EFFECTS OF LANDSCAPE STRUCTURE ON THE DISTRIBUTION OF MOUNTAIN VIZCACHA (Lagidium viscacia) IN THE PATAGONIAN STEPPE
By
REBECCA SUSAN WALKER
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UMVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
2001 ACKNOWLEDGMENTS
As this research spanned several years, several disciplines, and two continents, I have many, many people to thank. There are so many that I am quite sure I will
inadvertently leave someone out, and I apologize for that.
In Gainesville
First of all, 1 would like to thank my advisor, Dr. Lyn Branch, for her untiring and inspiring mentoring, and her many, many hours of thoughtful reading and editing of
numerous proposals and this dissertation. I also thank her for the financial support she
provided in various forms during my Ph.D. program. Finally, I thank her for her incessant demand for rigorous science and clear writing.
Dr. Brian Bowen got me started in molecular ecology and helped me to never lose
faith that it would all work out. He also provided important input into several proposals and the data analysis.
Dr. Bill Farmerie deciphered many molecular mysteries for me and helped make
my dream of Lagidium microsatellite primers a reality. I thank him for his crucial guidance in developing the microsatellite primers, for the use of his lab, and for his eternal good humor.
ii I thank the other members of my original committee, Dr. George Tanner, Dr. John
Eisenberg, and Dr. Colin Chapman, for their guidance and input into various stages of the
research over the years.
Ginger Clark taught me how to use a pipettor, to pour a gel, to run all the
machines, to use a Mac, to mix solutions, etc. I thank her for all of her teaching, for her
patience, and for all of the help, advice, and good company she provided during my many
months of lab work. I also thank her for storing my boxes while I was in the field.
Anna Bass, Regina Shaw, Alicia Francisco, Dr. Dan Brazeau, Rachel Wacker,
and Christina Romagosa also made the labwork much more pleasant with their help and
companionship. Regina helped me with more of those molecular mysteries, and Alicia
did many isolations, PCRs, and gels for me.
Dr. David Moraga Amador of the ICBR Education Core very kindly shared
equipment, space and supplies at some stages of the labwork. I also appreciate his
comments and advice during the development of the primers.
I thank Dr. Mel Sunquist for replacing John Eisenberg on my committee during
the last semester, and I thank him and Fiona Sunquist for the many curry dinners over the years and for letting me stay in their cabin during the final stages of the dissertation.
Many thanks go to Jay Harrison and Galen Jones of the IFAS Statistics department for answering my questions about statistics and SAS. Jay recommended the
use of GLIMMIX for the analysis in Chapter 3.
I thank Dr. Jim Nichols for advice on statistical analysis and for reviewing the appendix on the population size index. '
iii I would also like to thank the administrative and support staff of the Department
of Wildlife Ecology and Conservation, especially Cynthia Sain, Mindy Edwards (former
staff), Caprice McRae, and Monica Lindberg, who helped me in many ways over the
years. I also thank Polly Falcon and to Delores Tillman, who helped me with all the
details of the last semester.
Of course, over all of my years off and on at the University of Florida I greatly
benefited from my interactions with fellow students, both academically and personally. I
made many good friends who influenced me in many subtle and not so subtle ways.
There are so many that I couldn't begin to name all of them, and I offer my heartfelt
thanks to all. I specifically mention only Becky Ditgen and Marcela Machicote, who
provided crucial help and support when I had to be in Gainesville without my family to
do labwork. Becky also helped me in many other ways, including explaining some
geology terms and concepts to me. r i ,
The ICBR DNA Sequencing Core at the University of Florida did the sequencing
for the development of the microsatellite primers. The BEECS Genetic Analysis Core
and the Iowa State University DNA Sequencing Facility did the GENESCAN analyses of
microsatellite alleles.
Funding
During my coursework for the Ph.D. and while doing the labwork in Gainesville
I was funded by an IFAS research fellowship in 1995, teaching assistantships for
Conservation Biology provided by the Department of Wildlife Ecology and Conservation in fall 1996, 1997, and 1999, and the federal work-study program in spring and summer
iv of 1997. In addition, I thank the African Safari Club for the scholarship they awarded me
in 1997.
Fieldwork was funded by grants from the Lincoln Park Zoo Scott Neotropic Fund,
the American Society of Mammalogists, and Sigma Xi, The Research Society. I used a
vehicle belonging to a Wildlife Conservation Society (WCS) project in the field, and
WCS provided some support in the final stages of the research.
Lyn Branch underwrote the laboratory work in the early stages. The final year of
laboratory analysis was ftinded by a National Science Foundation Doctoral Dissertation
Improvement grant (DEB-9972717).
In Patagonia
I give many thanks to Gabriela Ackermaim, Veronica Pancotto, Judith Shachter
Broide, Maria Beatriz Bongiomo, Gladys Galende, Obdulio Monsalvo, Emiliano
Donadio, Sebastian DiMartino, and Mariana Aubone, who assisted me in the field in
various parts of the study.
I thank Alejandro del Valle, Martin Funes, and Obdulio Monsalvo of the Centro de Ecologfa Aplicada del Neuquen (CEAN) for helping with logistics and in so many
other ways over the years. Lodging during part of the study was provided by CEAN. I thank Carlos Rambeaud for sharing his extensive knowledge of the region and people of the region and for permission to survey his family's ranch at Campo Grande.
I thank the Guardafaunas of Junin de los Andes for the use of their trailer in 1996, and the Guardafaunas of San Martin de los Andes for the use of their GPS that same year.
I greatly appreciate the assistance, companionship, and use of facilities at the school in
Pilolil, provided by Miguel and Silvana Gomez. The study would not have been possible
V without the skills of Aldo Parra and Marcelo Prieto in capturing mountain vizcachas at
Pilolil.
I thank the following individuals, families, ranch owners, administrators, capataces, and others for permission to carry out the research on their land, or for assistance in obtaining permission or access to the numerous field sites:
Comunidad Mapuche de Malleo Comunidad Mapuche de Rucachoroi Comunidad Mapuche Namuncura Corfone Abra Ancha Corporacion Interestadual Pulmari Delegacion Regional Patagonia, Administracion de Parques Nacionales, especially Claudio Chehebar Familia Cordero Familia Criado, Valle Encantado Familia Creide, La Lipela
Familia Fernandez Beschtedt (I especially thank them for the hot baths!) Familia Gonzalez, Valle Encantado Familia Modarelli Familia Prieto, PiloHl Familia Quinas
G. Anz, Estancia Pilolil G. Bergese Gerardo Alinea, Estancia Catan Lil Guardafaunas of Alumine Javier Sanguinetti and the Guardaparques of Lanfn National Park Lorenzo Simpson Mariana Cordero of Estancia Santa Rosa Miguel Anz and the owners of Estancia Los Remolinos Owners and administrators of Estancia Los Ranchos Owners and administrators of Estancia Porvenir P. Larminat, Estancia Cerro de los Pinos Pacito Gonzalez, Estancia Nahuel Mapi Pedro Ochoa Roberto Iriarte, for Estancias Palitue, La Rinconada, and Quilachanquil S. Rambeaud, Estancia Rahue Sandro, at the campground in Alumine Santiago Rambeaud, Estancia Panqueco Sr. Van Ditmar, Valle Encantado V. Soleno, Estancia CoUon Co Victorino Rodriguez and Sra. Rosa, Pilolil Ward Lay and Boris Nunez, Estancia Alicura
VI During the final stages of the research, Ana Prieto and Sebastian DiMartino
helped me with many computer programs. Javier Ayessa and the Laboratorio de
Teledeteccion de INTA Bariloche provided the digitized geology map through an
agreement with CEAN for analysis of the distribution of wildlife species in the province
of Neuquen. I also give many thanks to Adriana Belloti for sharing her knowledge of the
geology of the region with me.
Personal
I offer a very special thanks to my daughter, Jimena Isabel Novaro, who made it
all so much more difficult but so much more interesting. She pleasantly tolerated being
dragged between continents and cliffs and thought we were on vacation during my
fieldwork.
Maria Elena Sepulveda, Graciela Ema, and Maria Beatriz Bongiomo helped me
with Jimena while I was doing coursework, writing proposals, doing the fieldwork, and
writing the dissertation. Knowing that Jimena was in good and loving hands made it
possible for me to dedicate myself to my work and I am eternally grateful to the three of
them. Maria Elena also helped with some of the mapping.
My parents, R.D. and Juanita Walker, also took care of Jimena in the final stages
of my writing of the NSF proposal and my dissertation, and during two of our
intercontinental moves. They helped in many ways with our various moves. They also
loaned vehicles and household supplies for our stay in Gainesville in 1999-2000.
My brother, Ray Walker, let me use his home office to finish and submit the NSF
proposal in 1998 and to print the dissertation in 2000. I also thank him for help with
software and hardware on several occasions.
vii Finally, I would like to thank with all my heart my husband, Andres Novaro. His unfailing belief in me and in this project sustained me through many periods of
exhaustion and discouragement. He worked as an untiring field assistant, helped with
Jimena, and handled all of the logistics of the project in the field. He helped in
innumerable ways that I couldn't begin to list. He made many personal sacrifices to help
me complete this project and I thank him for everything.
viii TABLE OF CONTENTS
ACKNOWLEDGMENTS ii
ABSTRACT xi
CHAPTER
1 INTRODUCTION 1
2 STRUCTURE AND SOCIAL ORGANIZATION OF A POPULATION OF MOUNTAIN VIZCACHAS IN FRAGMENTED HABITAT 6
Introduction 6 Methods 10 Results 13 Discussion 16
3 EFFECTS OF QUALITY, SIZE, AND ISOLATION OF PATCHES ON PATH OCCUPANCY OF MOUTAIN VIZCACHAS 21
Introduction 21 Methods 23 Study Area 23 Assessment of Habitat Quality within Cliffs 23 Results 26 Discussion 29
4 LANDSCAPE CONNECTIVITY AFFECTS DISTRIBUTION OF THE MOUNTAIN VIZCACHA, A HABITAT SPECIALIST IN FRAGMENTED HABITAT 33
Introduction 33 Landscape Connectivity 33 Background and Research Hypotheses 37 Methods 38 Study Area 38 Survey of Patch Occupancy 39 GIS Analysis of Landscape Structure 40
ix Genetic Analyses 41 Landscape Connectivity 44 Results 47 Patch occupancy and Landscape Structure 47 Genetic Structure 48 Landscape Connectivity 49 Discussion 50
5 CONCLUSION 61
APPENDIX
A EVALUATION OF FECAL-PELLET INDEX OF ABUNDANCE FOR MOUNTAIN VIZCACHAS IN PATAGONIA 64
B CHARACTERICATION OF MICROSATELLITE LOCI FROM THE MOUNTAIN VIZCACHA Lagidium viscacia AND THEIR USE FOR THE PLAINS VIZCACHA Lagostomus maximus 72
C TABLE OF ALLELES 76
LIST OF REFERENCES 83
BIOGRAPHICAL SKETCH 94
X Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy
EFFECTS OF LANDSCAPE STRUCTURE ON THE DISTRIBUTION OF MOUNTAIN VIZCACHA (Lagidium viscacia) IN THE PATAGONIAN STEPPE
By
Rebecca Susan Walker
May 2001
Chairman: Dr. Lyn C. Branch Major Department: Wildlife Ecology and Conservation
Metapopulation theory is used widely to analyze the dynamics of populations in fragmented habitats. Empirical studies testing predictions of metapopulation theory at
regional scales, which are appropriate for conservation, are still rare. Likewise, predictions resulting from modeling in landscape ecology, which emphasizes a more spatially explicit approach, have rarely been tested empirically on real landscapes at a large scale. The objectives of this study are to evaluate the role of dispersal in the distribution of a habitat specialist in fragmented habitat, and to determine the effects of the landscape structure on dispersal. The study focuses on the mountain vizcacha
{Lagidium viscacia), a rabbit-sized rodent restricted to patches of rocky habitat in the
Patagonian steppe of Argentina. I evaluate the relationship between habitat structure, population structure, and social system, and explore the role of dispersal at a local scale.
Then I examine the relative importance of patch-level and landscape-level factors in
determining the pattern of patch occupancy by mountain vizcachas. Finally, I determine
xi the role of landscape connectivity in the distribution of the species at a regional scale,
over a 12,000 km^-area. The approach I use to determine landscape connectivity combines molecular genetic estimates of gene flow as an analog of movement rates and cost-distance analysis with GIS to model landscape resistance and facilitation of movement.
The results indicate that spatial structure of habitat affects distribution of mountain vizcachas at several scales. Within patches, mountain vizcachas are more likely to be found where rock crevices are deep and numerous. The presence of mountain vizcachas in a particular patch with appropriate rock depends on the distance and connectivity between the patch and other occupied patches. Distribution at the regional
scale is a result of the pattern of connectivity at the landscape level. Connectivity for
mountain vizcachas is defined by the spatial pattern of surface geology and the barrier effect of some rivers. The dependence of distribution on connectivity implicates dispersal as the key process determining distribution at the regional scale. The study emphasizes the necessity of incorporating the concept of landscape connectivity into metapopulation analyses.
xii CHAPTER 1 INTRODUCTION
Early ecological theory was based largely on spatial homogeneity, although some
ecologists recognized that persistence time could be increased in spatially subdivided
populations with dispersal among local populations (Andrewartha and Birch, 1 954;
Huffaker, 1958; Den Boer, 1968). This concept was formalized mathematically in
Levins' metapopulation theory (1969). Metapopulation theory went largely unnoticed at
the time, perhaps because it followed closely the publication of MacArthur and Wilson's
(1967) island biogeography theory. This theory addressed the maintenance of biodiversity in islands of differing sizes and degrees of isolation, and was recognized quickly as being of great relevance to the design of reserves and parks that were intended to protect biodiversity. Because island biogeography theory had a Umited ability to explain the persistence of any particular species or population, ecologists eventually turned to metapopulation theory in search of guidelines for management and prediction of persistence of subdivided populations (Hanski and Gilpin, 1991; McCullough, 1996).
Both island biogeography and metapopulation theories consider relatively homogeneous islands or patches in a homogeneous matrix, but this is an oversimplification for most ecosystems and species. In recent years, the field of landscape ecology has begun to assess the effects of complex and changing landscapes on ecological processes (Wiens,
1995).
1 2
Currently, the spatial structure of populations and corresponding habitats is recognized as a primary influence on population dynamics, persistence, and species' distributions. Understanding the processes that affect the distribution and persistence of
populations in fragmented habitats is urgent because an increasing number of species is restricted to remnant patches of habitat in landscapes fragmented by humans. Wildlife managers are increasingly depending on population biology to provide the principles by which populations can be managed for their long-term survival.
Many theoretical advances have been made in recent years, but few empirical studies have tested the predictions of metapopulation theory at scales appropriate for conservation (Hanski and Gilpin, 1997). Likewise, predictions resulting from modeling in landscape ecology, which emphasizes a more spatially explicit approach, have rarely been tested empirically on real landscapes at a large scale. Species that evolved in naturally fragmented habitats provide models of how viable regional populations are maintained in small, discrete patches. However, studies of dispersal, local extinction, and colonization—the processes relevant to metapopulation dynamics and predictions from landscape ecology—are technically very difficult to undertake at the appropriate scales of space and time.
The objectives of this research were to evaluate the role of dispersal in the distribution of a habitat specialist in fragmented habitat, and to determine the effects of the landscape structure on that process. The study focused on the mountain vizcacha
(Lagidium viscacia or boxi. Family Chinchilhdae), a rabbit-sized rodent restricted to patches of rocky habitat in the Patagonian steppe of Argentina. The research employed recently developed techniques in geographic information systems (GIS) and molecular 3
genetics to test predictions from both metapopulation theory in population biology and
modeling in landscape ecology. I analyzed relationships among the species' distribution,
gene frequency distribution, and landscape features with GIS, and I inferred the history of
dispersal among local populations from allele frequency distributions of microsatellite
k)cL The GIS allowed analysis of patterns of distribution of mountain vizcachas and
landscape structure at a regional scale (12,000 km^). The use of molecular genetics
allowed inferences about historical movements among populations.
Little is known about the biology of the mountain vizcacha, a South American
caviomorph that lives m naturally fi^mented rock habitats along the Andean cordillera
in Peru, Bolivia, Chile and Argentina and m the Patagonian steppe of Argentina (Redford
and Eisenberg, 1992). Even the taxonomy of the genus is unclear, with some authorities
recognizing three species (Redford and Eisenberg, 1992; Woods, 1993) and others
recognizing four species (Pearson, 1995). The different species of the genus appear to be
similar ecologically, but only L. peruanum was the subject of a major ecological study in the highlands of Peru (Pearson, 1948).
In the northern Patagonian steppe, the resident species is either L. viscacia or L.
boxi. The only published studies of the species have described its diet, indicating high dietary overlap with an abundant and widespread introduced species, the European hare
(Lepus europeaus) (Galende and Grigera, 1998; Galende et al., 1998; Puig et al., 1998).
The status of the mountain vizcacha in most areas is unknown (Woods, 1993), although
L. wolfsohni in southern Patagonia and some subspecies of L viscacia in Chile are considered endangered (Glade, 1988). In the Patagonian steppe, several factors potentially place it at risk: 1) Populations are typically small and isolated. 2) Overgrazing 4
by livestock and introduced wildlife has seriously degraded the steppe habitat
surrounding the cliffs. 3) Mountain vizcachas are heavily hunted for food (Funes and
Novaro, 1999) and are easy to shoot because they are diurnal and relatively fearless when they are among the rocks. 4) They have a low potential for recovery from population reductions because females bear one young at a time and no more than two per year
(Weir, 1974). My preliminary research indicates that mountain vizcachas are declining in northwestern Patagonia. Because of the fragmented nature of mountain vizcacha habitat
in the Patagonian steppe, it is essential to consider the role of dispersal among habitat patches in the distribution and maintenance of the regional population (Hanski and
Gilpin, 1997).
Therefore, this research accomplishes two goals: 1) it provides basic information on the ecology of mountain vizcachas that can be used to evaluate the status of the
species in specific landscapes; and 2) it provides an empirical test of predictions derived from metapopulation theory and landscape ecology, which could result in a much broader application for conservation of other species in fragmented habitats.
The foUowmg chapters are written in a format for independent publications, but
each contributes to my general objective and broad goals. In Chapter 2, 1 evaluate the relationships among habitat structure, population structure, and social system; and explore the role of dispersal at a local scale. The study area, 30 km^ on the eastern side of the Alumine River, is within the study area for Chapters 3 and 4. Chapter 3 examines the relative importance of within-patch, patch-level, and landscape-level factors in determining the pattern of occupancy by mountam vizcachas within a 12,000-km^ area.
Finally, in Chapter 4 1 determine the role of landscape connectivity in the distribution of 5
the species at the regional scale. This multi-scale approach allows a more comprehensive evaluation of the processes affecting the dynamics and distribution of the species than could be achieved by study at any single scale. CHAPTER 2 STRUCTURE AND SOCIAL ORGANIZATION OF A POPULATION OF MOUNTAIN VIZCACHAS IN FRAGMENTED HABITAT
Introduction
In habitats distributed as patches in a matrix of unsuitable habitat, the structure of a population depends on the rate of movement of individuals among patches. If individuals move among habitat patches on a daily, seasonal, or regular basis, the demography and dynamics are those of a single large population (patchy population). If
movement is limited to occasional dispersal between habitat patches, the population may be a metapopulation, and local population dynamics cannot be fully understood without
reference to populations occupying other patches and the system as a whole (Stacey et al.,
1997). Finally, if there is no movement among habitat patches, populations inhabiting each patch are distinct, with independent dynamics. Populations of many species may
exhibit mixtures of these structures in different parts of their range (Stith et al., 1996), depending on the local configuration of habitat patches or the scale at which they are examined.
Social organization within a population may also be affected by a patchy distribution of habitat, if that habitat constitutes a limiting resource for the population
(Carr and MacDonald, 1986). According to Emlen and Oring (1977), when a single male is capable of monopolizing a resource patch, the population has a high potential for a harem-polygynous social structure. However, social relationships of males are determined by their strategies for obtaining mates, but the reproductive success of
6 7
females is not usually as limited by availability of males as by availability of other resources. Therefore, Wrangham (1980) suggests that selection pressures on female behavior, rather than on male behavior, ultimately determine the effect of resource distribution and availability on social structure.
For mammal species that are specialized for rocky habitats, rocks are critical resources, providing den sites and shelter from predation, and most rock-dwelling mammals have a harem-polygyny mating system (Mares and Lacher, 1987). Citing the theory of Emlen and Oring (1977), Mares and Lacher (1987) attributed this to the nature of the habitat because rock outcrops are easily monopolized and defended by a single male. They noted two rock-dwelling species that were exceptions and that did not have a harem polygyny system; moimtain vizcachas {Lagidium spp. Family Chinchillidae), and pikas {Ochotona spp.. Family Ochotonidae). In addition, the social unit of the yellow- bellied marmot (Marmotaflaviventris) originally was described as a polygynous harem,
but later analyses indicate that it is actually a matriline of related adult females with their young and yearlings. Yellow-bellied marmots are large-bodied ground squirrels that occupy rocky talus slopes in the western mountains ofNorth America. Males do not form harems by monopolizing critical resources—harems are formed when a male associates with one or more matrilines that already have monopolized rock patches
(Rayor and Armitage, 1991).
Mountain vizcachas are large rodents ( 1 to 3 kg) that live in rocky cUffs and outcrops from northern Peru (10°S) to southern Patagonia (52°S). The only species of the genus that has been the subject of a major ecological study is L. peruanum in the altiplano of Peru, and its social and mating systems are exceptions to those of most rock- dwelling mammals. L. peruanum lives in colonies containing from 4 to 75 individuals. 8
The colonies comprise smaller groups of two to five individuals, usually putative families of one adult male and female and one or more subadults, but other types of groups are also found (Pearson, 1948). Mares and Lacher (1987) hypothesized that the absence of the harem-polygyny system in L. peruanum was because in the area in Peru where they were studied, they inhabited huge expanses of rocky boulders. This habitat, imlike the small, isolated outcrops used by most other rock-dwelling species, was not easily defensible. Mares and Lacher predicted that in areas where mountain vizcacha habitat is more fragmented, rock patches would be more defensible, and moimtain vizcachas would exhibit a harem-polygyny system.
In rock-dwelling mammals, differences in social organization and mating systems have been noted between occupants of fragmented versus more continuous habitats. Pikas
(Ochotona princeps) are small lagomorphs restricted to talus slopes in western North
America, and they are monogamous (Smith and Ivins, 1984). Within continuous habitat,
mating is not random and is most common between intermediate relatives, while random mating occurs in fragmented habitat (Peacock and Smith, 1997). Rock hyraxes (Procavia johnstoni and Heterohyrax brmei) in the Serengeti plain of Africa which inhabit small rock outcrops live in groups comprised of 3 to 7 females with one adult male and juveniles of both sexes. Groups on large rock outcrops (more continuous habitat) also include peripheral adult males, which occasionally mate with young females from the
harems (Hoeck et al., 1982).
In addition to the mountain vizcachas, two other genera comprise the South
American family Chinchillidae, and both are colonial. The chinchillas {Chinchilla laniger and C. brevicaudus), which are also rock dwelling, are considered monogamous
(Walker, 1975), but are extremely endangered and their social organization has not been 9
studied in the wild. They live at high altitudes in the Andes of northern Chile and
Argentina. The plains vizcacha {Lagostomus maximus) lives in burrow systems that it
constructs in grasslands and shrublands. This species has a complex social organization
comparable to that of the North American prairie dogs, and a polygynous mating system
(Branch, 1993). Plains vizcachas have a very patchy distribution at the local level, but
this is a result of their social system, rather than a consequence of the distribution of their
habitat.
The purpose of this study was to investigate the role of habitat in structuring the
population and social organization ofLagidium viscacia in the Patagonian steppe. I used
surveys of available habitat, live captures, and genetic analyses to infer the structure and
social organization of a population in a 30-km^ area. In this area, rocky cliffs and outcrops are small (< ~lkm in length) and scattered (Fig. 2-1). Thus, the population could constitute a metapopulation, and according to the prediction of Mares and Lacher, could have a harem-polygyny mating system.
Under expectations of a metapopulation structure, genetic analyses would reflect
limited movement among colonies occupying different habitat patches, and not all habitat patches would be occupied due to extinction/colonization dynamics (Hanski, 1991). To
test these predictions, I surveyed habitat patches for the presence of mountain vizcachas, determined genetic population structure, and estimated gene flow among colonies. To
characterize the social organization, I estimated population sizes, compared genetic relatedness of males and females within and among colonies, and determined the
probability that adults captured with juveniles were parents. I predicted that if the population had a harem-polygyny mating system, a single male would fether most of the
offspring within a colony (Moritz et al., 1997). 10
Methods
The study was conducted during the Patagonian summer, from January to March
1996. The study area is mixed grass-shrub steppe, east of the Alumine River in the province ofNeuquen, Argentina (70° 56' W, 39° 36'S) (Fig. 2-1). Elevation is from 800
to 1 500 meters. Forty-one rocky cliffs and outcrops are scattered throughout the study area, and mountain vizcachas are found to the north, east and south of the study area.
Although similar rock outcrops are foimd on the western side of the Alumine, no
mountain vizcachas Uve on that side of the river. CUffs were digitized from a 1 :50,000
map of the study area with CARTALINX 1 .0 (Clark Labs, Worcester, MA) and the digitized map was imported mto ARCVIEW 3.2 (ESRI, Redlands, CA). I used
ARCVIEW to measure distance between cliffs and total length of cUffs. The longest straight-line distance without rock between two cliffs was 1667 m (mean: 721.5 m; standard deviation: 523.2; range: 150.5 to 1667 m). Average length of cliffs was 478 m
(standard deviation: 292; range: 76 to 1353 m).
I surveyed all cliffs in the study area to determine if they were occupied by
mountain vizcachas. The presence of mountain vizcachas is easily discerned by the presence of their fecal pellets, which are unlike those of any other species in the area. At cliffs where mountain vizcachas were captured, I estimated the height of the cliff at two randomly chosen points within every 100 meters of cUff Height was averaged and
multiplied by the length of the cliff to obtain an estimate of the surface area of the cliffs, which were mostly vertical.
To obtain samples for genetic analysis and to mark individuals for estimation of population size, mountain vizcachas were captured at sbc cliffs in three areas: Prieto (n =
4), Los Ranchos (n = 1 1), and Rodriguez (n = 8) (Fig. 2-1). No mountain vizcachas 11
entered traps placed below and near rocky cliffs during 140 trap nights. One mountain vizcacha was captured in a trap positioned to block the entrance to its den, and 22 were captured by hand by local assistants after being chased into rock crevices. Thus, captures were limited to those individuals that sought refuge in accessible crevices.
Captured mountain vizcachas were kept in covered cages and sedated with 20 mg/kg of ketamine hydrochloride injected intramuscularly before processing. While each
individual was under sedation it was sexed, measured, weighed, and marked with colored
and numbered rabbit ear-tags in both ears. For genetic analyses, I collected the ear tissue that was pimched out in order to place the tags. This tissue was stored in a DMSO-based buffer at room temperatiire. Mountain vizcachas were held overnight and released the
next day at the site of their capture. I used a mark-resight method to estimate population sizes with the unbiased Lincobi Peterson estimator at five of the six cliffs where mountain vizcachas were captured (Appendix A). As estimated popvilation sizes were small (Table 2-1), sample sizes for genetic analysis represented a large percentage (33 to
88%) of individuals from each cliff where mountain vizcachas were captured.
The DNA was isolated from tissue and blood using standard phenol-chloroform
extraction procedures (Sambrook et al., 1989). Eight microsatellite loci developed for mountain vizcachas (Appendix B) were amplified with fluorescently-tagged primers in
25 nL PGR reactions [10 x Assay Buffer A (Fisher Scientific), 200 mM dNTPs, 0.4 mM
primer, 1 U Taq DNA polymerase (Fisher Scientific)] under the following conditions:
93°C, 3 m; 7 cycles of 92°C, 30 s; optimal annealing temperature (52 to 58°C; see
Appendix B, Table B-1), 60 s; 72°C, 90 s; and 30 cycles of 89''C, 30 s; optimal annealing temperature minus one degree, 60 s; 72"C, 90 s (Gagneux et al, 1997). Alleles were resolved by electrophoresis using a 4.5% polyacrylamide gel on an automated sequencer 12
[Applied Biosystems Inc. (ABI, Foster City, CA)-Prism 377]. Data were collected and
analyzed using GENESCAN 2.0.2 and GENOTYPER 2.0 (ABI).
Samples from the six cliffs were grouped into three areas (Prieto, Rodriguez, and
Los Ranchos) for genetic analysis of population structure. Exact tests for Hardy-
Weinberg equilibrium (Guo and Thompson, 1992), linkage disequilibrium, and
population differentiation among the three areas were computed with GENEPOP 3.2
(Raymond and Rousset, 1995a; 1995b). Frequencies of null alleles were calculated with
CERVUS versionl.O (Marshall et al., 1998). The F-statistics were computed according
to Weir and Cockerham (1984), and R^ (Slatkin, 1995) was computed with standardized
alleles according to Goodman (1997). Probability of F-statistics was determined by 1000
permutations with the program FSTAT version 2.9. 1 (Goudet, 2000). Alleles were
permuted within and among populations for Fis and Fit, respectively, and genotypes were
permuted among populations for Fst.
For suspected parent-offspring pairs (n = 5), I used the program CERVUS to
calculate likelihood of parentage (Marshall et al, 1998). I also calculated the hkelihood
that each adult male in a cliff was the father of every other moimtain vizcacha from the
same cUff. Likelihood analysis takes the actual data as a starting point, and evaluates
hypotheses about probable parenthood given that data (e.g., HI = The candidate parent is
the true parent; Ha = The candidate parent is not the true parent). The hkelihood of each
hypothesis given the observed genotypes can be calculated from the probability of
obtaining the observed genotypes under that hypothesis. The hkelihood of one hypothesis
is always evaluated relative to another (i.e., likelihood ratio). In this case the likelihood ratio is the hkelihood that the candidate parent is the true parent divided by the likelihood that the candidate parent is not the true parent, given the observed genotypes. In 13
CERVUS, confidence levels for parentage assignment are obtained by comparison to results of a large number of simulated parentage tests using the same data. Parameters provided by the user for the simulations are the number of candidate parents, the proportion of those that were sampled, and the rate of typing error. Based on estimated population sizes at cliffs and the proportions of populations sampled, the parameters I used in the simulations w^ere 4 candidate parents, with 50% of those sanqjled, and a typing error of 5%.
Shared allele distance between individuals (Chakraborty and Jin, 1993, Estoup et
al., 1995) was calculated with the program by J. Brzustowski at
http://www.biology.ualberta.ca/jbrzusto/sharedst.htnil. I tested the null hypothesis that mean distance between individuals from different cliffs was no greater than mean distance between individuals from within cliffs with a paired-sample t-test.
Results
All 41 rock outcrops were occupied. The number of individiials at the five cliffs
where I was able to estimate population size were small (Table 2-1). Density averaged
1 1.2 mountain vizcachas per kilometer of linear cliff" (n = 5; 95% confidence intervals:
8.6 tol6.1/km), or 10.8 moimtain vizcachas per hectare of cliff surface (n = 4; 95% confidence intervals: 7.7 to 18.1 /ha; height estimates were not available for one cliff). As a total of approximately 17,912 meters of cliff occurred within the area, the population for the 30-km^ area was estimated as 201 mountain vizcachas (95% confidence intervals:
154 to 288), or 0.067 mountain vizcachas/ha (95% confidence intervals: 0.051 to 0.096).
Alleles for all individuals are presented in Appendix C. The sample as a whole
was not in Hardy- Weinberg equilibrium (Chi^ = 35.4; df = 16; P = 0.004). However, all 14
loci were in Hardy-Weinberg equilibrium within the three areas (P > 0.46), and there was no evidence of linkage disequilibrium among loci. I also considered that there was no
evidence for null alleles, as estimated frequencies ranged from -0.07 to 0.07. Fis and Fit
were not significantly different from 0 (Table 2-2). Fst (0.055) and Rst (0.052) were
similar, and Fst was highly significant (P = 0.007).
Table 2-1 . Numbers of mountain vizcachas captured, location of captures, estimated population sizes, and proportion of population sampled, in Neuquen Province, Argentina.
Estimated Proportion of Cliff surface Number population population Area Cliff Length (m) area (m^) captured size (95% CI) sampled
ROD B4 1,070 16,536 7 19(14-23) 0.42
ROD B20 420 NA 1 3 0.33 RAN B15 640 8,283 7 8 (7-13) 0.88 RAN B19 530 5,079 4 5 0.80 PRI B31 280 1,960 3 4 0.75
PRI B22 330 7,307 1 NA NA
95% CI = 95% confidence interval. ROD = Rodriguez; RAN = Los Ranchos; PRI = Prieto. NA = not enough data available to estimate.
An almost equal number of male (n = 1 1) and female (n = 12) mountain vizcachas were captured. Most (70%) were caught in the same rock crevice with one or two other
individuals. These included adult males caught with adult females (n = 3), and adult females caught with much smaller males or females (n = 5). Likelihood analysis ruled out the female as the mother of the smaller animal in all but one case (0.90 < P < 0.95).
On one occasion a large female and male were captured with a juvenile. The male could not be ruled out as the father (0.80 < P < 0.90), but the female was not the mother. 15
At the three cliffs where I caught two or more adult males, at least two males
sired offspring at each cliff. At two cliffs (B4 and B15), one male was identified as the probable father (P > 0.80) of one to two mountain vizcachas, but no probable fether was identified for 4 and 5 other vizcachas, respectively. At both cliffs a male was present for which no offspring were identified. At the third cliff, two probable father-offspring pairs were identified (P > 0.85).
Overall, the genetic distance between adults from different cliffs was greater than
that between adults from the same cliff (t = -1.862; df = 15; P = 0.041). Compared by
sex, however, genetic distance was as similar between cliffs as within cliffs for males (t =
= -0.974; df = 6; P = 0. 1 8), but for females it was greater between than within cliffs (t
-2.056; df = 8; P = 0.037).
Table 2-2. The F- statistics by locus and overall, calculated according to Weir and Cockerham(1984).
Locus F« F« Fi,
LVl 0.095 0.073 0.161 LV4 -0.181 0.154 0.001 LV5 0.051 0.13 0.175 LV8 -0.075 0.21 0.151 LV16 0.182 -0.042 0.148 LV17 -0.106 -0.024 -0.133 LV25 -0.16 -0.008 -0.169 LV33 0.022 -0.035 -0.012
All -0.012(0.049) 0.055 (0.033) 0.043 (0.054) Probability 0.58 0.007 0.197
Standard errors in parentheses after estimate. 16
Discussion
Gene flow between colonies of mountain vizcachas in different parts of the study
area was restricted, suggesting limited movement of individuals among areas. All cliffs
were probably accessible from other cliffs, as no two were more than 1 .7 km apart. No
information on dispersal distance of mountain vizcachas is available, but rock hyraxes,
rock-dwelling mammals of similar body size in the African savanna, are known to
disperse at least 2 km (Hoeck, 1987), and male plains vizcachas, over three times larger
than moimtain vizcachas, can disperse at least 8 km (L. Branch, personal
communication). Nevertheless, the population as a whole was not panmictic, as it did not
conform to Hardy-Weinberg expectations, and local populations within clusters of cliffs
did. This suggests that the departure from random mating within the whole population
was due to subdivision into genetically distinct groups, and indeed, the overall Fst was
significant. However, sample sizes within groups were small for testing for conformance
to Hardy-Weinberg expectations. Dispersal is probably male-biased, as females were
more closely related to other adults within their colonies than to those from other
colonies, but males were not. Therefore, although limited movement could be due to the
spatial arrangement of habitat patches, it could also be a consequence of the social
structure, with females more inclined than males to remain within their natal colonies
(Chesser, 1983; Surridge et al, 1999).
Although there was limited movement of moimtain vizcachas between cliffs, all
clififs were occupied, so the population does not appear to be exhibiting
extinction/reco Ionization dynamics as expected for a classical metapopulation (Levins,
1969). However, there could be a metapopulation maintained by rescue effect, in which migration prevents local extinction (Stacey et al., 1997). This type of metapopulation 17
structure is diflficuh to detect unless migration is disrupted or can be measured (e.g.,
Stacey and Taper, 1992; Stacey et al., 1997). In addition, the term "rescue effect" may
not appropriately describe the actual effect of immigration on local populations if the
effect is not rescue from imminent extinction. Male immigration may be irrelevant to
population persistence, and female immigration into occupied cliffs could be detrimental
due to competition between immigrants and kin groups of resident females (Clinchy,
1997).
Social structure of female mountain vizcachas consists of female-kin groups
occupying different habitat patches with limited exchange. Male social structure could
not be definitively characterized. An equal number of males and females were captured,
and there was no evidence that single males were fethering most of the offspring within a
cHff, which suggests multiple male-female pairs. However, 1 could not rule out a harem-
polygyny mating system with multiple harems per cliff and non-reproducing males
(Downhower and Armitage, 1971). Among plains vizcachas, more than one male often is
associated with a female kin group, but whether or not all males are reproducing is
unknown (Branch, 1993).
The largest group I found was composed of about twenty individuals, while in
Peru groups of up to 75 L. peruanum were observed (Pearson, 1948). Density for the entire study area (not just rocky habitat~0.067/ha) was also lower than in Peru
(0.099/ha), possibly due to a lower proportion ofrocky habitat. However, because mountain vizcachas in Patagonia were approximately twice the size of those in Peru (0.95 to kg), 1.5 biomass of mountain vizcachas in Patagonia (161 g/ha) was higher than in
Peru (122 g/ha). 18
The cliffs and outcrops in this study were probably too small to support
aggregations of 75 mountain vizcachas. Thus, habitat in the study area probably
approximates the more patchy distribution that Mares and Lacher (1987) predicted to
result in a harem-polygyny mating system. The sizes of the cliffs were very similar to the
size of rock outcrops inhabited by rock hyraxes in Africa, which do have a harem-
polygyny system (Hoeck et al, 1987). Within cliffs, rock crevices that provide suitable
shelter from predators and for den sites seem to be a critical and limiting resource for
mountain vizcachas in the steppe (Walker et al., 2000; Chapter 3). The spatial
distribution of suitable crevices within a cliff may make it difficult for a male to
monopolize more than one crevice. Thus, if the mating system depends on male ability
to monopolize resources, the distribution of crevices within cliffs may have a greater
effect on social organization than the distribution of habitat patches within the landscape.
However, even if the mating system is polygynous, it appears xmlikely to be due to male
monopolization of crevices. Male defense of crevices against other males would
probably not assure them of mating with the female occupants of the crevice, as mountain
vizcachas mix and forage in groups away from the crevices (personal observation).
I conclude that the relative importance of habitat structure and social structure in
determining patterns of movement in fragmented habitat merits further evaluation within
the context of metapopulation theory. The effect of variation in habitat structure on
metapopulation dynamics is an area of active investigation. The potential effects of conspecific attraction in highly social species, in which movement only occurs to habitat patches that are already occupied, have been discussed (Smith and Peacock, 1990).
However, the effects of male-biased dispersal and a social structure based on female kin groups need to be addressed theoretically and investigated empirically (Clinchy, 1997). 19
These phenomena could accentuate the effect of isolation on the probability of colonization of a patch or recolonization of patches where populations go extinct, because successful colonization depends on much rarer female dispersal. In addition, the probability of local extinction could depend largely on processes affecting female survival and reproductive success. 20
2 km
2-1 Figure . Cliffs in Pilolil, Neuqu^n Province, Argentina, and three areas where mountain vizcachas were captured. Inset map from htQ)://encarta,msn.com/maps. CHAPTER 3 EFFECTS OF QUALITY, SIZE, AND ISOLATION OF PATCHES ON PATCH OCCUPANCY BY MOUNTAIN VIZCACHAS
Introduction
For species with patchy distributions, presence in a habitat patch may be
determined by the quality of the habitat within the patch, the size of the patch, and
landscape-level factors that affect movement among patches or extinction probabilities
within patches (Mazerolle and Villard, 1999). Patches that can support only small
populations because of small area or poor habitat quality are less likely to be occupied
than large or high quality patches, because small populations have a higher probability of
extinction than do large populations (Wilcox, 1980). Small populations are highly
susceptible to extinction through natural stochastic phenomena, and very small
populations may be subject to extinction due to demographic stochasticity. In addition,
for many colonial species, demographic parameters such as survival and reproduction are
positively related to colony size (Courchamp et al., 1999; Stephens and Sutherland,
1999). If movement among patches is constrained by the composition and configuration
of the landscape (Wiens, 1997), isolated patches may be unoccupied because colonization
of empty patches and demographic rescue of isolated populations is limited.
Metapopulation processes refer to the effects of movement of individuals among habitat patches on local population dynamics and the regional persistence ofthe species
(Stacey et al., 1997). Habitat quality is commonly evaluated based on habitat
21 22
characteristics associated with the presence or absence of the species (Vemer et al.,
1986), and studies of metapopulation dynamics of subdivided populations are frequently
based on presence or absence of the species in a habitat patch (Hanski, 1994a, b). Effects
of habitat quality and metapopulation dynamics on patch occupancy may be confounded
in studies based on presence or absence of a species. Therefore, many researchers
include both habitat and landscape variables in their analyses to determine the relative
importance of local and landscape-level processes for the persistence and distribution of
subdivided populations (Mazerolle and Villard, 1999).
Moimtain vizcachas (Lagidium spp, family Chinchillidae) are large hystricomorph
rodents that live in rocky habitats in the southern Andes and the Patagonian steppe of
South America (Redford and Eisenberg, 1992). The taxonomy of the genus is not well
defmed and little research has been conducted on the ecology of the three or four species
(Galende et al., 1998; Pearson, 1948; Puig et al., 1998; Walker et al., 2000). Mountain
vizcachas are herbivorous, and all species appear to be colonial, living in kin groups
within colonies (Pearson, 1948; Chapter 2). The broadly homogeneous Patagonian
steppe is interspersed with rocky cliffs and outcrops that provide a patchily distributed
habitat for mountain vizcachas. These rocky habitats are widespread, but not all are
occupied by mountain vizcachas (Chapter 4). In some areas of Patagonia, anecdotal
evidence indicates that numbers of mountain vizcachas have declined greatly in recent
years. Because there are no data on abundance and distribution in the past, population
trends and potential causes for a decline are only speculative.
The purpose of this study was to identify factors at the patch and landscape level that affect the distribution and persistence of the mountain vizcacha in the Patagonian 23
steppe. Specifically, I identified characteristics of the rocky habitat patches that could
indicate habitat quality and compared the effects of habitat quality, patch size, and patch
isolation on patch occupancy. If isolation is an important determinant of patch
occupancy, movement among patches is probably limited, and the distribution and
persistence of the regional population should be analyzed in a metapopulation context
(Hanski and SimberlofF, 1997).
Methods
Study Area
The study was conducted in a 12,000-km^ area in southern Neuquen and western
Rio Negro provinces in Argentina (Fig. 3-1). This area is semi-arid, grass-shrub steppe
with an average ground cover of about 50% (Leon et al., 1 998). Cliffs and rock outcrops
are mostly of volcanic origin. Cliffs are often found at the tops of hills and have flat tops
and relatively vertical faces. They range in size from less than 100 m to more than 50 km
long.
Assessment of Habitat Quality within Cliflfs
To identify habitat characteristics associated with presence of mountain vizcachas
within cliffs, I used logistic regression to compare features of occupied and imoccupied
portions of occupied cliffs at two sites within the study area, Pilolil and La Rinconada
(Fig. 3-1). By working within occupied patches rather than comparing occupied and unoccupied patches, 1 avoided the potential confounding effects of landscape-level factors (e.g., patch isolation). Pilolil has numerous rock outcrops of various sizes, and La
I^conada is a single long (> 50 km) vertical cliff Eighteen cUffs were sampled at 24
Pilolil and La Rinconada was sampled with five random transects. Transects, which were
placed on the tops of cliffs and rock outcrops, were 400 meters long, except cliffs less
than 400 meters in length were sampled in their entirety.
The following habitat variables were measured at two randomly chosen points
within each 100 meters of transect: distance between, depth o^ and width of rock
crevices, and height and slope of the cliff Rock crevices are used for dens and as refuge
from predators, and height and slope of the cUff may influence predation risk. Crevice
measurements were log-transformed before analysis to meet the assumption of normality.
Each sampling point was also scored as occupied or unoccupied by mountain
vizcachas. The occupancy of a cliff or rock outcrop by mountain vizcachas is easily
discerned by the presence of their feces. Mountain vizcachas defecate in simning spots
near their dens in the rocks, and their feces are distinct from the feces of any other animal
in the region (Pearson, 1948). Sampling points with mountain vizcacha feces within ten
meters in either direction were considered occupied.
An assumption of logistic regression is indef>endence of san^ling units. This
assumption is violated when there is spatial autocorrelation among sampling units for the
variable being measured. Use of logistic regression in the presence of spatial
autocorrelation will result in incorrect standard errors and estimates of probability. By
using a mixed linear model with the appropriate covariance structure, spatial
autocorrelation can be incorporated into the logistic regression model (Gumpertz et al.,
2000).
The potential for spatial autocorrelation was high in my data set because sampling points were compared from within the same cliffs. To determine the pattern and extent of 25
this autocorrelation I used the program GEO-EAS (Englund and Sparks, 1988) to plot
semi-variograms of different independent variables against distance between points (Cliff
and Ord, 1973). I then did the logistic regression analysis using the GLIMMIX macro for
SAS (Littell et. al, 1996; SAS Institute, Inc., 1996). This macro uses General Linear
Models to accommodate fixed effects within a logistic regression model and allows
specification of the covariance function. I stratified the analysis by cliff, and modeled the
particular pattern of spatial autocorrelation revealed by the semi-variograms.
The semi-variograms indicated that the range of spatial autocorrelation of
sampling points along the same cliff was 100 to 120 meters for most variables.
Therefore, adjacent samplmg points were autocorrelated, as I sampled at two locations
within each 100 meters. Because the habitat is linear in form, the autocorrelation occurs
in only one dimension, and I used a Toeplitz covariance structure (Littell et al., 1996)
with 2 bands to incorporate spatial autocorrelation between neighboring points into my
logistic regression analysis. I also tested a model using a Toeplitz structure with 3 bands
to determine if extending the range of autocorrelation provided a better description of the
data.
Comparison of Factors Affecting Patch Occupancy
Analysis of the effects of habitat quality, patch size, and isolation on patch occupancy was based on sampling of 36 chffs that were selected from throughout the entire 10,000-km^ area (Fig. 3-1). These sites were surveyed in the same manner as the
PiloUl and La Rinconada sites, and habitat measurements were averaged for each
transect. In addition to the habitat variables measured within the cliffs, I included the length of the cliff as a measure of patch size and the distance between the cUff and the 26
nearest clifiF that was occupied by mountain vizcachas as a measure of patch isolation.
Lengths of cliffs and distances to nearest occupied neighbors were measured from a
digitized map of the study area using ARCVIEW (ESRI, Redlands, CA) GIS software
and log-transformed. I used logistic regression to test effects of habitat variables, cliff
length, and distance to nearest occupied neighbor on the probability of occupancy of
cliffs. I used discriminant fimction analysis to compare the ability of these variables to
predict patch occupancy.
As orientation of chffs has been considered an important factor for habitat
suitability for the closely related chinchilla, also a rock specialist (Jimenez, 1995), I also
measured the orientation of the cliff at each sampling point. Cliff orientation could not
be analyzed with logistic regression because it represents data from a circular
distribution. I calculated the mean angle of each cliff and compared mean angles of
occupied and unoccupied cliffs using Watson's test for two samples (Zar, 1984).
Results
Mountain vizcachas were more likely to be found at points along cliffs with more
and deeper crevices (Chi^ = 15.89; df = 2; P < 0.001) (Table 3-1). The depth of crevices was retained in the habitat model despite its probability > 0.05 because removing this
parameter decreased the model's fit (Chi^ = 6.12; df = 1; P = 0.01). Variables not included were the width of crevices (P = 0.45), and the height (P = 0.10) and slope (P =
0.27) of the cliff Inclusion of the spatial autocorrelation between neighboring points significantly improved the fit of the habitat model (Chi^ = 1 1.87; df = 1; P < 0.001), but inclusion of autocorrelation = = = beyond 100 meters did not (Chi^ 0.60; df 1 ; P 0.44). 27
Table 3-1 . Logistic regression model predicting presence of mountain vizcachas within patches, incorporating spatial autocorrelation, Neuquen Province, Argentina.
Mean Mean occ. unocc.
(cm) (cm) Coefficient S.E. t df Probability
Distance between crevices 60.9 117.0 -0.49 0.24 -2.04 129 0.044 Depth of crevices 30.3 22.1 0.42 0.23 1.87 129 0.063 Intercept 1.49 1.14 1.31 20 0.205
Mean occ. = mean for occupied points (n = 105). Mean imocc. = mean for unoccupied points (n - 47). S.E. = standard error of the coefficient.
The slope of the cliff as well as the distance between and depth of crevices were
significant patch characteristics predicting patch occupancy (Chi^ = 25.6; df = 3; P <
0.001) (Table 3-2). Mountain vizcachas were more likely to be found on patches with
more and deeper crevices and steeper slopes. However, when patch size and the distance
to the nearest neighbor were added to the analysis, the best logistic regression model
predicting patch occupancy included only the distance between and depth of crevices and
the distance between a cliff and its nearest occupied neighbor (Chi^ = 27.7; df = 3; P <
0.001) (Table 3-2). As in the other models, mountain vizcachas were more likely to be
found on patches with more and deeper crevices, and the probability of a patch being
occupied decreased with increasing isolation from other occupied cliffs. This model that
incorporated both habitat quality and patch isolation was significantly better than the model with only the habitat variables (Chi^ = 14.8; df = 2; P < 0.001). The length (P =
0.27) and height of the cliff (P = 0.1 1) and the width of crevices (P = 0.38) did not contribute significantly to predicting the probability of occupancy of a cliff. The mean orientation of occupied cliffs (38.5°) was not different from that of unoccupied cliffs
(32.5°) (U^7,i9 = 0.07; 0.20 < P < 0.50). 28
Table 3-2. Logistic regression models of probability of occupancy of cliffs in 10,000- km^ area according to patch characteristics and according to patch characteristics and isolation, Neuquen Province, Argentina.
Mean Mean
occ. unocc. Coefficient S.E. t Probability
Model based on patch characteristics:
Distance between crevices 90.9 cm 127.3 cm -2.3 0.96 -2.40 0.016 Depth of crevices 38.1 cm 24.9 cm 2.8 1.08 2.54 0.011 Slope 38.3° 32.8° 0.21 0.10 2.16 0.03 Intercept -5.75 4.28 -1.34 0.179
Model based on patch characteristics and patch isolation:
Distance between crevices 90.9 cm 127.3 cm -2.71 1.24 -2.18 0.037
Depth of crevices 38. 1 cm 24.9 cm 4.4 1.87 2.35 0.025 Distance to nearest
occupied cliff 2. 1 km 7.1 km -2.17 0.81 -2.67 0.012 Intercept 14.65 7.13 -2.05 0.048
Mean occ. = mean for occupied cliffs (n = 20). Mean unocc. = mean for unoccupied cliffs (n = 16). S.E. = standard error of the coefficient. Degrees of freedom = 31.
Seventy-five percent of the cliffs were correctly classified as occupied or not
occupied by mountain vizcachas based solely on the measures of habitat quality (distance
between and depth of crevices) (Wilks' Lambda: 0.68; approx. F2,32 = 7.63; P < 0.002).
Isolation alone was equally good at predicting patch occupancy, correctly classifying
77% of cUffs (Wilks' Lambda = 0.69; approx. Fi,33 = 14.69; P < 0.0005). Ninety-one
percent of cliffs were correctly classified when habitat quality, the slope of the cliff, and the distance to its nearest occupied neighbor were used as predictors (Wilks' Lambda;
0.39; approx. F4^9 = 1 1.19; P < 0.001). 29
Discussion
Cliffs with high quality habitat for mountain vizcachas contain abundant and deep
crevices. Crevices are used for denning and for escape from potential predators (personal
observation). Potential predators (puma, culpeo fox, buzzard eagles, humans) are
abimdant in the area but none are capable of removing a mountain vizcacha from a very
deep crevice. Food availability could also be an important factor determining habitat
quality for a species. However, food availability is difficult to assess for a generalist
herbivore (Sinclair et al., 1982). A previous study found no relationship between
vegetative cover and movements of mountain vizcachas away from the rocks (Walker et
al., 2000). The high success in predicting presence/absence of mountain vizcachas with
measurements of rock crevices indicates that rock crevices, rather than food or vegetation
cover, are the primary resources limiting for this species.
This high level of success in predicting occupancy only with data on rock also
may suggest that most unoccupied cliffs did not have adequate crevices. The nature of
the crevices in rocks of volcanic origin depends on the speed and temperature at which they cooled (Mottana et al, 1978). Many of the rocks in the area are products of the
same geological event, and should therefore have similar crevices as they formed under
similar environmental conditions. Rocks without appropriate crevices could have formed at different times, or could have had different local conditions during cooling.
Patch isolation also was an important predictor of patch occupancy by mountain vizcachas, but patch size was not. Patch isolation and size have been found to be
important predictors of patch occupancy for a variety of species (Micol et al., 1994; van
Apeldoom et al., 1994; Arnold et al., 1995; Hokit et al., 1999). The effect of patch size 30
on patch occupancy is attributed to the fact that smaller patches have smaller population sizes, and smaller populations have a greater risk of extinction (Wilcox, 1980). In this study the length of the cliff was not related to the probability that the cliff was occupied by mountain vizcachas. Therefore, either no cliffs were too small to support mountain
vizcachas, the length of the cliff is not correlated with mountain vizcacha population size, or some other factors are more important in determining a population's risk of extinction.
Population size of mountain vizcachas in the area has been found to be related to the
slope of a cliff (Walker et al., 2000). The slope of the cliff was a significant predictor of probability of patch occupancy in this study when only within-patch variables were considered. When patch isolation, measured as distance to nearest occupied neighbor, was added, the significance of the slope of the cliff disappeared. One possible
explanation is that small populations on patches with low slopes may have a greater risk
of extinction, but a less isolated patch is more likely to be rescued from extinction by immigration from a neighboring patch or recolonized after extinction than an isolated patch with low slope.
Habitat quality and patch isolation are both major factors affecting distribution of the mountain vizcacha in this Patagonian steppe landscape. Presence of mountain vizcachas in a particular patch with appropriate rock depends on the distance between the patch and other patches with mountain vizcachas. Therefore, movement among patches is an important process determining distribution, and analysis of persistence and distribution of the species in the region should be evaluated in a metapopulation context.
The persistence and distribution of mountain vizcachas cannot be fully understood 31
without considering the rate and distance of movement of individuals among patches
(Stacey et al., 1997). 32
Figure 3-1. Study area with locations where habitat quality for mountain vizcachas was evaluated (Pilolil and La Rinconada), and locations of 36 random transects in southern Neuquen Province, Argentina. CHAPTER 4 LANDSCAPE CONNECTIVITY AFFECTS DISTRIBUTION OF THE MOUNTAIN VIZCACHA, A HABITAT SPECIALIST IN FRAGMENTED HABITAT
Introduction
Landscape Connectivity
The potential impacts of spatial structure of habitats on the dynamics of populations have long been recognized (Andrewartha and Birch, 1954). Recently, however, increasing habitat fragmentation has introduced urgency into the analyses of factors affecting persistence and distribution of subdivided populations. Metapopulation theory has become the theoretical paradigm for these analyses (McCullough, 1996;
Hanski and Gilpin, 1997). The metapopulation concept has expanded greatly since its original formulation (Levins, 1969) to include different configurations of habitat patches and rates of movement among patches. Currently, the defining feature of a
metapopulation is dispersal connecting discrete local populations, rather than extinction/reco Ionization processes (Hanski and Simberloff, 1997). Metapopulation dynamics under this broader defmition may affect persistence of local or regional populations and/or the regional distribution of a species, according to the particular configuration of habitat patches available and the rate of movement of individuals among patches (Harrison and Taylor, 1997).
The metapopulation approach implies a sharp, binary contrast between suitable and unsuitable habitat, with the unsuitable habitat assumed to be a featureless matrix in
33 34
which patches of suitable habitat are embedded (Hanski and Gilpin, 1997). This idealized
representation of landscapes has found Umited applicability for explaining persistence
and distribution of real species in real landscapes (Harrison and Taylor, 1997). Dispersal
is the key process in metapopulation theory, but the ability of individuals to disperse
between any two habitat patches depends on the distance between the patches, the biotic
and abiotic characteristics of the matrix between the patches, and the biology and
behavior of the organism (Taylor et al., 1993). Modehng studies indicate that the
dynamics of subdivided populations are very sensitive to the rate of movement among patches (Lefkovitch and Fahrig, 1985; Fahrig, 1988, 1990; Fahrig and Paloheimo, 1988),
and that dispersal rates and pathways may be constrained by spatial features of the
landscape, especially in the case of matrbc-sensitive habitat-specialists (Stan:q)s et al.,
1987).
This interaction between attributes of the species and structure of the landscape in mediating dispersal among habitat patches is embodied in the concept of landscape
connectivity (Merriam, 1984; Tischendorf and Fahrig, 2000). Landscape connectivity is the degree to which the landscape facilitates or impedes movement among patches
(Taylor et al., 1993). The concept incorporates the combined effects on the movement rate among habitat patches of 1) the landscape structure and 2) the species' use of, ability to move through, and risk of mortality in different landscape elements. Therefore, connectivity is species- and landscape-specific, and in any particular situation, the landscape structure must be described from a species' point of view (Tischendorf and
Fahrig, 2000). The concept of connectivity provides an alternative to the straight-line, unimpeded dispersal distances usually implicitly assumed in metapopulation theory 35
(Hannson, 1995). A more comprehensive understanding of the role of dispersal in
determining the regional distribution of a species could result from empirical studies that
evaluate landscape connectivity and structure at a regional scale.
Evaluating the connectivity of a landscape for a given species requires some
measure of movement rates through different landscape elements at a spatial scale
sufficient to encompass movement capabilities of individuals over the entire landscape
(Merriam, 1995; Pither and Taylor, 1998). Likewise, resolving the effects of dispersal
on the regional distribution of a species requires study at an appropriate scale. Few
studies have attempted to en^irically evaluate connectivity, and these studies have used mark-recapture or radio-telemetry techniques to estimate rates and distances of dispersal in a population (reviewed in Tischendorf and Fahrig, 2000). These techniques are not feasible at a regional scale for many medium- or large-bodied species or for species or populations in which individuals are difficult to capture (Merriam, 1995).
Analysis of genetic structure and gene flow is an alternative method for estimating dispersal (Koenig et al., 1996). Gene flow measures more than dispersal, however, as it depends on successful reproduction at the new site (Barton, 1992).
Genetic differentiation of populations requires multiple generations of reproduction, so measures of gene flow reflect historical rather than current movement patterns. However, the recently introduced molecular markers that evolve very rapidly, such as microsatellites (Ashley and Dow, 1994), may resolve movement patterns on a shorter historical timeframe than did previous genetic markers (allozymes and mtDNA). Gene flow may be used to estimate the movement component of connectivity when the time scale of the existence of habitat patches is long relative to generation length of the species 36
under study and the mutation rate of the molecule marker. Recent advances in genetic
technology also allow indirect sampling of individuals by obtaining DNA from hair or
feces (Woodruff, 1993). For some species this type of sampling may allow assessment of
gene flow at the appropriate geographic scale for the study of landscape connectivity and
its effect on regional distribution. Although not framed in the context of metapopulation
theory or landscape connectivity, several studies have examined the effects of landscape
structure on the genetic architecture of populations (Templeton et al., 1990), and have
documented reduced gene flow across rivers (Pounds, 1981; King, 1987) and through
different habitat types (Arter, 1990; Keyghobadi et al., 1999).
In this study I empirically explore the relationship between landscape connectivity and the regional distribution of a rock-dwelling mammal, the mountain vizcacha (Lagidium viscacia. Family Chinchillidae), in naturally fragmented habitat. The approach I use to determine landscape connectivity combines molecular genetic estimates of gene flow as an analog of movement rates and cost-distance analysis with GIS to
model landscape resistance and facilitation of movement. Thus, I determine species-
specific connectivity at a regional scale in an explicit landscape. I test predictions about patterns of patch occupancy and genetic structure that would be expected under a series of alternative hypotheses about the mechanisms determining regional distribution of the species. My results provide key insights into how the effect of inter-population dispersal on regional distribution may be constrained by configuration of habitat patches, permeability of the matrix between patches, and the biology of the species. 37
Background and Research Hypotheses
Lagidium viscacia is distributed along the Andean cordillera of South America
from Bolivia to northern Patagonia and throughout the northern Patagonian steppe. It is a
social species, living in small kin groups within larger colonies (Chapter 2). The broadly homogeneous, mixed grass/shrub steppe of Patagonia is interspersed with rocky cliffs and outcrops, which provide a widespread but patchily distributed habitat for the mountain vizcacha. Local movements of mountain vizcachas away from the cliffs for foraging or
other activities are highly correlated with rocky substrate (Walker et al., 2000). Rock- dwelling mammals have morphological adaptations for Uving among rocks that make movement across other substrates difficult (Mares and Lacher, 1987). Studies of such species in Africa and Australia suggest that their movement through the non-rocky
habitat matrix is limited (Hoeck, 1982; Pope et al., 1996). Research on the pika
{Ochotona princeps), a small rock-dwelling mammal ofNorth America, has revealed relationships between dispersal, landscape, and mating systems (Peacock, 1997; Peacock and Smith, 1997), and highlighted the potential importance of spatially-autocorrelated processes in determining distribution (Smith and Gilpin, 1997).
My central hypothesis is that landscape connectivity and structure determine the regional distribution of mountain vizcachas. I simultaneously evaluated four hypotheses that embody alternative mechanisms to explain regional distribution: 1) dispersal capabilities of mountain vizcachas are so great relative to distances between habitat patches and characteristics of the matrix that distribution is determined only by availability of habitat; 2) movement among habitat patches is so limited that distribution is determined only by local population dynamics; 3) movement is not constrained by 38
patch configuration or characteristics of the matrix, and distribution reflects a dynamic
balance between local extinction of populations and recolonization of vacant habitat
patches (classical metapopulation); and 4) distribution is determined by spatially-
autocorrelated extinction events.
I tested predictions derived from these hypotheses based on patterns of patch
occupancy, genetic structure, and landscape structure (Table 4-1). The pattern of patch
occupancy reflects the history of colonization and/or extinction. In a highly social mammal such as the moimtain vizcacha, colonization of vacant patches may be much rarer than dispersal of individimls to occupied patches, due to conspecific attraction
(Smith and Peacock, 1990). The use of genetic analyses allowed me to evaluate the role
of dispersal among occupied patches (Stacey et al., 1997). Because of the strict
association of mountain vizcachas with rocks, I considered landscape structure from this species' point of view to be defined by the distribution of cliffs, rock outcrops, and rocky substrate. I also considered that rivers could be potential barriers to movement and thus, important features affecting landscape connectivity.
Methods
Study Area
The study was conducted in Patagonian steppe in the southern portion of the
province ofNeuquen, Argentina (39.5° S and 71° W) (Fig. 4-1). The study area is approximately 12,000 km^ of semi-arid mixed grass-shrub steppe (Leon et al., 1998) with numerous rock outcrops, many in the form of cliffs with vertical faces and flat tops. The geology of the area has been determined by volcanic activity along the Andean cordillera 39
during the last 20 million years and Pleistocene glaciations. Several rivers ranging from
10 to 100 meters wide traverse the area.
4-1 Table . Hypotheses about mechanisms determining distribution of mountam vizcachas in the Patagonian steppe and predictions that would support each hypothesis.
Hypothesis Predictions
1) Landscape connectivity Patch occupancy and genetic structure will be correlated with geological surface structure and/or rivers.
2) Availability of habitat All suitable habitat patches will be occupied.
3) Local population dynamics All suitable habitat patches may or may not be occupied.
If not all habitat patches are occupied, unoccupied patches will not be correlated with geological surface structure and/or rivers, nor will they be spatially autocorrelated. Genetic structure will be very high. 4) Classical metapopulation Patch occupancy and genetic structure will not be
correlated with geological surface structiire and/or rivers, nor be spatially autocorrelated. 5) Spatially correlated Patch occupancy and genetic structure will not be extinction events correlated with geological surface structure and/or rivers, but will be spatially autocorrelated.
Survey of Patch Occupancy
Analyses of patch occupancy were based on surveys of an 8,800-km^ area where I
was able to determine occupancy status for all cliffs (n = 208). Mountain vizcachas are
highly visible and their presence is known to people who live near the chflfs. In the
course of the study I conducted field surveys of 100 cliffs, and determined whether another 108 cliffs were occupied by mountain vizcachas based on interviews. Field 40
surveys were conducted for all cliffs where information on cliff occupancy was
questionable or unavailable. Not considering the cliffs where I had conflicting or
suspicious information, local residents were always correct about the occupancy status of
cliffs that I subsequently surveyed. Therefore, I have high confidence in the
classification of the 108 cliffs as occupied or iinoccupied by mountain vizcachas based on
interviews.
GIS Analysis of Landscape Structure
Cliffs and rivers were digitized from topographic maps (1 : 100,000) of the study
area. Straight-Une distances between cliffs used in genetic analysis were measured in
ARCVIEW (ESRI, Redlands, CA). Polygons delimiting different categories of surface
geology were digitized from a map (1 :500,000) created by the Department of Geology
and Mining of the Province ofNeuquen. ARCVIEW maps of cliffs, rivers, and geological
categories were rasterized in IDRISI 32 (Clark Labs, Worcester, MA). All cliffs were
classified as occupied or unoccupied, raster images of cliffs and geological categories
were overlaid, and all cliffs were assigned to one of the geological categories listed in
Table 4-2. Some geological categories contained no cliffs, and others contained no cliffs
occupied by mountain vizcachas. I tested for association between occupancy by mountain
vizcachas and geological category with a chi-square test. By overlaying the river vector
map on the cUff image of occupied and unoccupied cliffs, I determined whether or not a
river separated each cliff from the nearest occupied cliff I used a chi-square contingency
table to test the relationship between occupancy of a chfif by mountain vizcachas and the
presence of a river between that cliff and its nearest occupied neighbor. Spatial
autocorrelation of cliff occupancy and geological category was tested with Moran's I 41
(Cliff and Ord, 1973) in IDRISI. Images were contracted by pixel thinning to create
different lags to test for autocorrelation (Eastman, 1999).
Table 4-2. Geological categories where cliffs were located in southern Neuquen Province, Argentina, and the percent of these cliffs occupied by mountain vizcachas.
Geological % Category Lithology # Cliffs Occupied
3* Unstratified and stratified drift (till), ritmite 27 0 4 Conglomerates, gravel, blocks, sand 27 7
6 Andesite, basandesite, basalt, tuff, breccia, fine- 60 72 grained ash conglomerate, sandstone 8 Basalt, andesite, ignimbrite 28 82 10 TufiT, ignimbrite and basalt 17 71
11 Andesite, basandesite, basalt, tuff, conglomerates, 9 33 sand, clay
20 Breccia, volcanic agglomerate, ignimbrite, tuff 35 57 22 Granite, granodiarite and tonaUte, sienite and 6 33 migmatite
Missing numbers represent geological categories within the area that did not contain cHfifs.
Genetic Analysis
Samples for genetic analysis were collected at 28 cliffs, chosen to provide
samples from both sides of several rivers, and to encompass the entire study area.
Mountain vizcachas are difficult to capture in the study area because their rocky refiiges
are usually in crevices on the face of vertical cliffs and they are hesitant to enter traps
away from the rock (Chapter 3). Therefore, 1 obtained samples for genetic analyses from feces. Feces have been used for sampling DNA in wild populations in a variety of studies (Constable et al., 1995; Gerlofif et al., 1995; Kohn et al., 1995; Paxinos et al.. 42
1997; Taberlet et al., 1997). Mountain vizcachas defecate in piles at sunning spots near
the tops of cliffs, and feces dry quickly in the arid environment. Fresh fecal pellets are
easily collected and discrete piles come from a single individual.
Studies using hair and feces as sources of DNA have encountered allelic dropout,
in which samples are falsely determined to be homozygous because only one of the two
alleles present amplified (Gerloff et al., 1995; Taberlet et al., 1996; Gagneaux et al,
1997). This problem seems to be due to the minute quantities of DNA sometimes
obtained from this type of sample or to degradation of the DNA. I tested for allelic
dropout by comparing genotypes obtained from tissue and fecal samples from a captive
mountain vizcacha, and by comparing genotypes from repeated amplifications of DNA
from fecal samples. DNA amplified from tissue and feces of the captured moimtain
vizcacha produced the same genotype at five microsatellite loci. Repeated amplifications
of 20 samples from fecal DNA indicated a rate of allelic dropout of 11%. I accepted this
rate of potential error because I was comparing populations, not genotyping individuals
for comparison, and I assvuned that the rate would be similar across populations.
DNA was isolated from tissue using standard phenol-chloroform extraction
procedures (Sambrook et al., 1989). Fecal samples collected in the field were kept in
individual paper bags, which were stored in boxes containing silica gel to maintain dryness. Two methods were used for DNA extraction from fecal pellets, a CTAB (Cetyl- trimethyl ammonium bromide) method (n = 39) and a magnetic bead (Dynabeads DNA
Direct, Dynal AS, Oslo, Norway) method (n = 395). For the CTAB method, I incubated
3 to 4 pellets overnight at 65°C in a 1 .5-mL eppendorf tube with 800 |aL 2% CTAB buffer (Hillis et al., 1996) and 2 20 mg/ml Proteinase K. Incubation was followed by 43
a standard phenol-chloroform extraction that was precipitated in 100% ETOH overnight at -20°C. The pellet was spun down, washed twice with 70% ETOH, and resuspended in
50 |j,L TE. This solution was cleaned up with a QIAQuick PGR cleanup kit (QIAGen,
Valencia, CA), resuspended in 30 |aL Tris HCl (10 mM), and the entire process was repeated.
For the magnetic bead method, three to four fecal pellets were gently washed in a
1.5-niL eppendorf tube containing approximately 600 |xL phosphate-buffered saline
(PBS). The supernatant was transferred to a clean tube and 200 p,L (1 U) of Dynabeads were added. Extraction was performed according to the manufacturer's protocol.
Extracted DNA was eluted in 40 jiL TE for 5 min at 65°C (Flagstad et al. 1999).
I amplified four microsatellite loci (LVl, LV4, LV5, and LV8) with primer sets I developed for this species (Appendix B). Primers were fluorescently labeled with dyes 6-
FAM and HEX. PGR thermocycling was performed with 0.5 ten^late in 12.5 |iL reactions [10 x Buffer with 1.5 mM MgGl (Finnzymes), 100 to 200 dNTPs, 0.1 to 0.2
primer, 0.2 U DyNazyme^" Thermostable DNA Polymerase (Finnzymes)]. Gycling
conditions were 95°G for 3 min, 94"G for 1 5 s, optimal annealing temperature of the
primer (50° to 58°G:Appendix B) for 30 s, 72°G for 45 s, for 50 cycles (Ernest et al.,
2000). Alleles were resolved by electrophoresis using a 4.5% polyacrylamide gel on an automated DNA sequencer [Applied Biosystems Inc. (ABI, Foster Gity, GA)-Prism 373].
Data were analyzed using GENESGAN 2.0.2 and GENOTYPER 2.0 (ABI) software.
Ninety percent of isolations from feces amplified at least once with either microsatellite primers or "universal" cytochrome b primers. However, the rate of amplifications per locus was low (28% for LVl, 30% for LV4, 64% for LV5, 31% for 44
LV8). This resuhed in greatly reduced sample sizes per cliff, so cliffs were grouped into
seven areas, one each to the east (Alumine, n = 46) and west (Abra Ancha, n = 28) of the
Alumine River, two to the east (Catan Lil, n = 49 and CoUon Cura, n = 19) and one to the
west (Quilquihue, n = 14) of the CoUon Cura River, and one each to the north (Traflil, n =
13) and south of the Limay River (Rio Negro, n = 17) (Figure 4-2). For two of these
areas, CoUon Cura and Traful, data on only two of the four loci was obtained. Alleles
obtained for all samples that amplified are listed in Appendix C.
I tested for Hardy-Weinberg equilibrium per locus and area using randomization
of alleles among individuals within areas with FSTAT version 2.9.1 (Goudet, 2000).
Fisher's exact test using a Markov chain was used to test for linkage disequilibrium
among loci with GENEPOP 3.2 (Raymond and Rousset, 1995a; 1995b). I tested overaU
genetic structure by assessing deviations from random mating with Weir and
Cockerham's (1984) F statistics with FSTAT and R,, with RSTCALC (Goodman, 1997).
The Rst is based on variance m allele size between populations, and is specifically
applicable to microsateUite data (Slatkm, 1995). Probability of each statistic was
determined by 5000 permutations. For Fis and Fit, alleles were permuted within and
among areas, respectively, and for Fs, and R^ genotypes were permuted among areas.
The rejection threshold was adjusted with Bonferroni corrections for all multiple tests
(Rice, 1989).
Landscape Connectivity
My estimation of landscape connectivity encompasses the two key components of landscape connectivity: landscape structure and the species' ability to successfully move through different elements of the landscape. To analyze the component related to 45
landscape structure, I constructed potential models of movement costs for mountain
vizcachas throughout the entire study area and determined the cost-distance between each
set of sampling sites (Knaapen et al., 1992). For each model, I created a friction surface
with IDRISI by assigning friction values to geological surface categories and rivers,
based on the estimated difficulty for mountain vizcachas of moving through these
landscape features relative to movement through rocky cliffs.
For all models, the actual cliffs were assigned a friction value of 1, and lakes,
which I assumed to be absolute barriers, were assigned a value of 10,000. Assuming that
mountain vizcachas are at much greater risk of predation and have more difficulty
traveling through the matrix away from rocky cliffs, I assigned much higher friction
values to all non-cliff" (matrix) areas than to cliff areas. For the first model (1), I assigned
increasingly greater values to areas comprised of geological categories that were most
used by mountain vizcachas, that were sometimes used, that had no cliffs used by
mountain vizcachas, and that had no cliffs at all (Tables 4-2 and 4-3). This model also
included high costs for travel across medium and large rivers. For the second model (2),
I assigned the same high friction value for all areas comprised of geological categories
that had cliffs (whether used or not), a higher value for those with no cliffs, and included
the rivers. In the third model (3), friction values for areas of different geological
categories varied as in the first model, but I did not include any extra cost for crossing a river. In the fourth model (4), I assigned the same, relatively low friction value to matrix areas of all geological categories and gave high friction values only to medium and large rivers. Using the friction surfaces in a pushbroom algorithm, I created a separate cost- distance image from each area where I did genetic sampling (Eastman, 1989). Each point 46
in this image is assigned the lowest possible "cost" of travel to that point from the area of
genetic sampling. The cost incorporates both distance and the frictions (resistance)
encountered in movement from the sampling area to that point. I overlaid vector maps of
all genetics sampling areas with each cost-distance image to determine the cost-distance
between each set of sampling sites.
Table 4-3. Friction values for different models of cost of movement for mountain
vizcachas and probabilities for Mantel regressions between model images and Fst's among areas.
Models
1 2 3 4 5 Cliffs within geological classes
used by mountain vizcachas 1 1 1 1 0 Non-cliff in geological classes most often used 50 500 50 100 0 Geological classes sometimes used by mountain vizcachas 100 500 100 100 0 Geological classes with chffs, but cliffs not used 500 500 500 100 0 Geological classes with no cliffs 1000 1000 1000 100 0 Medium rivers (Lower Alumin^, lower Catan Lil) 500 500 100 500 0 Large rivers (Lower CoUon Cura, lower Chimehuin and Limay) 1000 1000 100 1000 0 Lakes 10,000 10,000 10,000 10,000 0 Probability of Mantel regression with Fst's 0.003* 0.05 0.008* 0.053 0.1
*Significant after Bonferroni's adjustment. The models are as follows:
1) Difficulty of movement varies according to geological class, rivers included as barriers.
2) All geological classes with cliffs equally difficult to move through, rivers included. Difficulty 3) of movement varies according to geological class, rivers not barriers. 4) Rivers only barrier. Distance 5) only. Difficulty of movement does not vary by geological class, rivers not barriers. 47
To represent the component of landscape coimectivity based on the movement
abilities of the species, I used a pairwise matrix of Fst's (Weir and Cockerham, 1984)
between populations (Table 4-4). Fs,-based estimates, based on identity or nonidentity of
allelic states, are less biased than Rst's for microsatellite data when sample sizes are small
and the number of loci scored is low (Gaggiotti et al., 1999). I used Fa /I- Fst and the
natural log of the distance between populations as recommended by Rousset (1997). I
tested the different models of potential connectivity with Mantel regressions (Mantel,
1967) of the Fst matrix and pairwise matrices of straight-line distance and cost-distance
according to the five models outhned above using the ISOLDE subprogram of
GENEPOP.
Table 4-4. Matrix of pairwise Fst's among populations in the seven areas of genetic sampling in Neuquen Province, Argentina.
CatanLil CollonCura AbraAncha Alumine Quilquihue Traful CoUon Cura 0.002 Abra Ancha 0.025 0.017 Alumine 0.042 0.008 0.003 Quilquihue 0.062 0.056 0.076 0.117 Traful 0.006 0.018 0.036 0.029 0.241 Rio Negro 0.044 -0.006 0.060 0.073 0.086 -0.038
Results
Patch Occupancy and Landscape Structure
pattern The of patch occupancy by mountain vizcachas was related to landscape structure. Fifty percent of the cliffs (n = 208) in the 8,800 km^-study area were occupied
(Fig. 4-3). Cliffs were less likely to be occupied by mountain vizcachas if there was a 48
river between them and the nearest occupied cliff (Chi^ = 18.01; df = 1; P < 0.0001).
Most occupied cliffs were to the east of the Alumine and CoUon Cura rivers. The
presence of mountain vizcachas at cliffs was also strongly associated with specific
geological categories (Chi^ = 68.87; df = 8; P < 0.0001). Most cliffs occupied by
mountain vizcachas (98%) were of four categories (Table 4-2). Many unoccupied cliffs
(40%) were also of these categories. Most of the unoccupied cliffs of these categories
were to the west of the Alumine and CoUon Cura Rivers, and were on the same
formations as occupied cliffs on the other side of the rivers (Fig. 4-4). Finally, the spatial
pattern of occupancy of cliffs by mountain vizcachas was highly autocorrelated (Moran's
I = 0.32; z = 18.91; P < 0.0001 at the 3rd lag). Geological categories were also spatially
autocorrelated (Moran's I = 0.64; z = 65.21 ; P < 0.0001 at the 4th lag).
Genetic Structure
Genetic differentiation (Fa and Rst) among the sampling areas was significant
(Table 4-5). The Rst value was greater than the Fst. This is consistent v^th an
underestimation of population subdivision when Fst is used for microsatellite data. By
considering differences in the sizes of alleles, Rst incorporates information on the
mutational history of alleles in the population (Slatkin, 1995). However, Gaggiotti et al.
(1999) found that in simulations, Rst overestimated population subdivision when sample sizes were small and the number of loci examined was low (< 20).
Allelic dropout could result in a general heterozygote deficiency, and, indeed, Fis
(which estimates heterozygote deficit within areas) and F^ (which estimates heterozygote deficit overall) were also significant (Table 4-5). Some of the within-area deficit (Fis) could be due to the Wahlund principle (Hartl and Clark, 1997), if sampling areas were 49
encompassing more than one randomly mating unit (Giles and Goudet, 1997), and to
inbreeding within relatively isolated areas (Chapter 2). When examined individually,
some areas had high Fjs and low probabilities of Hardy Weinberg equilibrium at some
loci (Table 4-6). However, after Bonferonni adjustment to significance level for multiple
tests (P < 0.0018), the hypothesis of Hardy Weinberg equilibrium was rejected for only
one population at one locus. Gene diversity per locus and population was high, ranging
from 0.67 to 0.97. There was no evidence of linkage disequilibrium (P = 0.45 to 0.99).
Table 4-5. The F-statistics (Fi,, Fst, and Fjs), Rst, and number of alleles per locus for mountain vizcachas from seven sites in southern Neuquen, Argentina.
Locus F„ Fs, Fis Rst Alleles
LVl 0.36 (.03) 0.04 (.03) 0.33 (.03) 0.20 13 LV4 0.18 (.04) 0.01 (.02) 0.17 (.04) -0.01 12 LV5 0.11 (.06) 0.02 (.01) 0.10 (.06) 0.10 13 LV8 0.50 (.08) 0.09 (.04) 0.45 (.07) 0.17 10 Overall 0.29* 0.04* 0.26* 0.12*
Standard errors in parentheses. Probability < 0.01
Landscape Connectivity
Genetic structure was related to landscape structure in the form of surface geology
(Table 4-3). Genetic structure was not correlated with straight-line distance between areas (P = 0.1 1). However, cost distance based on different levels of difficulty of
movement according to geological category (models 1 and 3) was significantly correlated with genetic structure. Removal of rivers as barriers in model 3 did not change the probability of the regression much from model 1. Rivers alone (model 4) and the model 50
that treated all geological categories with cliffs equally (model 2) were only weakly correlated with genetic structure, and the correlations were not significant after
Bonferroni adjustment for experimentwise error rate.
Discussion
The regional distribution of the mountain vizcacha in the Patagonian steppe is strongly influenced by the pattern of landscape connectivity (Table 4-7). As some
suitable habitat patches were not occupied, distribution is not determined only by availability of habitat. Distribution also could not be explained solely by local population dynamics. The pattern of patch occupancy was spatially autocorrelated and clearly related to landscape structure (rivers and geology). In the presence of environmental phenomena or deterministic processes that are spatially autocorrelated or correlated to landscape features, this pattern of occupancy could be exhibited even if there was no movement of individuals among populations (Clinchy, 1997). In that case, however, very high genetic structure would be expected, with no evidence of gene flow among areas.
The fact that genetic structure was also correlated with landscape structure indicates that
gene flow is occurring within the constraints unposed by geological features.
The evidence also did not support the classical metapopulation model as an explanation for the observed distribution pattern. If every population has an equal probability of extinction and every patch an equal probability of colonization, as assumed by classical metapopulation theory, the pattern of patch occupancy should be random.
Patch occupancy was not random, but was spatially autocorrelated and correlated to landscape structure. Finally, spatially correlated extinction events (Smith and Gilpin,
1997) appear unlikely to be the main mechanism determining distribution of mountain <' I
51
1
^ O — 3 rs o — CM o o o o o ^ o >0 r<^ O i O O C>
00 fs ON O
Z o o 5: .a ^ "?f u-i ro 0\ O rS O O C> c5 J ex. o o m in (N o o c5 c» o o t-~ t-~ •— <— — 1 O c5 o o c> 3 oo fS Tj- vo p<-i ^ o r- o o c5 o ^ f- -H •— o o oON o ' ^ ^ Vw' »— fn i m^ o § en U < o o d < VO "ou d c> Tf /--V ^—^ <— O OO 00 o o ^ O o o m vo m r ^ oo — tJ- m 00 § i > > > > J J »_) hJ 52 vizcachas. Patch occupancy was spatially autocorrelated, but the strong relationship between patch occupancy and geology suggests that spatial autocorrelation of mountain vizcachas reflects spatial autocorrelation of important landscape features, not of population events. Table 4-7. Summary of hypotheses, evidence that was found in reference to predictions in Table 4-1, and conclusions regarding hypotheses. Hypothesis Evidence found Conclusion 1) Landscape connectivity Patch occupancy and genetic Cannot reject structure related to landscape hypothesis structure. 2) Availability of habitat Habitat in some parts of study Reject hypothesis area not occupied. 3) Local population Not all habitat occupied. Reject hypothesis dynamics Unoccupied patches related to landscape structure and spatially autocorrelated. Evidence of gene flow among areas. 4) Classical Patch occupancy and genetic Reject hypothesis metapopulation structure related to landscape structure and spatially autocorrelated. 5) Spatially correlated Patch occupancy and genetic Reject hypothesis extmction events structure related to landscape structure as well as spatially autocorrelated. The negative correlation between occupancy of a cliff by mountain vizcachas and the presence of a river between that cliff and its nearest occupied neighbor indicates that movements by mountain vizcachas across rivers are limited. Examination of the distribution of unoccupied cUfiFs on preferred geological categories (Fig. 4-4) suggests 53 that the lower Alumine and the CoUon Cura rivers are "hard" barriers for mountain vizcachas (Merriam, 1995). These rivers either prevented colonization of appropriate habitat to the west or prevented successful reco Ionization once populations to the west went extinct. Not all cliffs found in the preferred geological categories are suitable for habitation, but some cliffs to the west of the Alumine surveyed during a study of mountain vizcacha habitat (Chapter 3) provided suitable crevices for mountain vizcachas. In spite of the evidence for rivers as barriers based on patch occupancy, I did not fmd strong genetic evidence of limited movement across rivers. This could be because of low statistical power due to small sample size and small number of loci. The probability of the Mantel regression based only on cost of movement across rivers (model 4) was low (0.053), although not statistically significant. It is also possible that I was only able to sample from both sides of rivers in places where the river was not a barrier, because there were simply no moimtain vizcachas to sample where the river was a strong barrier. Several explanations are possible for the variable connectivity that different geological categories provide for mountain vizcachas. One possible explanation is greater ease of movement across substrate provided by some geological categories. Movements of moimtain vizcachas away from cliffs are strongly associated with the amount of rocky substrate (Walker et al., 2000). Alternatively, survival when traveling through steppe habitat on these types of geological surfaces could be greater if they provide greater availability of rocks for shelter. Presence of mountain vizcachas away from cliffs is associated with presence of rocks taller than a mountain vizcacha (Walker et al., 2000), but whether or not availability of these rocks differs by geological category is not known. Finally, as the number of successful dispersers that reach a patch is a 54 product of the probability of successful movement between patches and the mmiber of individuals attempting dispersal, increased gene flow within some geological categories could be due to population size effects—some areas contain more habitable cliffs and therefore produce a greater number of dispersing individuals. If this is the case, connectivity in this landscape would be more dependent on the quality (Chapter 3), abundance, and spatial arrangement of habitat patches than on the "friction" of the matrbc. With et al. (1997) found that the spatial arrangement and abundance of habitat were important determinants of cormectivity and distribution of species in simulation studies. A few other studies have explored how gene flow can be constrained by elements of the landscape and a species' biology (Arter, 1990; King, 1987; Keyghobadi et al., 1999). These studies are very relevant to the discussion of landscape connectivity, but are not treated in a recent review (Tishendorf and Fahrig, 2000), nor is genetic analysis discussed as a potential aid in measuring connectivity. Perhaps this is because population geneticists have rarely addressed the relevance of their findings to population persistence and distribution in the context of landscape ecology. Mantel regressions between matrices of genetic and geographic distance are commonly used to test for isolation by distance (Rousset, 1997). These regressions can be used, as they were in this study, to test for isolating effects of parameters other than straight-line geographic distance. Arter (1990) tested for correlation between genetic distances and geographic distances based on paths that would have to be taken by alpine snails restricted to following the drainage system. King (1987) tested the barrier effect of rivers for a beetle species by doing a Mantel regression between genetic distance and a matrbc of ones and zeros representing 55 the presence or absence of a river between two sampling sites. Keyghobadi et al. (1999) tested for effects of distance through different habitat types. Neither gene flow nor dispersal alone measures connectivity. Mantel regressions can be used to test the effects of different landscape structures or features on gene flow, incorporating both a measure of movement and how the landscape impedes or facilitates it. Thus, the technique provides a comprehensive test for hypotheses about different patterns of landscape connectivity. This study demonstrates how the connectivity of a landscape from a species' perspective can determine the distribution of the species. Connectivity can be influenced by the spatial arrangement of habitat patches relative to each other and to features in the matrbc that inhibit dispersal, and/or to characteristics of the matrix that limit its permeability. The particular features of the matrix that determine permeability are specific to the species. For example, experiments have shown that understory birds of the south-temperate rainforest will not enter open pasture (Sieving et al, 1996), and that Florida scrub lizards have restricted ability to move through thick ground cover (Hokit et al, 1999). Genetic data indicate that alpine butterflies move more easily through forest than open meadows (Keyghobadi et al, 1999). This study also demonstrates the necessity of incorporating connectivity into metapopulation analyses. Theoretical refinements to metapopulation theory implicate limited dispersal among local populations on discrete habitat patches as the critical and defining process, v^rith primary consequences for the regional distribution of the species (Harrison and Taylor, 1997). The concept of connectivity encompasses dispersal, but also incorporates factors that limit or enhance dispersal~the biology of the species under 56 consideration and important elements of the structure of the landscape as perceived by that species. In this study distribution of mountain vizcachas can be explained by their inferred movements among habitat patches, but only in the context of the explicit geological structure of the landscape. j , 57 BOLIVIA South [ BRAZIL Pacific Ocean /- PARAGUAY I San Miguel de Tucuman of San Santa ^Juan C6rdoba^ e, pg^y .Mendoza ^ Rosario „ ^ , URUGUAY. ^ RioCuarto BUENOS AIRES ^i:-v^ ^ ^ La Plata^ v ^- ^Rfode / la Plata Bahia Blanca Mar del Plata Study £ reai South Atlantic Ocean Falkland Islands (lsl»s Malvms) AdminstcfedbyiiK claiiiiedbyARGEMTINA. Cape Horn 200 <00lgn 200 400 mi Figiire 4-1. Approximate location of study area in Neuquen Province, Argentina. Map from http://www.lib.utexas.edu/Libs/PCL/Map_collection. 58 Figure 4-2. Sampling sites for genetic analysis of mountain vizcachas in Neuqu^n and Rio Negro Provinces, Argentina. 59 Cura 50 km Figure 4-3. Cliffs occupied and not occupied by mountain vizcachas in Neuquen Province, Argentina. 60 Figure 4-4. Unoccupied cliffs (black dots) on four geological categories that contained most cliffs used by mountain vizcachas in Neuquen Province, Argentina. Magenta = category 6. Blue = category 8. Green = category 10. Orange = category 20. See Table 4-2 for explanation of categories. CHAPTER 5 CONCLUSION My research indicates that spatial structure of habitat affects the distribution of mountain vizcachas at several scales. Within habitat patches, mountain vizcachas are more likely to be found where rock crevices are deep and numerous. However, both habitat quality and patch isolation are related to the presence of mountain vizcachas in a patch, and more isolated patches are less likely to be occupied (Chapter 3). Distribution of mountain vizcachas at the regional scale is a result of the pattern of connectivity at the landscape level. Connectivity is determined by the spatial pattern of surface geology and the barrier effect of some rivers (Chapter 4). The dependence of distribution on connectivity implicates dispersal as the key process determining distribution at the regional scale. Any analysis of the distribution or persistence of the species should take into consideration the rate and distance of dispersal of individuals among patches (Chapter 3). However, dispersal appears to be male-biased, and the social structure is based on female km groups (Chapter 2). This social system could accentuate the effect of isolation on the probability of colonization of a patch or recolonization of patches where populations go extinct. Although male dispersers may temporarily mhabit rock patches without females, colonization depends on much rarer female dispersal. Persistence of local populations probably also depends disproportionately on processes affecting female survival and reproduction. Chapters 2 and demonstrate that 3 the relative importance of habitat structure and social structure in 61 62 determining patterns of movement among patches in fragmented habitats merits fiirther evaluation within the context of metapopulation theory. In Chapter 4, 1 introduced a new method for analyzing landscape cormectivity that incorporates its two key components: landscape structure and the species' ability to successfiilly move through different elements of the landscape. To analyze the component related to landscape structure, I constructed potential models of movement costs for mountain vizcachas throughout the study area and determined the cost-distance between each set of sampling sites. I used a measure of gene flow to estimate the rate of movement of moimtain vizcachas among patches. I tested the relationship between these two parameters using Mantel regressions of pairwise matrices of Fst's with matrices of cost-distance. This method provides a comprehensive test for hypotheses about different patterns of landscape connectivity that may be used for other species as well. Mantel regressions between matrices of genetic and geographic distance are commonly used to test for isolation by distance. Parameters other than straight-line geographic distance can be used to test for isolating effects of features in the landscape that may affect movement abilities of the species under study. This research demonstrates the power and utility of an interdisciplinary approach to enq)irically examine questions about the distribution of populations in fragmented habitats at an appropriate scale. My analysis of landscape connectivity was only possible through a combination of molecular genetic techniques that allowed amplification of useful DNA markers from feces and GIS techniques for estimating landscape resistance based on the biology of the species. Through integration of molecular ecology, population genetics, metapopulation theory, and landscape ecology, I was able to provide 63 insight into how the distribution of a little-known species is affected by landscape structure in fragmented habitat. APPENDIX A EVALUATION OF A FECAL-PELLET INDEX OF ABUNDANCE FOR MOUNTAIN VIZCACHAS IN PATAGONIA Introduction Indices of animal abundance based on signs are widely used, but there are relatively few reports in the literature describing evaluation of these methods (Vincent et al., 1991; Southwell, 1994; Mallick et al., 1997; Hubbs et al, 2000), and existing evaluations are often incomplete. To be valid as a measure of relative abundance, an index should be related to true abundance as in the equation E(n)=pN, with E(n) the expected value of the index, N the true population size, and p the constant proportion of the population size which is represented by the index (Lancia et al, 1994). The index must be calibrated to the true population size in order to estimate p (Eberhardt and Simmons, 1987). This may be done by obtaining unbiased estimates of population size and values of the index for several time periods or locations, and doing a regression of population estimates on index values. In a true constant-proportion index, there is a linear relationship between population size and index values, a zero intercept, and the slope of the regression provides an estimate of p (Lancia et al, 1994). Mountain vizcachas (Lagidium spp.) are large caviomorph rodents ofthe Andean region and the Patagonian steppe of South America, with a distribution from central Peru to southwestern Argentina (Redford and Eisenberg, 1992). Although there are studies of other species of the genus (Pearson, 1948), there is little information available on the 64 65 ecology and behavior of Lagidium viscacia (Molina, 1782), the species of the southern Andes and the northern Patagonian steppe (Galende et al., 1998, Puig et al, 1998). Its status throughout its range is unknown, but preliminary work indicates that its densities are declining in northwestern Patagonia (R. S. Walker et al, unpublished data). Therefore, it is important to determine current population sizes and monitor changes. Mountain vizcachas live in deep crevices in large rocks (Pearson, 1948). In my study area rocks used by mountain vizcachas are mostly in the form of linear chffs with vertical faces. Mountain vizcachas may pass many hours a day sunning on rocks near their dens, usually near the tops of the cliffs. They are coprophagic, and defecate in their sunning spots (personal observation). Thus the nature of their habitat in my area, combined with their behavior, results in piles of feces laid out near the tops of the cliffs. Mountain vizcachas tend to use the same sunning spots day after day, and with observation, it is often possible to learn how many inhabit a site (personal observation). On occasion, the number of individuals at a given site may be counted in a single visit. More often, it takes many days or even weeks of observation, depending on weather conditions, history of hunting by humans, and other factors that affect their visibility. Therefore, an indirect method is more suitable as a quick method for estimating population sizes at numerous sites. The objective of this study is to evaluate the use of fecal-pellet counts as an index of abundance for mountain vizcachas in the Patagonian steppe. Because this method is quick and inexpensive, it could make it feasible to monitor populations over large areas in little time. . 66 Methods The study was carried out in the vicinity of Pilolil (39° 35' S, 70° 80' W), Province ofNeuquen, Argentina, in a habitat of grass/shrub steppe (Leon et al., 1998). In siunmer (February-March) of 1996, 1 estimated the number of mountain vizcachas at six cliffs by capture and marking of animals and by direct observations. Each cliff averaged from 10 to 15 m high and had an almost vertical face. Cliffs ranged in length from 260 to 1090 m. Mountain vizcachas were captiired by local assistants who chased them into holes or crevices and grabbed them by hand. I kept captured animals in covered cages and before handling, sedated them with 20 mg/kg of ketamine hydrochloride injected intramuscularly. I attached a colored plastic disc approximately 2.5 cm in diameter in each ear with metal rabbit ear-tags. Each individual could be identified from a distance by the unique combinations of colored discs in their ears. Animals were held overnight and released at the point of capture the next day. At each cliff, observations were carried out repeatedly over several days, at different times of day. At sites where marked animals were resighted I used the unbiased Lincoln-Peterson estimator (Lancia et al, 1994) to estimate the population size. I estimated sighting probability as the ratio of the number of mountain vizcachas sighted at one time to the Lincohi-Peterson estimate of population size, and averaged these over various occasions and sites. To estimate population size at sites where I had no marked animals, but many hours of observations, I divided the maximum number of individuals seen by the average sighting probability. The index I used was the number of fresh piles of fecal pellets encountered in a transect walked along the upper edge ofthe cliff. I counted each group of feces which 67 was still in the form of a pile, assuming that these Ire recently deposited, as the incessant Patagonian winds quickly disperse the fecal pellets when they dry in the sun. Transects were done in the morning hours, and not on rainy days or days with high winds. By counting only intact piles of feces under these conditions, 1 avoided the necessity for estimating rate of decay of feces, as is done in many fecal-based indices for other species (Taylor, 1956; Neflf, 1968). I did a least-squares regression of estimated population size on the number of fresh piles per cliff To determine the 95% confidence intervals for the slope and the intercept, I bootstrapped the regression 1000 times, using the program Resampling Stats (Bruce et al., 1995). Count data are not expected to be normally distributed, and my sample size was small. By bootstrapping, I determined the probability that the intercept and the slope were different from zero without makmg assumptions about fit of the data to any theoretical distribution fimction (Sokal and Rohlf, 1995). To evaluate whether the number of piles of fecal pellets is a constant-proportion index, I repeated the transects in May (fall) 1996. I assumed that the number of individuals inhabiting the cliffs was the same because of the short time (approximately two months) between the transects. I was unable to repeat the transect at cliff #20 because of weather conditions, but I was able to add a cliff (#32) for which I did not have an estimate in the summer. To determine whether the relationship between the number of piles of pellets and the number of individuals differed between seasons, I did t-tests of differences between the slopes and intercepts (Zar, 1996) for the regressions with the fall and summer counts. For my fmal estimate of the proportion represented by p, I repeated the regressions for the two sets of counts, but this time using an intercept of 0. I then 68 calculated a weighted regression coefficient to incorporate both sets of estimates into a single estimate of p (Zar, 1996) with 95% confidence intervals calculated with resampling as described above. Results and Discussion Population sizes at cUffs were small (3 to 19 individuals) (Table A-1). I used the Lincoln-Peterson estimator at sites #4, 15, 19, 20, and 32. At sites 19, 20, and 32, either all marked animals were observed or all observed animals were marked. This results in a variance of zero for the Lincoln-Peterson estimates of population size at these sites, due to zeros in the numerators of the variance equations. The 95% confidence intervals at the two cliffs where I was able to estimate a non-zero variance (cliff #4 and cUff #15) were small and non-overlapping (Table A-1), supporting the rank order of the population sizes. The average sighting probability was 0,61 (standard error = 0.08). The population-size estimates at #2 and #3 were calculated with the sighting probability. At sites #2, #3, and #20 my estimates of population size were corroborated by people living nearby, who had observed the cliffs over longer time periods. My results indicate that transects to covmt fresh piles of fecal pellets are valid as an index of abundance of mountain vizcachas under the conditions of my study. There was a significant linear relationship (Table A-2) between the number of piles of fresh pellets and the estimated number of moimtain vizcachas in both the summer (P = 0.006) and the fall (P < 0.001). There was no difference between the least-squares estimates of the intercepts for the summer and fell regressions (t = 0.31, df= 9, P = 0.76), and both 95% confidence intervals for the intercepts encompassed zero. Therefore there was no evidence that the intercepts are different from zero. Finally, the slopes for the two 69 seasons were not different (t = 1.57, df = 8, P = 0.16), which suggests that this is a constant-proportion index. The weighted slope estimate was 10.6 (95% confidence interval: 8.9 to 12.3). Table A-1. Numbers of mountain vizcachas based on captures and sightings (No. ind.) and number of fresh piles of pellets encountered in summer and fall at cliffs of different lengths (in meters) in southern Neuquen Province, Argentina, 1996. # Piles of Pellets CUfif Length No. ind. 95% CI Summer Fall 2 280 3 NA 38 15 3 270 5 NA 83 42 4 1090 19 14-23 235 161 15 640 8 7-13 68 97 19 590 5 NA 59 65 20 420 3 NA 49 32 280 3 NA 20 95% CI = lower and upper limits of 95% confidence intervals for the estimation of number of individuals. NA = not available. The apphcation of this mdex to estimate absolute numbers should be restricted to the conditions under which it was calibrated (Eberhardt and Simmons, 1987). I tested the index only in simmier and Ml, and studies evaluating counts of deer pellets as indices of abundance have found seasonal differences m the numbers of pellets produced (Dzieciolowski, 1976). I beUeve that seasonal variation in numbers of pellets produced should not greatly affect my index, because I am counting piles of pellets, rather than quantities of individual pellets. However, the index has not been tested for winter and spring, and behavioral differences of mountain vizcachas in the production and depositing of feces may vary among seasons. 70 Table A-2. Intercepts and slopes estimated with least-squares regression of numbers of mountain vizcachas on fresh piles of fecal pellets in summer and fall, 1996, in southern Neuquen Province, Argentina. Intercept Slope Slope without intercept Summer 4.6 (-15.5 to 44.8) 11.7 (3.6 to 13.9) 12.1 (10 to 14.1) Fall 4.3 (-40.2 to 20.7) 8.7 (7.6 to 18.7) 9.08 (8.4 to 11.7) Combined 10.6 (8.9 to 12.3) 95% confidence intervals in parentheses after estimates are based on bootstrapped regressions. The index should be used for mountain vizcachas only m areas where the habitat is in the form of low (<15 to 20 m high) linear cliffs with a vertical face, resultmg in feces being concentrated near the edge of the cUff. Two other cUffs in my area have vertical faces, but sections that are very high (>25 m), providing many sunning spots that are not visible from the top of the chflF. At those sites the index provided an estimate much lower than the number of mountain vizcachas known to inhabit those sites. Also, the method could not be used at some sites in my study area where the top edge of the cUflf was not distinct, and I could not determine where to walk the transect. In other areas, moimtain vizcachas may live in talus slopes and other rock formations with crevices, and the method cannot be used in those types of habitat either. However, the index is potentially valid for a large proportion of mountain vizcacha habitat in Patagonia, as vertical cliffs on the edge of plateaus are widespread throughout the Patagonian steppe. If the index is used outside of the study area, it should be used as an index of relative abundance rather than to estimate absolute numbers (Mallick et al., 1997). Its use outside the study area implicitly mvolves the assumption that local conditions do not 71 affect the relationship between number of fresh fecal piles and number of mountain vizcachas that I found in this study. Behavioral differences among populations of mountain vizcachas could have a strong impact on the functioning of the index, though, so w^henever possible, calibration should be attempted within the area where the index is to be used. Without a validation study done locally, this assumption is untested, and results must be interpreted with caution. APPENDIX B CHARACTERIZATION OF MCROSATELLITE LOCI FROM THE MOUNTAIN VIZCACHA Lagidium viscacia AND THEIR USE FOR THE PLAINS VIZCACHA Lagostomus maximus The Chinchillidae is a family of hystricomorph rodents indigenous to southern South America. There are three genera, chinchillas {Chinchilla spp.) and mountain vizcachas {Lagidium spp.), which live in crevices in rocky habitat, and plains vizcachas {Lagostomus maximus), which construct underground burrow systems in open grasslands and shrublands. Some species of all three genera have been heavily persecuted and are threatened or endangered. I developed microsatellite markers to study movement among local populations of mountain vizcachas, Lagidium viscacia, in Argentina. I also tested these microsatellite markers on the plains vizcacha. Genomic DNA was isolated from heart tissue of two mountain vizcachas using standard phenol/chloroform extraction procedures (Sambrook et aL, 1989). Twenty micrograms of pooled DNA were digested with Sau3Al and fractionated by agarose gel electrophoresis. DNA fragments 300 to 900 bp long were recovered by electrophoresis onto diethylaminoethyl (DEAE) paper and ligated to SAULA/SAULB oligonucleotide linkers (Armour et al., 1994). SAULA primer was used to ampliiy ligation products in a 100-nL PCR reaction [10 x Assay Buffer A (Fisher Scientific), 200 [iM dNTPs, 10 ^iM SAULA primer, 2.5 U Taq DNA polymerase (Fisher Scientific)] for 30 cycles (94"C, 45 s; 57*'C, 45 s; 72''C, 90 s) in a Biometra Uno-Thermoblock thermocycler. Fragments containing (CA)n repeats were captured by hybridization of l)j,g of the amplification 72 73 product to (GT)io oligonucleotide bound to nylon membranes, as described by Karagyozov et al. (1993), using only a single round of hybridization. The DNA enriched with (CA)n repeats was recovered from the filters and ethanol-precipitated overnight at -20°C using 1 (iM SAULA as carrier. The precipitate was amplified in a 100-jiL reaction for 25 cycles as described above, with no additional primer, as the pellet contained SAULA primer. One \i\ of the amplification product was hgated into pCR™2.1 vector (Invitrogen) and transformed into JM109 (Promega). Plasmids were prepared from 81 transformed colonies (WizardPlus Miniprep system, Promega), dotted on a nylon membrane, and screened by hybridization with an alkaline-phosphate linked (CA)n probe and detected using the Quick-Light™ Hybridization Kit (FMC). Fifty percent of the colonies (n = 41) screened by hybridization were positive for CA repeats. Twenty-eight of these were DNA-sequenced, and were found to represent 27 distinctive (CA)n locL PCR primers were designed in the flanking region of ten of the bci using the program OLIGO (National Biosciences Inc., Version 4.0) and tested on DNA obtained from ear tissue of 24 Lagidium viscacia and 45 Lagostomus maximus and feces of three Chinchilla lanigera in 12.5-nL PCR reactions [10 x Assay Buffer A (Fisher Scientific), 200 ^iM dNTPs, 0.4 ^iM primer, 0.5 U Taq DNA polymerase (Fisher Scientific)] using cycling parameters described by Gagneux et al (1997). Of the ten primer sets, eight amplified a band of the appropriate size in Lagidium viscacia samples (Table B-1), and four did so for Lagostomus maximus (Table B-2). In addition, primers for loci LVl, LV4, LV5, LV17, LV25, and LV33 amplified bands of the appropriate size from the three Chinchilla samples. ( 74 o\ o .1 r- 00 oo 00 00 oo 00 00 00 <^->O CO O g o C o o 00 O 1 a o *J O U 00 CO 12 '3 00 00 -H in _2 u u I OS 1-H so o ON SO t— OS tN 1 1 1 1 1 1 I o Wl ON 00 o Vi3 < so o 00 o > > > > > > > > h-l 75 Alleles were resolved for Lagidium and Lagostomus samples by electrophoresis using a 4.5% polyacrylamide gel on an Applied Biosystems Inc. (ABI, Foster City, CA)- Prism 373 automated DNA sequencer. Data were analyzed using GENESCAN 2.0.2 and GENOTYPER 2.0 (ABI) software. All loci amplified in both Lagidium and Lagostomus were polymorphic (Tables B-1, B-2). These microsatellite loci provide useful markers for population studies for two of the three genera of the family Chinchillidae, and their utility in the third genus. Chinchilla, warrants fiirther investigation. Table B-2. Results ofLagidium primers tested on Lagostomus and Chinchilla samples. Lagostomus Size range Number of Observed Locus Chinchilla Lagostomus (bp) alleles heterozygosity LVIA Amplified Amplified 157-175 10 0.84 LV4 Amplified Amplified 184-204 8 0.62 LV5A Amplified Didn't amplify LV8 Didn't amplify Didn't amplify LV16 Didn't amplify Amplified 245-275 11 0.84 LV17 Amplified Didn't amplify LV25 Amplified Amplified 178-200 6 0.64 LV33 Didn't amplify Didn't amplify APPENDIX C TABLE OF ALLELES Table C-1 . Alleles obtained from individual piles of mountain vizcacha fecal pellets collected in southern Neuquen Province, Argentina, for four microsatellite loci. Locus and Allele Fecal Pile Number LVl-1 LVl-2 LV4-1 LV4-2 LV5-1 LV5-2 LV8-1 LV8-2 Catan Lil 1 1 1811 O 1 181 NA NA 101 110Syj NA NA 2 1 O-J NA NA 1 01 1 00yy zto948 Zto748 1811 O 1 1811 O 1 NA NA 1 01 1Vy071 NA NA 4 NA NA NA NA 1 70 1lyjOS NA NA 1 55 1 SO NA NA 1811 o 1 lyj1 OS NA NA 1^1 1^1 1 on lol 1lolfil 7/17 7 1 55 161 NA NA 1lolR1 1lol81 NA NA 8 155 159 NA NA 187 189 230 246 12 163 181 NA NA 191 193 232 234 13 183 183 NA NA 189 193 NA NA 14 NA NA NA NA 181 181 230 230 15 NA NA NA NA 183 193 NA NA 17 163 181 NA NA 191 193 NA NA 19 157 193 NA NA 189 199 NA NA 41 163 179 NA NA NA NA NA NA 42 NA NA 182 190 NA NA 244 246 43 NA NA 190 190 189 191 NA NA 44 161 161 182 190 NA NA 234 238 45 NA NA 186 190 NA NA NA NA 46 NA NA 190 190 NA NA NA NA 47 159 161 180 192 193 195 NA NA 49 161 161 NA NA 191 193 NA NA 50 159 163 NA NA 193 195 NA NA 53 NA NA NA NA 197 199 NA NA 54 159 183 NA NA 183 197 NA NA 55 NA NA 192 192 187 189 NA NA 56 NA NA NA NA 189 191 234 234 57 155 155 NA NA 191 191 232 232 58 NA NA NA NA NA NA 232 232 76 77 Table C-1 —continued. Fecal Pile Number LVl--1 LVl-2 LV4-1 LV4-2 LV5-1 LV5-2 LV8-1 LV8-2 62 159 163 NA NA 191 193 232 232 63 159 167 192 192 191 193 234 244 64 177 177 NA NA 191 197 232 232 65 163 163 NA NA 191 193 232 246 66 157 157 188 190 189 189 232 232 67 163 163 NA NA 181 181 220 232 68 155 157 NA NA 189 189 232 232 69 161 163 NA NA NA NA 244 246 71 161 161 NA NA 193 195 NA NA 73 NA NA 180 190 NA NA NA NA 74 NA NA 188 188 187 189 NA NA 75 NA NA 188 190 NA NA NA NA 76 NA NA 188 190 NA NA NA NA 77 NA NA 186 190 NA NA NA NA 78 NA NA NA NA 185 187 NA NA 79 NA NA NA NA NA NA 232 236 A XT A XT A 80 NA NA XTNAA XTNA 199 201 NA NA 81 NA NA NA NA 187 189 NA NA 83 NA NA 184 184 197 199 232 246 84 159 163 NA NA NA NA NA NA 89 NA NA NA NA 187 187 232 232 90 157 159 NA NA NA NA NA NA 91 157 163 NA NA NA NA 232 232 93 155 155 NA NA NA NA NA NA 431 NA NA NA NA 191 191 NA NA Abra Ancha 20 NA NA 192 192 189 191 NA NA 22 NA NA NA NA 189 191 NA NA 23 161 161 NA NA 191 193 NA NA 24 NA NA 188 192 193 195 NA NA 26 161 161 NA NA 189 191 236 236 27 161 161 190 190 193 195 NA NA 28 163 179 190 190 NA NA 232 234 30 NA NA 188 190 187 191 234 234 31 NA NA 182 182 195 197 234 238 33 173 173 190 190 NA NA NA NA 34 NA NA 192 192 NA NA NA NA 35 NA NA 190 192 NA NA NA NA 36 163 181 192 192 201 203 232 238 38 159 161 NA NA NA NA 236 236 140 NA NA NA NA 191 193 NA NA 78 Table C-1 —continued. Fecal Pile Number LVl -1 LVl-2 LV4-1 LV4-2 LV5-1 LV5-2 LV8-1 LV8-2 141 159 161 190 190 195 199 NA NA 142 155 161 NA NA 191 193 NA NA 143 155 161 190 190 189 191 234 236 144 NA NA NA NA 193 195 NA NA 145 NA NA NA NA 183 183 NA NA 148 NA NA NA NA 201 203 NA NA 149 NA NA NA NA 193 195 NA NA 150 NA NA XT A 1 fiO 1 Q1 XT A XTJNAA 1 CI 1 CO 1 jl ijy lol NA NA 193 195 234 234 152 159 161 NA NA 193 195 NA NA 153 NA NA NA NA 177 187 NA NA' 216 157 157 NA NA NA NA 220 234 222 NA NA NA NA 185 189 NA NA 223 NA NA 188 190 NA NA NA NA 224 NA NA 190 190 NA NA 220 232 225 NA NA NA NA NA NA 232 236 Alumine 96 157 159 NA NA NA NA NA NA 97 NA NA 190 192 191 193 NA NA 98 NA NA NA NA 189 191 NA NA 99 159 161 192 202 NA NA NA NA 100 NA NA NA NA 191 199 NA NA 101 159 161 NA NA 189 193 NA NA 102 161 163 190 190 185 189 NA NA 103 NA NA NA NA 189 191 NA NA 104 161 161 190 190 NA NA NA NA 105 NA NA NA NA 189 191 NA NA 106 NA NA 186 192 189 191 NA NA 107 NA NA 184 192 181 183 NA NA 108 NA NA NA NA 189 191 NA NA 109 NA NA NA NA 189 191 NA NA 115 NA NA 184 190 189 191 NA NA 116 NA NA NA NA 189 189 NA NA 117 NA NA 186 186 NA NA NA NA 118 NA NA 184 210 NA NA NA NA 121 159 159 190 202 193 197 220 220 123 NA NA 190 192 187 189 NA NA 124 NA NA 190 192 183 203 NA NA 125 NA NA 182 186 181 183 NA NA 126 NA NA NA NA 189 189 NA NA 128 NA NA 184 192 201 203 NA NA 79 Table C-1 —continued. Fecal Pile Number LVl -1 LVl-2 LV4-1 LV4-2 LV5-1 LV5-2 LV8-1 LV8-2 129 NA NA 184 190 191 191 NA NA 131 NA NA 182 192 189 191 NA NA 133 NA NA 190 190 NA NA 232 232 134 NA NA NA NA NA NA 234 248 135 NA NA 188 190 NA NA NA NA 136 NA NA 190 190 NA NA NA NA 137 NA NA 184 190 NA NA NA NA 138 NA NA 182 184 NA NA NA NA 139 NA NA 184 190 NA NA 234 234 184 NA NA NA NA 195 199 NA NA 187 161 161 NA NA NA NA 234 234 192 NA NA NA NA NA NA NA NA 194 NA NA 190 190 NA NA NA NA 195 NA NA 190 190 NA NA 232 232 196 NA NA 188 190 NA NA NA NA 202 NA NA NA NA 187 187 NA NA 203 NA NA NA NA 183 183 NA NA 205 159 159 NA NA 189 189 NA NA 206 NA NA NA NA NA NA 234 234 207 NA NA 190 190 NA NA NA NA 208 NA NA 190 190 NA NA NA NA 209 NA NA 190 192 191 191 NA NA 210 NA NA 192 192 183 187 NA NA 211 NA NA 190 190 189 193 NA NA 212 NA NA 180 192 185 187 NA NA 213 NA NA NA NA 179 191 NA NA 235 NA NA NA NA 191 195 NA NA Collon Cura 369 NA NA 184 190 NA NA NA NA 371 NA NA 192 192 NA NA NA NA 373 NA NA 188 190 NA NA NA NA 374 NA NA 188 190 NA NA NA NA 376 NA NA 188 192 NA NA NA NA 377 NA NA 180 184 NA NA NA NA 379 NA NA 184 188 NA NA NA NA 385 NA NA 184 188 NA NA NA NA 386 NA NA 180 192 NA NA NA NA 389 NA NA NA NA 187 187 NA NA 391 NA NA NA NA 187 195 NA NA 392 NA NA 182 188 NA NA NA NA 393 NA NA 188 188 NA NA NA NA 80 Table C-1 —continued. Fecal Pile Number LVl--1 LVl-2 LV4-1 LV4-2 LV5-1 LV5-2 LV8-1 LV8-2 394 NA NA 186 190 NA NA NA NA 395 NA NA 190 192 NA NA NA NA 396 NA NA 186 186 187 189 NA NA 398 NA NA 1 M A MAJNA XTJNAA XTJNAA 400 XTNAA XTNAA NA NA 189 189 NA NA 401 NA NA 184 190 189 189 NA NA 402 NA NA 190 190 NA NA NA NA 403 NA NA NA NA 191 191 NA NA 404 NA NA 188 188 NA NA NA NA 406 NA NA 212 212 NA NA NA NA 407 NA NA NA NA 179 185 NA NA 408 NA NA 192 192 NA NA NA NA Ouilquihue 155 NA NA NA NA 189 191 266 266 156 NA NA NA NA 185 191 NA NA 157 NA NA NA NA 189 191 NA NA 159 NA NA NA NA NA NA 236 242 160 NA NA NA NA 197 199 NA NA 163 NA NA NA NA NA NA 220 220 165 NA NA NA NA 197 199 NA NA 166 NA NA NA NA 191 193 246 246 168 NA NA NA NA 183 187 NA NA 169 NA NA 192 192 NA NA NA NA 170 NA NA 1184QA 1 oo XT A XTJNAA XTJNAA XTInAA 1/1 XTNAA XTNAA 192 192 NA NA NA NA 172 NA NA 192 192 NA NA NA NA 173 NA NA 190 192 185 187 NA NA 174 NA NA NA NA 185 187 NA NA 177 NA NA NA NA NA NA 234 238 178 NA NA NA NA 187 187 NA NA 180 155 157 NA NA NA NA 232 238 181 153 153 NA NA NA NA 230 236 Traful 270 NA NA NA NA 179 179 NA NA 271 NA NA NA NA 191 191 NA NA 273 NA NA NA NA 179 191 NA NA 275 NA NA NA NA 195 195 NA NA 280 NA NA NA NA 185 189 NA NA 284 NA NA NA NA 183 189 NA NA 285 NA NA 190 190 NA NA NA NA 81 Table C-1 —continued. Fecal Pile Number LVl -1 LVl-2 LV4-1 LV4-2 LV5-1 LV5-2 LV8-1 LV8-2 1 on 1 XT A XT A 287 NA NA loo lOJ lo / JNA JNA Z6S XTNAA XTNAA 188 190 NA NA NA NA 293 NA NA NA NA 191 191 NA NA 435 NA NA NA NA 181 181 NA NA 445 NA NA NA NA 187 189 NA NA 447 NA NA NA NA 187 191 NA NA 454 NA NA NA NA 189 191 NA NA 459 NA NA NA NA 187 191 NA NA 463 NA NA NA NA 185 187 NA NA Rio Negro 256 159 161 NA NA NA NA 236 236 258 159 159 NA NA NA NA 236 236 259 NA NA NA NA 183 183 232 232 261 NA NA 190 190 187 187 NA NA 265 NA NA NA NA 189 189 NA NA 295 NA NA 192 192 193 197 NA NA 299 NA NA 190 192 NA NA NA NA 300 NA NA NA NA 191 195 NA NA 307 165 165 190 192 187 197 NA NA 321 NA NA 180 190 NA NA NA NA 322 NA NA 182 186 NA NA NA NA 323 NA NA 190 190 189 189 NA NA 324 NA NA 184 190 187 191 NA NA 325 NA NA 202 206 187 191 NA NA 326 NA NA 190 190 185 189 NA NA 327 NA NA 188 192 187 189 NA NA 328 NA NA 182 182 189 191 NA NA 330 NA NA 184 184 NA NA NA NA 334 NA NA 188 190 NA NA NA NA NA = not available. ' 82 I > *^ • t « •a I > U .3 .a c > o .3 c ca > 1 00 6 s 0^0^0^0^0\0^a^O^C^a^O^O^O^O^C^O^O^O^O^O^O^O^a^ O^O^O^O^O^C^O^O^C^O^00O^00(3^O^O^<^O^C^^^O^a^O^ o dsa\0\oosooNOsONOOOONaNC3\ONONOO\ N > .op Q Q Q Q Q Q ^ Q o S 2 <^ 2 ^ 999999S O a. Oh K 0. W K o; pei Qt^ pej Ou, O r<> w-> vo ao On o O LIST OF REFERENCES Andrewartha, H.G. and L. C. Birch. 1954. The distribution and abundance of animals. University of Chicago Press, Chicago. Armour, J. A. L., R. Neumann, S. Gobert and A. J. Jeffreys. 1994. Isolation of human simple repeat loci by hybridization selection. Human Molecular Genetics 3: 599- 605. Arnold, G. W., J. R. Weeldenburg, and V. M. Ng. 1995. Factors affecting the distribution and abundance of Western grey kangaroos {Macropus fuliginosus) and euros (M robustus) in a fragmented landscape. Landscape Ecology 10: 65- 74. Arter, H. E. 1 990. Spatial relationship and gene flow paths between populations of the alpine snail Arianta arbustorum (Pulmonata: Helicidae). Evolution 44: 966-980. Ashley, M. V. and B. D. Dow. 1994. The use of microsatellite analysis in population biology: background, methods and potential applications. Pp. 185-201 in B. Schierwater, B. Strict, G. P. Wagner, and R. DeSalle, (eds.). Molecular ecology and evolution: Approaches and applications. Birkhauser Verlag, Basel. Barton, N. H. 1992. The genetic consequences of dispersal. Pp. 37-59 in N. C. Stenseth and W. Z. Lidicker, (eds.). Animal dispersal: Small mammals as a model. Chapman and Hall, London. Branch, L. C. 1993. Social organization and mating system of the plains viscacha (Lagostomus maximus). Journal of Zoology, London 229: 473-491. Bruce, P., J.Simon, and T. Oswald. 1995. Resampling Stats user's guide. Resampling Stats, Inc. Arlington, Virginia. Carr, G. M. and D. W. Macdonald. 1986. The sociality of solitary foragers: a model based on resource dispersion. Animal Behaviour 34: 1540-1549. Chakraborty, R., and L. Jin. 1993. A unified approach to study hypervariable polymorphisms: statistical consideration of determining relatedness and population distances. Pp. 153-175 m S.D. J. Pena, R. Chakraborty, J. T. Epplen, and A. J. Jeffreys, (eds.), DNA fmgerprinting: State of the science. Birkhauser Verlag, Basel, Switzerland. 83 : 84 Chesser, R. K. 1983. Genetic variability within and among populations of the black- tailed prairie dog. Evolution 37: 320-331. Cliff, A. D. and J. K. Ord. 1973. Spatial autocorrelation. Pion Limited, London. Clinchy, M. 1997. Does immigration "rescue" populations from extinction? Implications regarding movement corridors and the conservation of mammals. Oikos 80:618-622. Constable, J. J., C. Packer, D. A. CoUings, and A. E. Pusey. 1995. Nuclear DNA from primate dung. Nature 373: 393. Courchamp, F., T. Clutton-Brock, and B. Grenfell. 1999. Inverse density dependence and the Allee effect. Trends in Ecology and Evolution 14: 405-410. DenBoer, P. J. 1968. Spreading of risk and stabilization of animal numbers. Acta Biotheorologica 18: 165-194. Downhower, J. F., and K. B. Armitage. 1971. The yellow-bellied marmot and the evolution of polygamy. American Naturalist 105: 355-370. Dzieciolowski, R. 1976. Roe deer census by pellet-group counts. Acta Theriologica 21 351-358. Eastman, J.R., 1989. Pushbroom algorithms for calculating distances in raster grids. Autocarto 9: 288-297. Eastman, J. R. 1999. Guide to GIS and image processing. Clark Labs, Clark University, Worcester, MA. Eberhardt, L. L. and M. A. Simmons. 1987. Calibrating population indices by double sampling. Journal of Wildlife Management 51: 665-675. Emlen, S. T., and L. W. Oring. 1977. Ecology, sexual selection, and the evolution of mating systems. Science 197: 215-233. Englund, E. and A. Sparks. 1988. GEO-EAS User's guide. EPA600/4-88/033. Environmental Monitoring Systems Laboratory, Office of Research and Development, U.S. Environmental Protection Agency, Las Vegas, Nevada. Ernest, H. B., M. C. T. Penedo, B. P. May, M. Syvanen, and W. M. Boyce. 2000. Molecular tracking of mountain lions in the Yosemite Valley region in California: genetic analysis using microsatellites and faecal DNA. Molecular Ecology 9- 433-441. 85 Estoup, A., L. Gamery, M. Solignac, and J-M Comuet. 1995. Microsatellite variation in honey bee (Apis mellifera L.) populations: Hierarchical genetic structure and test of the infinite allele and stepwise mutation models. Genetics 140: 679-695. Fahrig, L. 1988. A general model of populations in patchy habitats. Applied Mathematics and Computation 27: 53-66. Fahrig, L. 1990. Interacting effects of disturbance and dispersal on individual selection and population stability. Communications in Theoretical Biology 1 : 275-297. Fahrig, L. and J. Paloheimo. 1988. Determinants of local population size in patchy habitats. Theoretical Population Biology 34: 194-213. Flagstad, 0., K. Roed, J. E. Stacy, and K. S. Jakobsen. 1999. Reliable noninvasive genotyping based on excremental PCR of nuclear DNA purified with a magnetic bead protocol. Molecular Ecology 8: 879-883. Funes, M. C, and A. J. Novaro. 1999. El rol de la fauna silvestre en la economia del poblador rural en la provincia del Neuquen, Argentina. Revista Argentina de Produccion Animal 19: 265-271. Gaggiotti, O. E., O. Lange, K. Rassman, and C. Gliddon. 1999. A comparison of two indirect methods for estimating average levels of gene flow using microsatellite data. MolecularEcology 8: 1513-1520. Gagneux, P., C. Boesch, and D. S. Woodruff 1997. Microsatellite scoring errors associated with noninvasive genotypmg based on nuclear DNA amplified from shed hair. Molecular Ecology 6: 861-868. Galende, G. I., and D. Grigera. 1998. Relaciones alimentarias de Zag/c/mw vwcac/a (Rodentia, Chinchillidae) con herbfvoros introducidos en el Parque Nacional Nahuel Huapi, Argentina. Iheringia, Serie Zoologica, Porto Alegre 84: 3-10. Galende, G. I., D. Grigera, J. von Thungen. 1998. Composicion de la dieta del chinchillon {Lagidium viscacia, Chinchillidae) en el noroeste de la Patagonia. Mastozoologia Neotropical 5: 123 -128. Gerloff, U., C. Schlotterer, K. Rassmann, I. Rambold, G. Hohmann, B. Fruth, and D. Tautz. 1995. Amplification of hypervariable simple sequence repeats (microsatellites) from excremental DNA of wild living bonobos {Pan paniscus). Molecular Ecology 4: 515-518. Giles, B. E., and J. Goudet. 1997. A case study of genetic structure in a plant metapopulation. Pp. 429-454 in I. Hanski and M. E. Gilpin, (eds.), Metapopulation biology: Ecology, genetics, and evolution. Academic Press, San Diego. 86 Glade, A. A. 1988. Red list of Chilean terrestrial vertebrates. Proceedings of the Symposium on the conservation status of Chilean terrestrial vertebrate fauna. Chilean Forest Service (CONAF), Santiago, Chile. Goodman, S. 1997. Rst Calc: a collection of computer programs for calculating estimates of genetic differentiation from microsatellite data and determining their significance. Molecular Ecology 6: 881-885. Goudet, J. 2000. FSTAT, a program to estimate and test gene diversities and fixation indices (version 2.9.1). Available from http://www.unil.ch/izea/softwares/fstat.html. Gumpertz, M. L., C. Wu, and J. M. Pye. 2000. Logistic regression for southern pine beetle outbreaks with spatial and temporal autocorrelation. Forest Science 46: 95- 107. Guo, S. W. and E. A. Thompson. 1992. Performing the exact test of Hardy-Weinberg proportion for muhiple alleles. Biometrics 48: 361-372. Hannson, L. 1995. Development and application of landscape approaches in mammalian ecology. Pp. 20-39 m W.Z. Lidicker, (ed.), Landscape approaches in mammalian ecology and conservation. University of Minnesota Press, MinneapoUs. Hanski, L 1991. Single-species metapopulation dynamics: concepts, models and observations. Biological Journal of the Linnean Society 42: 17-38. Hanski, L 1994a. A practical model of metapopulation dynamics. Journal of Animal Ecology 63: 151-162. Hanski, L 1994b. Patch-occupancy dynamics in fragmented landscapes. Trends in Ecology and Evolution 9: 131-135. Hanski, L and M. E. Gilpin. 1991. Metapopulation dynamics: brief history and conceptual domain. Biological Journal of the Linnean Society 42: 3-16. Hanski, L and M. E. Gilpin. 1997. Metapopulation biology: Ecology, genetics, and evolution. Academic Press, San Diego. Hanski, L and D. SimberlofiF. 1997. The metapopulation approach, its history, conceptual domain, and application to conservation. Pp. 5-26 in L Hanski and M. E. Gilpin, (eds.), Metapopulation biology: Ecology, genetics, and evolution. Academic Press, San Diego. 87 Harrison, S. and A. D. Taylor. 1997. Empirical evidence for metapopulation dynamics. Pages 26-42 in I. Hanski and M. E. Gilpin, (eds.), Metapopulation biology: Ecology, genetics, and evolution. Academic Press, San Diego. Hartl, D. L., and A. G. Clark. 1997. Principles of population genetics. Third Edition. Sinauer Associates, Inc., Sunderland, Massachusetts. Hoeck, H. N. 1982. Population dynamics, dispersal and genetic isolation in two species of hyrax {Heterohyrax brucei and Procaviajohmtoni) on habitat islands in the Serengeti. Zeitschrift fur Tierpsycho logic 59: 177-210. Hoeck, H. N., H. Klein, and P. Hoeck. 1987. Flexible social organization in hyrax. Zeitschrift fiir Tierpsychologie 59: 265-298. Hokit, D. G., B. M. Stith, and L. C. Branch. 1999. Effects of landscape structure in Florida scrub: a population perspective. Ecological Applications 9: 124-134. Hubbs, A. H.; T. Karels, and R. Boonstra. 2000. Indices of population size for burrowing mammals. Journal of Wildlife Management 64: 296-301. HuflFaker, C. B. 1958. Experimental studies of predations: Dispersal factors and predator-prey oscillation. Hilgardia 27: 343-383. Jimenez, J. E. 1995. Conservation of the last wild chinchilla {Chinchilla lanigera) archipelago: a metapopulation approach. Vida Silvestre Neotropical 4: 89-97. Karagyozov L, Kalcheva ID, Chapman VM (1993) Construction of random small-insert genomic Ubraries highly enriched for simple sequence repeats. Nucleic Acids Research 21: 3911-3912. Keyghobadi, N., J. Roland, and C. Strobeck. 1999. Influence of landscape on the population genetic structure of the alpine butterfly Parnassus smintheus (Papilionidae). Molecular Ecology 8: 1481-1495. > King, P.S. 1987. Macro- and microgeographic structure of a spatially subdivided beetle species in nature. Evolution 4 1 : 40 1 -4 1 6. Knaapen, J. P., M. Schefifer, and B. Harms. 1992. Estimating habitat isolation m landscape planning. Landscape and Urban Planning 23: 1-16. Koenig, W. D., D. Van Vuren, and P. N. Hooge. 1996. Detectability, philopatry, and the distribution of dispersal distances in vertebrates. Trends in Ecology and Evolution 11: 514-517. : 88 Kohn, M., F. Knauer, A. Stoffella, W. Schroeder, and S. Paabo. 1995. Conservation genetics of the European brown bear - a study using excremental PGR of nuclear and mitochondrial sequences. Molecular Ecology 4: 95-103. Lancia, R. A., J. D. Nichols, and K. H. Pollock. 1994. Estimating the number of animals in wildlife populations. Pp. 215-253 m T. A. Bookhout, (ed.). Research and management techniques for wildlife and habitats, 5th edition. The WildUfe Society, Bethesda, MD. Leflcovitch, L. P. and L. Fahrig. 1985. Spatial characteristics of habitat patches and population survival. Ecological Modeling 30: 297-308. Leon, R. J. C, D. Bran, M. Collantes, J. M. Paruelo, and A. Soriano. 1998. Grandes unidades de vegetacion de la Patagonia extra andina. Ecologia Austral 8: 125- 144. Levins, R. 1969. Some demographic and genetic consequences of environmental heterogeneity for biological control. Bulletin of the Entomological Society of America 15: 237-240. Littell, R. G. Milliken, W. Stroup, and R. Wolfmger. 1996. SAS system for mixed models. SAS Institute, Gary, NG. MacArthur, R. M., and E. O.Wilson. 1967. The theory of island biogeography. Princeton University Press, Princeton. Mallick, S. A., M. M. Driessen, and G. J. Hocking. 1997. Diggmgs as a population index for the eastern barred bandicoot. Journal of Wildlife Management 61 1378-1383. Mantel, N. 1967. The detection of disease clustering and a generalized regression approach. Gancer Research 27:209-220. Mares, M. A., and T. E. Lacher. 1987. Ecological, morphological, and behavioral convergence in rock-dwelling mammals. Pp. 307-348 m H. Genoways, (ed.), Gurrent mammalogy, vol. 1. Plenum Press, New York. Marshall, T. G., J. Slate, L. Kruuk, and J. M. Pemberton. 1998. Statistical confidence for likelihood-based paternity inference in natural populations. Molecular Ecology 7: 639-655. MazeroUe, M. J., and M.-A. Villard. 1999. Patch characteristics and landscape context as predictors of species presence and abundance: a review. Ecoscience 6:117- 124. 89 McCullough, D. R. 1996. Metapopulations and wildlife conservation. Island Press, Washington, D.C. Merriam, G. 1984. Connectivity: a fundamental ecological characteristic of landscape 1'' pattern. Pp. 5-15 in J. Brandt and P. Agger, (eds.), Proceedings of the international seminar on methodology in landscape ecological research and planning. Roskilde Univiversity, Denmark. Merriam, G. 1995. Movement in spatially divided populations: responses to landscape structure. Pp. 64-77 m W.Z. Lidicker, (ed.). Landscape approaches in mammalian ecology and conservation. University of Minnesota Press, Miimeapolis. Micol, T., C. P. Doncaster, and L. A. Mackinlay. 1994. Correlates of local variation in the abundance of hedgehogs Erinaceus europaeus. Journal of Animal Ecology 63: 851-860. Moritz, C, A. Heideman, E. GeflFen, and P. McRae. 1997. Genetic population structure of the Greater Bilby Macrotis lagotis, a marsupial in decline. Molecular Ecology 6: 925-936. Mottana, A., R . Crespi, and G. Liborio. 1978. Guide to rocks and minerals. Simon and Schuster Inc., New York. Neflf, D. J. 1968. The pellet-group coimt technique for big game trend, census, and distribution: a review. Journal of Wildlife Management 32: 597-614. Paxinos, E., C. Mcintosh, K. Ralls, and R. Fleischer. 1997. A noninvasive method for distinguishing among canid species: amplification and enzyme restriction of DNA from dung. Molecular Ecology 6: 483-486. Peacock, M. 1997. Determining natal dispersal patterns in a population ofNorth American pikas (Ochotona princeps) using direct mark-resight and indirect genetic methods. Behavioral Ecology 8: 373-412. Peacock, M. M., and A. T. Smith. 1997. The effect of habitat fragmentation on dispersal patterns, mating behavior, and genetic variation in a pika {Ochotona princeps) metapopulation. Oecologia 112: 524-533. Pearson, O. 1948. Life history of mountain viscachas in Peru. Journal of Mammalogy 29: 345-374. Pearson, O. P. 1995. Annotated keys for identifying small mammals living in or near Nahuel Huapi National Park or Lanin National Park, southern Argentina. Mastozoologia Neotropical 2: 99-148. 90 Pither, J. and P. D. Taylor. 1998. An experimental assessment of landscape connectivity. Oikos 83: 166-174. Pope, L. C, A. Sharp, and C. Moritz. 1996. Population structure of the yellow-footed rock-wallaby Petrogale xanthopus (Gray, 1 854) inferred from mtDNA sequences and microsatellite loci. Molecular Ecology 5: 629-640. Pounds, J. A. and J. F. Jackson. 1 981 . Riverine barriers to gene flow and the differentiation of fence lizard populations. Evolution 35: 516-528. Puig, S., F. Videla, M. Cona, S. Monge, and V. Roig. 1998. Diet of the moimtain vizcacha (Lagidium viscacia Molina 1 782) and food availability in northern Patagonia, Argentina. Zeitschrift fur SSugetierkunde 63: 228-238. Raymond, M. and F. Rousset. 1995a. GENEPOP (version 1.2): population genetics software for exact tests and ecumenicism. Journal of Heredity 86: 248-249. Raymond, M. and F. Rousset. 1995b. An exact test for population differentiation. Evolution 49: 1280-1283. Rayor, L. S., and K. B. Armitage. 1991 . Social behavior and space-use of young of groimd-dwelling squirrel species with different levels of sociality. Ethology, Ecology, and Evolution 3: 185-205. Redford, K., and J. F. Eisenberg. 1992. Mammals of the Neotropics: the Southern Cone. Vol. 2. University of Chicago Press, Chicago, IL. Rice, W. R. 1989. Analysing tables of statistical tests. Evolution 43: 223-225. Rousset F, 1997. Genetic differentiation and estimation of gene flow from F-statistics under isolation by distance. Genetics 145: 1219-1228. Sambrook J, E. F. Fritsch, and T. Maniatis. 1989. Molecular cloning: A laboratory manual. 2nd ed. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. SAS Institute, Inc. 1996. Language guide for personal computers. Version 6.10 Editioa SAS Institute Inc., Gary, N.C. Sieving, K. E., M. F. Willson, and T. L. De Santo. 1996. Habitat barriers to movement of understory birds in fragmented south-tenqjerate rainforest. Auk 113: 944-949. Sinclair, A. R. E., C. J. Krebs, and J. N. M. Smith. 1982. Diet quaUty and food limitation in herbivores: the case of the snowshoe hare. Canadian Journal of Zoology 60: 889-897. 91 Slatkin, M. 1995. A measure of population subdivision based on microsatellite allele frequencies. Genetics 139: 457-462. Smith, A. T. and M. E. Gilpin. 1997. Spatially correlated dynamics in a pika metapopulation. Pp. 407-428 m I. Hanski and M. E. Gilpin, (eds.), Metapopulation biology: Ecology, genetics, and evolution. Academic Press, San Diego. Smith, A. T., and B. Ivins. 1984. Spatial relationships and social organization in adult pikas: A facultatively monogamous mammal. Zeitschrift fur Tierpsychologie 66: 289-308. Smith, A. T. and M. M. Peacock. 1990. Conspecific attraction and the determination of metapopulation colonization rates. Conservation Biology 4: 320-323. Sokal, R. R., and F. J. Rohlf 1995. Biometry. Third Ed. Freeman and Company, New York. Southwell, C. 1994. Evaluation of walked transect counts for estimating macropod density. Journal of Wildlife Management 58: 348-356. Stacey, P. and Taper. 1992. Environmental variation and the persistence of small populations. Ecological Applications 2: 18-29. Stacey, P. B., M. Taper, and V. A. Johnson. 1997. Migration within metapopulations: the impact upon local population dynamics. Pp. 267-291 in I. Hanski and M. E. Gilpin, (eds.), Metapopulation biology: Ecology, genetics, and evolution. Academic Press, San Diego. Stamps, J. A., M. Buechner, and V. V. Krishnan. 1987. The effects of edge permeability and habitat geometry on emigration from patches of habitat. American Naturalist 129: 533-52. Stephens, P. A., and W. J. Sutherland. 1999. Consequences of the AUee effect for behaviour, ecology and conservation. Trends in Ecology and Evolution 14: 401- 405. Stith, B. M., J. W. Fitzpatrick, G. E. Woolfenden, and B. Pranty. 1996. Classification and conservation of metapopulations: a case study of the Florida scrub jay. Pp. 188- 215 in D. R. McCuUough, (ed.), Metapopulations and wildlife conservation. Island Press, Washington, D.C. Surridge, A. K., K. M. Ibrahim, D. J. Bell, N. J. Webb, C. Rico, and G. M. Hewitt. 1999. Fine-scale genetic structuring m a natural population of European wild rabbits {Oryctolagus cuniculus). Molecular Ecology 8: 299-307. 92 Taberlet, P., J. J. Camarra, S. Griffin, E. Uhres, O, Hanotte, L. P. Waits, C. Dubois- Paganon, T. Biirke, and J. Bouvet. 1997. Noninvasive genetic tracking of the endangered Pyrenean brown bear population. Molecular Ecology 6: 869-876. Taberlet, P., S. Griffin, B. Goossens, S. Questiau, V. Manceau, N. Escaravage, L. P. 1996. Reliable genotyping of samples with very low DNA quantities using PGR. Nucleic Acids Research 16:3189-3194. Taylor, P. D., L. Fahrig, K. Henein, and G. Merriam. 1993. Connectivity is a vital element of landscape structure. Oikos 68: 571-573. Taylor, R. H. and R. M. Williams. 1956. The use of pellet counts for estimating the density of populations of the wild rabbit, Oryctolagus cuniculus. New Zealand Journal of Science and Technology, Sec. B, Vol. 38, No. 3: 13-25. Templeton, A. R., K. Shaw, E. Routman, and S. K. Davis. 1990. The genetic consequences of habitat fragmentation. Annals of the Missouri Botanical Garden 77: 13-27. Tischendorf, L. and L. Fahrig. 2000. On the usage and measurement of landscape connectivity. Oikos 90: 7-19. Van Apeldoom, R. C., C. Celada, and W. Nieuwenhuizen. 1994. Distribution and dynamics of the red squirrel {Sciurus vulgaris L.) in a landscape with fr^mented habitat. Landscape Ecology 9: 227-235. Vemer, J., M. L. Morrison, and C. J. Ralph. 1986. Wildlife 2000: Modeling habitat relationships of terrestrial vertebrates. The University of Wisconsin Press, Madison, WI. Vincent, J. P., J. M. Gaillard, and E. Bideau. 1991. Kilometric index as biological indicator for monitoring forest roe deer populations. Acta Theriologica 36: 315- 328. Walker, E. 1975. Mammals of the world. Johns Hopkins University Press, Baltimore, Maryland. Walker, R. S., G. Ackermann, J. Schachter-Broide, V. Pancotto, and A. J. Novaro. 2000. Habitat use by mountain vizcachas {Lagidium viscacia Molina, 1782) in the Patagonian steppe. Zeitschrift fiir Saugetierkunde 65: 293-300, Weir, B. 1974. Reproductive characteristics of hystricomorph rodents. Symposia of the Zoological Society of London No. 34: 265-301. Weir, B. S., and C. C. Cockerham. 1984. Estimating F-statistics for the analysis of population structure. Evolution 38: 1358-1370. 93 Wiens, J. A. 1995. Landscape mosaics and ecological theory. Pp. 1-26 in L. Hansson, L. Fahrig, and G. Merriam, (eds.). Mosaic landscapes and ecological processes. Chapman and Hall, London. Wiens, J. A. 1997. Metapopulation dynamics and landscape ecology. Pp. 43-68 in I. Hanski and M. E. Gilpin, (eds.), Metapopulation biology: Ecology, genetics, and evolution. Academic Press, San Diego. Wilcox, B. A. 1980. Insular ecology and conservation. Pp. 95-1 17 in M. E. Soule and B. A. Wilcox, (eds.), Conservation biology: An evolutionary-ecological perspective. Sinauer Assoc., Inc., Sunderland, Massachussetts. With, K. A., R. H. Gardner, and M. G. Turner. 1997. Landscape connectivity and population distributions in heterogeneous environments. Oikos 78: 1 5 1 - 1 69. Woodruff D. S. 1993. Non-invasive genotyping of primates. Primates 34: 333-346. Woods, C. A. 1993. Suborder hystricognathi. Pp. 777-778 in D. E. Wilson and D. M. Reeder, (eds.). Mammal species of the world: A taxonomic and geographic reference. 2nd Ed. Smithsonian Institution Press, Washington. Wrangham, R. W. 1980. An ecological model of female-bonded primate groups. Behavior 75: 262-300. Zar, J. H. 1996. Biostatistical analysis. Prentice Hall, Upper Saddle River, NJ. BIOGRAPHICAL SKETCH Rebecca Susan Walker was bom in Huntsville, Alabama. Before studying wildlife she received a B.A. in anthropology from the University of Tennessee and an M.S. in nursing from Pace University in New York. She worked as a waitress in Alaska, as a maternity nurse in a public hospital in the Bronx, as an English teacher in a Berlitz School in Japan, and as a neonatal intensive care nurse in Tennessee. After assisting on some wildlife research projects she decided to study wildlife ecology, and earned an M.S. in Wildlife and Range Sciences at the University of Florida. She has done ecological fieldwork in Belize, Guatemala, Thailand, Florida, and Argentina, in habitats ranging from tropical rainforests to Patagonian steppe. She has lived in Junin de los Andes, Argentina, for the last eight years, and plans to remain there after graduation and to continue to work on wildlife conservation in the Patagonian steppe and forests. 94 I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is folly adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy Lyn C. Branch, Chair Associate Professor of Wildlife Ecology and Conservation I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is folly adeqiiate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. George Tanner Professor of Wildlife Ecology and Conservation I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is folly adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosopljy. J /"f^ Melvin E. Sunquist Associate Professor of Wildlife Ecology and Conservation I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is folly adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. Bjfan W. B^ven Assistant Professor of Fisheries and Aquatic Sciences I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is folly adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. Colin A. Chapman Associate Professor of Zoology This dissertation was submitted to the Graduate Faculty of the College of Agricultural and Life Sciences and to the Graduate School and was accepted as partial fulfillment of the requirements for the degree of Doctor of Philosophy. ' May 2001 X^^^ ( , ^ Dean, College of Agricultural a^oSi^ife Sciences Dean, Graduate School