. 43
Figure 2-11. Monthly percent of time spent in mixed species associations for redtails....44
Figure 2-12. The proportion of observations in which redtails in unlogged and
heavily logged areas. ............. ................ .............................................. 45
Figure 3-1. The study site is located in the Kanyawara field site in the northern part
of Kibale National Park in western Uganda........................................ ................ 72
Figure 3-2. Patterns of habitat utilization by redtail monkeys in unlogged and heavily
log g ed forest area. .................................................................................................... 7 3
Figure 3-3. Surface maps for six of the habitat variables measured in an area of
u n lo g g ed fo rest ......................................................................................................... 7 4
Figure 3-4. Surface maps for six of the habitat variables measured in an area of
heavily logged forest.. ............. ................ .............................................. 75
Figure 3-5. Path diagram (Model 1) describing habitat utilization by redtails in unlogged
and heavily logged areas ..................................................................... ................ 76
Figure 3-6. Path diagram (Model 2) describing habitat utilization by redtails in an
u n lo g g ed are a .......................................................................................................... 7 7
Figure 3-7. Probability maps of trees potentially producing unripe and ripe fruits in
unlogged and heavily logged forest of Kibale National Park, Uganda. ..............78
Abstract of Thesis Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Master of Science
THE EFFECTS OF LOGGING ON PRIMATE-HABITAT INTERACTIONS: A CASE
STUDY OF REDTAIL MONKEYS (Cercopithecus ascanius) IN KIBALE NATIONAL
Claudia Margret Stickler
Chair: Colin A. Chapman
Major Department: School of Natural Resources and Environment
The scale and intensity of timber exploitation in the tropics have increased in recent
decades, leaving degraded forests and deforested areas in their wake. The impact of these
activities on forest-dependent vertebrates is varied-ranging from loss of habitat and
resources to population isolation or extinction. In recognition of the inadequacy of
existing protected areas to support viable populations in the long-term, the focus of
wildlife conservation in tropical forests is shifting to identifying and supporting more
ecologically sound timber management strategies. This study focuses on the impact of
logging on primate behavior and habitat use because primates play an important role in
tropical forest communities, serving as seed dispersers and as indicators of overall
mammal species richness. Primates may also face a greater threat from broad-scale
habitat loss and hunting than most birds do, because they are constrained to travel
arboreally or terrestrially, and may not be able to move readily between areas to take
advantage of greater resource availability or other suitable habitat.
To better understand how logging affects primates, I first focused on contrasting
the behavior of two habituated groups of redtail monkeys in unlogged and heavily logged
forest in Kibale National Park, Uganda. Monkeys in the heavily logged area were found
to use larger home ranges, travel further and spend more time foraging and less time
feeding than conspecifics in the unlogged areas. They appear to be constrained by
resource and habitat limitations. Specifically, a reduced tree density in the heavily
logged area leads to a lower abundance of potential food trees and reduced arboreal
pathways. Microhabitat characteristics in the heavily logged area also appeared to
constrain the animals' movements and foraging ability.
I also examined habitat selection by redtails in the heavily logged and unlogged
areas to determine whether the behavioral differences observed in the first part were due
to habitat limitations or to differences in habitat preferences. I evaluated a set of
topographic, forest structural, and resource attributes for each home range to determine
the spatial structure of each. Tree basal area showed the least amount of patch structure
in both areas and operated on the small scale, in the range of 20 to 40 m. In both areas,
the majority of remaining variables grouped together with slope. In the unlogged area,
the scale of these variables was relatively large, whereas in the heavily logged area, the
variables showed an intermediate scale of patch structure. These results suggest that
there is a common organizing structure to both areas, but that the specific characteristics
of each area are a result of scale. Habitat selection by the monkeys was similar in both
areas; they were primarily influenced by basal area, but understory vegetation also played
an important role.
The scale and intensity of timber exploitation in the tropics has increased in recent
decades, leaving degraded forests and deforested areas in its wake (Fimbel et al. 2001,
Weber et al. 2001). The impact of these activities on forest-dependent vertebrates are
varied-ranging from loss of habitat and resources to population isolation or extinction
(Fimbel et al. 2001). In recognition of the inadequacy of existing protected areas to
support viable populations in the long-term, the focus of wildlife conservation in tropical
forests is shifting to identifying and supporting more ecologically sound timber
management strategies (Frumhoff 1995). This shift is accompanied by a heightened
awareness of the role vertebrates play in community processes. For example, in tropical
areas, vertebrates' role in seed dispersal is notable, with up 45 to 90% of forest trees
thought to be adapted for vertebrate dispersal (Jansen and Zuidema 2001). As a result,
the challenge for managers and researchers is to help in designing timber management
plans that do not compromise these interactions and processes. A prerequisite for
developing such suggestions, however, is a deeper understanding of how animals respond
to the habitat changes caused by logging. Unfortunately, these are still not well-known.
This study focuses on the impact of logging on primate behavior and habitat use
for several reasons. Primates play an important role in tropical forest communities.
Primates are thought to be important as seed dispersers, consuming a wide range of seeds
(Chapman 1995, Garber and Lambert 1998). As a result of the high percentage of
biomass they represent in many tropical rain forests (Terborgh 1983), primates' services
as seed dispersers for a great variety of fruiting trees suggests that their net impact on
forest regeneration may be high (Jansen and Zuidema 2001). Thus changes in primate
behavior and ranging due to logging and forest fragmentation may alter or disrupt seed
dispersal processes (Chapman and Onderdonk 1998, Clark et al. 2001). Primates also
serve as good indicators of mammal species richness in most tropical areas (Emmons
2000), indicating that forest management plans based on a better understanding of
primate response to logging could serve a wider animal community, as well. Finally,
primates may face a greater threat from broad-scale habitat loss and hunting than most
birds do. Because they are constrained to travel arboreally or terrestrially, and since
many are territorial, they may not be able to move readily between areas to take
advantage of greater resource availability or other suitable habitat (Cannon and Leighton
1994, Putz et al. 2001). Changes in the forest structure due to logging may create barriers
for primates where none exist for some bird species or larger mammals.
Among vertebrate taxa, primates are one of the best-studied. The majority of
research examining the effects of logging on primates focus on relationships between
population density and broad vegetation changes (Skorupa 1988, Johns and Skorupa
1987, Plumptre and Reynolds 1994, Chapman et al. 2000). However, these studies have
yielded few generalizations that contribute to the formulation of specific strategies for
sustainable managing forests (Plumptre and Grieser-Johns 2001). As a result, interest in
understanding primate behavior and the habitat changes caused by logging at a finer scale
is growing. Ample evidence exists to suggest that behavioral responses to resource
availability and habitat quality are important in dictating population density of organisms
(Krebs and Kacelnik 1991, Lima and Zollner 1996, Pulido and Diaz 1997). Among
primates, numerous studies have shown that seasonal differences in food availability
result in differences in diet composition (Van Schaik et al. 1993, Kaplin et al. 1998,
Poulsen et al. 2001), activity budgets, and travel patterns (Van Schaik et al. 1993,
Poulsen et al. 2001) for different species. Similarly, group sizes of foraging primates can
also be influenced by the size and spatial distribution of food patches (Leighton and
Leighton 1982). Animals may also alter travel patterns or fruit consumption rates as a
function of resource patch size and availability (Leighton and Leighton 1982, Chapman
1988). Those studies comparing primate behavior between logged and unlogged forest
show that primates do make some adjustments in their behavior or group size to
compensate for habitat changes (Olupot 2000, Clarke et al. 2002, Fairgrieve and
Muhuzoma 2003), but they are too few in number to allow generalization from the
variety of responses they describe.
I addressed these issues by quantifying the differences in behavior and habitat and
resources available to redtail monkeys (Cercopithecus ascanius) in heavily logged and
unlogged areas of Kibale National Park, Uganda. Broadly, I focus on the following
* Does the behavior of primates differ in heavily logged and unlogged forest?
* Does this reflect limitations posed by the habitat or resource availability?
* Do logged and unlogged forests differ in their structure and composition?
* Are such differences relevant to primate habitat selection?
* Which habitat attributes are important to primates in logged and unlogged areas in
selecting suitable habitats?
To better understand how logging affects primates, I first focused on contrasting
the behavior of two habituated groups of redtail monkeys in unlogged and heavily logged
forest. This species is of interest because redtails, and similar species across Africa, tend
to adapt well to conditions of disturbed forest (Cords 1986). At Kibale, however, regular
population censuses since logging show that redtail densities continue to decline 28 years
after logging (Chapman et al. 2000). I collected behavioral data on each group during
all-day follows, approximately 3 times a week over a 5 month time period, using 15-min
scan sampling. At each scan interval, I recorded the group's activities, diet, and location.
These data were analyzed to determine if and how the behavioral patterns of redtails in
heavily logged and unlogged areas differs. To evaluate the observed patterns in relation
to resource availability in each area, I inventoried the tree community above 10 cm
diameter at breast height (DBH). Each tree was identified and mapped, and subsequently
categorized as a "food" or "nonfood" tree based on the feeding observations collected
from each group. In addition, I assessed the density of the understory vegetation and
vertical stratification in each home range, to determine the availability of different
microhabitat types. I compared these measurements with redtails' canopy height
preferences and associations with understory density categories. This study allowed an
assessment whether redtails in the heavily logged areas were constrained by the
abundance and distribution of resources and preferred microhabitats.
To further understand how observed behavioral differences might be a response to
differences in habitat structure, I examined patterns of habitat utilization by the monkeys
relative to the habitat available through a spatial analysis. The major of objectives of this
second study were to identify the variables defining areas of suitable habitat and to
develop predictive models of habitat selection based on important ecological variables. I
analyzed the spatial structure of a suite of variables describing the topography and forest
structure and composition, to determine if the measured variables operate at the same
scale, and at the same scale in each home range. The variables, including slope, aspect,
canopy height, basal area, understory height, and the presence of several understory
vegetation types were sampled from a set of 50 m2 plots, spatially stratified in each
group's home range. Food tree density was determined from the tree map completed in
the first study. I identified which of these factors explained the most variation in each set
of data through a principal components analysis (PCA). In combination with correlation
analyses, the PCA helped to determine which variables to enter into a conceptual model
of the interacting variables determining redtail habitat use, using path analysis. The same
conceptual model was applied to the habitat attribute and monkey use data from each
area, to determine if the same set of variables and interactions explains redtail habitat
selection in both the logged and the unlogged area. Finally, I developed surface models
predicting redtail habitat utilization in each area based on the variables determined to be
important by the path analysis, using spatial interpolation.
The main objective of these two studies was to link patterns of resource
availability and suitable habitat to population density by examining redtail behavioral
patterns. Since the focus of this objective was to examine the relationships between
different variables on the basis of spatial patterns, the research relied on a spatially
explicit approach. This is based on the key tenet of landscape ecology-that ecological
pattern and process are linked (Turner 1989). A spatial approach has the potential to
uncover important relationships that might otherwise have gone unnoticed or remained
uninterpretable, by providing information about the spatial scales of interacting factors.
An important part of this is the ability to evaluate variability within a dataset for the
information it may reveal about the fundamental structure organizing the system (Forman
and Godron 1986). The focus on the interaction of spatial patterns allows for more
accurate description of the system, thus leading to more accurate models (Turner et al.
2001). Most importantly, from a management perspective, this suggests the potential to
predict and thus avoid undesirable spatial configurations created by logging activities,
based on a quantitative understanding of the spatial scales of basic attributes of the forest.
Equally important is the fact that this type of research builds the foundation for
developing predictive models over a much larger spatial scale by relating variable
measurements at different scales.
COPING WITH LOGGING: BEHAVIORAL MECHANISMS OF REDTAIL
MONKEYS (Cercopithecus ascanius) IN KIBALE NATIONAL PARK, WESTERN
Currently, tropical forests are being degraded and fragmented at increasing rates,
primarily due to logging-associated activities (Chapman and Peres 2001). Wildlife may
be directly affected by habitat degradation, or indirectly through increased human
accessibility to forests, which can lead to increased hunting or habitat loss through land
conversion (Frumhoff 1995). Although most wildlife populations dependent on closed-
canopy mature forests decline as a result of logging activities, the long-term ecological
responses of most vertebrates are still not well understood (Fimbel et al. 2001).
The majority of studies examining the effects of timber extraction on primates
compare population densities in logged and unlogged areas of forest (Skorupa 1988,
White 1992, 1994, Plumptre and Reynolds 1994, Grieser-Johns and Grieser-Johns 1995,
Chapman et al. 2000). Most research suggests that vegetation changes resulting from
logging are responsible for differences in primate population densities, although
individual species responses sometimes differ greatly (Plumptre and Reynolds 1994;
Struhsaker 1997, Chapman et al. 2000). Studies are often not comparable due to
differences among the methods employed or time allowed after the logging event
(Chapman et al. in prep). Additionally, numerous confounding factors (e.g., hunting) can
affect primate densities (Wilkie et al. 1992, 1998, Oates 1996, Struhsaker 1997, Chapman
et al. in prep). Thus, although primates are one of the best-studied taxa of tropical forest
vertebrates, few generalizations about their responses to the effects of logging have
This lack of generalizability could be the result of a high degree of intraspecific
variation in responses to logging among sites. In the Budongo Forest Reserve, Uganda,
the increase in blue monkey (Cercopithecus mitis) populations in logged areas appears to
be the result of increased fruit availability associated with a change in forest composition
following logging (Fairgreave and Muhuzuma 2003). In contrast, in Kibale National
Park, Uganda, less than 200 km from Budongo, blue monkey populations decrease after
logging. Even among folivorous species, the variation in responses is high, with some
responding positively and others negatively to the habitat changes caused by logging, due
to group size requirements for resources (Chapman et al. in prep). Although research
regarding primates' behavioral responses to logging is growing (Clarke et al. 2000,
Olupot 2000), more is needed to understand how changes induced by logging affect
Here, I quantify some aspects of the behavior of redtail monkeys (Cercopithecus
ascanius schmidtii Matschie) in logged and unlogged forest in Kibale National Park,
Uganda. This species is of particular interest because redtails and other members of their
lineage (e.g., C. cephus) are generally the most adaptable of the cercopithecines, often
thriving in disturbed secondary forest and on forest edges, as well as in undisturbed forest
(Struhsaker and Leland 1979, Cords 1986, Butynski 1990, Thomas 1991, Naughton-
Treves 1996). However, primate censuses conducted nearly 30 years after logging in
Kibale indicate a continued decline in redtail population density in the heavily logged
areas (Chapman et al. 2000). Redtail density is significantly lower in the logged areas
(1.04 groups/km2) than in the unlogged areas (11.48 groups/km2; Chapman et al. 2000).
The objective of this study was to determine whether redtails exhibit different
behavioral patterns in logged and unlogged forest. To meet this objective, I contrast the
ranging behavior and foraging ecology of two groups of redtail monkeys in logged and
unlogged forest of Kibale. Furthermore, I assess the quality of habitat and abundance of
resources in each area to develop possible explanations for observed differences in
behavior. This type of information is important for constructing informed forest
Kibale National Park (766 km2) is located in western Uganda (0 13'-0o 41'N and
300 19'-300 32'E), near the base of the Ruwenzori Mountains (Fig. 2-1). It is classified as
moist evergreen forest, transitional between lowland rain forest and montane forest
(Struhsaker 1997, Chapman and Lambert 2000). Situated at 1500 m elevation, the
Kanyawara study site in the northern part of the park, receives an average of 1741 mm of
rainfall per year, distributed over distinct biannual wet and dry seasons (Chapman and
Chapman, unpublished data). Kibale's logging history, coupled with a lack of hunting,
provides an ideal environment for examining primate response to logging.
I contrasted the behavior of a group of redtails in a heavily logged area, to one in
an unlogged area. The heavily logged area was harvested in 1969 (Struhsaker 1997), and
approximately 7.4 stems/ha or 21m3/ha were removed (Skorupa 1988; Struhsaker 1997).
However, as a result of high incidental damage associated with the felling, Skorupa
(1988) estimates that approximately 50% of all trees in the area were destroyed. Since
then, many areas have become dominated by dense, thorny herbaceous vegetation,
frequently dominated by semi-woody Acanthaceae species (primarily, Acanthus
arborescens), which surrounds emergent colonizer or remnant trees (Struhsaker 1997,
Paul et al. in press). The area used by the focal monkey group also abuts smallholder
agricultural fields and tea plantations. The unlogged area has never been commercially
harvested. A few large stems (0.03-0.04 trees/ha) were removed before 1970; however,
this low level of extraction appears to have had little effect on the structure of the forest
(Skorupa and Kasenene 1984, Skorupa 1988, Struhsaker 1997, Chapman et al. 2000).
There is an extensive, gridded trail system in both areas (Struhsaker 1997).
Diet and behavioral data were collected from one group of redtails in each site
between August and December 2002. Both groups have been previously studied and
were accustomed to human presence; nevertheless, to ensure that both were adequately
habituated, a field assistant and I followed each group for two months prior to collecting
quantitative data. The group in the unlogged forest, group UL, consisted of 24
individuals, while the group in the heavily logged area, group HL, had 16 individuals.
These numbers are representative of the mean group sizes of groups in unlogged and
heavily logged forest at Kibale (unlogged: 27.70.9, n= 3, heavily logged: 15.00.5, n=
3; K. Rode, unpublished data).
Behavioral scores were recorded during all-day follows using a 15-minute scan
sampling regime. We attempted to carry out follows in blocks of 3 days, spaced at 1-
week intervals, for each group. However, due to the often extremely dense and
impassable vegetation in the heavily logged area, we often followed group HL for 4 days
and group UL for 2 days to obtain comparable numbers of observations per week (UL:
2.40.76 days/week, 20.7+7.05 hours/week; HL: 2.40.70 days/week, 18.1+6.88
hours/week). Scans lasted 10 minutes or until 5 individuals had been sampled, whichever
came first. Observers recorded the time and activity maintained by an individual for at
least 5 sec. Individuals were never scored more than once during a scan, and we tried to
avoid scoring the same individual in two consecutive scans. Twenty-one different
behavioral categories were scored, but subsequently collapsed into six categories for
analyses: feed, search for insects, rest, move, social, and other (which included drinking,
urinating, and defecating). If an individual was feeding, observers recorded the food part
(e.g., insect, fruit, leaf) and food species, if known (less than 0.01% of species could not
Habitat use was quantified by estimating each individual's height in the canopy
and ranking the density of the understory vegetation below the individual. Height in the
canopy was scored on a scale of 1 to 5, with the scale representing the following height
divisions 1 = 0-7 m (4.72.9); 2 = 7-14 m (12.94.6); 3 = 14-18 m (16.74.7); 4 = 18-23
m (20.43.6); 5 = 23 m or higher (24.82.2). Understory vegetation was also scored on a
scale of 1 to 5, which ranged from 1 = very open (no or almost no vegetation, generally
only tree seedlings or saplings) to 5 = very dense (generally herbaceous, often a
combination of Graminaceae and Acanthaceae species, generally impassable for
The presence of other primate species, and other groups of redtails or potential
predators within 50 m of the group was recorded at the end of each scan period. Group
spread was estimated every 15 minutes as the length of the axes of an ellipse whose
center was the perceived center of mass of the group. The surface area of the ellipse was
calculated and averaged over all scans. I calculated the square root of this measure to
obtain mean group spread.
We recorded the location of the center of the group every 15 min by measuring
the distance and angle to the nearest point with known coordinates. Reference points
were located at 20 m intervals on the gridded trail system in each group's home range.
Reference points were recorded using a hand-held global positioning system.
Coordinates for group locations were calculated and plotted in the ArcView 3.2 and
ArcGIS 8.3 geographic information systems (ESRI 2002). This method allows more
exact measurement of day range and home range estimation because the measurements
are based on geographically accurate positions. Moreover, recording monkey locations
in a GIS database enables a richer and more quantitative analysis of habitat and space use
by facilitating the integration of animal movements with multiple habitat layers (Hooge et
Resource and Habitat Assessment
Resources and habitat quality were assessed for an area at least as large as that
considered to constitute each redtail group's stable home range, based on observations of
these groups for 12 months prior to the current study (K. Rode, unpublished data). In the
areas used by the focal groups, all trees above 10 cm diameter at breast height (DBH)
were identified to species, measured, and plotted in the GIS. Less than 1 % of all stems
mapped could not be identified to species. We measured the distance and angle to each
tree from the reference points located throughout each study area and calculated
geographic coordinates for each tree (UL: 15,489 trees over 48.4 ha; HL: 8433 trees over
40.1 ha). The availability of plant food resources was assessed by categorizing each tree
as "food" or "nonfood" based on the feeding observations. For example, in the unlogged
area only those tree species that monkeys fed on between August and December 2002 in
this area were scored as "food" species. Stem density and total basal area were calculated
for all trees, all food trees, and specific species. Shannon diversity and evenness were
calculated for the entire tree community and for the suite of potential food trees in each
To assess other habitat characteristics likely to be important to redtails, such as
vertical complexity and density of understory vegetation, I established a set of 50 m2
plots in each home range, spatially stratified by spacing plots every 20 m along the trail
grid in each area. Trails were usually no more than 50 to 100 m apart, resulting in 1484
plots in the unlogged area and 1362 plots in the heavily logged area. The plots assessed
7.42 ha (37%) and 6.81 ha (32%) of the total home range of the focal group in the
unlogged and heavily logged forest. Vertical stratification (i.e., number of canopy layers)
was assessed by scoring presence or absence of vegetation in each of the height
Home range size was estimated in two ways. I used the Animal Movement
extension for ArcView (Hooge and Eichenlaub 1997) to estimate total and monthly area
used by the minimum convex polygon method. The MCP method estimates the area both
used and probably traversed by each group, but likely contains areas not used by the
animals (Burt 1943). Although this method generally overestimates home range size, it is
the most commonly employed and therefore facilitates comparison with other studies
(Hooge et al. 2001). I also estimated home range sizes using grid-cell analysis (White
and Garrot 1990), which has been employed in other studies of primate ranging (Kaplin
2001, Di Fiore 2003). The GCA method reflects only the area the group used during
observation periods, and thus, provides a conservative estimate of home range size
(White and Garrott 1990). However, it is highly dependent on the size of grid cell
utilized (White and Garrott 1990, Hooge et al. 2001). In this case, a 30 x 30 m cell size
was determined from the observed mean group spread of each group (unlogged: x =
29.60.16; heavily logged: x = 31.40.22), and from the groups' 30 m movement
distance between observation intervals during foraging. The best estimate of home range
area for each group is likely to fall in between the values calculated by the MCP and
GCA methods. For each method, mean monthly home range sizes were compared using
Home range use was measured by superimposing a grid of 30 x 30 m cells over the
plotted group locations. To assess home range use, I counted the number of different grid
cells the troop entered each day, the total number of cells entered per day, and the
cumulative number of different cells entered per month per troop. In addition, I
calculated the number of times a grid cell was entered during the study for each group.
Daily travel distances were calculated, by summing the straight line distances
between successive location points for each day. I used only full observation days in
evaluating and comparing average daily travel distances. Full days were defined as
beginning by at least 0800 h and lasting until at least 1800 h. This also allows
comparison with previous studies of redtails at Kibale that used the same time periods
(Lambert 1997, K. Rode unpublished data). Mean daily travel distances were compared
between groups using t-tests and within groups using Kruskal-Wallis tests.
For each site, the entire group served as the unit of analysis. Therefore, the
proportion that each activity type and each food part contributed to each scan was
calculated. For understory vegetation density and height in canopy, scores for individual
observations within each scan period were averaged. To evaluate the appropriate
sampling interval, I contrasted monthly proportion of scans that each activity comprised
using 15- and 60-min scan intervals for each group. Based on our observation, redtails
rarely maintained the same activity for 60 minutes. There was no significant difference
using either scan interval (15 or 60 min) for any of the activity categories for either group
(Table 2-1). As a result, each 15-min scan interval was considered to be an independent
observation, and scans were pooled to determine overall and monthly activity budgets,
diet composition (by food part) and overall habitat use. Monthly activity budgets
between groups were compared using Mann-Whitney tests.
I determined the contribution of various plant species to the monthly diets of the
two groups by calculating their simple proportional representation among the set of all
feeding records for each group. Diet composition and selection were compared using a
variety of indices. Dietary diversity and evenness was calculated for both monthly and
overall diets using the Shannon-Wiener (H') index and Shannon evenness (E). Shannon
t-tests were used to compare diversity of diets by calculating a variance of H' and degrees
of freedom based on the proportional abundance of individual food species in each
group's overall diet (Magurran 1988). Overlap between the diets was calculated using
the Holmes-Pitelka index, Di = ZSi, where Di is the total percent overlap and Si is the
percent overlap between shared food items (Holmes and Pitelka 1968; Struhsaker 1975).
Finally, selectivity of food items relative to their availability was assessed using the
Strauss selectivity index (L): L = ri pi, where ri and pi equal the percent of the food
item in the diet and the environment, respectively (Strauss 1979). This index is less
sensitive than many other commonly used selectivity indices to the presence of rare food
items. Preference values range from -1 to +1 and are centered on zero, which indicates
random feeding; negative values indicate avoidance and positive values indicate
In cases where nonparametric statistical tests were used, equivalent parametric
tests were carried out and found to show the same general trends (i.e., significant or non-
significant probabilities). Therefore, even in cases where non-parametric tests are
performed, parametric means are presented in figures because they are more comparable
to other studies.
The UL group spent more time engaged in feeding than the HL group (Mann-
Whitney: p = 0.008). However, HL spent significantly more time searching for insects,
moving, and resting than group UL (search: p = 0.016; move: p = 0.008; rest: p = 0.016).
Overall, the UL group spent nearly 50% of its time feeding and 20% of its time searching
for insects (Fig. 2-2). The group in the heavily logged area devoted approximately 30%
of its time to feeding and 30% to searching for insects. Time spent resting differed
markedly between the two groups, with UL group spending about 15% of its time resting
and HL group spending 25% of the time resting. HL group was engaged in moving 3
times more frequently than UL group. Time spent engaged in social activities, such as
grooming, did not differ between the groups (p = 0.69), and comprised 7.31.80% and
7.94.14% of the UL and HL groups' activity budget, respectively.
Home Range Size and Use
The mean monthly home range size of UL group based on the minimum convex
polygon (MCP) method was twice as large than when grid-cell analysis (GCA) was used
(sign test paired by month, Z = 1.788, p < 0.01, Table 2-2). The total area used by the
UL group was 32.5 ha, based on MCP, and 20.2 ha, based on GCA. The group used 40-
57% of the total home range based on GCA in any one month, and 48-76% of total MCP
calculated home range. The mean monthly home range area of the HL group based on
the MCP method was more than three times larger than the GCA estimate (sign test, Z =
1.788, p < 0.01, Table 2-2). The total area used by the HL group was 49.0 ha, based on
MCP, and 21.3 ha, based on GCA. The group used 49-63% of total home range based on
MCP in any one month and 33-46% of total GCA-calculated home range. Monthly home
range size was larger for HL group than the UL group when estimated using MCP (Mann
Whitney: U = 1.0, df= 8; p = 0.016). When monthly home range size was estimated
using GCA, there was no difference between the groups (U = 5.0, df= 8, p = 0.117).
To explain these inconsistencies between the two methods of estimating home
range size, I examined the pattern of home range use of each group (Fig. 2-3). Both
groups entered new quadrats throughout the study, suggesting that the home range of
each group will be larger than what was recorded over 5 months (Fig. 2-3). Nevertheless,
the pattern of home range use between groups appears consistent over that time.
Generally, UL group entered more grid cells per month, but HL group entered slightly
more new cells per month (UL: 46.6 cells per month 49.26; HL: 48.4 cells per month
35.01). This result is predicted by an examination of spatial patterns of home range use
of each of the groups (Fig. 2-4). Overall, more contiguous grid cells in the UL home
range were entered at least once during the period (i.e., there are very few unshaded
quadrats in UL's range use map within the outer boundary of the recorded home range),
indicating that the entire bounded area was used more intensively (Fig. 2-4b). In
contrast, group HL's home range use map shows a larger number of quadrats not entered,
with more widely dispersed areas of use (Fig. 2-4a).
The mean distance traveled each day was greater for the group in the logged area
than the group in the unlogged forest (UL group: 1111 m + 265.0; HL group: 1332 m
387.8; t = -2.677, df= 61, p = 0.01). Mean daily travel distance for group UL did not
differ among months (Kruskal-Wallis: x2 = 2.728; df = 4; p = 0.604), but varied among
months for the HL group (X2 = 10.692; df= 4; p = 0.03; Fig. 2-5).
The majority of feeding observations for both groups were on insects, but a
significantly greater proportion of feeding observations were made on insects for the HL
group than the UL group (UL: 62.3%; HL: 75.5%; t = 7.042; df= 2221; p <0.001).
Although insects made up the bulk of feeding observations for both groups, they
consumed a wide variety of other food types, as well (Fig. 2-6). A greater proportion of
feeding observations of the UL group was comprised of mature leaves, young leaves,
petioles, and ripe fruits (mature leaves: t = 3.389; df = 2221; p=0.001; young leaves:
t=4.806; df= 2221; p<0.001; petioles: t= 4.245; df= 2221; p<0.001; ripe fruit: t = 8.596;
df = 2221; p <0.001), but this group spent significantly less time feeding on flowers than
group HL (t = -6.036; df= 2221; p< 0.001). Whereas more than a third of the UL
group's diet was composed of fruits, leaves, seeds, and flowers, less than 25% of the HL
group's overall diet was composed of these plant foods.
Dietary Selection and Diversity
Consistent with previous studies (Gautier-Hion 1988), both group's plant diets
were dominated by relatively few plant species. Monkeys in the UL group selected plant
foods from 70 different species, but 81.8% of all feeding observations came from only 12
species (Table 2-3). Of these, one was a terrestrial herb (Marantochloa leucantha),
which was not common in the heavily logged area and one was a liana, while the
remaining are trees. Three of the remaining 10 tree species were small trees (mean DBH
< 15cm). In contrast, the plants comprising 80.9% of the HL group's plant diet were all
medium to large trees (mean DBH > 15cm). The HL group's plant diet was selected
from 31 species, but only 10 tree species comprised 80% of the observations.
UL group's overall plant diet, over the study, was more diverse than that of HL
group (Shannon t = 5.24; df= 808; p < 0.001; Table 2-4). Mean dietary overlap across
months between the groups in the different areas was 21.7% (+6.45 s.d.), but ranged from
15.6 to 30.2%. Redtails in the unlogged area also consumed fruit from eight different
Ficus species, whereas monkeys in the heavily logged area fed from only two fig species.
The tree community above 10cm DBH in the unlogged area was less diverse
(Shannon t = -37.30; df = 20641; p<0.001) than that of the heavily logged area, yet had a
50% higher stem density (Table 2-5). When I considered only food species, I found that
nearly 90% of the individuals inventoried in the unlogged area were potential food trees,
whereas fewer than 60% of individuals fell in this category in the heavily logged area
(Table 2-5). Food tree diversity also differed significantly between the two areas
(Shannon t= 4.67, df= 13838, p<0.001). Stem density of potential food trees was almost
2.5 times greater in the unlogged area in comparison to the heavily logged area.
The Strauss selectivity index (L) showed that, of the most preferred food species,
only UL monkeys' choice ofFicus exasperata was not different from random, and only
Diospyros abyssinica was selected less than it was available (Fig. 2-7a). The remaining
species were selected in greater proportions than they were available in the forest.
However, since the herb M. leucantha and the vine of the Piper sp. were not inventoried,
it was difficult to judge how much the monkeys selection of these species differed from
their respective distributions in the unlogged forest. Similarly, Rothmannia urcelliformis,
Craterispermum laurinum, and Tarennapavioides are small trees, so they were likely to
be underrepresented in the inventory if the mean DBH for each species' population is not
at least 10cm. However, it was also clear that the monkeys were not only selecting
species that were among the most dominant in the forest. Linocierajohnsonii and
Macaranga /I ei/ lu thii i/il, among the most preferred food species for UL group, did not
count among the most dominant species in the tree community in the unlogged forest
(Table 2-3; Fig. 2-7).
In the heavily logged area, monkeys selected eight of their ten most preferred
food species more than they were available in the forest (Fig. 2-7b). This was
particularly true for Symphonia globulifera and D. abyssinica. They selected Celtis
durandii less than expected based on its availability, and did not select Celtis africana.
Of the 10 most preferred food species, all were among the most abundant in the forest.
Redtails in the heavily logged forest occupied the highest part of the canopy (22 m
or higher) 86% of the time (Fig. 2-8). They were never observed to descend to the
ground and were observed in lower/emerging understory vegetation (up to 7 m) less than
1% of the time. The animals in the unlogged area spent 85% of their time in a
combination of three layers: the upper canopy (43.0%), the lower-canopy (11.8%) and
lower/emerging understory (25.6%; Fig. 2-8). This difference is unlikely to be due to the
relative availability of the different canopy layers, since an assessment of the available
number of canopy layers in each area showed a similar distribution (Fig. 2-9; statistics
are presented in the figure legend). Both groups occupied the two upper canopy layers
more than expected and the lower layers less than expected, given the proportional
distribution of each height category in the home ranges.
The two areas differed in the density of understory vegetation. The heavily
logged area was dominated by very dense herbaceous growth and very little open
understory, whereas the unlogged area showed a more even distribution of understory
vegetation density (Fig. 2-10). The groups differed in their association with different
classes of understory vegetation. In the unlogged area, redtails were associated with the
majority of understory vegetation classes in proportion to their distribution. The
monkeys were observed to associate with the most dense vegetation less than expected,
given the proportional distribution of vegetation density classes in the home range (x =
26.89, df = 1, p<0.001). In contrast, redtails in the heavily logged area associated with
the two most open understory vegetation classes more than expected and with the more
closed understories less than expected (Fig. 2-11). Open understory was rarer in the
heavily logged site, so it appears that HL redtails were selecting areas of less dense
vegetation and that this habitat attribute may be more important in explaining redtail
presence than the number of canopy layers.
Mixed Species Associations
The UL group spent more of its time (65.7%) in mixed-species associations than
the HL group (36.2%) (x2 = 503.93; df= 1; p<0.001). Of the time UL group spent in
mixed-species associations, 72% was with red colobus, 31% with blue monkeys, and
20% with black-and-white colobus (Fig. 2-12). HL group spent 52% of its time in
association with red colobus, 24% with blue monkeys, 24% with black-and-white
colobus, and 20% with grey-cheeked mangabeys. Because the monkeys sometimes
associate in groups of more than two species, these percentages do sum to 100. When
percent of time observed in association was compared with the percent each species
comprises of the primate community in each area (derived from Chapman et al. 2000),
redtails in the both areas were found to associate with red colobus (UL: 2 = 1181.65, df
= 1, p<0.001; HL: x2 = 137.16, df =1, p<0.001) and blue monkeys (UL: x2 = 1424.92, df
= 1, p<0.001; HL: 2 = 1059.88, df= 1, p<0.001) more than expected. In the heavily
logged area, redtails associate with black-and-white colobus less than expected (x =
224.83, df= 1, p<0.001), whereas associations between these two species occur
proportional to the black-and-white colobus' density in the unlogged area (x2 = 224.83,
df=l, p=0.021). Association with mangabeys is more than expected for redtails in the HL
group (x2= 189.41, df =1, p<0.001) and less than expected in the unlogged area (2 =
53.08, df = 1, p<0.001).
Redtail monkeys inhabiting heavily logged forests used larger home ranges, had
lower dietary diversity and spent less time feeding than conspecifics inhabiting unlogged
forest habitats. Despite having a smaller group size, redtail monkeys in logged areas
travel further and spend more time moving than redtails in unlogged habitats. These
results suggest that redtails in logged forests are spending more time and expending more
energy in search of food (Milton 1984; Janson 1988). The results of my habitat
assessment further indicate that reduced abundance of resources and suitable habitat
constrain redtails in the heavily logged forest, which may explain lower population
The quality and quantity of resources available to the monkeys is likely to strongly
influence behavioral patterns and reproductive success (Janson and Chapman 2000). To
compensate for the greater energy expenditure by redtails in the heavily logged area, I
would predict higher feeding and foraging rates. My observations confirm that although
the HL group spent more time searching for insects than group UL, it did not spend more
time feeding. I explain this pattern by the decrease in potential plant food sources in the
heavily logged area. Our inventory of trees indicates that there are fewer potential food
trees in the heavily logged area. This corresponds with an inventory conducted in 1980-
81, 11 years after logging operations ended (Skorupa 1988). In addition, the heavily
logged area had a lower density of of figs, which appear to be important to redtails in
both areas (Struhsaker 1997). In other forests, this loss of food trees appears to be
counterbalanced by the colonization of gaps and cleared areas by important fruit-
producing pioneer species, such as Musanga sp. (Johns and Skorupa 1987). At Kibale,
however, pioneer tree species that colonize after large disturbances are not important
food sources for frugivores (Struhsaker 1997). Finally, trees in heavily logged areas may
also be less productive. For example, Skorupa (1988) reports a 25% reduction in fruiting
and young leaf production in one heavily logged part of Kibale. It appears likely, then,
that redtails in the heavily logged area feed less on plant foods because the latter are both
less abundant and less productive.
Redtails may compensate for increased energy expenditure by feeding more on
insects. Skorupa (1988) provides indirect evidence that insect abundances could be
higher in heavily logged areas at Kibale. He suggests that the lower production of young
leaf and fruit in heavily logged areas at Kibale may be due in part to increased pressure
from insects (Skorupa 1988). A better assessment of insect resources is necessary to
determine to whether lack of plant resource availability leads to increased insect feeding
or if increased insect abundance results in the higher number of observations of insect
The habitat structure in the heavily logged area may also contribute to the
behavioral patterns observed in the focal group. The results indicate that habitat may be
of lower quality for redtails in the heavily logged area. Redtails exploit numerous canopy
strata, frequently foraging in the mid-canopy layers and even descending to the ground to
forage (Struhsaker and Leland 1979, Gebo and Chapman 1995). It is likely that the
insect resources comprising the bulk of redtails' diets are located throughout the canopy
and that their vertical locations within the canopy may vary daily or seasonally (Terborgh
1993). This may partially explain redtails' general preference for mid-canopy layers in
the unlogged area. Our habitat assessment shows a distinct difference in amount of mid-
canopy available to redtails and in the density of understory vegetation in logged areas.
Since the vegetation in many parts of the heavily logged areas consists primarily of dense
thickets of herbaceous vegetation and semi-woody plants dominated by Acanthaceae
species, descending to the ground or the lower canopy appears to be physically
prohibitive, limiting monkeys to the upper strata. Moreover, the understory plant M
leucantha appears to be important for redtails in the unlogged areas. This plant is often
found in large patches throughout the undisturbed area. By contrast, very few isolated M.
leucantha plants were observed in the heavily logged area. I propose that redtails in
unlogged areas can exploit smaller areas-and coexist in higher densities-at least
particially because the vertical component of their environment provides more resources
and habitat than in the heavily logged areas.
Overall, I conclude that both resource abundance and the amount of suitable habitat
are reduced, causing redtails to expend greater effort in searching for food in the heavily
logged areas of Kibale. The fact that redtail population densities in these logged areas
were still declining 28 years after logging (Chapman et al. 2000), suggests that redtails in
heavily logged areas are still above the carrying capacity of the forest. Whereas
colobines at Kibale appear to compensate for ecological differences resulting from
habitat disturbance by adjusting their group size, a reduction in redtail group sizes in
logged areas has still not resulted in a stable population density. Other studies suggest
that parasite loads (Gillespie et al., submitted) and predation risk (Struhsaker and Leakey
1990, Mitani et al. 2001) could also be contributing to this decline. However, the results
presented here suggest that constraints on the particular habitat and resource requirements
of redtails are the primary drivers of population decline in redtails.
This study points to the importance not only of structure and composition of
regenerating forest in predicting the response of primates to disturbance, but of the
specific habitat requirements of different species. Members of the C. cephus lineage are
generally assumed to be highly flexible in their dietary requirements and highly adaptable
to regenerating and other disturbed forests (Thomas 1991, Kingdon 1997). However,
their preference for mid-canopy vegetation seems to make them more vulnerable to the
effects of logging than other species that prefer upper canopy strata. Conventional
logging generally alters the character of the understory through incidental damage
(Thiollay 1992, Laurance and Laurance 1996). Where timber stands are managed, plans
often call for the removal of vines, which normally serve as access routes and food
sources for nonvolant arboreal animals (Galletti et al. 1994, Putz et al. 2001). When
these vines are removed, animals suffer the loss of functional habitat areas. For the large
number of primate and other species that occupy the mid- and understory niche within a
community (Thomas 1991, Boinski and Sirot 1997, Putz et al. 2001), these management
strategies are especially harmful.
Management objectives focused on both timber harvest and maintenance of
healthy wildlife populations will have to mitigate the impacts of this disturbance. In the
case of Kibale, restoration efforts would have to focus not only on replanting food trees
and improving accessibility to those trees, but also on improving mid-canopy and
understory habitat. In another example, although reduced impact logging prescribes vine
cutting for target trees, the number of trees removed and the overall damage to the forest
is reduced from that caused by conventional logging practices, indicating that the amount
of habitat loss might be reduced as well (Holmes et al. 2002). This scheme appears to be
more cost effective than conventional logging, so it presents an interesting possibility for
compromise between the interests of foresters and wildlife conservationists. However,
research on the effects of the disturbance caused by reduced impact logging is still too
limited to unequivocally advocate the approach.
At Kibale, as in many sites, long-term impacts of logging on population densities
vary widely among primate species, but these differences may not be apparent when
behavioral responses are not understood. Thus, the ability of management or restoration
plans to take into account different species' resource and habitat needs is likely to be
critical in maintaining viable populations following logging disturbance.
Table 2-1. Test of independence for group activity samples pooled by 60-min or 15-min
sample for redtail groups in unlogged and heavily logged forest at Kibale
National Park, Uganda.
j Test of Independencea
Unlogged Heavily Logged
z p z p
Feeding 0.008 1.0 0.104 0.99
Searching 0.008 1.0 0.012 1.0
Traveling 0.32 0.99 0.32 0.99
Resting 0.014 1.0 0.083 0.99
Social 0.019 1.0 0.051 1.0
Other 0.21 0.995 0.66 0.96
adf = 4
Table 2-2. Monthly and mean home range size (ha) for redtail groups in logged and
unlogged forest in Kibale National Park, Uganda between August and
December 2002 (UL = unlogged group, HL, heavily logged group).
Home range area (ha) MCP1 Home range area (ha) GCA2
Month UL HL UL HL
August 21.13 23.94 11.43 9.18
September 17.23 30.39 8.19 9.90
October 22.89 30.00 10.44 8.73
November 24.60 30.80 9.45 8.37
December 15.54 29.11 9.99 7.02
Mean (+/-SD) 20.28 (+/-3.81) 28.85 (+/-2.81) 9.90 (+/-1.20) 8.64 (+/-1.07)
TOTAL 32.49 48.97 20.16 21.33
1MCP=minimum convex polygon method; 2GCA=grid-cell analysis, based on 30x30 m grid cell
Table 2-3. Time spent eating different types of foods by two redtail groups in unlogged
and heavily logged forest of Kibale National Park, Uganda, from tree species
comprising 80% of their respective feeding time (UL (unlogged) = 81.84%;
IL (heavily logged)= 80.9%).
4.5 Celtis Africana
Ripe fruit 2.5
Ripe fruit 6.7
Table 2-3. Continued
Group UL Group HL
Species (Family), Part consumed % in Species Part % in
density diet (Family), density consumed diet
Ficus exasperata Ripe fruit 1.7 Ficus Ripe fruit 2.3
(Moraceae), 7.20 Unripe fruit 0.1 sansibarica Unripe fruit 0.4
Piper gnensis Ripe fruit 1.8
Tarenna Mature leaves 1.1
pavetoides Young leaves 0.2
(Rubiaceae), Unripe fruit 0.4
Total 81.8 80.9
Table 2-4. Dietary diversity and overlap across months for redtail groups in logged and
unlogged forest of Kibale National Park, Uganda.
Month Dietary Diversity (H') % Diet Overlap
Unlogged Heavily Logged
August 3.04 2.77 30.2
September 2.84 2.32 15.6
October 2.54 2.30 20.0
November 2.54 2.21 26.5
December 2.58 1.28 16.3
21.7 (+ 6.45)
Table 2-5. Characteristics of the tree community in home range areas of focal redtail
groups in unlogged and heavily logged forest areas of Kibale National Park,
Uganda, for all trees above 10cm DBH. Food trees species were identified
based on feeding observations for each group during the study period from
August to December 2002.
Unlogged Heavily Logged
All Trees Food Trees All Trees Food Trees
Species richness (S) 106 49 (46.23%)' 110 24 (21.82%)1
Individuals (N) 13542 11958 (88.30%)2 8433 4868 (57.73%)2
Simpson (1/D) 13.085 10.29 27.852 11.324
Shannon (H') 3.131 2.704 3.752 2.638
Shannon evenness 0.6714 0.6984 0.7982 0.8413
Stem Density (trees/ha) 334.4 295.3 210.3 121.4
Total Basal Area (m2) 4141.86 3700.36 2645.04 1905.02
1 Figure in parentheses indicates percent of total species
2 Figure in parentheses indicates percent of total number of individuals
M '0 5 10 kilometers
Figure. 2-1. The Kanyawara study site is located in Kibale National Park, western
Uganda. Maps are adapted from MUBFS 2003.
8 30 *
Feed Forage Travel Rest Social Other
Figure 2-2. Percentage of time engaged in different activities by redtails in unlogged
unshadedd) and heavily logged (shaded) forest areas of Kibale National Park,
Uganda. Bars indicate standard error, and asterisks indicate significantly
greater proportions of time devoted to different activities.
August September Ocbber November December
Figure 2-3. Monthly total grid cells entered and cumulative grid cells entered by redtails
in unlogged unshadedd) and heavily logged (shaded) areas of Kibale National
Park, Uganda. The bars indicate the number of different grid cells entered per
month by each group unshadedd = unlogged; shaded = heavily logged). The
lines show the cumulative number of new grid cells enter over the study
period (unlogged = solid line; heavily logged = dashed line).
percent of Observations
0 250 500 Meters
Figure 2-4. Pattern of home range use by redtail gri (b) a (a) heavily logged and (b)
unlogged forest of Kibale National Park, Uganda. Darker shading indicates a
greater percentage of times that the group was present in a given 30x30 m grid
cell. The relative position of the grids to each other reflects the true spatial
relationship of the two home ranges.
Figure 2-5. Daily distance traveled by redtail monkeys in logged (shaded) and unlogged
unshadedd) areas of Kibale National Park, Uganda. Boxes indicate the
interquartile range of distances per month per group. The line inside each box
indicates the mean distance traveled each month and the bars show the
extreme values obtained each month.
E- 300 -
ML YL FL P SD RF URF Insect Other
Figure 2-6. Average percent of monthly feeding observations on different items by redtail
monkeys in unlogged unshadedd) and heavily logged (shaded) forest in Kibale
National Park, Uganda. Lines above bars indicate standard error. ML =
mature leaves; YL = young leaves; FL = flowers; P = petioles; SD = seeds;
RF = ripe fruit; URF = unripe fruit. The group in the unlogged forest was
observed to consume seeds on a few occasions, but the amount was negligible
relative to other food items.
0 0.05 0.1 0.15
-0.05 0 0.05 0.1 0.15 0.2 0.25 0.3
Figure 2-7. Strauss selectivity index (measures selectivity vs. dominance) for top 80% of
food species selected by redtails in (a) unlogged and (b)heavily logged areas
of Kibale National Park, Uganda. Abbreviations: BlUn = Blighia unijugata;
BoPh = Bosqueia phoberos; CeAf= Celtis africana; CeDu = Celtis durandii;
CrLa = Craterispermum laurinum; DiAb = Diaspyros abyssinica; FiBr =
Ficus brachyleps; FiEx = Ficus excelsa; LiJo = Linocierajohnsonii; MaLe =
Marantochloa 'let i/lha, MaSc = Macaranga % /I ei//n thlii, MiDu =
Millettia dura; MiBa = Mimusops bagshaweii; PaEx = Paranari excelsa;
PiGn = Piper gnensis; RoUr = Rothmannia urcelliformis; SyGl = Symphonia
globulifera; TaPa = Tarenna pavioides.
C 40- 43
S- 26 GROUP
5 4 3 2 1
Height in canopy
Figure 2-8. Height in canopy of redtails in unlogged and heavily logged forest at Kibale
National Park, Uganda, by percent of observations in each category. Height
category scale: 1 = 0-7m (x=4.72.9); 2 = 7-14m (x = 12.94.6); 3 = 14-18m
(x = 16.74.7); 4 = 18-23m (x= 20.43.6); 5 = 23m or higher (x = 24.82.2).
40- 41 42
1 2 3 4 5
Number of vegetated height categories
Figure 2-9. Distribution of number of canopy layers observed in home ranges of redtails
in heavily logged and unlogged areas of Kibale National Park, Uganda. Plots
were scored as having vegetation in from 1 to all 5 of the possible height
categories. Height categories were as follows: 0-7m; 7-14m; 14-18m; 18-
23m; 23m or higher. Height in canopy of redtails relative to vegetated height
categories in each area: Unlogged: Class 1: 2 = 5379.033; df =l1; p < 0.001,
Class 2: 2 = 239.889; df= 1; p < 0.001, Class 3: 2 = 104.21; df= 1; p <
0.001, Class 4: X2 = 273.489; df= 1; p < 0.001, Class 5: X2 = 140.688; df= 1;
p < 0.001; Heavily Logged: Class 1: x2 = 80,004.19; df= 1; p < 0.001, Class
2: X2 =13.2129; df= 1; p < 0.001, Class 3: 2 = 769.0676; df= 1; p < 0.001,
Class 4: 1265.837; df= 1; p < 0.001, Class 5: n/a (0 observations).
1 2 3 4 5
Understory vegetation density ranking
Figure 2-10. Understory vegetation associated with redtail location in unlogged
unshadedd) and heavily logged (shaded) forest at Kibale National Park,
Uganda, by percent of observations in each category. Understory vegetation
was scored on a scale of 1 to 5, where 1 = very open (no or almost no
vegetation, generally only tree seedlings or saplings) and 5 = very dense
(generally herbaceous, often a combination of Graminaceae and Acanthaceae
species). Lines indicate the distribution of understory vegetation density
across categories, observed in the unlogged (solid line) and heavily logged
(dashed line) home ranges. Association of redtails with understory vegetation
classes in each area: Unlogged: Class 1: 2 = 0.194; df =l; p < 0.95, Class 2:
2 = 1.817; df= 1; p > 0.10, Class 3: 2 = 3.613; df= 1; p = 0.06, Class 4: 2
3.466; df= 1; p = 0.07, Class 5: X2 = 26.893; df= 1; p < 0.001; Heavily
Logged: Class 1: x2 = 994.73; df=l; p < 0.001, Class 2: 2 = 168.04; df= 1; p
< 0.001, Class 3: X2 = 5.825; df= 1; p = 0.04, Class 4: X2 = 5.825; df= 1; p
0.04, Class 5: 2 = 86.331; df= 1; p < 0.001. Asterisks indicate that redtails'
association with an understory vegetation class was different than expected.
RC BM BWC MB
Figure 2-11. Monthly percent of time spent in mixed species associations for redtails in
unlogged unshadedd) and heavily logged (shaded) areas of Kibale National
Park Uganda. RC = red colobus (Procolobus badius), BM = blue monkey
(Cercopithecus mitis), BWC = black and white colobus (Colobus guereza),
MB = grey-cheeked mangabey (Lophocebus albigena).
(b) Heavily Logged
Figure 2-12. The proportion of observations in which redtails in (a) unlogged and (b)
heavily logged areas of Kibale National Park, Uganda were associated with
each of four other prominent species in the diurnal primate community
unshadedd), compared with each species' proportional representation in the
community (shaded). RC = red colobus (Procolobus badius), BM = blue
monkey (Cercopithecus mitis), BWC = black and white colobus (Colobus
guereza), MB = grey-cheeked mangabey (Lophocebus albigena).
THE EFFECTS OF PATCH STRUCTURE ON REDTAIL MONKEY (Cercopithecus
ascanius) HABITAT USE IN UNLOGGED AND HEAVILY LOGGED AREAS OF
KIBALE NATIONAL PARK, UGANDA
The timber industry is of central importance to many tropical countries, providing
sources of income and links to regional and global markets, among other things
(Naughton-Treves and Weber 2001). However, logging practices frequently damage and
degrade forests to the point of compromising their ability to provide important ecological
services, such as nutrient and water cycling and habitat for wildlife (FAO 2003). As a
result, recent years have seen a dramatic rise in efforts toward more sustainable logging
practices (Frumhoff 1995, Lindenmayer et al. 1999). However, neither the impact of
these schemes, nor the more conventional plans, on forest wildlife is not well-known. In
general, logging primarily affects habitat structure and the resource base (Fimbel et al.
2001), but it can also lead to forest fragmentation, which results in reduced landscape
connectivity, smaller patch sizes, increased edge effects, and population isolation
(Laurance and Bierregaard 1997).
The occurrence and the magnitude of fragmentation effects, as well as their effect
on wildlife, are determined by the spatial extent and intensity of the logging operations.
Clear-cutting is generally the most detrimental to forest species, but conventional
selective logging operations can often have significant negative impacts, as well (Fimbel
et al. 2001). In a comparison of the simulated effects of alternative timber harvest plans
on landscape structure, Gustafson and Crow (1996) found that a plan prescribing group
selection (the removal of small groups of trees) created nearly as much edge as a plan
calling for clear-cutting of a much larger area of forest. The more limited size of harvest
areas in the group selection plan affected forest spatial pattern more than the greater
intensity of harvest in the clearcut plan, implying that the selective logging plan may
have more negative consequences for forest dwelling animals. In an example from
Venezuela, enrichment strips (a silvicultural treatment in which strips are cleared and
replanted with commercial seedlings following selective logging) negatively affected
understory and terrestrial birds (Mason 1996). Strip enrichment was more damaging than
selective logging because it opened larger areas of canopy and created barriers within the
Such spatial discontinuity in forested landscapes can have a number of
consequences for wildlife, including local extinction and changes in species composition.
The effects can extend to community and ecosystem processes insofar as animals play an
important role in seed dispersal and pollination and serve other functions. In the tropics,
up to 90% of all plant species are adapted for seed dispersal by vertebrates (Howe and
Smallwood 1982, Jansen and Zuidema 2001). As a result, the ability of vertebrates to
persist in and move around tropical forests is of great importance for natural regeneration
processes. There is, thus, a need to carefully assess the effects of the spatial pattern of
logging on forest continuity. In developing a forest management plan that supports such
different vertebrates, it is necessary to determine the amount and configuration of suitable
habitat that is important for different animal species. This is often not well known in the
tropics (Fimbel et al. 2001), so a first step must be to identify the combination of
attributes that defines suitable habitat for different species.
In this study, I examine the degree to which an analysis of habitat use creates a
better understanding of the variables that are important to forest-dwelling vertebrates, and
evaluate whether this information can contribute to the formulation of timber extraction
plans that are more compatible with the persistence wildlife populations. This study
focuses on the response of redtail monkeys (Cercopithecus ascanius schmidtii Matschie)
to habitat change as a result of logging 30 years earlier, in Kibale National Park, Uganda.
Redtails are part of a rich primate community in the park, consisting of 13 species and
one of the highest densities of primates in Africa (Struhsaker 1997). Periodic population
censuses since the time of the logging indicate a continued decline in redtail population
density in the heavily logged areas since the time of logging (Chapman et al. 2000).
Redtail density is significantly lower in the logged areas (1.04 groups/km2) than in the
unlogged areas (11.48 groups/km2; Chapman et al. 2000). Kibale provides a good model
for testing primate response to logging alone, since direct human disturbance through
other factors such as hunting is not a factor.
The heavily logged site in Kibale has half the stem density and basal area of trees
of the unlogged area (Stickler and Chapman, in prep); thus, total basal area will likely be
important in dictating redtail habitat use. Moreover, in the heavily logged area, 60% of
the trees were potential food sources for the monkeys, whereas nearly 90% of the trees in
the unlogged area were potential food sources (Stickler and Chapman, in prep). Thus the
spatial distribution of potential food trees will likely influence differences in habitat use.
I also predict that the type and quality of the understory will influence redtail habitat use.
Specifically, I expect that monkeys will avoid areas dominated by a dense understory
characterized by thorny vines or herbs because it reduces the vertical foraging area
(Fimbel et al. 2001), and since they exploit both the understory and the higher canopy
within forests (Struhsaker and Leland 1979; Cords 1986), I expect an intermediate
amount of understory vegetation will be preferred.
To identify the factors important in redtail habitat selection, I examined habitat use
by redtail monkeys in heavily logged and unlogged areas and examined the role of spatial
structure of habitat attributes and resources in determining behavioral differences.
Additionally, I compared habitat use models for monkeys in unlogged and heavily logged
areas to examine the extent of overlap in critical habitat characteristics.
Kibale National Park (766 km2) is located in western Uganda (0 13'-0o 41'N and
300 19'-300 32'E), near the base of the Ruwenzori Mountains (Fig. 3-1). It is classified as
moist evergreen forest, transitional between lowland rain forest and montane forest
(Struhsaker 1997, Chapman and Lambert 2000). Situated at 1500 m elevation, the
Kanyawara study site in the northern part of the park receives an average of 1741 mm of
rainfall per year, distributed over distinct biannual wet and dry seasons (Chapman and
Chapman, unpublished data). I contrasted habitat use by a group of redtails in a heavily
logged area (16 individuals), with that of a group in an unlogged area (24 individuals).
The heavily logged area was harvested in 1969 (Struhsaker 1997). Approximately 7.4
stems/ha or 21 m3/ha were removed (Skorupa 1988, Struhsaker 1997). However, as a
result of high incidental damage associated with the felling, Skorupa (1988) estimates
that approximately 50% of all trees in the area were destroyed. Since then, dense, thorny
herbaceous vegetation has grown in many areas, which are frequently dominated by
semi-woody Acanthaceae species (primarily, Acanthus arborescens) (Struhsaker 1997,
Paul et al., in press). The area used by the focal group in the logged forest also abuts
smallholder agricultural fields and tea plantations. The unlogged area has never been
commercially harvested. A few large stems (0.03-0.04 trees/ha) were removed before
1970; however, this low level of extraction appears to have had little effect on forest
structure (Skorupa and Kasenene 1984, Skorupa 1988, Struhsaker 1997, Chapman et al.
2000). There is an extensive, gridded trail system in both areas.
Behavioral and Habitat Attributes
Behavioral data were collected on one group of redtails in each site between
August and December 2002. The group in the unlogged area had 24 individuals and the
group in the logged area had 16. Group locations were recorded every 15 minutes during
all-day follows, in 3-day blocks at 1-week intervals, yielding 1563 observations for the
group in the unlogged area and 1301 observations for the group in the heavily logged
area. We recorded the location of the center of the group at every scan interval by
measuring the distance and angle to the nearest point with known coordinates. Reference
points were located at 20 m intervals on the gridded trail system in each group's home
range. Reference points were recorded using a hand-held global positioning system
(Garmin 12 XL). Coordinates for group locations were calculated and plotted in the
ArcView 3.2 and ArcGIS 8.3 geographic information systems (ESRI 2002). Further
details concerning behavioral data collection are included in Stickler and Chapman (in
To measure habitat characteristics, a set of spatially stratified set of 50 m2 plots
were selected by spacing plots every 20 m along the trail grid in each area (Table 3-1).
The plots assessed 37% (n = 1484) and 32% (n = 1362) of the area in unlogged and
heavily logged forest, respectively. In each plot, slope and canopy height were measured
using a clinometer. Three categories of understory vegetation were scored as being
present or absent. The understory type was scored as being present if it was present in at
least 25% of the plot. The categories were selected based on their association with
different forest structures and/or their importance to the monkeys as food sources or as
obstacles. In the areas used by the focal groups, all trees above 10 cm diameter at breast
height (DBH) were identified to species, measured, and plotted in the GIS. The
geographic coordinates for each tree were calculated from the distance and angle from
the reference points (unlogged: 15,489 trees in 48.4 ha; heavily logged: 8433 trees in 40.1
ha). The availability of plant food resources was assessed by categorizing each tree as
"food" or "nonfood", and as providing ripe fruits, unripe fruits, flowers or young leaves,
based on the feeding observations.
Variograms were constructed for group locations to determine the distance at which
observations of redtail group presence were spatially independent. The variogram gives a
geostatistical measure of the spatial scale over which patterns are dependent, defining a
distance or spatial extent (= range) over which spatial dependence can be dependent
(Isaaks and Srivastava 1988). Spatial independence occurred after 32 m in the unlogged
area and after 30 m in the heavily area. Therefore, to derive appropriate and comparable
sampling units for the two areas, I superimposed a grid of 30 x 30 m cells over the group
locations, habitat plots, and tree locations in the GIS. The grid contained 531 cells for the
unlogged area and 466 cells for the heavily logged area. The number of times the group
entered a cell was used to calculate probabilities of redtail use. Home range size was
calculated using an adaptive kernel function for the point distribution in each area, using
the Animal Movement extension for ArcView 3.2(Hooge and Eichenlaub 1997). I
assessed the spatial pattern of habitat use with a coefficient of dispersion (CD) statistic
(Upton and Fingleton 1985). The CD calculates the ratio of the variance against the
mean of the number of times the group entered a grid cell. A value less than one
indicates a uniform dispersion, a value greater than one indicates a clumped dispersion,
and a value near one indicates a random pattern.
To determine the scale of variation for each variable, spatial structure was analyzed
using variogram analysis in the Geostatistical Analyst extension for ArcGIS 8.3 (ESRI
2002). For each variable, a semivariogram produced a range value, which specifies the
distance at which spatial independence is achieved, and a nugget value, which represents
independent error, measurement error or microscale variation at scales too fine to detect
(Isaaks and Srivastava 1988). Nugget values closer to 0 indicate that the amount of error
and/or undetected patch structure in the dataset are low (Isaaks and Srivastava 1988).
The spatial stationarity of each variable was evaluated using variogram analysis (Cressie
1993). As the assumption of stationarity was met for each variable, Moran's I statistic
was calculated as a measure of global autocorrelation for each variable over the extent of
each home range. Moran's I can be thought of similarly to Pearson's r (Cressie 1993).
Typically, Moran's Ivaries on a scale from -1 to +1. Occasionally, the statistic can take
on a value outside this range, due to an unusual weight matrix or averaging within a
distance class (Fotheringham et al. 2000). A large I value indicates that the variable
exhibits spatial dependence (Cressie 1993).
To characterize the gradients of variation in each area and evaluate if they were the
same, I conducted a principal components analysis (PCA). PCA assesses the
relationships within a set of variables, in this case the set of ecological variables (Table 3-
1) to identify the most important sources of variability within the set (McGarigal et al.
For further analyses of the relationships between ecological variables, I assigned
values derived from the habitat plots to the grid in the GIS. Canopy and understory
vegetation heights, slope, and aspect were averaged for each cell, and basal area was
summed. For the understory vegetation types, scores of presence or absence were
summed over each cell to derive a count of the number times a vegetation type occurs in
a cell. Similarly, the number of trees (in all categories) was summed by grid cell. For
both home range datasets, I calculated probabilities of such count variables' occurrence
in each grid cell. Variables were transformed when necessary to ensure normality. Tree
counts and probabilities of understory vegetation types, group presence, and activity site
selection were overdispersed, and thus log-transformed.
Habitat selection models were created and evaluated using path analysis in the
AMOS 5 software (Arbuckle 2003). Path analysis is a subset of structural equation
modelling that allows more than one dependent variable and the effects of dependent
variables on one another to be considered (Everitt and Dunn 1991). This method is
appropriate when correlations that are not easily explained exist among the dependent
variables and an experimental approach is not possible. Moreover, it is often more
interpretable than a partial correlation matrix because the relationships are expressed in
terms of a path diagram. In path analysis, the model specified by the user (default model)
is compared to a saturated model (the most general model, which fits any dataset) and an
independence model (in which the observed variables are assumed to be completely
uncorrelated) (Mitchell 2001). Model fit is assessed by comparing the default model with
the saturated model using a 2 goodness-of-fit index. If the expected correlations (default
model) do not deviate significantly from the saturated model, the hypothesized model is
judged to fit well (Shipley 1997). As a result, the best path analysis model typically has a
p-value higher than 0.05, unlike standard regression models where the best model has a
p-value below 0.05. Path analysis assumes linearity among variables, large sample size,
and the inclusion of all important variables in the model (Everitt and Dunn 1991).
I used the results of patch structure and principal components analyses, as well as
previous research and my understanding of the system, as a guide in determining the
components and likely structure of the hypothetical model. I constructed several default
models testing different hypothesized relationships among the independent and
dependent variables. Each of the default models was compared to the saturated model to
determine which best represented the relationships among the variables. The ability of
different models to accurately reflect system interactions was also evaluated by
examining individual relationships within each model on the basis of coefficient value
significance. To avoid problems resulting from potential remaining spatial
autocorrelation or non-normally distributed data, I report as significant only those
relationships for which the conventionally reported p-values were <0.001. Here, the
desirable p-value for showing that a significant relationship between two variables exists
is again standard, p<0.05. Each variable in the model also has direct and indirect effects
on the response variable, described by the correlation coefficient for each pair of
variables (Mitchell 2001). Because some variables have no direct connection to the
response variable, the direct effect in these cases is 0. The total effect is simply the sum
of the direct and indirect effects, and gives an indication of the importance and overall
direction of correlation of the variable in the system as a whole (Mitchell 2001). Each
model also contains a set of error terms (or disturbance terms) that reflect the unexplained
variance and the measurement error resulting from the regression of each set of
independent variables on their respective dependent variables, as specified in the default
model (Shipley 1997). Each variable's error term is assumed to be uncorrelated with the
other variables internal to the default model.
Once the best model to explain general habitat use was identified, I also
constructed hypothetical models to examine habitat effect on feeding, foraging (searching
for insects), and resting site selection in each site. These models aided in understanding
the relationship between habitat attributes and the probability of any one of the redtails'
major activities occurring, to determine if different activities could be explained by
different habitat qualities.
In both the unlogged and logged areas, redtails exhibit a clustered pattern of habitat
use (unlogged: CD = 11.12; heavily logged: CD = 9.36). However, an examination of the
spatial distribution of grid cell use suggests that redtails in the unlogged area use their
home range more uniformly than their counterparts in the heavily logged area (Fig. 3-2).
In addition, some areas within the outer bounds of the home range are used infrequently
or not at all in the heavily logged area, whereas in the unlogged area, almost all areas are
used at least once. Overall, the range of frequencies of grid cell use was greater for the
group in the unlogged area (0 to 55 observations) than in the heavily logged area (0 to 33
observations). The area described by the 95% probability contour is 29.24 ha in the
unlogged area and 39.76 ha in the heavily logged area (Fig. 3-2).
Spatial Structure of Habitat Attributes
In the unlogged area, Moran's I statistic shows that slope is strongly autocorrelated,
and that this pattern deviates significantly from random (Fig. 3-3). The remaining
variables also exhibited positive autocorrelation, but the results of z-tests do not allow us
to infer that the respective spatial patterns are non-random. The spatial scale of the
habitat attributes shows a great range of variation among variables. Analysis with
semivariograms showed that Marantochloa leucantha (21 m) and basal area (38 m) had
the smallest patch sizes, and that both had nugget (intercept) values close to 0, suggesting
that these variables were sampled at an appropriate scale (Fig. 3-3). By contrast, slope
(17.41) and canopy height (32.75) had high nugget values, indicating that there may be
some degree of variation that the sampling does not detect. Understory height, A.
arborescens, and sapling presence scaled between approximately 180 and 230 m.
In the heavily logged area, the majority of measured variables exhibited strong and
highly significant autocorrelation, and thus, patchiness (Fig. 3-4). As in the unlogged
area, basal area exhibited the smallest patch size (range = 29 m) and was sampled at a
fine enough scale to detect most spatial structure. Slope and A. arborescens scaled
similarly, 88 m and 85 m, respectively. However, whereas A. arborescens presence
appears to have been measured at a suitable resolution, slope may not have been (nugget
= 23.09). As in the unlogged area, canopy height in the disturbed area had a high nugget
value (23.12). Understory height and sapling presence scaled at 115 m and 111 m,
respectively, and had nugget values close to 0. M. leucantha scaled at nearly 250 m, but I
attribute this result, and the low Moran's I statistic value, to the extreme rarity of this
vegetation type in the understory (i.e., M. leucantha was scored as present in 5 sample
In each area, there appeared to be three different scales of patchiness for habitat
attributes: (1) 20 to 40 m; (2) 85 to 135 m; and (3) 180 to 250 m. In both areas, basal
area falls in the first category, as does M. leucantha in the unlogged forest only. In the
unlogged forest canopy height falls in the second category. In the heavily logged area,
slope, canopy height, understory vegetation height and A. arborescens and sapling
presence fall in the second category. Finally, in the unlogged area, slope, understory
vegetation height, and A. arborescens and sapling presence group in the third category.
In both areas, the first four variability gradients extracted in the PCA described
approximately 80% of the variation in the set of habitat measurements (Table 3-2).
Furthermore, the variables or groups of variables contributing most to each of these
components was similar among between the two areas. In each case, component 1
appeared to describe high, open forest (described by positive relationships between basal
area, saplings and canopy height), whereas component 2 described a low understory, as
well as a dense, high A. arborescens-dominated understory. All of the variables loaded at
least moderately on component 1 in the unlogged area, indicating that the habitat
structure may be more variable than in the heavily logged area. Slope had the highest
loading on component 3, and component 4 described high forest, with a sapling-
dominated understory. In the heavily logged area, basal area and A. arborescens-
dominated understory loaded heavily on components 5 and 6, respectively. The opposite
was true for the unlogged area, with the A. arborescens understory defining component 5,
and canopy height and basal area defining component 6.
When the same basic conceptual model (Model 1) was applied to both cases, the
path analysis reveals both similarities and differences in the drivers of redtail habitat use
(Fig. 3-5). The variables and design of Model 1 were based on the analyses of spatial
structure and the principal components analysis, and on a general understanding of the
system Model 1, including average percent slope, total basal area and mean canopy and
understory vegetation heights, fits well in the heavily logged case; the hypothesized
model does not differ significantly from the saturated model (Fig. 3-5). When this Model
1 is applied to the unlogged area, the path analysis shows that the default model does
differ significantly from the saturated model (Fig. 3-5). In both the heavily logged and
the unlogged area, total basal area of trees above 10 cm DBH was important in explaining
patterns of habitat use. Basal area was a stronger predictor of habitat use in the unlogged
area than in the heavily logged area ((Table 3-3). Slope had a strong indirect effect on
habitat use, and had a significant effect on all other explanatory components of the model
in the unlogged area. Only its relationships to canopy and understory height were
significant in the heavily logged area. Understory vegetation height had a strong effect
on habitat use in the heavily logged area, but not in the unlogged area. For both models,
canopy height has neither a strong nor a significant effect on monkey occurrence, but
when links from basal area or understory height to canopy height are removed, model fit
decreases in both cases (e.g., when link from basal area to canopy height is removed:
unlogged: x2 = 144.95, df= 3, p < 0.001; heavily logged: x2 = 182.59, df= 3, p < 0.001).
Since Model 1 deviated significantly from the observed correlations in the
unlogged area, I developed a second model (Model 2) that better explained monkey
occurrence. The best model of general habitat variables included probability of A.
arborescens presence instead of understory vegetation height, although A. arborescens
did not have a significant effect on habitat use within the default model (Fig. 3-6). Slope
remained an important factor in influencing both canopy height and total basal area.
Thus, the results show that total basal area and slope are the most important factors
explaining redtail habitat used in the unlogged area.
I determined whether this combination of variables also explained each of the
groups' major activities. In the unlogged area, model fit improved markedly over the
general habitat use model when feeding, foraging, and resting were included as the
response variables, respectively (Table 3-4). In each case, slope and basal area had the
only significant effects. The significance of A. arborescens as an explanatory variable
decreased, although replacing this variable with understory height yielded even poorer
models. When the general habitat use model was applied to explaining feeding, foraging,
and resting site choice in the heavily logged area, models improved when abundance of
food trees and A. arborescens were included, even though A. arborescens had no direct
effect on the activity site selection. The best models, however, were produced when
abundance of potential fruit trees was entered into the model with A. arborescens (Table
3-4). Since the abundance of fruit trees appeared to be important, I examined their spatial
structure in the two home range areas with respect to fruit trees (Fig. 3-7). Trees
potentially offering ripe fruits occur exhibit a weaker patch structure and occur at a larger
scale in the unlogged than the heavily logged area. This is consistent with the basal area
of trees in both areas.
Aspect, presence of saplings and M. leit nilith, and the density of trees providing
young leaves and flowers as a potential preferred food did not have a strong direct or
indirect effect on any of the other variables, so they did not enter into the final models.
Patterns of habitat use by redtail monkeys in unlogged and heavily logged areas of
Kibale National Park differed in their spatial scale, but could be explained by a similar
suite of habitat attributes. Further analysis of the spatial structure of the habitat
characteristics revealed that these patterns of use are a reflection of the underlying local
environment. I compared the spatial structure of habitat characteristics within and
between the two areas. My analysis showed that heavily logged forest exhibited a higher
degree of patchiness for most habitat attributes over the spatial extent examined. In
general, most variables in the unlogged area tended towards a more random spatial
pattern, with the exception of slope. Nevertheless, positive autocorrelation was apparent
for the other variables, including cumulative basal area, understory and canopy height,
and two understory vegetation types, A. arborescens and saplings. In the disturbed
forest, only basal area and the understory herb M. leucantha lacked significant spatial
The variables appeared to reflect three scales of spatial structure among both areas.
Basal area varied over the shortest distance in both areas. The scale of variation appears
to be consistent between the areas, despite the fact that the basal area in the heavily
logged area is less than 65% that of the unlogged area (Stickler and Chapman, in prep).
This suggests that changes due to logging do not alter the fundamental pattern of how
tree size varies with distance. In both areas, slope, understory vegetation height, and the
presence of A. arborescens and saplings group together in the same scale range.
However, this group of variables fell into the large scale category in the unlogged area, as
opposed to the middle distance category in the heavily logged area. Finally, canopy
height scaled with slope and the other variables in the heavily logged area, while in the
unlogged area, the distance over which canopy height showed patchiness was
intermediate relative to basal area and slope. The principal components analysis
corroborates the suggestion that there is an underlying structure common to both areas.
In both areas, the analysis extracted four distinct gradients of variation describing habitat
Factors Explaining Habitat Use Patterns
Path analysis showed that redtails are most strongly influenced by basal area in
both areas. Consistent with my expectations at the outset of the study, this indicates that
the animals directly respond to the fine-scale patch structure represented by the
combination of all trees in their environment.
Similarly, monkeys were also negatively influenced by the height of the understory
vegetation in both areas, although this relationship was not statistically significant in the
unlogged area. Particularly in the heavily logged area, understory height was as an
important factor as basal area in determining the extent to which redtails used an area. In
the unlogged area, it was understory dominated by A. arborescens that was an important
negative estimator of reduced habitat use. Understory height and A. arborescens are
often positively correlated, particularly in the heavily logged area (Stickler, unpublished
data). However, the composition of the understory is more diverse in the unlogged area
than the heavily logged, which may explain why redtail habitat use is more influenced by
the presence of A. arborescens than understory in general. This provides support for my
initial prediction that redtails would prefer areas with intermediate density of understory,
since this increases the vertical foraging area within the canopy. This result also
emphasizes the importance of the type of understory vegetation. This is demonstrated in
the heavily logged area where A. arborescens and similar plants are largely dominant.
For redtails and some other members of the Kibale primate community, such areas are
largely inaccessible due to their growth form and thorny leaves (Struhsaker 1997).
The least expected result was the finding that slope plays such an important role in
habitat use. Slope was a critical indirect estimator of habitat use in both areas. In the
unlogged area, it had a strong positive effect on basal area and canopy height. The only
part of this area that had a relatively flat slope was in the valley bottom at the north end
of the home range. This area is characterized by wet, swampy conditions, a
predominance of A. arborescens and a lower tree density, than in higher areas-a
combination that appears to be less preferred by redtails. As a result, it is not unusual
that in this area, slope can be used as an indicator of monkey habitat use. Moreover,
examining the habitat characteristics' patch structure demonstrates that slope operates at
a larger scale than the other variables, revealing its importance in the system as a whole.
In the heavily logged area, slope had a strong positive influence on canopy height
and understory height, but a non-significant positive effect on basal area. I explain these
relationships both by the natural characteristics of this area and the effect of logging. As
in the unlogged area, slopes are flatter in two swampy valleys. Throughout the
Kanyawara site, tree cover in such areas tends to be sparse, and the understory tends to be
dense and dominated by A. arborescens and herbs and vines. Average canopy height also
tends to be lower. These valleys bound a higher, flatter ridge that might be predicted to
have similar characteristics to the upland areas in the unlogged area. However, this area
was subjected to intensive logging in the early 1970s. As a result, tree density is low.
Perhaps because of this, canopy height is sometimes higher in areas with steeper slopes
than in the flatter areas. Nevertheless, because of the often very steep slopes between the
high ridge and the valley bottoms, tree cover is higher in flatter areas than on the slopes.
This may explain why the effect of slope on basal area is not so strong. It may also
explain the positive relationship between slope and understory height; steeper slopes may
encourage more dense, taller herbaceous growth if trees cannot establish or survive as
well on these slopes.
Results suggest fruiting trees are important determinants of redtail habitat use.
When feeding, foraging, and resting site selection were examined, trees potentially
providing ripe or unripe fruits in the heavily logged areas significantly improved the path
analysis models. Potential fruit trees have a restricted and highly clumped distribution in
the heavily logged area that coincides with the areas of highest basal area. This is
probably related to the fact that only 60% of all trees in the heavily logged area were
potential food trees. In the unlogged area, by contrast, fruit trees (which represent nearly
90% of all trees in the area) appear to be much more uniformly distributed, although the
relationship to high basal area is also strong. This result provides support for the
importance of high basal area in the monkeys' overall habitat use patterns because areas
with a higher density and size of trees are more likely to have a higher number of fruit
A combination of natural and logging-induced factors contribute to differences in
behavior and differences in population density among redtails at Kibale. The availability
of suitable habitat appears to be more restricted and more patchy in the logged than in the
unlogged area. The differences in activity patterns, diet, and ranging behavior between
the groups observed in a previous study (Stickler and Chapman, in prep) appear to be a
function of habitat constraints. It seems that the interaction of logging with the local
environmental conditions has lowered the carrying capacity of the affected forest. As a
result, I suggest that redtails reduce their group size and use larger areas to compensate
for the reduction and fragmentation of preferred habitat. At the population level, this
likely translates to the decline in numbers observed by Chapman et al. (2000).
The example of redtails in Kibale National Park illustrates the importance of
including landscape characteristics in models describing small-scale habitat preferences.
Typically, explanations of habitat selection focus on vegetation-related components as
these variables are hypothesized to directly affect species abundance at larger spatial
scales (Morrison et al. 1998). However, wildlife-habitat relationships are often
hierarchical and thus inconsistent across spatial scales (Schulz and Joyce 1992, Cushman
and McGarigal 2002, Johnson et al. 2002). There is good evidence that both small-scale
and landscape-scale variables are important in the movement and distribution of a variety
of animals (Kotliar and Wiens 1990, Mazerolle and Villard 1999, Johnson et al. 2002,
Thompson and McGarigal 2002). The current study provides further support for the
importance of variables at multiple scales in explaining and predicting habitat
preferences, even over a relatively small spatial extent and fine resolution at the
organismal level. These findings emphasize the importance of a multi-scale approach in
order to better interpret the seemingly conflicting results of habitat studies conducted at
different scales (Thompson and McGarigal 2002).
Equally important is the spatial distribution of resources and habitat (Fahrig and
Paloheimo 1988). In this study, a comparison of the spatial structure of neighboring
forest stands that were subjected to different management plans highlights the importance
of habitat availability and distribution for primates. At Kibale, as in many other sites, the
underlying topography appears to be critical in structuring vegetation-related elements of
habitat (Bormann and Likens 1979, Forman and Godron 1986). This observation
suggests that slope may play an important role in forest regeneration (expressed by finer-
scale variables), and that the combination of this interaction with the effect of disturbance
caused by logging may partially explain the fact that more than 30 years has not been a
sufficient time period for the heavily logged forest to return to its pre-harvest structure
(Paul et al. in press). As a consequence, the amount of suitable habitat for redtails in
heavily logged areas is lower than in the unlogged area. Moreover, the spatial pattern of
potential habitat is such that some resources and habitat are inaccessible, reducing
functional habitat even further (Orians and Wittenberger 1991). Thus, from a
management or restoration perspective, management interventions to increase the
abundance of high-quality habitat may not automatically lead to an equal increase in
availability of such habitat (Gustafson and Crow 1994 Storch 2002).
These conclusions have important implications for forest management and
conservation strategies and policies. They provide further support for the need for
management plans to take into account the spatial structure of ecosystem characteristics
and the spatial pattern of disturbance (Poiani et al. 2000, Putz et al. 2000, Sanderson et al.
2002). Some timber extraction schemes, such as reduced impact logging, already
incorporate spatial considerations to lower the damage done to the forest (Holmes et al.
2002). It seems that such an approach could be effective if it took into account the spatial
requirements of resident wildlife. This could be done relatively easily, since it already
takes into account topographic and forest structural variables in determining where roads
and skid trails should be placed to decrease damage and increase efficiency. In this case,
for example, the spatial discontinuities caused by logging should avoid segregating areas
such that they become inaccessible to the primates. In addition, gaps or other damage
caused by logging must be at scales smaller than the home range size and location
constraints imposed on organisms by their social ecology or their physical size and
requirements. Although Putz et al. (2001) recommend silvicultural enrichments as one
possible way of restoring habitat or mitigating the impacts of logging, they also point out
that some of these methods are incompatible with maintaining the pathways and patch
sizes necessary to support nonvolant arboreal animals and, temporarily, for birds. For
example, when a forest stand is thinned, by removing competing trees and vines or
understory trees and vegetation (Smith 1986), the size of gaps caused by the removal of
target trees alone may increase to the point of being insurmountable by some species of
animals (Putz et al. 2001). Where conventional extraction is followed by strip
enrichment (Mason 1996), strips must be planned to be short enough in length that
nonvolant animals can circumvent them and still access resources and habitat on the other
side. In addition, such strips must be far enough apart so as not to create functionally
isolated forest patches because they are not large enough to accommodate a group of
foraging primates, for example.
In countries dependent on forestry but interested in maintaining viable and diverse
wildlife communities, policies regarding harvestable size and harvest intensities must be
altered to accommodate the habitat and space use requirements of forest wildlife. The
integration of spatial and behavioral ecology provides an important tool for modeling the
effects of different management schemes on vertebrates, providing for the basis for
scientifically-sound decisions regarding both the economic and ecological future of an
Table 3-1. Habitat attributes examined for two redtail monkey home ranges in unlogged
and heavily logged forest in Kibale National Park, western Uganda, and
expected relationships. The attributes were measured in a set of spatially
stratified plots. Measures of Resource Availability are derived from an
inventory of all trees above 10cm diameter at breast height (DBH) in each
Total Basal Area
Description of attribute and relationship to other forest
Direction of slope (in degrees)
Maximum height (m) estimated in each habitat plot. Higher canopy
height is related to a better developed arboreal habitat
A function of abundance and size (diameter at breast height) of each
tree, measured in cm2. A higher total basal area is related to a more
connected canopy and a higher resource base.
The average height (m) of the understory vegetation. A higher
understory height is related to greater density of the understory.
Acanthus arborescens and related plants constitute a dense,
herbaceous vegetation that often dominates regenerating or swampy
areas. This vegetation is avoided by redtails.
Marantachloa leucantha is an understory herb characteristic of
upland forest light gaps and is a preferred food of redtails
Saplings and/or seedlings are typically present in closed forests with
an open understory.
The density (individuals/m2) of tree species fed on by redtails in
each area. A higher density of redtail food items indicates a higher
carrying capacity of the forest.
The density (individuals/m2) of tree species from which monkeys
were observed to feed on unripe fruits
The density (individuals/m2) of tree species from which monkeys
were observed to feed on ripe fruits
The density (individuals/m2) of tree species from which monkeys
were observed to feed on young leaves
The density (individuals/m2) of tree species from which monkeys
were observed to feed on flowers
Table 3-2. The six components produced in Principal Components Analysis describing
the major variation in habitat structure in areas of unlogged (UL) and heavily logged
(HL) forest in Kibale National Forest, Uganda. The loading values describe the
correlation of the habitat variables with the respective component.
Component 1 Component 2 Component 3 Component 4 Component 5 Component 6
UL HL UL HL UL HL UL HL UL HL UL HL
Importance 39.85 27.41 21.04 25.43 14.27 16.99 9.81 12.31 8.72 10.99 6.29 6.86
Cumulative 39.85 27.41 60.89 52.84 75.16 69.83 84.98 82.14 93.70 93.13 100.0 100.0
Basal Area 0.454 0.578 -0.333 -0.215 0.327 0.153 -0.470 -0.278 0.662 0.239 0.389
Slope 0.306 -0.186 -0.918 -0.969 -0.139 0.187 0.171
Understory -0.315 -0.116 -0.628 -0.658 -0.118 -0.108 0.335 0.692 -0.420 0.114 0.501
Canopy 0.515 0.534 -0.299 -0.155 -0.144 -0.699 -0.448 -0.783
Acanthus -0.326 -0.124 -0.605 -0.700 0.357 -0.127 -0.624 0.270 -0.102 -0.637
Sapling 0.478 0.593 0.200 -0.149 0.787 0.597 0.224 -0.277 0.239 -0.433
Table 3-3. Standardized direct, indirect and total effect coefficients for path analysis
models (Model 1) examining habitat variables on probability of occurrence of redtails in
unlogged and heavily logged areas of Kibale National Park for general habitat utilization
model (Fig. 3-2).
Unlogged Heavily Logged
Habitat Utilization Habitat Utilization
Variable Direct Indirect Total Direct Indirect Total
Effect Effect Effect Effect Effect Effect
Slope 0.000 0.144 0.144 0.000 -0.064 -0.064
Basal Area 0.362 0.004 0.366 0.278 -0.075 0.203
(p = 0.001) (p = 0.009)
Canopy Height -0.003 0.012 0.010 -0.053 -0.083 -0.136
Understory -0.044 0.000 -0.044 -0.249 0.000 -0.249
Height (p = 0.001)
r2 0.140 0.105
Table 3-4. Effect coefficients for path analysis models examining the effect of unripe
fruit trees on feeding site selection by redtails in unlogged and heavily logged
areas of Kibale National Park. Overall model fit and coefficient significance
and values are similar in each to models examining the effect of unripe or ripe
fruit trees on the following response variables: feeding site selection, foraging
site selection, and resting site selection.
Unripe food trees
Unripe food trees
2 = 13.671
p = 0.001
X2 = 0.080
p = 0.961
Figure 3-1. The study site is located in the Kanyawara field site in the northern part of
Kibale National Park in western Uganda. Map of Uganda and Kibale
National Park area adapted from MUBFS (2003).
Unlogged Heavily Logged
j Paer,:ert Uhelikehocf c e Use
JK E35 9S
(c) + (d) +
Figure 3-2. Patterns of habitat utilization by redtail monkeys in unlogged (a, c) and
heavily logged (b, d) forest areas of Kibale National Park, western Uganda.
Graphs (a) and (b) depict the number of times the focal monkey group entered
a grid cell in study area. Units are in meters, in the Universal Transverse
Mercator (UTM) coordinate system, Zone 36 North. Graphs (c) and (d) show
the probability distribution of habitat use by groups in each area. Outer
bounds describe the 95% confidence limit of groups' space use and are shown
in light blue (unlogged: 29.24 ha; heavily logged: 39.76 ha). Bright pink areas
indicate most intensively used areas.
Range = 26 8 m
Nugget= 17 407
I = 0 265
Range = 178.27 m
Acanthus presence Canopy Height /
1 I 0.233 =0.238
Range = 231.72m Range= 98.24
Nugget= 0.0746 Nugget= 32749
Basal Area Sapling presence ......... +
S -0.061 I 0.183
Range= 38.15 m Meters Range 229.09 mrn
Nugget < 0.oo 1 0 175 350 525 700 Nugget= 0.1359
Figure 3-3. Surface maps for six of the habitat variables measured in an area of unlogged
forest in Kibale National Park, western Uganda. Darker shading corresponds
to a higher value of the variable. Cutoff values for shading levels were chosen
to produce approximately equal frequency quintiles. The Moran's I statistic
describes the degree of spatial autocorrelation in the variable; significant
(p<0.001) autocorrelations are marked with an asterisk. The range describes
distance over which the variable show autocorrelation, and a nugget value
close to 0 indicates that most of the spatial variation in the variable values has
been captured by the sampling design.
Slope I= 1.516B Understory Height I= 1.491m
Range = 88.0 m Range= 114.48 m
Nugget= 23.089 ...... Nugget = 1.2246
Acanthus presence 1=0.700- Canopy Height I= 1 474*
Range =84 67 m Range =134 17 rnm
S\ Nugget= 0.1345 .. Nugget= 23 122
Basal Area I= -0.036 Sapling presence I=0 572"
Range = 28 81 m .-. Range = 111 15 n
.. Nugget < 00001 Nugget=0.1251
0 175 350 525 700
Figure 3-4. Surface maps for six of the habitat variables measured in an area of heavily
logged forest in Kibale National Park, western Uganda. Darker shading
corresponds to a higher value of the variable. Cutoff values for shading levels
were chosen to produce approximately equal frequency quintiles. Moran's I
statistic describes the degree of spatial autocorrelation in the variable;
significant (p<0.001) autocorrelations are marked with an asterisk. The range
describes distance over which the variable show autocorrelation, and a nugget
value close to 0 indicates that most of the spatial variation in the variable
values has been captured by the sampling design.
p = 0.025
Figure 3-5. Path diagram (Model 1) describing habitat utilization by redtails in unlogged
and heavily logged areas of Kibale National Park, western Uganda.
Standardized regression weights are shown for each pair of variables.
Significant correlations (p<0.01) are indicated by thick arrows. Model
goodness-of-fit is described by the X2 statistic and associated p-value. A p-
value above 0.05 indicates that the model does not differ significantly from
the saturated model, which indicates that the model has a good fit.
Components denoted Es, Et, Eu, Ec, and Em are error terms corresponding to
each variable in the model.
p = 0.049
Figure 3-6. Path diagram (Model 2) describing habitat utilization by redtails in an
unlogged area of Kibale National Park, western Uganda. Standardized
regression weights are shown for each pair of variables. Significant
correlations (p<0.01) are indicated by thick arrows. Model goodness-of-fit is
described by the X2 statistic and associated p-value. A p-value above 0.05
indicates that the model does not differ significantly from the saturated model,
which indicates that the model has a good fit. Components denoted Es, Et, Eu,
Ec, and Em are error terms corresponding to each variable in the model.
Range = 220.74 m Range = 212.2 m
Nugget 0.1468 Nugget= 0.146
Unripe Fruit Ripe Fruit
Range = 147.19 m Range = 149.37 m
Nugget= 0.1079 Nugget= 0 1641
Heavily Logged Heavily Logged
Unripe Fruit Ripe Fruit
Figure 3-7. Probability maps of trees potentially producing unripe and ripe fruits in
unlogged and heavily logged forest of Kibale National Park, Uganda, during
the period from August to December, 2002. Tree species were identified as
potential fruit trees in each area on the basis of feeding observations on two
groups of redtail monkeys whose home ranges coincided with these areas.
Darker shading corresponds to a higher number of fruit trees. Range indicates
the scale of each variable's patch size. Nugget values close to zero indicate
that the variable has been sampled at the appropriate spatial resolution.
SUMMARY AND CONCLUSIONS
This study contributes to our understanding of the variables important in
determining the behavioral patterns of one primate species, redtail monkeys
(Cercopithecus ascanius) in response to logging as a link to population density. Most
importantly, however, it serves as a model for applying a spatial approach to
investigating primate behavior in relationship to the habitat changes caused by
anthropogenic disturbance to gain a better understanding of the mechanisms underlying
population change in relation to large-scale disturbance, such as timber extraction. This
is critical for planning ecologically sound conservation strategies.
In the first study, I contrasted the behavior of redtail monkeys in unlogged and
heavily logged forest in Kibale National Park, Uganda. My results indicate that redtails
in the heavily logged area use larger home ranges, travel further and spend more time
foraging and less time feeding than conspecifics in the unlogged areas. They appear to be
constrained by resource and habitat limitations. Specifically, a reduced tree density in the
heavily logged area leads to a lower abundance of potential food trees and reduced
arboreal pathways. Microhabitat characteristics in the heavily logged area also appeared
to constrain the animals' movements and foraging ability. Using measurements obtained
from a set of sampling plots spatially stratified throughout the home range of each focal
group, I determined that understory density was high in a much larger proportion of the
heavily logged area than the unlogged area. In cases where density was high, the
understory tended to be dominated by a thorny, herbaceous plant surrounding emergent
trees. Vertical stratification was also reduced over much of the heavily logged area.
These results suggest that the forest in the heavily logged area is structurally limiting to
redtails. Since they typically forage in both the upper and mid-canopy layers and
sometimes descend to the ground, as well, having access to understory vegetation that is
passable and provides resources may be critical. Moreover, such understory helps to
create pathways throughout the forest. The finding that fewer canopy layers and a denser
understory dominate the heavily logged area, suggests that travel for redtails is
constrained by patches of inaccessible vegetation, limiting both the vertical and
horizontal foraging space of the monkeys. These differences likely contribute to the
lower group sizes, higher travel distances and larger home ranges of redtails in the
heavily logged areas, and may explain the reduction in population density of redtails in
logged areas, as well.
In the second study, I examined habitat selection by redtails in the heavily logged
and unlogged areas to determine whether the behavioral differences observed in the first
part were due to habitat limitations or to differences in habitat preferences. I evaluated a
set of topographic, forest structural, and resource attributes for each home range to
determine the spatial structure of each. In particular, I was interested in determining
whether the variables showed distinct patch structures and whether variables operated on
similar spatial scales. I found that the variables tended to arrange themselves spatially,
but that these scales varied among the two areas. Basal area showed the least amount of
patch structure in both areas and operated on the small scale, in the range of 20 to 40 m.
In both areas, the majority of remaining variables grouped together with slope. In the
unlogged area, the scale of these variables was relatively large, whereas in the heavily
logged area, the variables showed an intermediate scale of patch structure. These results
suggest that there is a common organizing structure to both areas, but that the specific
characteristics of each area are a result of scale. Conducting a path analysis to test the
relationship between these variables and to the response variable (redtail habitat use)
revealed that slope is most likely the major structuring variable in each area. Both the
location of disturbance from logging (typically in high flat areas) and the nature of
regeneration (dominated by an aggressive, colonizing herb: Acanthus arborescens) in the
heavily logged study appears to interact with this structuring variable to define the
characteristics of the area.
Habitat selection by the monkeys was similar in both areas, but reflected this
difference in the specific characteristics of both areas. That is, redtails were primarily
influenced by basal area, but understory vegetation also played an important role. In the
heavily logged area, the height of the understory (which is correlated with density) was
the most critical direct factor, in addition to basal area, negatively correlating with
monkey habitat use. In the unlogged area, A. arborescens, contributed to a better model
of habitat selection. I interpret this result as a reflection of the more diverse understory in
the area, where the correlation between understory height and understory density is not as
strong as in the heavily logged area. A. arborescens is far less common in the unlogged
than the heavily logged area; thus, the results of this analysis suggest that the monkeys
avoid areas dominated by A. arborescens, which are also low in basal area. I also
interpret this result as an indication that monkeys in the unlogged area have greater
flexibility in selecting suitable habitat, since it appears to be more common and more
accessible than in the unlogged area. When feeding, foraging, and resting site selection,
were examined, potential fruit trees were the strongest estimator of these response
variables only in the heavily logged area. A subsequent analysis of fruit tree spatial
structure showed that fruit trees are patchier in the heavily logged area than the unlogged.
Most importantly, however, slope had a strong indirect effect on the monkeys in both
areas through its relationship to the other variables.
This study highlights the complex interaction of variables characterizing forest
habitats. In particular, it shows how the spatial structure of each variable is important in
defining its relationship to the other variables, and how these interactions lead to the
unique characteristics defining different areas. This analysis allowed me to begin to
separate the effects of underlying topography from those of disturbance from logging on
primate response to timber harvesting. In a broader sense, this highlights the importance
of structuring variables in determining ecosystem trajectories. This combination of
variables is then reflected in the distribution and abundance of species over the larger
landscape. Thus, sustainable forest management plans must identify the general set of
variables describing the system and determine the relationships between them as a
prerequisite for deciding the level and spatial pattern of timber extraction that will best
maintain the integrity of the system. Using quantitative spatial methods, it should be
possible to model the cascading effects of different plans and make informed decisions
about which plan best fits economic and ecological goals.
Considering the important role primates are thought to play in seed dispersal, this
study also has implications for forest regeneration. In the particular case of Kibale, better
planning prior to logging might have considered the way in which timber extraction
reconfigured primate foraging and travel spaces, and thus the spatial pattern of
regeneration. Where primates cannot access trees, the suite of dispersers for an
individual tree is decreased and the likelihood that seeds remains closer to the parent tree
and become more vulnerable to density-dependent mortality increases. Moreover,
because of the recolonizing vegetation of the area, even those trees that do manage to
attract dispersers or those that rely on autodispersal will have to contend with
inhospitable conditions for germination, seedling growth, and sapling survival. Thus, the
relationship between primates, logging, and seed dispersal using spatial analysis is an
important avenue for further research.
Furthermore, this research also reinforces the importance of planning
conservation and forest management at multiple scales. Current wisdom holds that
biodiversity conservation plans must integrate human settlements, natural resource
exploitation, and protection across the landscape (Poiani et al. 2000, Sanderson et al.
2002). Because of the large home range needs of some vertebrates, focusing on the
maintenance of population viability of these species has been suggested as a way of
ensuring for the protection of large, ecologically diverse areas (Wilcox 1984, Lambeck
1997, Caro and O'Doherty 1999, Miller et al. 1999). In light of the range of scales over
which organisms interact with the environment, more recent plans expand this concept to
account for the scale and grain of habitat requirements of these large vertebrates
(Sanderson et al. 2002). However, as the current study shows, organisms with relatively
small, restricted, and stable home ranges can be negatively affected by resource
extraction that occurs on a scale that coincides with their habitat requirements and
territorial constraints, even if this level of extraction does not affect larger vertebrates. In
Kibale, although elephants and chimpanzees are affected by logging-induced habitat
changes (Struhsaker 1997), they are not arboreally restricted and can travel further to
seek out suitable resources and habitat. Thus, vertebrates with smaller spatial
requirements should form part of the suite of "landscape species" (Sanderson et al. 2002)
chosen for conservation planning because they may be more directly affected by finer-
scale changes in forest spatial structure caused by logging.
LIST OF REFERENCES
Arbuckle, J. 2003. AMOS 5 Path Analysis Software. Chicago, IL: Sweetwaters Corp.
Boinski, S. and Sirot, L. 1997. Uncertain conservation status of squirrel monkeys in
Costa Rica, Saimiri oerstedi citrinellus. Folia Primatologica 68: 181-193.
Burt, W.H. 1943. Territoriality and home range concepts as applied to mammals.
Journal of Mammalogy 24: 346-352.
Butynski, T.M. 1990. Comparative ecology of blue monkeys (Cercopithecus mitis) in
high- and low-density subpopulations. Ecological Monographs 60(1): 1-26.
Bormann, F.H., and Likens, G.E. 1979. Pattern and Process in a Forested Ecosystem.
New York: Springer Verlag.
Cannon, C.H. and Leighton, M. 1994. Comparative locomotor ecology of gibbons and
macaques: selection of canopy elements for crossing gaps. American Journal of Physical
Anthropology 93: 505-524.
Caro, T.M. and O'Doherty, G. 1999. On the use of surrogate species in conservation
biology. Conservation Biology 13: 804-814.
Chapman, C.A. 1988. Patterns of foraging and range use by three species of neotropical
primates. Primates 29:177-194.
Chapman, C.A. 1995. Primate seed dispersal: coevolution and conservation
implications. Evolutionary Anthropology 4: 74-82.
Chapman, C.A. and Onderdonk, D.A. 1998. Forests without primates: primate/plant
codependency. American Journal ofPrimatology 45: 127-141.
Chapman, C.A., Balcomb, S.R., Gillespie, T.R., Skorupa, J.P. and Struhsaker, T.T. 2000.
Long-term effects of logging on African primate communities: a 28-year comparison
from Kibale National Park, Uganda. Conservation Biology 14: 207-217.
Chapman, C.A. and Lambert, J.E. 2000. Habitat alteration and the conservation of
African primates: case study of Kibale National Park, Uganda. American Journal of
Primatology 50: 169-185.
Chapman, C.A. and Peres, C. 2001. Primate conservation in the new millennium: the
role of scientists. Evolutionary Anthropology 10: 16-33.
Chapman, C.A., M.D. Wasserman, and T.R. Gillespie. Behavioural patterns of colobus
in logged and unlogged forests: The conservation value of harvested forests. In V.
Reynolds, H. Notman, and N. Newton-Fisher (Eds.). Primates of Uganda. (Submitted).
Clarke, M.R., Collins, D.A., and Zucker, E.L. 2002. Responses to deforestation in a
group of mantled howlers (Alouattapalliata) in Costa Rica. International Journal of
Primatology 23: 365-381.
Clark, C.J., J.R. Poulsen and V.T. Parker. 2001. The role of arboreal seed dispersal
groups on the seed rain of a lowland tropical forest. Biotropica 33: 606-620.
Cords, M. 1986. Interspecific and intraspecific variation in diet of two forest guenons,
Cercopithecus ascanius and C. mitis. Journal of Animal Ecology 55: 811-827.
Cowlishaw, G. and Dunbar, R. 2000. Primate Conservation Biology. Chicago: The
University of Chicago Press.
Cressie, N.A.C. 1993. Statistics for Spatial Data. New York: John Wiley and Sons, Inc.
Cushman, S. A., and K. McGarigal. 2002. Hierarchical, multi-scale decomposition of
species-environment relationships. Landscape Ecology 17:637-646.
Di Fiore, A. 2003. Ranging behavior and foraging ecology of lowland woolly monkeys
(Lagothrix lagotrichapoeppiggi) in Yasuni National Park, Ecuador. American Journal
of Primatology 59: 47-66.
Emmons, L.H. 2000. Of mice and monkeys: Primates as predictors of mammal
community richness. In J.G. Fleagle, C.H. Janson, K.E. Reed (Eds),
Primate Communities. Cambridge: Cambridge University Press. Pp. 171-188.
Environmental Systems Research Institute (ESRI). 2002. ArcView 3.3 and ArcGIS 8.3
Geographic Information Systems Software. Redlands, CA.
Everitt, B. S. and Dunn, G. 1991. AppliedMultivariate Data Analysis. London: Edward
Fahrig, L. and Paloheimo, J. 1988. Effect of spatial arrangement of habitat patches on
local population size. Ecology 69: 468-475.
Fairgrieve, C. and Muhumuza, G. 2003. Feeding ecology and dietary differences
between blue monkey (Cercopithecus mitis stuhlmanni Matschie) groups in logged and
unlogged forest, Budongo Forest Reserve, Uganda. African Journal of Ecology 41: 141-
Fimbel, R.A., Grajal, A., and Robinson, J.G, eds. 2001. Logging-wildlife issues in the
tropics: an overview. In R.A. Fimbel, A. Grajal, and J.G. Robinson (Eds.), The Cutting
Edge: Conserving Wildlife in Logged Tropical Forests, Eds. New York: Columbia
University Press. Pp. 3-9.
Food and Agriculture Organization (FAO). 2003. State of the World's Forests 2003.
Rome: FAO Information Division.
Forman, R.T.T., and Godron, M. 1986. Landscape Ecology. New York: John Wiley and
Fotheringham, A.S., Brunsdon, C., and Charlton, M. 2002. Quantitative Geography:
Perspectives on Spatial Data Analysis. London: Sage Publications.
Frumhoff, P.C. 1995. Conserving wildlife in tropical forests managed for timber.
Bioscience 45: 456-464.
Galletti, M., Pedroni, F., and Morrellato, L.P.C. 1994. Diet of the brown howler monkey
(Alouattafusca) in a forest fragment in southeastern Brazil. Mammalia 48: 111-118.
Garber, P.A. and Lambert, J.E. 1998. Primates as seed dispersers: Ecological processes
and directions for future research. American Journal of Primatology 45: 3-8.
Gautier-Hion, A. 1988. Food niches and co-existence in sympatric primates in Gabon.
In D.J.Chivers and J.Herbert (Eds.), Recent Advances in Primatology. London:
Academic Press. Pp. 269-286.
Gebo, D. and Chapman, C.A. 1995. Positional behavior in five species of old world
monkeys. American Journal of Physical Anthropology 97: 49-76.
Gillespie, T.R., Chapman, C.A. and Greiner, E.C. Long-term effects of logging on
parasite dynamics in African primate populations. Unpublished manuscript.
Grieser-Johns, A. 1997. Timber Production and Biodiversity Conservation in Tropical
Rain Forests. Cambridge: Cambridge University Press.
Grieser-Johns, A. and Grieser-Johns, B. 1995. Tropical forest primates and logging:
long term co-existence? Oryx 29: 205-211.
Gustafson, E.J. and Crow, T.R. 1996. Simulating the effects of alternative forest
management strategies on landscape structure. Journal of Environmental Management