Citation
Metapopulations of marsh rabbits

Material Information

Title:
Metapopulations of marsh rabbits a population viability analysis of the lower keys marsh rabbit (Sylvilagus palustris Hefneri)
Creator:
Forys, Elizabeth A., 1966-
Publication Date:
Language:
English
Physical Description:
viii, 244 leaves : ill. ; 29 cm.

Subjects

Subjects / Keywords:
Cats ( jstor )
Copyrights ( jstor )
Female animals ( jstor )
Juveniles ( jstor )
Keys ( jstor )
Marshes ( jstor )
Metapopulation ecology ( jstor )
Mortality ( jstor )
Rabbits ( jstor )
Species ( jstor )
Cottontails -- Florida ( lcsh )
Dissertations, Academic -- Wildlife Ecology and Conservation -- UF
Habitat conservation -- Florida ( lcsh )
Sylvilagus -- Florida ( lcsh )
Wildlife Ecology and Conservation thesis Ph. D
City of Key West ( local )
Genre:
bibliography ( marcgt )
non-fiction ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 1995.
Bibliography:
Includes bibliographical references (leaves 222-243)
General Note:
Typescript.
General Note:
Vita.
Statement of Responsibility:
by Elizabeth A. Forys.

Record Information

Source Institution:
University of Florida
Rights Management:
The University of Florida George A. Smathers Libraries respect the intellectual property rights of others and do not claim any copyright interest in this item. This item may be protected by copyright but is made available here under a claim of fair use (17 U.S.C. §107) for non-profit research and educational purposes. Users of this work have responsibility for determining copyright status prior to reusing, publishing or reproducing this item for purposes other than what is allowed by fair use or other copyright exemptions. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder. The Smathers Libraries would like to learn more about this item and invite individuals or organizations to contact the RDS coordinator (ufdissertations@uflib.ufl.edu) with any additional information they can provide.
Resource Identifier:
021857874 ( ALEPH )
33472892 ( OCLC )

Downloads

This item has the following downloads:


Full Text










METAPOPULATIONS OF MARSH RABBITS: A POPULATION VIABILITY ANALYSIS OF THE LOWER KEYS MARSH RABBIT
(SYLVILAGUS PALUSTRIS HEFNERI)















By

ELIZABETH A. FORYS











A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE
UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 1995













ACKNOWLEDGMENTS

I wish to thank Steve Humphrey for his advise and untiring support

during all of the phases of my Ph.D. work. Discussions with Buzz Holling and John Eisenberg gave me new perspectives and insights about the science of conservation biology. George Tanner and Lyn Branch provided constructive comments on both my proposal and this manuscript.

This research would not have been possible without the funding from the U. S. Navy (USN). Don Wood of the Florida Game and Fresh Water Fish Commission (FGFWFC) effectively and efficiently administered the grant. Additional financial support was provided by the U.S. Fish and Wildlife Service (USFWS) and the Nature Conservancy (TNC). Many individuals provided additional logistic support in the Lower Florida Keys: Annie Simpkins and Arnim Sheutz (USN), Tom Wilmers, Stuart Marcus, John Andrew, Mark Yanno, and Jane Tutton (USFWS), Randy Tate (TNC), Phil Frank (FGFWFC), Bill Keogh (wildlife photographer), and Lee Irby (long suffering husband). My friends in both Gainesville and the Keys remained faithful throughout my multiple moves across Florida. Many fellow graduate students helped during the planning, analysis, and writing of my dissertation.




11








TABLE OF CONTENTS

pge

ACKNOWLEDGMENTS .................................................. .................. ii

A B ST R A C T ..... ........................................... ........................................... vii

CHAPTERS

1 INTRODUCTION .....................................................................1.....

Population Viability Analysis ........................................ .............
Spatial Structure ......................................................4
M etapopulations ...............................................................................6
Metapopulation Persistence ......................................... .............
Dissertation Structure ..................................................9

2 THE LOWER KEYS OF FLORIDA ............................................... 12

Geologic History ............... ............................... 12
C lim ate ................................................. ......................... ... ..13
Flora and Fauna....................................................................... 14
D evelopm ent ................................................. .... .................. ....... 16
Study Sites ...... ............................................................................... 18

3 POPULATION BIOLOGY (INTRA-PATCH DYNAMICS) ............22

Introduction .................................................................................... 22
M ethods ....................................................... .......... ............26
Trapping Girds ..............................................................................26
Genetic Analyses ..................................................27
Radio-telem etry ..................................... .............. 28
N atality .................................................................................... 29
M ortality ...................................................... ............... .. ........... 30
R esults. ........................................................ .......... ............32
Growth and Morphology ....... ....................32
G en etics ............................................. ........................... ......... 34
Demographics ...................................................35
N atality ..................... .. ............................................................. 36
M ortality ............................................................................. .... .. 37
D iscussion ................................................................................. ..... 39



iii








Genetic Threats to Persistence...............................................42
Demographic Threats to Persistence ..........................................42

4 SPATIAL ORGANIZATION (INTER-PATCH MOVEMENTS) .....55

Introduction ...................... ................................................ .... .........55
M ethods ....................................................... .......... ............60
Radio-telemetry ................................................61
H om e R ange ............................................................................62
Spacing B ehavior ................................................................ 63
D ispersal .................................................................... ............. 64
C orridor U se ............................................................................ 65
R esults ......................................... ........................................... 66
H om e R ange ........................................................................ 67
Spacing Behavior ..................................................................68
Home-range Features ........................................ ......69
Dispersal ........ ..............................70
Corridor U se .............................................................71
D iscu ssion ...................................................................................... 72
Spatial Organization.........................................73
Conclusions .................................................... 76

5 METAPOPULATION DYNAMICS: PATCH OCCUPANCY
AND HABITAT QUALITY ......................................... ........... 92

Introduction ...................................................................................... 92
Metapopulation Structure ...........................................................92
Habitat Use ......................................................94
Methods ....................................................... ......................95
Pellet Grids .....................................................................................95
Pellet Size ...... ................................ 96
Metapopulation Structure ......................................................... 97
Dietary Analysis .................................... ...... ....... 98
Microhabitat Use................................................................... 100
Macrohabitat ....................................................................... 101
R esults. ........................................................................................... 103
Pellet Degradation Rate and Pellet Size............................... 103
Metapopulation Structure ....................................................... 104
Dietary Analysis ..................................... 106
Resource Use and Availability ................................................. 107
H abitat M odel ................... ....................................................... 108
Potential Reintroduction Sites .................................................. 109



iv








Discussion ............................................................................ 110
Metapopulation Structure ......................................................... 110
H abitat U se .............................................................................. 112

6 POPULATION VIABILITY ANALYSIS ............................... 135

Introduction.................................................................................. 135
Multiple Levels of Population Viability Analysis ..................... 135
Population Viability Analysis Models ...................................... 138
Costs and Benefits of Using a PVA Model........................... 139
M ethods ......................................................................................... 14 1
Population-level Parameters .................................................. 141
Inter-population Level Parameters ............................................ 147
Metapopulation Parameters .................... .......... ........... ........... 148
The M odel ............................................................................... 148
Scenarios................................................................................. 149
Scenario #1 Decrease Predation ................................... 149
Scenario #2 Decrease Road-kills ..................................... 150
Scenario #3 Reintroduce Rabbits ..................... ............. 150
Scenario #4 Disease ..................................................... 150
Scenario #5 Hurricanes.................................................. 151
Scenario #6 Corridor Destruction .......................................... 152
Scenario #7 Habitat Destruction ...................................... 152
R esults .........................................................................153
Population Parameter Estimations ............................................ 153
Density Estimation .......... ................................................ 155
Between Population and Metapopulation Parameters ............ 156
Simulation Results ....................................... 157
D iscussion ..................................................... ............................ ..... 160
Model Validation .............. ......................... 162
Minimum Viable Population, Area, and Number of Patches..... 167

7 CONCLUSIONS AND MANAGEMENT
RECOMMENDATIONS ................................................................ 193

Population Viability Analysis (PVA) and Metapopulation
D ynam ics ....................................................................................... 193
Management for the Recovery of the Lower Keys Marsh Rabbit.... 196
Primary Recovery Actions ........................................ 196
Secondary Recovery Actions............................................... 198
Future Research ..................................................201




V










APPENDICES

A DESCRIPTIVE INFORMATION ABOUT THE
HABITAT PATCHES AND MAPS .......................................203

B VARIABLES USED IN THE DISCRIMINANT FUNCTION
ANALYSIS ......................................................................... 213

C MARSH RABBIT DENSITY ESTIMATES ...................................216

D MARSH RABBIT PATCH OCCUPANCY ..................................2..19

REFERENCES ..........................................................................................222

BIOGRAPHICAL SKETCH ....................................................................244


































vi













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

METAPOPULATIONS OF MARSH RABBITS: A POPULATION VIABILITY ANALYSIS OF THE LOWER KEYS MARSH RABBIT
(SYLVILAGUS PALUSTRIS HEFNERI)

By

Elizabeth A. Forys

May, 1995

Chairman: Stephen R. Humphrey
Major Department: Wildlife Ecology and Conservation (Forest Resources and Conservation)

The Lower Keys marsh rabbit (Sylvilagus palustris hefneri) is a state and federally endangered subspecies that historically ranged from Big Pine Key to the southernmost of the Florida Keys, Key West. Lower Keys marsh rabbits inhabit the marsh transition zone, an area that is currently highly fragmented due to development. This dissertation incorporates data collected during a 2.5-year study of the population biology, habitat requirements, and spatial structure of the populations of the Lower Keys marsh rabbit.

Within and between patch dynamics were studied by live-trapping

marsh rabbits and fitting them with radio-collars at 6 habitat patches. Genetic




vii








variability, growth and morphology, natality, mortality, home range size, and dispersal were estimated from this portion of the research. To study metapopulation dynamics, all patches of marsh rabbit habitat in the Keys were located and sampled with fecal-pellet grids during the 2.5 years to determine presence/absence and density of marsh rabbits at each patch.

Fifty-three marsh rabbits were live-trapped during the study and 43 of were fitted with radio-collars. Compared to other species of rabbits, the Lower Keys marsh rabbits were found to have average genetic variation, lower natality, and higher mortality. Subadult males were the primary dispersers; all but one male left its natal patch at the onset of sexual maturity and moved 180

- 2050 m. Due to the amount of movement between habitat patches, the marsh rabbits appear to exist in a metapopulation.

Fifty-nine patches of transition-zone habitat were located throughout the Lower Keys. Twenty of these patches had pellets present during all of the surveys, 22 had pellets present during at least one of the surveys, and 17 never had any pellets present. Habitat patches that were close to other patches and that had dense vegetation were most likely to be inhabited. When all 59 habitat patches and the information about the population biology of S. p. hefneri were combined in a population viability model, the model predicted that the rabbit would go extinct in the next 20-30 years. Reducing mortality in each life stage had the greatest effect on overall metapopulation persistence.



viii













CHAPTER 1
INTRODUCTION


The primary objective of this dissertation is to develop a population viability analysis (PVA) for the endangered Lower Keys marsh rabbit (Sylvilagus palustris hefneri). The Lower Keys marsh rabbit was first described as a distinct subspecies by Lazell (1984). Lower Keys marsh rabbits have a shorter molariform tooth row, a higher and more convex frontonasal profile, a broader cranium, and a longer dentary symphysis than mainland and Upper Keys rabbits.

Historically, the range of the Lower Keys marsh rabbit extended from Big Pine Key to the southernmost of the Florida Keys, Key West (dePourtales 1877). During the 1970s and 1980s, a decline in Lower Keys marsh rabbits was reported (J. Lazell, pers. comm.) and a study of the rabbit's status was commissioned by the Florida Game and Fresh Water Fish Commission (Howe 1988). Howe (1988) recorded marsh rabbits present at 13 of J. Lazell's (unpubl. data) original 17 sites and absent from 4 of the sites. The Lower Keys marsh rabbit was listed as endangered by the Florida Fresh Water Fish Commission in 1989 (F.A.C. 39-27) and by the U.S. Fish and Wildlife Service in 1990 (U.S. Fish and Wildlife Service 1990).


1





2


Three subspecies of marsh rabbits are recognized, S. p. paludicola, S. p. palustris (Chapman and Willner 1981), and S. p. hefneri (Figure 1.1). The marsh rabbit is found in lowlands from the Dismal Swamp of Virginia into the Florida Keys. Little is known about the biology of the Lower Keys marsh rabbit (Chapman and Flux 1990, Wolfe 1992). Compared to other species of cottontails, the biology and ecology of marsh rabbits in general (Sylvilagus palustris spp.) is poorly understood (Chapman et al. 1982).

Tomkins (1935) and Carr (1939) commented on marsh rabbit (S. p. paludicola) behavior in North Carolina and north-central Florida marshes. Blair (1935, 1936) studied the diet and habits of marsh rabbits near Gainesville, Florida. The reproduction of the marsh rabbit (S. P palustris) was studied by Holler and Conaway (1979) in Belle Glade, Florida.

Despite these studies, large gaps in the knowledge of the population ecology, habitat use, and risk of extinction for S. p. hefneri remain. Extrapolation from these studies to the Lower Keys marsh rabbit may not be accurate. The Lower Keys marsh rabbit inhabits a unique island ecosystem and is subject to different pressures and resources.



Population Viability Analysis

Population viability analysis (PVA) is a comprehensive examination of the interacting factors that put a population (or species) at risk of extinction





3


(Gilpin and Soul6 1986). In small populations, both stochastic and deterministic phenomena can affect the persistence time of a population. Perturbations that are stochastic include variation in the environment (climatic variation, natural catastrophes), genetic composition (loss of genetic variability, inbreeding depression), or the demographic composition (sex, age ratios) of a population (Shaffer 1981). Demographic and genetic stochasticity are most important in small populations. Environmental stochasticity is important for all populations, and its importance decreases only slightly with increasing population size.

Deterministic causes of extinction include habitat destruction, invasion by introduced predators, disease, and climatic change (Nunney and Campbell 1993). Deterministic extinction can occur regardless of population size, if all of the population is affected. Both stochastic and deterministic phenomena may interact via feed back loops leading to potential extinction via "extinction vortices" (Gilpin and Soul6 1986), and differ in magnitude and importance for species with different life-history attributes. Because knowledge about intrinsic population dynamics of a species is often limited and the occurrence and impact of extrinsic factors is uncertain, PVA can only make probabilistic predictions about a species' future (Ginzburg et al. 1982, Shaffer 1990).





4


Spatial Structure

Most endangered species live in habitats that have become fragmented (checkerspot butterfly: Murphy et al. 1990; forest-dwelling raptors: Thiollay and Meyburg 1988, spotted owl: see Lamberson et al. 1994). For these species, the spatial structure of the habitat patches may have large effects on the species' persistence. A PVA for a species existing in a fragmented habitat must look at several spatial and temporal scales of resolution (Gaines et al. 1992, Lacy 1993).

Populations occurring in habitat patches may be locally subject to higher risks of extinction than in continuous habitat due to factors (nest predation, increased exotics, microclimate changes) that occur more frequently in smaller areas that have high perimeter-to-area ratios (Lovejoy et al. 1986, Loiselle and Hoppes 1983). These "edge effects" may be intensified in smaller habitat patches, where nearly all of the patch is edge and little is interior (Wilcox and Murphy 1985). Small patches will support fewer individuals, increasing the rate of extinction caused by stochastic events (Gilpin and Soul6 1986).

For populations of species with poor dispersal ability relative to the

distance between patches, the time to extinction will be equal to the time that the last local population goes extinct (Hanski and Gilpin 1991). Each patch of habitat represents a population of these species; surviving populations may be said to be relictual (Berry 1986). Harrison (1991, 1994) uses the term "non-





5

equilibrium" to describe metapopulations where movements between patches are not great enough to increase persistence time. This population configuration has been seen in species isolated by climatic change (Brown 1971), clear-cutting of forests (Leck 1979, Laurence 1982), and man-created islands (Willis 1974, Karr 1982). These relictual species may lack the locomotive ability, evolutionary history, and/or behavioral plasticity to allow them to move between patches, or the configuration of patches may prohibit movement. Some species may exhibit "conspecific attraction" and preferentially colonize patches where conspecifics are present (Smith and Peacock 1990, Ray et al. 1991). These species may be less likely to

(re)colonize empty habitat patches than other species.

If individuals of the species are exceptionally mobile, relative to the

distance between habitat patches, then the individuals within the patches may frequently interact and form one large demographic entity inhabiting a patchy environment (Harrison 1991). Many highly mobile species of birds (Blake and Karr 1987, Rolstad 1991) use forest patches, wetlands, continental shelf, and oceanic islands in this manner. Other species that are adapted to highly variable environments such as disturbance or "r-selected" species also inhabit temporally and spatially variable habitats but remain a single population (Edwards et al. 1981, Adler and Wilson 1987).





6


Between these 2 extremes are species that spend at least a portion of their life in 1 habitat patch but are capable of moving to other habitat patches during their lifetime. For populations of these species, local extinctions may be counteracted by colonizations from nearby patches, and the overall persistence time of the species may be collectively longer than the persistence time of any one patch (Levins 1969). Populations of species that exhibit this dynamic are said to exist together as a metapopulation (Levins 1970).



Metapopulations

Recently the term "true or classic metapopulation" has been used to distinguish between this type of population configuration and others that outwardly appear to be metapopulations such as mainland-island and sourcesink configurations (Harrison 1991). In the mainland-island configuration, most movement occurs from a larger patch to smaller patches. These metapopulations do not function in the "true" reciprocal colonization pattern unless the mainland population goes extinct (Thomas and Jones 1993) and is recolonized by the smaller "islands." In a source-sink metapopulation, the higher quality source habitat patch produces a surplus of descendants (rate of population growth or X > 0), whereas the low quality patches may be sinks that produce a deficit (X < 0) (Pulliam 1988, Pulliam and Danielson 1991).





7


Debate currently exists over how many examples of true

metapopulations exist in nature (Harrison 1994). This may be due in part to the lack of studies that embody the spatial and temporal scales necessary for studying metapopulations.



Metapopulation Persistence

For a metapopulation to persist, the rate of patches being colonized and established must exceed the rate of patches going extinct (Levins 1970, Hanski 1989). A metapopulation is said to be at equilibrium when these rates are equal. Departures from equilibrium towards increasing extinction rate may ultimately result in a regional extinction. There may be a threshold number and configuration of occupied patches, below which the species is not likely to rebound (Hanski 1991), similar to the concept of a minimum viable population of individuals (MVP; Shaffer 1981).

Persistence of individual patches is determined by dynamics that occur both within the patch and by movements between patches. Population dynamics, internal to patch dynamics, such as natality, mortality, sex ratio, and age structure, interacting with patch size, habitat quality, and predator density determine the size of the local population and its variability.

Between-patch dynamics are shaped by a species' home range, spacing and movement patterns, especially dispersal. Movements between patches are





8


obviously important for patch colonization and establishment. These movements also may stabilize local population variability of occupied patches and increasing the persistence time for the local population through the rescue effect (Brown and Kodric-Brown 1977). Patch configuration and the type of habitat between the patches (including developed land) may ultimately have a large impact on the individual's movements.

At the scale of the entire metapopulation, the correlation in

environmental variation among patches and the frequency of large-scale environmental catastrophes, is vitally important (Quinn and Hastings 1987, Gilpin 1988, Hanski 1989, Stacey and Taper 1992). If all patches are subjected to the same adverse environmental conditions at the same time, then patch extinctions may be correlated, leaving few or no source populations to colonize the extinct patches.

Additionally, the cause of local extinctions may be important to the

overall metapopulation dynamics. It has been suggested (Thomas 1994) that most local extinctions are due to deterministic changes in the environment of a patch that render a patch unsuitable. If deterministic, these conditions are likely to persist following the local extinction leaving the patch empty, but unsuitable. Therefore, most empty patches are currently unsuitable for occupancy and few suitable patches become empty. Some populations, either because of their small size or inherently high variability, may be subject to





9

stochastic as well as deterministic forces of extinction. In these "turnoverprone" species (Schoener and Spiller 1987; Harrison 1991), empty patches of suitable habitat may be relatively common.



Dissertation Structure

This dissertation incorporates data on the Lower Keys marsh rabbit at 3 spatial levels of observation: within-patch population dynamics, inter-patch between population dynamics, and metapopulation dynamics. Using simulation models that incorporate these three spatial scales, predictions are made for a range of temporal scales.

Chapter 3 investigates the dynamics of 5 populations of Lower Keys marsh rabbits living in habitat patches. Estimates of natality, mortality, demographic structure, genetic and morphological variation were made at this scale. In Chapter 4, the home range size, movement patterns, and dispersal ability of the marsh rabbit were measured. This information about marsh rabbit inter-patch use provided the basis for testing the hypothesis that the marsh rabbits are existing in a true metapopulation configuration.

Chapter 5 explores patch occupancy, extinction, and (re)colonization at the metapopulation scale. Marsh rabbit diet, microhabitat, and macrohabitat selection were combined with the physical attributes of the patch to determine if these parameters influenced patch occupancy. Chapter 6 incorporates the





10


conclusions drawn from chapters 3-5 into a PVA to test the hypothesis that the metapopulation was in equilibrium. Predictions about the future for the Lower Keys marsh rabbit were made under a number of different scenarios.

In the final chapter (Chapter 7), conclusions and recommendations are made for the management and conservation of the Lower Keys marsh rabbit and other endangered species inhabiting patch/fragmented landscapes. Predator control, species reintroductions, the impact of more habitat loss, diseases, and the impact of hurricanes, are examined in light of the new research on metapopulation dynamics.

























S. p. palustris S. p. paludicola S. p. hefneri






Figure 1.1--Distribution of the 3 subspecies of marsh rabbit (Sylvilagus palustris).













CHAPTER 2
THE LOWER KEYS OF FLORIDA

Geologic History

The Lower Keys of Florida are the terminal portion of an archipelago of islands extending westward from the mainland of Florida (Figure 2.1). For the purposes of this dissertation, the Lower Keys includes all of the islands from Key West to Noname Key (inclusive). The Lower Keys are separated from the Upper Keys by nearly 20 km of water (from Noname Key to Vaca Key).

The Keys are derived from coral reefs that formed during the Sangamon interglacial period (340,000-100,000 BP). During the subsequent Wisconsin glacial period, the Lower Keys were covered with oolitic sands and were gradually exposed as sea level began to fall. At the height of the Wisconsin glaciation (20,000 15,000 BP) the Keys were points of high relief amid the large land mass of Florida (Shinn 1988, Mueller and Winston 1991). At this point in time, Florida had nearly twice the exposed land as today (Webb 1990). Sea level remained low until nearly 15,000 BP, when the climate began to warm. Sea level rose rapidly from 15,000 10,000 BP, reaching a maximum of 7 feet/year, and then the rate slowed. During the past 2,000 years sea level has risen only 6 feet.



12





13

Today, the Florida Keys are composed of two formations of limestone. The Upper Keys (Soldier Key to the southeastern corner of Big Pine Key) are composed of the Key Largo limestone and the Lower Keys (the majority of Big Pine Key to Key West) are from the Miami limestone. The Miami limestone is composed of small ovoid pellets of calcium carbonate (ooids; Hoffmeister 1974) that hardened when sea level fell and the land surface was exposed to air.



Climate

Although several degrees north of the Tropic of Cancer, the close

proximity of the Gulf Stream and maritime influences produce a subtropical climate in the Lower Keys (Chen and Gerber 1990). In 50 years of data collection, the temperature in the Lower Keys has never fallen below freezing, and the record low temperature is 410 F (50 C). Winter cold fronts are buffered by the warm ocean water before proceeding to the Keys. Monthly average temperatures differ only 150 F (90 C) from January to July (Figure 2.2). Daily temperatures rarely vary more than 100 F (120 C) between the high and low.

There are distinct dry and wet seasons in the Lower Keys, although this area is generally slightly drier overall than the Upper Keys and mainland Florida (Chen and Gerber 1990). The dry season spans November through April and accounts for less than a third of the annual precipitation. The





14

remaining two-thirds of the rainfall occurs during the wet season which begins in May and ends in October (Figure 2.2). Relative humidity remains high throughout the year, averaging nearly 75% (NOAA 1993).

Hurricane season occurs mainly during the wet season, from June until November. The chance that a hurricane (winds >74 mph) will hit the Lower Keys is slightly less than 10% each year. Historically, most hurricanes have occurred during the months of September and October. Tropical storms (i.e., winds 39-74) occur more frequently, but do not produce the large storm surges and destruction that hurricanes are capable of producing.



Flora and Fauna

The flora of the Lower Keys is derived from 4 sources: the Caribbean, the eastern U.S. coastal plain, endemic taxa evolved in place, and exotic taxa from cosmopolitan sources (Long 1974). Four plant associations predominate: mangrove, transitional salt-marsh, pineland, and hardwood hammock.

Mangrove community is dominated by 3 saline-tolerant, unrelated species: red mangrove (Rhizophora mangle), black mangrove (Avicennia germinans), and white mangrove (Laguncularia racemosa). These trees occur in areas that are either continually submerged or tidally inundated.

Transitional salt-marsh (also called the transition zone) is a grassy,

nearly treeless marsh area that generally occurs from 1 to 3 m above sea level.





15

The majority of transition zone marshes are subject to predictable flooding by spring lunar high tides (Williams 1991). Transition zone marshes often occur between the mangrove community and the upland hardwood or pine hammocks. The transition zone can be further divided into two components: an open saltmarsh and at slightly higher elevations, a more forested area dominated by buttonwood (Conocarpus erectus). Several species of mammals, including the Lower Keys marsh rabbit inhabit this area.

On land where there are well-developed fresh water lenses, the pineland community can occur. These areas are rarely inundated by salt water and are maintained through periodic burns. Slash pine (Pinus elliottii) is the dominant tree species, although various species of palms and ferns are also abundant (Snyder et al. 1990). At the highest elevations (>3 m) the diverse hardwood hammock occurs (Carlson et al. 1992). This is the climax community in the Keys and occurs on nearly every large key. Nearly 10% of all tree species found in this area are endemic (Long and Lakela 1971).

Although less common, fresh water wetlands dominated by sawgrass (Cladium jamaicensis) do occur on the few keys that have a fresh water lens. A few keys also have limited amounts of beach and dune habitat, but most of the keys have exposed limestone rock on their coasts.

All of the native terrestrial mammals are derived from populations of the continental United States (Layne 1974). Currently, only 5 species of native





16

terrestrial mammals and perhaps 1 bat are found in the Lower Keys. All of the terrestrial mammals have been cited as being endemic in the literature at either the species or subspecies level (see Lazell 1989), although the accuracy of these taxonomic claims has been debated (see Humphrey 1994). The paucity of mammals in the Lower Keys may be related to the current isolation and small area of the Keys, combined with a lack of fresh water (Layne 1974).

Herpetofaunal diversity is greater than the mammals, although a large number of reptiles and amphibians in the Lower Keys are exotics (Wilson and Porras 1983). Of importance to the Lower Keys marsh rabbit, the eastern diamondback rattlesnake (Crotalus adamanteus) is common throughout the Lower Keys and the alligator (Alligator mississippiensis), although rare, does occur. Avian diversity, both breeding and over-wintering is fairly high in the Lower Keys although lower than mainland Florida (Robertson and Kushlan 1974). Special references to birds and reptiles as predators will be made in Chapter 3.



Development

By the 1890s Key West was the largest city in Florida and one of the

largest in the United States. The remainder of the Keys were unpopulated until the completion of the overseas highway in 1938 and the first water pipeline in 1942 (Gallagher 1991). Today, over 78,000 people live in Monroe county,





17

nearly half of them in the Lower Keys (Shermyen 1993). The populations has increased nearly 65% since 1960. Tourism, which was also high during the late 1800s, has also resurged in the Lower Keys. At the 2 National Park beaches in the Lower Keys, over 500,000 tourists were recorded in 1993. The majority of these visitors came to the Lower Keys and Key West via the Overseas highway (Shermyen 1993).

Human impact on the Keys' wildlife was noticed as early as 1908, when the Key West National Wildlife Refuge was established to provide habitat for migratory birds. The Great White Heron National Wildlife Refuge was established in 1938, encompassing a collection of small islands north of the main chain of keys. In 1957, the National Key Deer Refuge was established; its main mission was to protect the Key Deer (Odocoileus virginianus clavium) and other wildlife. The largest tracts of land for this refuge are on Big Pine Key and Noname Key, but other areas on the Torch Keys are currently being added (John Andrew, National Key Deer Refuge Manager, pers. comm., 1993).

The United States military presence began in the Keys during the Civil War, and waxed and waned throughout the past century (Hambright 1991). Due to the Lower Keys' proximity to Cuba, the Caribbean, and open water ideal for training aviators, the U.S. Navy currently maintains a base in the Keys. The Navy owns a large amount of land on Key West, Boca Chica Key,





18

and North Saddlebunch Key. The U.S. Air Force manages the northernmost portion of Cudjoe Key.



Study Sites

This study attempted to encompass all possible marsh rabbit habitat throughout the Lower Keys, including saltwater transition-zone habitat and fresh water marsh areas. Habitat areas were located using information from Howe (1988), aerial photographs and ground survey. Initially, a patch of habitat was considered isolated from another patch if the 2 areas were divided by a major road, airplane runway, or large body of water. As a general rule, patches of habitat <0.5 ha were not sampled, although some exceptions were made. A previous study of marsh rabbit home ranges (Payne 1975) found that no adult home range was <0.5 ha.

Fifty-nine habitat patches were identified throughout the Lower Keys (Appendix A); no habitat was found on Key West, the most densely humanpopulated Key. Twenty-seven habitat patches were found on Boca Chica, 25 of which were owned by the U.S. Navy. Four patches were examined at the Saddlebunch Keys; 2 of which were owned by the U.S. Navy. On Sugarloaf Key, 9 patches of transition zone habitat were found; 1 of which is owned by the U.S. Fish and Wildlife Service (USFWS), and other is owned by Monroe County. Two privately owned patches were found on Cudjoe, and 1 site





19

partially on U.S. Air Force land. One privately owned site was found on each of Summerland and Ramrod Keys. Big Torch Key had 2 privately owned sites and one site owned by the USFWS. Middle Torch had 2 sites, both of which were owned by the USFWS. Little Torch had 1 habitat patch and this patch was owned by the Nature Conservancy (TNC). Big Pine had 7 habitat patches: 3 owned by the USFWS, 2 privately owned, and 2 owned by the South Florida Water Management District (SFWMD). Noname Key had 1 USFWS-owned habitat site.

Additionally, habitat was found on several of the smaller islands that are part of the Lower Keys chain but that are only accessible by boats. Habitat patches were found on Annette and Porpoise Keys, both of which are small islets northeast of Big Pine Key. These islets are part of the Great White Heron National Wildlife Refuge. Transition-zone patches were observed on Big Munson Key and Saddlehill Key. Big Munson Key is south of Big Pine and is owned by the Boy Scouts of America, Inc. Saddlehill is southeast of Boca Chica Key and is privately owned. The search for habitat on other isolated islets was not complete; the existence of small patches of habitat on more distant keys is a possibility.












Middle Torch

Big Torch Big Pine Cudjoe Noname




Saddlebunch Summerland Ra USI Little Torch
... ....... ......... ..okm

SSugarlof 5


Key West







Figure 2. 1--The Lower Keys of Florida. The solid line connecting the Keys is highway US-i.





21







88 10

9
84


80 C 7 0

6 a
E 76
-5
CM 72 4 0

68- 3 <


64


60 -... 0
1 2 3 4 5 6 7 8 9 10 11 12 Month









Figure 2.2--The average temperature (line, left axis) and average precipitation (bars, right axis) during the past 25(1965-1990) years measured at Key West airport, Key West, Florida.













CHAPTER 3
POPULATION BIOLOGY (INTRA-PATCH DYNAMICS)


Introduction

The first step in developing a population viability analysis (PVA) is determining the demographic and genetic composition of a population, how these compositions vary, and how large a role stochastic and environmental variability play (Salwasser et al. 1984, Gilpin and Soul6 1986). Demographic and genetic compositions of a population can influence population growth, variability in growth, and the ultimate population size, all of which can in turn influence the populations chance for persistence (Goodman 1987).

The demographic composition of a populations is determined by

variation in the birth and death rates and in the sex ratio of the new recruits. Some variation in these parameters will be purely random, other sources of variation may be from external sources (i.e., differential mortality by sex or age group, biased sex ratios at birth ), and therefore this variability may be correlated throughout the population. In a large, continuous population, random variations will contribute little to the overall population variability but environmental variability may be important; in very small, fragmented




22





23

populations, both stochastic and environmental variation can substantially alter sex and age ratios.

Genetic variability, at both the individual and population level, may also be important. Highly fragmented, isolated populations are at risk of loosing both types of variation. When a population is small and isolated, the chance of a mating between close relatives increases. These matings can produce young with a higher proportion of homozygous loci, thus potentially decreasing the number of heterozygous loci and increasing the number of deleterious recessive genes (Packer 1979). Certain heterozygous loci have been correlated with greater tolerance to environmental variations in some species (Lerner 1954).

This increase in homozygous deleterious genes, combined with a decrease of heterozygote loci, is believed to cause greater mortality, and reduced fecundity, creating a phenomena called "inbreeding depression" (see Lacy 1993). Loss of genetic variability among individuals or between populations may contribute to inbreeding depression, and may decrease the overall chance that the population can evolve to meet new environmental conditions. Although loss of genetic variability in insular populations is well documented (Kilpatrick 1981, Berry 1986) its overall effect on population persistence is less known.

Inbreeding depression was documented early in domesticated animals (Wright 1977, Falconer 1981), laboratory animals (Strong 1978) and more





24

recently in captively breed animals from wild stock (Ralls and Ballou 1983, Ralls et al. 1988). Evidence for inbreeding depression in natural populations is less clear (Lacy 1993), and in need of further study. Most examples of possible instances for inbreeding depression in the wild come from populations of large, long-lived species such as the grizzly bear (Ursus arctos; Harris and Allendorf 1989) and the Florida panther (Felis concolor coryi; Seal and Lacy 1989). These species have existed at extremely small numbers in the wild for several decades and have largely overlapping generations.

For smaller, shorter-lived species, it is currently hypothesized that

before inbreeding depression can have an effect in the wild, habitat loss and demographic problems may have already caused the demise of the wild populations (Lacy 1993). Management for these problems may also help in preventing inbreeding depression and overall loss of genetic variability in the population, although more intense management may be needed for populations that have recently undergone severe population bottlenecks (Fuerst and Maruyama 1986). Inbreeding depression for these species may still be an important consideration in PVA, if captive breeding and reintroductions are available as management options (Haig et al. 1990).

The sensitivities of demographic and genetic compositions to random and environmental variation are dependent on the life history and population dynamics of the species in question. Further, they may be affected by





25

idiosyncratic factors such as the habitat available and climatic variation, and therefore require in-depth study.

Lagomorph population biology is characterized by high reproductive rates and high rates of mortality. All lagomorphs are entirely herbivorous, although they feed on a diversity of vegetation. Due to their high abundances and intermediate size, lagomorphs are the base of many predator-prey systems involving small to medium-sized predators (Chapman and Flux 1990).

Most research on rabbits has been on the most abundant species (e.g., eastern cottontail, Sylvilagus floridanus), primarily in relation to hunting. There is little information on the management of rare or endangered lagomorph species. Additionally, most research on endangered species (in general) has historically concentrated on larger species that are long-lived and have longer generation-times (Murphy et al. 1990). Threats to the persistence of smaller species with higher reproductive rates, but shorter life-spans (r-selected) will differ. These "r" selected species may be more habitat specific than "K" species and may experience higher variability in population numbers (Pimm 1991).

The objective of this portion of the research was to determine the demographic and genetic compositions of the Lower Keys marsh rabbit. Special emphasis was placed on parameters that affect the intrinsic rate of growth of the marsh rabbit populations (natality, mortality, sex and age ratios),





26

the variability of these parameters, and the extrinsic factors that influence this variability. This information about the population biology was used in the final model predicting the future of the Lower Keys marsh rabbit presented in Chapter 6.



Methods

Two methods were employed to study the population biology of the Lower Keys marsh rabbit: live-trapping and radio-telemetry. Marsh rabbits were studied at a subsample of 6 habitat patches. All of the patches chosen were on Navy-owned land; a diversity of patch size and shape was sampled. Five habitat patches were selected on Boca Chica and 1 on Saddlebunch Key. Additional sites throughout the Lower Keys were used for the portion of the study that examined the genetic composition of the marsh rabbits.



Trapping Grids

Individual marsh rabbits were examined by trapping the 6 main sites. Trapping occurred twice during the wet season (June November), and twice during the dry season (December May). For the 5 sites on Boca Chica Key (Figure 3.1), trapping occurred from June 1991 to May 1993 (8 trapping sessions). On Saddlebunch, trapping was conducted from June 1992 to May 1993 (4 trapping sessions). Trapping grids were placed on each site, using





27

unbaited collapsible National live traps (80 x 30 x 30 cm), placed in a 6 x 6 array, spaced approximately 25 m apart. Each trapping session consisted of 5 nights where the traps were open, 2 nights with the traps closed and another 5 nights with traps open. Traps were checked twice daily, once in the morning and once in the evening and were covered in burlap for shade.

All rabbits caught were sexed, weighed, and tagged (Monel no. 3,

National Band and Tag, Newport, KY). Measurements were made of the right rear foot, ear length from notch, and total body length from nose to (and including) tail. Marsh rabbits are relatively easy to sex (Nagy and Haufler 1980), but an aging method has not been determined. An attempt to correlate weight, total length, ear length, and rear right foot length with age was made. Rabbits were classified as juveniles (not sexually mature), subadults (entering sexual maturity), or adults (fully sexually mature).



Genetic Analysis

Blood samples were obtained from all animals caught on the 6 trapping grids between June 1992 and May 1993. In addition, blood was taken from animals caught during preliminary trapping on Sugarloaf and Big Pine Keys. Blood samples were obtained by lancing the ear with a hypodermic needle and collected using heparinized capillary tubes. Blood was stored on wet ice for <1 hour and was centrifuged into plasma and hemolysate at 5,000 rpm for 10 min





28

(Scribner et al. 1983). The plasma and hemolysate were stored in liquid nitrogen and were analyzed used starch-gel electrophoresis within 1 year.

Eleven presumptive genetic loci were scored for the samples. Standard procedures for starch-gel electrophoresis were used (Harris and Hopkinson 1976). Locus nomenclature followed McAlpine et al. (1985) for mapped human genes. Two buffer systems were used. Tris-citrate, pH 6.7 was used for the following enzymes: esterase, E.C. 3.1.1.1 (EST-1); glucose phosphate isomerase, E.C. 5.3.1.9 (GPI); isocitrate dehydrogenase, E.C. 1.1.1.42 (ICD-1);

lactate dehydrogenase, E.C. 1.1.1.27 (LDH-1), malate dehydrogenase, E.C.

1.1.1.37 (MDH-1); mannose phosphate isomerase, E.C. (MPI); and superoxide dismutase, E.C. 1.15.1.1 (SOD-1). Tris citrate, pH 8.0 was used for the following enzymes: beta hemoglobin (PHb); glutamate oxaloacetic transaminase, E.C. 2.6.1.1 (GOT-l); peptidase, E.C. 3.4.13 (PEP-1); and

phosphoglucomutase, E.C. 2.7.5.1 (PGM-1).



Radio-telemetry

Each rabbit weighing >1,000 grams trapped at the 5 Boca Chica sites

and the Saddlebunch site was fitted with a radio-collar (weight <3 grams) and a transmitter with an estimated 10-month operational life (Telonics, Inc., Mesa, AZ). Smaller rabbits (300-1,000 g) were fitted with a similar radio-collar, but with a velcro break-away device added to allow the animals to lose the





29

equipment as their bodies grew larger. Collars were replaced as the animals aged. Previous studies on cottontail rabbits (S. floridanus) found that mortality rates were not statistically different for radio-telemetered rabbits and rabbits marked using other methods (Trent and Rongstad 1974, Rose 1977).

Collared rabbits were located on separate days three times a week, once in the early morning (7-9 a.m.), once at mid-day (11 a.m. 1 p.m.), and once in the evening (4-6 p.m.). Signals were followed until the animal could be seen or the exact location was found. All locations were made >24 hours apart to ensure independence of observations. Locations were plotted on 1:2400 aerial maps with 10-m2 grid overlays. Because the 5 sites on Boca Chica were studied for 2 years and the Saddlebunch site was only studied during the final year, natality and mortality portions of the study only used data from the 5 sites on Boca Chica Key (Figure 3.1).



Natality

Natality was studied by examining females during each trap session for pregnancy (by palpating the uterus) and lactation. Nesting was determined by following collared female rabbits until they centered their activities around one small area. Nest confirmation was made by observing the young through a tunnel in the grass or by finding deposits of juvenile pellets near the nest area. Where possible, the number of young observed in or fleeing the main chamber





30

of the nest was recorded. Although this method is approximate, it provided an estimate of the number of individuals surviving at the time of nest discovery. Number of litters for each female was calculated for both the wet and dry seasons and compared using a likelihood-ratio Chi-square (G2). The nest area was mapped, and the dominant vegetation was identified.



Mortality

Percent mortality was calculated for 5 age classes: nestling, juvenile, subadult, first-year adult, and second-year adult. Nestling mortality (0-3 months) was calculated by comparing the number of young observed in nests to the number of juveniles caught on the trapping grid. Juvenile mortality (4-7 months) was estimated using telemetry for individuals >300 g. Subadult (8-10 months), first-year and second-year adult mortality was determined from the telemetry data alone. First-year adults were rabbits that had been followed as an adult for 1 year; most had been collared as juvenile or subadults. Secondyear adults were rabbits that had been followed as adults for 2 years. First- and second-year adult classes may have contained individuals older than 2 and 3 years old.

All mortalities were located within 2 days of death. When possible, field necropsies were performed as outlined by Wobeser and Spraker (1980),





31

with special attention to the liver for signs of tularemia (Francisella tularensis). Site description, carcass condition, and position was recorded for each death.

All rabbit carcasses thought to have been preyed upon were examined for signs of trauma. The head, throat, and neck were examined for puncture wounds. Hemorrhaging, particularly the presence of blood in the mouth, nares, trachea or neck region indicated the rabbit was alive at the time of attack (Hawthorne 1980). Feeding pattern on the carcass was also examined: hindquarter feeding probably indicated scavenging, whereas feeding on the shoulders and neck indicated possible predation (Hawthorne 1980).

When predation was suspected, the site was examined for sign on and around the carcass. Tracks and scats were identified for potential information on the predator or scavengers. Potential predators found in the Lower Keys included Bald Eagles (Haliaeetus leucocephalus), Red-shouldered Hawks (Buteo lineatus), eastern diamondback rattlesnakes (Crotalus adamanteus), feral cats (Felis catus), raccoons (Procyon lotor) and possibly Black Vultures (Corgyps atratus) and domestic house-based dogs (Canis familiaris).

Birds of prey tend to capture rabbits in the middle of the back, and will kill using deep puncture wounds to the back and head (Hawthorne 1980). They may take their prey back to a nesting area. Feral cat predation may be assumed if the carcass has been dragged, eviscerated, or covered in dirt. Often cats will leave tooth marks on every exposed bone of their prey (Anderson





32

1969). Raccoons are more likely than cats to eat the breast, crop, and entrails of their prey. They may also carry portions of the prey to water (Anderson 1969).



Results

A total of 54 Lower Keys marsh rabbits was caught and examined

during this study (Table 3.1). Forty-three were caught on the 6 main grid sites (41 on Boca Chica, 2 on Saddlebunch), 5 rabbits were caught while trying to recapture collared dispersed rabbits, and 6 were caught on other sites in the Lower Keys. Rabbits were only examined once per trapping period. Data from the 54 rabbits was recorded 130 times during 8 trapping sessions (June 1991 May 1993).

Forty-three (28 male, 15 female) of the rabbits caught were fitted with radio-collars, and 7 of these were juveniles. More juveniles were caught before the break-away collar technology was complete. Forty-one of the radiocollared rabbits were on Boca Chica Key and two were on Saddlebunch Key.



Growth and Morphology

Mass of the 54 marsh rabbits ranged from 100 to 1400 grams (including pregnant females). It was judged that the 100-gram rabbits had just left the nest and were approximately 1 month old. Several young rabbits were causght,





33

but only 3 rabbits (1 male, 2 females) were caught during this stage and survived to adulthood. Marsh rabbit age was correlated with external measurements using these 3 individuals.

Both body mass and total body length were found to significantly (P

<0.05) predict marsh rabbit age for all 3 rabbits (Table 3.2). Ear length and the length of the right hind foot were not significant predictors in one of the female rabbits. Body mass has been used as an indicator of age in cottontails (Lord 1963), but the length of the right rear foot is used more often (Bothma et al. 1972). Because measuring the length of a non-sedated marsh rabbit is difficult, the marsh rabbit body mass was used as an indicator of age in conjunction with appearance of external sexual organs. Body mass can be influenced by nutritional and physiological variations but was the most objective and accurate age indicator available. Body mass of the marsh rabbits increased almost linearly with age for the 3 rabbits and plateaued at approximately 1,100 grams (Figure 3.2) as the rabbits approached 1 year.

External measurements collected from 13 non-pregnant adult female rabbits and 19 adult male rabbits were compared for differences using a Wilcoxon test. In an attempt to use only measurements on fully grown rabbits, only individuals that had been caught more than once as an adult were used. No significant differences were found in any of the measurements between sexes. Approximately half of all cottontial species are sexually dimorphic





34

(females larger) and the other half exhibit no dimorphism (Champman et al. 1982). The average measurements and standard deviations of 29 marsh rabbits were as follows: mass 1224.1 g. (80.9), total length 339.3 mm (24.9), ear from notch 52.7 mm (3.4), and right hind foot 73.6 mm (3.7).



Genetics

Two (EST-1 and PGM-1) of the 11 loci sampled from the 19 rabbits

were polymorphic, indicating that there is genetic variation in the Lower Keys marsh rabbit (Table 3.4). This proportion of variable loci is only slightly less than the proportions seen in studies of cottontail rabbits in Texas (Scribner and Warren 1986). Deviations from Hardy-Weinberg expectations were not observed (P > 0.05) at either of the polymorphic loci, but this test may not be valid on a so small sample. Differences in allelic proportions for each locus appeared to vary among keys, but sample sizes were too small for statistical comparison. Heterozygous loci and variation among individuals occurred at Boca Chica, Geiger, and Sugarloaf Keys. The 2 rabbits sampled on the Navy land on Saddlebunch Key were homozygous and monomorphic at all loci. Only 1 individual was sampled from Big Pine Key; this individual was heterozygous only at the EST-1 loci and contained an allele at the EST-1 locus that was not seen at any of the other keys.





35

Demographics

The number of marsh rabbits caught on the 5 main sites varied from a high of 18 rabbits during November 1991 and February 1992 to a low of 9 during November 1992 (Figure 3.3). The small number of rabbits at each site precluded the use of capture/recapture statistics (Pollock et al. 1990); population number at each site was estimated using the minimum number known alive (MNA; Hilborn et al. 1976).

When all of the male and female individuals caught over the 8 trapping sessions at the 5 Boca Chica sites were counted, there appeared to be more male than female rabbits on the 5 sites during the study (Table 3.3). The sex ratio significantly male-biased in only the subadults; nearly equal numbers of male and female juveniles and adults were captured on the 5 grids.

Trappability (the probability that a rabbit was captured more than once during the 10-day trapping session) differed among the demographic groups (Table 3.5). Male and female trappability was high and not statistically different. Juvenile trappability was lower, and females were more likely to be retrapped during a trap session than were males. Subadult trappability was high for the males, and very low for females. Sample sizes for the trappability comparisons were small.





36

Natality

Eleven adult female marsh rabbits were radio-collared, followed for >1 month, and used in this portion of the analysis. Length of time followed ranged from 2 to 22 months (X = 9.09 months, SE = 1.94 ) for a total of >100 months. Thirty-one nesting events were recorded; all females were observed to nest and produce a litter at least once. The number of young observed per nest ranged from 1 to 3 with an average of 1.77 kittens per nest (SE = 0.09).

All nests were made in clump grasses (thick grasses and sedges), with 22 of the nests predominately in Spartina spartinae and the other 9 in Fimbrystylis castanae. In general, the nests consisted of a main chamber with several smaller chambers and exit/entry routes. None of the nests were obviously lined with fur as reported in the northern subspecies (Tomkins 1935). Only 2 of the females used the same nesting area more than once, and none of the female rabbits used the nest of another rabbit during this study.

There was little apparent seasonal pattern in the reproduction of the marsh rabbit (Figure 3.4). Combining data from the 2 years of the study, the proportion of rabbits that produced litters each month ranged from 0-56%. The highest proportion of females with litters was seen in March and September; the lowest proportion was seen in April and December. The average number of litters produced during the wet and dry seasons did not differ significantly (G2 = 0.15, df = 1, P > 0.05). Although reproduction in most cottontail species is





37

highly seasonal (Chapman et al. 1982), reproduction in marsh rabbits in southern Florida (Holler and Conaway 1979), southern Texas (Bothma and Teer 1977), and southern California (Ingles 1941) exhibits only year-round slight seasonal fluctuations.

Time between litters ranged from 1 to 5 months (X = 2.45, SE = 0.30). Combining the results from the 11 rabbits, an average of 3.7 litters per year was produced. Although their research methods differed from those reported here, Holler and Conaway (1979) measured a higher fecundity rate (5.7 litters per year) for marsh rabbits (S. p paludicola) living in southern Florida.



Mortality

Fiftythree rabbit kittens observed in the nests, but only 15 young rabbits were trapped on the grids. The survival rates (from one developmental stage to the next) at the 5 main study sites ranged from 13 to 50%. Forty-one individuals (26 male, 15 female) were trapped and collared at the 5 sites on Boca Chica Key. Twenty seven (19 male, 8 female) died during the study. Because of these small sample sizes in some of the demographic groups statistical tests were not used. Percent mortality was highest for the second year adults (X = 90%) and lowest for the juveniles (X = 25%). Male mortality

(%) was higher than female mortality for each age class (Figure 3.5). None of the females in the study died during their subadult period, but nearly half of the





38

subadult males were killed. One female was caught as a subadult and then followed until the study ended, establishing a maximum longevity of 3 years for this study.

Cause of mortality was determined for 24 rabbits (Figure 3.6). None of the juvenile mortalities were included because their bodies were never found. Presence of blood and fur on the break-away collar and subsequent disappearance from the trapping grid were assumed to mean the animals had been killed. It was possible to determine the cause of mortality for all of the subadults (all male) and 19 of the adults. Only 1 of the adults disappeared during the study.

Cause of mortality was organized into 6 classes: cat/raccoon, vehicle, rattlesnake, raccoon, cat and poaching. The cat/raccoon class was used for mortalities where it was not possible to distinguish between predation by a feral cat or raccoon. Sign from both predators was abundant at the kill sites. In nearly all of the kills the rabbit was dragged and partially buried, indicating cat predation. In 3 cases the cat or raccoon was seen with the rabbit, and it was assumed that the predator had killed the animal. In addition, Arnim Sheutz (the Natural Resource Manager for the Navy on Boca Chica Key, pers. comm.) reported an observation oft a Bald Eagle grasping a medium-sized marsh rabbit.





39

Full necropsies were only possible on 2 of the rabbits. The remainder of the carcasses were entirely eaten, eviscerated, or decomposed. Neither of the necropsies revealed any sign of disease.

The most subadult and adult mortalities were caused by vehicles (Figure

3.6). It was followed closely by the number of individuals killed by feral cats and raccoons. Three rabbits were eaten by rattlesnakes (the radio-collar transmissions were traced to the rattlesnake) and one rabbit was shot. Both sexes appeared to be susceptible to predation by cats and raccoons, but more males were killed by vehicles. Seven males were killed by vehicles on the road, 4 of them subadult males. The only female road-kill was actually killed in the marsh adjacent to the road. The vehicle had apparently been driven through the habitat. Mortality on and off the base on Boca Chica Key was similar. Road-kills, feral cat predation, and raccoon predation occurred on both on and off the base on Boca Chica Key. All of the predation by rattlesnakes and the poaching occurred on Navy land off the Boca Chica base.



Discussion

The high recapture rate of adult marsh rabbits indicates that most adult rabbits living on a grid were probably captured during the study. The lower trappability percentages for juveniles and female subadults suggests that some of these individuals might have been missed. Because of this difference in





40

trappability, most notably in the female subadults, the apparent male bias in the sex ratio may be due to differences in the trappability. Both male and female biased sex ratios are common in the literature (Chapman et al. 1982). Speculation about the role of biased sex ratios of litter and of adult rabbits has been made in relation to mating systems (Chapman et al. 1977), population density, precipitation, and success of hunters (Edwards et al. 1981).

Average productivity for female Lower Keys marsh rabbit (6.36

young/year) was slightly lower than for marsh rabbits in southern Florida (Holler and Conaway 1979). However, the productivity measure in this study only accounts for young seen in the nest during a 2-week period after birth. It is possible more young were born and not seen in the nest or that some young died before, during, or shortly after birth. Productivity estimates are substantially lower than for eastern cottontail rabbits (Chapman et al. 1982). Lord (1960, 1963) noted that a decrease in the litter size of cottontails was correlated with a decrease in latitude. As the size of the litter decreases, the potential length of the breeding season increases. The breeding season and litter size of the Lower Keys marsh rabbit are consistent with this general observation.

Breeding was year-round in the Lower Keys. Cottontail breeding and reproduction are generally tied to temperature and precipitation. In the Keys, these climatic changes may be more subtle. Although the Florida Keys lie a





41

few degrees north of the Tropic of Cancer, the climate can be considered tropical (Walter et al. 1985). The temperature in the Lower Keys has not fallen below freezing in historic times. Rainfall is seasonal, but the Keys do receive one-third of their rainfall during the "dry" season. Slight variations in marsh rabbit reproduction might be related to inter-year variation in precipitation.

Marsh rabbit mortality was high for nearly every age class and sex. The estimate of nestling mortality, however, may be negatively biased. The probability of catching a rabbit during the first 3 months of its life is unknown and it is possible that some young rabbits survived the first 3 months but were killed later as juveniles. It seems unlikely that the young rabbits left the habitat patch. Cottontail rabbits are extremely altricial, and very young rabbits appear to be only capable of moving short distances, especially without cover.

Subadult males appeared to have a particularly high mortality rate,

mainly due to the number killed by vehicles. Second-year adults also have a high mortality rate, indicating that adult marsh rabbits may become more vulnerable to predation as they age. In general, fewer females were killed than males. This difference in mortality data may be due to the male-skewed sex ratio of the radio-collared subadult rabbits.

Vehicles killed nearly a quarter of radio-collared marsh rabbits in this

study, including the subadult dispersing males. Because marsh rabbits are most active from dusk until dawn, night-time road traffic appears to be a threat to





42

marsh rabbit survival. Predation by cats has not previously been reported for marsh rabbits, but cats were a major predator of adult and juvenile rabbits of other species in several other studies (Fitzgerald and Karl 1979, Jones and Coman 1981, Liberg 1985). The cat population of the Lower Keys is large and includes feral and house-based cats.



Genetic Threats to Persistence

The Lower Keys marsh rabbit had a level of variation similar to large populations of cottontails (Scribner and Warren 1986). Inbreeding depression does not appear to be a threat for most of the Lower Keys marsh rabbits. Although past levels of genetic variation are unknown, it does not appear that these marsh rabbits are currently deficient in genetic variation. Future threats may be dependent on the population size and structure.



Demographic Threats to Persistence

Fecundity of the Lower Keys marsh rabbit is lower than other subspecies of marsh rabbits and other species of cottontails. Possible explanations for this lower fecundity include: reabsorption of fetuses (Holler and Conaway 1979), insufficient food or nutrients for reproduction (Cheeke 1987), adaptive pressures that have resulted in longer periods between pregnancies, and lack of suitable mates at the habitat patch.





43

Although the other hypotheses cannot be eliminated, it is highly

probable that some females were without mating opportunities during part of the study. The low population density, combined with stochastic variation in the sex ratio of the adults at a habitat patch, left at least 2 females (sites 8 and 9, trapping period one) without mates. These females did not produce a litter during this period. The ability of males and females to travel between habitat patches will be examined in Chapter 4. If these habitat patches are separate populations, it appears that stochastic demographic variation can have a large impact on the intrinsic rate of growth of the population (r).

This lower fecundity was matched with a mortality rate higher than most non-hunted populations of wild rabbits (Trent and Rongstad 1974, Chapman et al. 1982). The majority of the mortality was from anthropogenic sources (vehicles, house cats), that are not generally density-dependent.





44

Table 3.1--Number of Lower Keys marsh rabbits (Sylvilagus palustris hefneri) trapped at each site in the Lower Keys of Florida from June 1991 to May 1993.



Site Ownership Male Female Total


3 Navy 1 0 1 4 Navy 2 1 3 7a Navy 4 3 7 8a Navy 8 4 12 9a Navy 6 3 9 10a Navy 4 1 5 13a Navy 4 4 8 16 Navy 1 0 1 19b Navy 2 1 2 22 Private 1 2 3 34 USFWSc 1 2 3 a One of the 5 main study sites.
b The Navy-owned Saddlebunch site. c United States Fish and Wildlife Service.








Table 3.2--Correlation coefficients between body mass, body, ear, and right foot length and rabbit age for three Lower Keys marsh rabbits (Sylvilagus palustris hefneri). In parenthesis are the number of times each rabbit was examined.


Trait Female #1 (N=4) Female #2 (N=5) Male #1 (N=5) R2 F P R2 F P R2 F P Body mass 0.97 70.1 0.01 0.97 84.1 0.01 0.90 18.6 0.05 Body length 0.93 27.0 0.04 0.94 49.7 0.01 0.94 29.5 0.03 Ear length 0.80 8.2 0.10 0.96 71.8 0.01 0.91 19.4 0.05 Right foot length 0.82 8.9 0.10 0.93 41.1 0.01 0.92 22.6 0.04





46

Table 3.3--The number of sexually mature Lower Keys marsh rabbits (Sylvilagus palustris hefneri) at the 5 main study sites on Boca Chica Key. Session Site

7 8 9 10 13 Total


1 Male 0 0 0 1 1 2

Female 0 1 1 1 2 5 2 Male 0 1 0 1 1 3

Female 0 1 1 1 2 5 3 Male 2 2 1 1 1 7

Female 2 1 1 1 2 7 4 Male 2 1 1 1 1 6

Female 2 2 1 1 1 7 5 Male 2 1 3 1 2 9

Female 1 1 1 0 1 4 6 Male 1 2 1 1 1 6

Female 1 2 2 0 1 6 7 Male 1 1 1 1 1 5

Female 0 1 1 0 1 3 8 Male 1 0 1 1 1 4

Female 1 0 2 0 2 5





47


Table 3.4--Allele frequencies for all variable locia among 5 marsh rabbit populations (Sylvilagus palustris hefneri) from the Lower Keys of Florida, June 1992 May 1993. Individuals were trapped on sites 7, 8 and 9 from Boca Chica, sites 10 and 13 from Geiger Key (the southeastern portion of Boca Chica), site 30 on Saddlebunch, site 33 on Sugarloaf, and site 53 on Big Pine.



Site
(N)

Locus Allele Boca Chica Geiger Saddlebunch Sugarloaf Big Pine
(9) (5) (2) (2) (1)


EST-1 A 0.389 0.200 1.000 0.250
B 0.611 0.800 0.750
C 1.000

PGM-1 A 0.778 0.500 1.000 0.500 1.000
B 0.222 0.500 0.500



aThe following loci were assayed but were monomorphic: GOT, GPI, Hb, ICD-1, LDH-1, MDH-1, MPI, PEP-i, SOD-1.







Table 3.5--The sum of all marsh rabbits (Sylvilagus palustris hefneri) trapped at the 5 main sites on Boca Chica during the 8 trapping sessions (June 1991 to May 1993), and the proportion of these rabbits that were trapped twice during a 10-day trapping session. The G2 for the "number" indicates a test to determine if the sex ratio of the trapped rabbits differed from 1:1. TheG2 for the "trappability" indiciates a test to determine if the trappability differed between sexes.





Age Male Female Male:Female

Number Trappability Number Trappability Number G2 Trappability G2 Juvenile 8 50% 6 67% 1.0 11.6* Subadult 14 71% 4 25% 14.3* 59.6** Adult 42 83% 42 88% 0.0 0.6


* P < 0.05
** P <0.001























SN meters
0 400





Figure 3.1--Five main habitat patches on Boca Chica used for trapping and radio-collaring of Lower Keys marsh rabbits (Sylvilagus palustris hefneri). The solid lines are runways and taxiways. The dashed line are roads.
r~rr 0/







1400
Male 1200 U Female Male 1000

800


0 600

400 200


0
0 2 4 6 8 10 12 14 Month


Figure 3.2--Growth curves for 3 (1 male and 2 female) Lower Keys marsh rabbits (Sylvilagus palustris hefneri).





51







20
4

18" 16


o 14

E
= 12Z

10



1 2 3 4 5 6 7 8
May 1991 May 1992 Session








Figure 3.3--The number of Lower Keys marsh rabbits (Sylvilagus palustris hefneri) caught during each trap session at the 5 main sites.





52










60 9

8
50


..J 40 8
C 11
.* 6 9 9
10 7
t) 30


00

0 9
o 10


0 7
1 2 3 4 5 6 7 8 9 10 11 12

Month







Figure 3.4-Proportion (bar) and number (above bar) of Lower Keys marsh rabbits (Sylvilagus palustris hefneri) that produced a litter during each month. Data from 2 years were combined and averaged. (January = month 1, December monthl2).





53








100 Male
WM Female
5
80


60
2 10 13


40- 8
7
5
20

0- 5

Juv. Sub. Adult-1 Adult-2 Age Class










Figure 3.5--Percent mortality of Lower Keys marsh rabbits (Sylvilagus palustris hefneri) by sex and age class. Sample size of each age/sex cohort is indicated above each bar.





54







10 Male

cn 9 l Female
(D
8

07 0 6

.05

Z 4

3






Vehicle Cat Cat/Rac Snake Rac Poach Cause of Mortality









Figure 3.6--Cause of death for 24 Lower Keys marsh rabbits (Sylvilagus palustris hefneri). (Vehicle car, truck, or plane; Cat = domestic cat; Cat/Rac = unable to distinguish if predator was a cat or a raccoon; Snake = eastern diamond back rattlesnake; Rac = raccoon; Poach = death by gun)













CHAPTER 4
SPATIAL ORGANIZATION (INTER-PATCH MOVEMENTS)


Introduction

To apply knowledge about the Lower Keys marsh rabbit's population biology in a PVA, it is necessary to define the unit of individuals that comprises a local population. A local population can be loosely defined as a set of individuals that has a high probability of interaction (Hanski and Gilpin 1991). Local populations may be part of a greater metapopulation where interaction occurs via individuals moving among populations (Levins 1970).

The unit of the local population can be determined by studying the

spatial structure and movements of individuals at different spatial scales. An individual's home range, spacing behavior, and movements (including dispersal) will determine the spatial scale of interaction with others. These behaviors may vary among individuals of different sexes and ages.

The home range of an individual is the area used for foraging, mating

and caring for young (Burt 1943, White and Garrott 1990). It does not include areas used in migrations, or other occasional round-trip excursions. Home range is calculated using a variety of methods that take periodic readings on an individual's location.


55





56

Home-range size varies greatly among individuals, populations and species of the genus Sylvilagus (see Chapman et al. 1982); Only 1 study has attempted to estimate home-range size for the marsh rabbit (S. palustris). Blair (1936) estimated a "linear home range" based on a maximum home range width of 183 m from trapping data.

It is generally assumed that most cottontails are not territorial (Chapman et al. 1982). This assumption is based largely on studies of the eastern cottontail (S. floridanus; Haugen 1942; Chapman and Trethewey 1972, Dixon et al. 1981) and the brush rabbit (S. bachmani; Shields 1960, Chapman 1971). Yet, Trent and Rongstad (1974) reported that female eastern cottontails did not have overlapping home ranges during the breeding season. Jurewicz et al. (1981) found small amounts of overlap in the nocturnal home ranges of breeding female eastern cottontails, and suggested that a spacing mechanism was in operation. Male swamp rabbits (S. aquaticus) have exhibited linear dominance hierarchies involving non-overlapping, defended home ranges in several studies (Marsden and Holler 1964, Sorenson et al. 1968, Holler and Sorenson 1969). Additionally, Kjolhaug and Woolf (1988) found no overlap between individuals of the same sex in both male and female swamp rabbits. Information on the spacing behavior of the other 10 species of cottontails (genus Svlvilagus) is lacking.





57

In addition to home-range size and spacing, it is also important to

determine which habitat features a rabbit incorporates into its home range and which features act as barriers. Rabbits generally move while under vegetative cover (Chapman 1971, Trent and Rongstad 1974), probably to avoid predation. Large open areas are generally seen as barriers or at least deterrents to rabbit movements. Although the response of cottontail rabbits to roads and vehicular traffic has not been directly studied, field studies have found that improved roads (2-lane and larger) can be a barrier to small mammal and hare movements (Oxley et al. 1974, Mader 1984). Other possible barriers to rabbit movements include disturbed or developed areas with no natural vegetation, vegetated areas with insufficient cover, and open bodies of water. The marsh rabbit and the swamp rabbit are both known to be good swimmers (Svihla 1929, Tomkins 1935), although the frequency of swimming is not known.

While most rabbits stay within their home range and make few longdistance movements, young, subadult male rabbits have been documented traveling longer distances (Shields 1960, Chapman and Trethewey 1972). It is believed that these males are natal dispersers. Natal dispersal is male-biased in most polygynous mammals (Dobson and Jones 1985). Determining which sex disperses in a species and what factors motivate these movements are essential to understand a species' spatial organization.





58

Currently there is debate over the causes of natal dispersal (Dobson and Jones 1985, Pusey 1987, Wolff et al. 1991). Most research has been directed on determining the costs and benefits of dispersal to the dispersing individual. Three main hypothesis have been proposed: inbreeding avoidance, competition for mates, and competition for environmental resources (Packer 1979, Greenwood 1980, Dobson 1982, Greenwood and Harvey 1982). Different predictions for each hypothesis have been made, but research on a diverse sample of mammalian species has not been able to falsify any of the hypotheses. Two of the hypotheses, inbreeding avoidance and competition for mates are not mutually exclusive in their predictions. It has been proposed that the causal agents for dispersal may differ among species (Dobson and Jones 1985).

Recently, an alternative hypothesis has been proposed. Anderson

(1989) has suggested that the focus should be on the fitness the resident (nondispersing) adults receive by forcing the young to leave. In most polygynous mammals generally the father would receive the greatest benefit by coercing his son to leave. Although some evidence of dominants evicting subordinates exists in the literature (Moore and Ali 1984), other hypotheses can not be eliminated (Pusey 1987, Wolff et al. 1991).

Determining what direction and what type of habitat dispersing

individuals will traverse are especially important in fragmented environments.





59

Two general theories have been proposed to predict the direct an individual will leave a patch. The random walk theory (Berg 1983) predicts that individuals will disperse across random locations along the edge of a patch, irrespective of habitat features. Patch geometry models (Stamps et al. 1987, Buechner 1989) predict that individuals will exit patches at more "permeable" areas. A landscape corridor may increase the patch edge permeability by extending patch habitat (La Polla and Barrett 1993), and allowing individuals to move from one habitat patch to another. The geometric and habitat features that constitute a "corridor" from the animal's perspective must be determined.

The objective of this part of the study was to use the information about the spatial organization of S.p. hefneri and movements to determine how they use the fragmented habitat in the Keys. Evidence supporting the hypothesized patterns of spatial use will be evaluated. Three possible scenarios of habitat use will be considered:

1. The Lower Keys marsh rabbit is confined to one habitat patch and is incapable of moving between patches. All home ranges occur strictly within the patch. Dispersal is within the patch, does not occur, or is unsuccessful (rabbit leaves a patch but does not reach another patch and dies or fails to reproduce). This will be called a "relictual" (Berry 1986) population.

2. Lower Keys marsh rabbits spend most of their lifetimes in a patch but are capable of moving between patches. All home ranges occur strictly





60

within the patch and individuals only interact with others living at the same patch. There may be occasional movements between patches and/or some of the dispersing individuals will successfully move to new patches. The areas surrounding the patches act as a barrier to most movements and is not included in the home range of the individuals. These patches contain sub or local populations that exist within a greater metapopulation (Levins 1970). These patches are not necessarily genetically distinct but are demographically isolated. Matings between individuals at different patches does not occur.

3. Lower Keys marsh rabbits move regularly between patches and may use several patches at once. The home ranges encompass several patches at one period of time and individuals interact with others from other patches regularly. All of the patches are part of one large population, which inhabits a highly fragmented environment.



Methods

Marsh rabbits were trapped at the 5 main sites on Boca Chica. Trapping occurred twice during the wet season (May October), and twice during the dry season (November April) from May 1991 to May 1993 (8 trapping sessions). Trapping grids were placed on each site, using unbaited collapsible National live traps (80 x 30 x 30 cm), placed in a 6 x 6 array, spaced approximately 25 m apart. Each trapping session consisted of 5 nights where





61

the traps were open, 2 nights with the traps closed and another 5 nights with traps open. Traps were checked twice daily, once in the morning and once in the evening and were covered in burlap for shade. All rabbits caught were sexed, weighed, and tagged (Monel no. 3, National Band and Tag, Newport, KY).



Radio-telemetry

Each rabbit weighing >1,000 grams was fitted with a radio-collar and a transmitter with an estimated 10-month operational life (Telonics, Inc., Mesa, AZ). Smaller rabbits (300-1,000 g) were fitted with a similar radio-collar, but with a velcro break-away device added to allow the animals to lose the equipment as their bodies grew larger. Collars were replaced as the animals aged.

Collared rabbits were located on separate days 3 times a week, once in the early morning (7-9 a.m.), once at mid-day (11 a.m. 1 p.m.), and once in the evening (4-6 p.m.). Signals were followed until the animal could be seen or the exact location was found. All locations were made >24 hours apart to ensure independence of observations. Locations were plotted on 1:2400 aerial maps with 10-m2 grid overlays. All road and water crossings were recorded and the distance of the water crossing was measured.





62

Home Range

All juvenile rabbits that were followed for more than 1 month (12

locations) were used for analysis of home ranges. The 12-location criterion was chosen because few juveniles were followed for more than 1 month. Attempts to determine a minimum number of locations for the adult homerange analysis by fitting the data to a negative exponential equation failed. None of the individuals had home-range sizes that reached an asymptote, despite the fact that some rabbits were followed their entire lives (Figures 4.1 and 4.2). Therefore, a minimum number of 30 locations (2.5 months) was chosen based on other studies of similar cottontail rabbits (Dixon and Chapman 1980; Kjolhaug and Woolf 1988). Computer program HOME RANGE (Ackerman et al. 1990) was used to determine sizes of minimum home ranges and to plot movements.

Because of the nature of Lower Keys topography and patterns of home ranges of marsh rabbits, several common methods of home-range estimation were excluded. The minimum convex polygon method (Hayne 1949) was rejected because it assumes that all areas within the perimeter of the outermost locations are used by the animal (White and Garrott 1990). Most of the marsh rabbits' home ranges contain unused areas, such as bodies of water, roads, or other man-made structures. One of the most statistically rigorous methods of calculating home-range size is the 95% probability ellipse (Jennrich and Turner





63

1969). It assumes that there is always only one center of activity (Harris et al. 1990), an assumption that was frequently violated in this study.

Two related methods that have worked well for analysis of rabbit home ranges are the harmonic-mean and core-area methods (Dixon and Chapman 1980). The harmonic mean is a nonparametric method based on a volume under a fitted 3-dimensional use distribution. It relates well to the actual distribution of locations. The core area (as calculated by program HOME RANGE), is the maximum area where the observed use distribution exceeds a uniform use distribution. Thus, it shows areas of particularly high home-range use and is relatively unaffected by outliers and sample size (Harris et al. 1990).

Seasonal differences in the home range size (95% harmonic mean and core area) were compared for each individual that was located at least 30 times in each season. The dry season locations occurred between November and April; the wet season was from May to October. Shifts in home range between season were also examined. Arithmetic centers of the dry and wet season home range of each individual were determined and the distance these centers shifted between seasons was calculated.



Spacing Behavior

The amount of overlap between the 95% harmonic mean and core areas of each individual was compared to other individuals living in the same habitat





64

patch. The first set of comparisons looked at all of the individuals of the same sex present at a patch during the same period of time. Percent overlap was determined by plotting all the home range contours together using Program HOME RANGE, and overlapping a 10 m grid. For each pair of individuals the amount of overlap was calculated as a percentage of the smaller home range.

This percent of overlap was compared to overlaps of same-sexed

individuals living at the same patch but at different times, and pairs of opposite sexed individuals. A t-test was used for normally distributed data and when the variances between the 2 groups were equal. A Mann-Whitney U test was used when these conditions were not met.



Dispersal

If an animal appeared to be making a long-distance movement, it was located at least once every day. These locations were not used in any home range calculations. A rabbit that made a one-way long-distance movement was said to have dispersed. The criterion for dispersal as a movement was the diameter of the average marsh rabbit home range (Ribble 1992).

Dispersing animals were classified by sex and age class. Locations from the daily radio-telemetry session were plotted and straight lines were drawn between them to approximate the minimum distance traveled. Data on the demography of dispersers was used to test hypotheses on the cause of





65

dispersal. A hierarchical hypothesis design was used (Table 4.1) to attempt to narrow the potential cause of dispersal.

First, a binomial probability test was used to determine if the sex ratio of the dispersers was significantly different from unity. If this sex-ratio showed a significant bias, then trapping and radio-telemetry data were examined to determine if other adult males inhabited the disperser's natal patch. Trapping records of the adult males were further examined to determine if these males could potentially be the father of the dispersers. Potential fatherhood was assumed if the adult male was present at the disperser's natal patch around the time when the dispersers were believed to have been conceived.



Corridor Use

To determine if dispersing individuals were randomly leaving a patch or if they were influenced by potential corridors, paths of all individuals making one-way long distance movements were plotted. The proportion of each mover's trip (measured in meters) that covered different habitat types and the number of roads, runways, and bodies of water they crossed was measured. Table 4.2 lists the habitat types and features compared.

Using this distance as a radius with the arithmetic centerpoint of the

natal home range as the center, a circle was drawn for each moving individual





66

(Figure 4.3). Aerial photographs were digitized (ARC/INFO 1990) to determine the amount of each habitat type and length of all roads, runways, and linear bodies of water was determined within this circle. The Johnson method (Johnson 1980) was used to determine if habitat use significantly deviated from the amount of each type of habitat available. A Waller-Duncan multiple comparison procedure (Waller and Duncan 1969) was used to determine differentially selected habitats. The Johnson method was chosen because it looked for overall differences in habitat use, not at individual habitat-use differences (Alldredge and Ratti 1992).



Results

A total of 54 Lower Keys marsh rabbits were caught and examined

during this study. Forty-three were caught on the 6 main grid sites (41 on Boca Chica, 2 on Saddlebunch), 5 rabbits were caught while trying to recapture collared dispersed rabbits, and 6 were caught on other sites in the Lower Keys. Rabbits were only examined once per trapping period. Data from the 54 rabbits was recorded 130 times during eight trapping sessions (May 1991 May 1993).

Forty-three (28 male, 15 female) of the rabbits caught were fitted with radio-collars, and 7 of these were juveniles. More juveniles were caught but at the time of their capture the break-away collars were not available. Forty-one





67

of the radio-collared rabbits were on Boca Chica Key and 2 were on Saddlebunch Key.



Home Range

Sufficient data were available for seven juveniles, 13 adult males, and 10 adult females for the home-range analysis (Table 4.3). Adult males were followed an average of>8 months (98 locations) and adult females were followed an average of>10 months (124 locations). Comparisons using the Wilcoxon test for differences between adult males and females did not show significant differences between the average distance moved between locations (Z = -1.71, P = 0.09), the 95% harmonic-mean home range (Z = -1.33, P =

0.18), or the core area (Z = -1.21, P =0.23). Resident adult males and females were similar with respect to their home-range shape and size (Figure 4.4). Both sexes appeared to establish a permanent home range shortly after maturity (9-10 months), and both resided in that area for the duration of their lives. Combining both sexes, the average 95% harmonic mean was 3.96 ha (E = 0.65) and the average core area was 1.21 ha (SE = 0.87). Home-range size estimates for juvenile rabbits were not significantly different from adults, but 3 of the juveniles did not develop core areas. The lack of core area formation may be an attribute of juvenile home ranges or an artifact of the low number of locations used in the analysis.





68

There did not appear to be any significant changes in home-range size or shifts between the dry and wet seasons (Table 4.4, Figure 4.5). Results of a Mann-Whitney U test, used because the variances of the home range size were not equal, failed to find a significant result comparing either the wet and dry 95% harmonic home range (T = 251, P = 0.45) or core areas (T=245.5, P =0.60). None of the shifts in home range between the wet and dry seasons exceeded the radius the marsh rabbit's home range. However, ANOVAs testing the effect of time on home-range size were significant for both the 95% harmonic mean ( = 6.25, df = 22, P = 0.02) and the core area (F = 5.55, df = 22, P = 0.03). Animals increased their home-range size with time (and number of locations), irrespective of season.



Spacing Behavior

Figures 4.6 and 4.7 show the duration of each rabbit's radio-telemetry records and which rabbits were contemporaneous. The average amount of overlap between same-sexed individuals occupying a patch at the same time was significantly less than same-sexed groups occupying a site at different times (Table 4.5). This relationship held true using both the 95% harmonic mean home-range estimate (L= -2.17, df= 12, P = 0.05) and the core area (t = -3.29, df = 12, P = 0.006). When a rabbit died, same-sexed individuals living in adjent areas expanded their home ranges (Figure 4.8). Similarly, the





69

overlap between these contemporary same-sex individuals was significantly less than male/female overlap for both the 95% harmonic mean (T = 79.5, P = 0.05) and core area (C = 45, P = 0.02). In most cases there was little overlap between adults of the same sex (Figure 4.9).

Most males' home ranges overlapped 1 female at a time; 3 males

overlapped 2 females simultaneously. During a portion of their lives, some males did not overlap with a female. Only 1 female overlapped with more than

1 male; this occurred for less than 1 month. Changes in overlap generally occurred when a rabbit died and another rabbit expanded his/her home range.



Home-range Features

All 43 rabbits, including the 7 juveniles, were used in the road- and water-crossing index. Only individuals that were currently in their natal or adult home ranges were used. Data from individual making long-distance movements were analyzed separately.

Home ranges of most of the marsh rabbits did not incorporate roads or large bodies of water. No water or road crossings were recorded for any of the juvenile rabbits. None of the rabbits crossed the major highway in the Keys (US-1). Only 1 adult female and 1 adult male crossed paved roads during the study. Rabbit A53F crossed a 2-lane several times. This same rabbit was the only adult female to cross a body of water; she regularly swam in shallow





70

water between mangrove prop roots while being radio-tracked. Male A5 1M also crossed a 2-lane road four times. Neither rabbit crossed the road to visit another "patch". Female A53F crossed the road to swim in the water adjacent to her home "patch"; male A5 1M crossed the road when construction was being completed in the corner of his home patch.



Dispersal

Seventeen rabbits (11 male, 6 female, all subadults) made permanent one-way movements (Table 4.6). The minimum dispersal distance was calculated using the diameter of the combined adult core-area size. Assuming a circular home-range shape, the average diameter was 124 m.

Eleven of the subadults made movements >124 m. Ten of the dispersers were male and one was female. Only 1 of the males failed to meet the criterion for dispersal, and whereas 5 of the females did not exceed the distance. In general, the males made long-distance movements far in excess of the criterion, including a 3-day movement that placed a male over 2 km from his natal range. Most of the females settled near their natal ranges, including the female that was classified as a disperser.

The sex ratio of the dispersers was significantly male biased (binomial P <0.001). Six of the males that dispersed left patches where there was another adult male present, but 4 left patches where there were no adult males. Five of





71

the males that dispersed left patches were the male present could have been their father, but the other 5 left patches were there either were no males present, or a male too young to be their parent was present. The 1 subadult male that did not disperse occupied a patch where there were no other adult males.



Corridor Use

Subadult rabbits traveled through a variety of habitats between their

natal and permanent home ranges. Three rabbits crossed a dirt road, 7 crossed a 2-lane road and 3 rabbits were observed crossing taxiways and runways. None of the rabbits crossed the 4-lane highway (US 1), but none of the dispersal radii encompassed or were adjacent to the highway. Two rabbits swam across ditches, 1 across a canal, and 1 crossed a (12 m) body of water.

In general, most of the subadult rabbits traveled through areas with dense ground cover. These marsh rabbits were recorded traveling through mangroves, upland hardwood hammocks, and in the vegetation between the shoulder of the road and the water. The narrowest strip of plant cover used (corridor) by a dispersing marsh rabbit was 3-5 m wide. The Johnson test that compared the amount of each habitat traveled through compared to its use was significant (F = 6.62, df = 3,14, P = 0.005). A Waller-Duncan comparison found that rabbits used areas of mangrove, hardwood hammock, and transition





72

zone significantly more than expected and used disturbed areas significantly less.



Discussion

The data suggest that members of S. p. hefneri spend most of their lives in 1 patch, but can move to other patches. Marsh rabbits are born in a patch of transition-zone habitat and remain there until they reach sexual maturity. At sexual maturity most rabbits make a relatively long, one-way movement. Male marsh rabbits may move a great distance away from their natal range; females are more likely to remain in the same patches where they were born. If a patch is relatively small, this movement may mean leaving the habitat patch. When these subadult marsh rabbits leave their natal ranges, they establish adult home ranges that they maintain for their lifetime. These adult home ranges do not incorporate any roads (2-lane and larger) bodies of water, or other types of habitat. All mating appears to occur within the same habitat patch. These results are consistent with the predictions of the second hypothesis, that S. p. hefneri exists in a metapopulation (Levins 1970, Hanski 1991).

The Lower Keys marsh rabbit's ability to exist as a metapopulation and disperse over relatively long distances may be a product of inhabiting a naturally patchy environment (Opdam 1991, Thomas 1994). Species that exist in temporally or spatially patchy environments must be able to deal with





73

isolation at a more regional scale than those species that occupy more permanent, interconnected habitats. Marsh rabbits throughout their range occupy upper marshes along the coasts and interior wetlands, much of which is patchily distributed. Before European colonization in the Keys, the transition zone was probably more contiguous. Most of the habitat types in the Keys exist in contiguous, concentric rings; habitat type is largely determined by elevation.



Spatial Organization

Marsh rabbit home-range size was well within the range for cottontail species (Chapman et al. 1982). The 95% harmonic mean estimates represent a conservative estimate of the amount of habitat used; the core area estimate delineated the area needed for more intense use. Despite strong seasonally of the rainfall in the Keys, home ranges were consistent throughout the year.

Similar to the eastern cottontail and the swamp rabbit, there is little overlap in home ranges of members of the same sex. Marsh rabbits may be territorial within their sex. Most current definitions of territoriality specify that an area must be used exclusively (at least within the sex) and that it must be actively defended (Eisenberg 1981, Begon et al. 1990). To be truly territorial, marsh rabbits would have to defend their home ranges. Territorial defense is difficult to observe in the field, especially in a secretive, crepuscular species





74

like the marsh rabbit. Some circumstantial evidence of physical defense was apparent in the males; 76% of the subadult and adult males had scratch marks and scars on their face, ears and back of their heads. Boxing, scratching, and paw-raking with the feet has been observed in other species of cottontails (Marsden and Holler 1964) and may be responsible for the scratch marks. Generally these behaviors are related to competition for mates (Eisenberg 1981). Female marsh rabbits showed no signs of scratch marks. It is not readily apparent why it would be advantageous for the females to be territorial, however there is some evidence that nesting sites may be limited (see Natality chapter 3). Fecal pellet marking is well-developed in other rabbit species (Teft and Chapman 1987) and may be used by male and female marsh rabbits to mark the boundaries of their home ranges.

Both males and females increased their home ranges following the death of a conspecific. Over the lifetime of a rabbit this was seen as an increase in home range with age. This phenomenon provides additional support that some spacing behavior is partially determining home-range size. This spacing behavior may have an impact on population density and total abundance of rabbits in a habitat patch.

Most of these home ranges consisted almost exclusively of transitionzone habitat. Physical features such as canals, roads, and runways were rarely crossed by any of the adult marsh rabbits, despite the proximity of the features.





75

The definition of a habitat patch put forth in the methods section appears to be valid.

Subadult marsh rabbits were more likely to cross these barriers and use alternate types of habitat. However, their movements did appear to be influenced by the surrounding habitat matrix. Concordant with other studies on rabbit movements (Chapman 1971, Trent and Rongstad 1974), marsh rabbits were more likely to cross the more densely vegetated native habitats (transition zone, hardwood hammock, and mangrove) than the more open, disturbed areas. Presence of these habitats around the natal patch appeared to facilitate movement between patches of transition zone habitat, acting as corridors in the highly fragmented landscape (Wilson and Willis 1975). These results support the patch geometry models (Buechner 1987, Stamps et al. 1987) rather than the random walk theory (Berg 1983).

Dispersal has not been well studied in other species of cottontails. Results from the telemetry indicate that most long-distance dispersers were young males, similar to most polygynous mammalian species (Dobson 1982). The small patch size and low density of rabbits in this study provided a unique opportunity to investigate the cause of dispersal. Because some of the males dispersed despite the lack of other adult males at their natal patch, the competition-for-mates hypothesis cannot entirely explain all of the movements. Similarly, lack of potential fathers at most of the natal patches excludes the





76

resident fitness hypothesis as being the sole explanation. Only the inbreeding avoidance hypothesis is consistent with all of the data collected. It is possible that dispersal may be motivated by several factors (Dobson and Jones 1985) and that there may be variance between individuals. This interpretation is based only on current dispersal patterns. Current dispersal behavior may be a response to population structure in evolutionary time. The marsh rabbit population structure has probably undergone dramatic changes since the colonization of man in the Keys.



Conclusions

These results imply that the S. p. hefneri should be managed as a metapopulation. Each local population is socially isolated from the other populations. Interchange generally occurs by movement of subadult males and this movement is facilitated by the occurrence of habitat corridors. The impact these conclusions has on the persistence of S..p. hefneri will be addressed in chapter 6.







Table 4.1--Hierarchical tests and predictions of the proximate cause of dispersal.



Prediction Cause of Dispersal



I. An equal number of males and females disperse Competition for resources II. Only one sex disperses.

A. The dispersers are all female. Inbreeding avoidance

B. The dispersers are all male.

1. A male disperses only if another adult male is present
at the habitat patch. Competition for mates

2. A male disperses only if other males are present and
could potentially be their father. Resident fitness hypothesis

3. All males disperse regardless of the presence or
absence of other males. Inbreeding avoidance





78

Table 4.2--Habitat types and possible physical barriers encompassed by Lower Keys marsh rabbit (Sylvilagus palustris hefneri) dispersal movements.



Feature Mode of measurement

Unimproved road (dirt, gravel) meters of length 2-land road meters of length 4-land highway meters of length

Mosquito drainage ditch (water < Im) meters of length Canal (1 10 m) meters of length Major body of water (water > O1m) meters of length

Disturbed or barren habitat ha Mangrove ha Hardwood hammock ha Pineland ha





79


Table 4.3-Home-range data on Lower Keys marsh rabbits (Sylvilagus palustris hefneri) observed for >1 month for juveniles, >30 locations for adults.

Distance 95%
between harmonic Core
Number of consecutive mean home area Rabbit Classification locations locations (m) range (ha) (ha)

J55M Juvenile 30 69.58 13.83 4.44 J99M Juvenile 18 47.40 2.49 0.64 J101F Juvenile 16 22.99 0.10 J102F Juvenile 21 25.81 1.11 0.38 J103M Juvenile 14 22.39 0.30 J199M Juvenile 14 40.60 0.81 x 38.13 3.11 1.82 SE (7.55) (2.17) (1.31)

A51M Adult male 122 48.34 11.94 3.15 A55M Adult male 70 30.81 2.73 0.89 A59M Adult male 79 28.71 2.13 0.93 A63M Adult male 217 25.33 5.81 1.58 A64M Adult male 61 17.87 0.36 0.02 A67M Adult male 99 42.10 3.47 1.10 A68M Adult male 142 35.30 3.63 1.24 A69M Adult male 138 50.88 7.61 2.60 A70M Adult male 88 36.85 2.99 0.86 A71M Adult male 67 37.47 3.18 1.22 A73M Adult male 60 38.24 4.23 1.20 A75M Adult male 74 28.07 4.35 1.18 A85M Adult male 54 52.36 2.81 1.03 x 36.33 4.25 1.31 SE (2.85) (0.80) (0.22)

A50F Adult female 252 28.14 6.52 2.00 A52F Adult female 127 34.31 10.32 3.17 A53F Adult female 201 59.76 8.60 2.09 A57F Adult female 220 21.90 1.67 0.53 A58F Adult female 90 20.90 0.72 0.24 A72F Adult female 107 26.15 2.49 1.02 A74F Adult female 75 27.89 1.27 0.46 A76F Adult female 81 22.12 1.53 0.53 A86F Adult female 44 32.68 1.72 0.53 A176F Adult female 45 30.52 0.98 0.25 x 30.44 3.58 1.08 SE (3.56) (1.12) (0.32)







Table 4.4--A comparison between the wet and dry season home ranges (ha.) of Lower Keys marsh rabbits (Sylvilagus palustris hefneri). The distance between the centers of each home range is compared to the radius of the 95% harmonic mean and core area estimates.

Wet season Dry season
ID #locals 95% Core #locals 95% Core 95% Core Meters Core radius harmonic area harmonic area harmonic radius moved meters moved mean mean radius


A70M 36 0.81 0.24 52 2.79 0.83 50.79 27.65 15.52 12.13 A71M 36 2.99 0.90 31 1.31 0.54 97.58 53.54 24.50 29.04 A68M 64 3.79 0.90 78 1.37 0.47 109.86 53.54 38.11 15.43 A67M 45 4.03 0.96 54 0.56 0.14 113.29 55.29 37.00 18.29 A63M 124 2.28 0.57 93 5.74 1.60 85.21 42.61 14.14 28.47 A69M 60 6.31 2.04 78 3.61 1.42 141.76 80.60 13.01 67.59 A59M 43 1.95 0.87 36 1.48 0.46 78.80 52.64 45.45 7.19 A55M 36 2.16 0.80 34 0.91 0.40 82.94 50.48 24.02 26.46 A51M 73 0.53 0.33 49 0.45 0.28 41.08 32.42 5.00 27.42 A72F 36 1.15 0.37 71 3.26 1.23 60.52 34.33 39.11 -4.78 A53F 81 10.18 2.51 120 6.50 1.72 180.06 89.41 45.17 44.24 A50F 132 3.49 1.04 120 5.60 1.75 105.43 57.55 12.17 45.38 A58F 48 1.46 0.47 42 0.43 0.20 68.19 38.69 28.07 10.62 A57F 106 1.38 0.45 114 0.78 0.27 66.29 37.86 9.84 28.02 A52F 78 6.21 2.21 49 7.15 2.31 140.63 83.89 4.24 79.65







Table 4.5--The amount of overlap between same and opposite sexed individuals occuping a site during the same time and between same-sexed individuals that occupied the same site during different times.



Same sex-same time Same sex-different time Opposite sex-same time
Pair 95% Core area Pair 95% har. Core area Pair 95% har. Core area


A59M/A55M 34% 12% A51M/A69M 100% 43% A51M/A52F 100% 67% A63M/A67M 30% 3% A55M/A75M 82% 44% A59M/A57F 68% 40% A70M/A71M 85% 42% A59M/A55M 45% 20% A59M/A58F 45% 22% A50F/A74F 9% 2% A59M/A75M 91% 72% A60M/A53F 44% 22% A53F/A76F 100% 21% A53F/A76F 100% 100% A63M/A50F 54% 32% A57F/A58F 16% 0% A57F/A86F 71% 74% A64M/A53F 8% 0% A72F/A65F 0% 0% A58F/A86F 45% 20% A67M/A50F 100% 100% A67M/A74F 85% 21%
A68M/A76F 100% 90%
A70M/A65F 8% 2%
A70M/A72F 82% 48%
A71M/A72F 90% 84%
A75M/A57F 100% 100%
A75M/A86F 100% 86%
Average 39% 11% 76% 53% 70% 51%





82

Table 4.6--Data on radio-collared marsh rabbits that made permanent, one-way movements. Rabbits were caught on Boca Chica, Geiger, and Saddlebunch Keys between June 1991 to May 1993.



Rabbit Classification Distance between natal Body mass at beginning site and last location (m) of movment (g)



A48M Subadult male 550 1,000 A49M Subadult male 550 1,050 A54M Subadult male 920 1,050 A55M Subadult male 1,100 800 A56M Subadult male 1,800 1,000 A60M Subadult male 510 1,000 A66M Subadult male 60 800 A68M Subaudlt male 180 1,000 A69M Subadult male 980 800 A84M Subadult male 2,050 1,050 A197M Subadult male 400 900 x 827 950 SE 191 32

A3F Subadult female 60 850 A65F Subadult female 40 900 A74F Subadult female 70 950 A76F Subadult female 80 900 A86F Subadult female 90 1,000 A176F Subadult female 150 900 x 82 917 SE 15 21





83







1.25 1.00



S0.75S0.50 0.25



0.00 II I
0 10 20 30 40 50 60 70 80 Number of Locations










Figure 4.1--The cumulative core area measurements of 8 radio-collared male Lower Keys marsh rabbits (Sylvilanus palustris hefneri).





84







2.00 1.75 1.50

- 1.252 1.00o 0.75

0.50 0.25 0.00
0 10 20 30 40 50 60 70 80 Number of Locations









Figure 4.2--The cumulative core area measurements of 6 radio-collared female Lower Keys marsh rabbits (Sylvilagus palustris hefneri).








"It. ..:. ..3 !-.








I-
'I /. "I". ....... ;... .. I....:




I-- --.-k.











Figure 4.3--To determine if a dispersing Lower Keys marsh rabbit (Sylvilagus palustris hefneri) used certain habitat as a corridor, the habitat type the rabbit moved through was compared to the habitat available. Using the centerpoint of the rabbit's natal home range and the distance dispersed as a radius, a circle of available habitat was drawn. The arrow
indicates the actual patch the rabbit took.
...........~.I. .~r.
..... .. ... ...I - .. --- --- -J .... ..

..~r- y...


..5:..........
..~:.. ..:k..--- ---- ... ....i..


Fiur 43-T dtemneifadipesigLoerKesmashrbbt Syvlauspautrs enei se erai hbta a




rabbit's natal home range antediac dispers ed aoerKes aas raius, a circle of available) usd ceta habitat wa ran Tearo indicates the actual patch the rabbit took.





86










./'A70M
/




I m









rabbit (S ilaus alustris heferiatsite#7onBocaChicaKe
\ \



A72F
10 m








Figure 4.4--Home ranges using the 95% harmonic mean (outer boundary) and core areas (inner boundary) of a sympatric male and female Lower Keys marsh rabbit (Sylvilagus palustris hefneri) at site #7 on Boca Chica Key.





87










A69M
wet



\ ,, ., idry


I / I / % i







10 m






Figure 4.5--A comparison of the wet and dry season home ranges using the 95% harmonic mean (outer boundary) and the core area (inner boundary) for an adult male at site 10.





88

Site 7

A70M 0 0 0 A72F - -A71M
A65F A182M
A185F

1 2 3 4 5 6 7 8

Site 8

A50F A63M A67M
A74F-


1 2 3 4 5 6 7 8 Site 10

A52F -A51M --- -A69M - -1 2 3 4 5 6 7 8 Trap Session

Figure 4.6--The duration that radio-collared Lower Keys marsh rabbits
(Sylvilagus palustris herneri) were followed at sites #7, #8, and #10.





89


Site 9

A53F 0
A60M A64M *-A66M
A68M - ---A76F 0 0 A176F

1 2 3 4 5 6 7 8


Site 13

A57F
A58F
A59M
A55M
A75M --
A86F
A186M

1 2 3 4 5 6 7 8 Trap Session






Figure 4.7-- The duration that radio-collared Lower Keys marsh rabbits (Sylvilagus palustris herneri) were followed at sites #9 and #13.






90







A. Both males alve (T = 3-6).


A63M
A67M















B. After A67M was killed (T = 6-7).

A63M
A67M


I I










10 m
\ --














Figure 4.8--The home ranges using core area, of 2 sympatric males telemetered between session 3 and 6 (A.) and the expansion of A63M's core area after the death of A67M.





91













/ A55M I /



\ /
A59M

A58F
10 m
















Figure 4.9--The core areas of 4 sympatric Lower Keys marsh rabbits (Sylvilagus palustris hefneri). Rabbit's A55M and A59M are adult males, A57F and A58F are adult females.














CHAPTER 5
METAPOPULATION DYNAMICS: PATCH OCCUPANCY AND HABITAT QUALITY


Introduction

In Levins' (1969) metapopulation model, all habitat patches were

identical in size and quality. More recent models (Taylor 1991, Thomas et al. 1992, Hanski 1994, Hanski et al. 1994), have taken an incidence-function approach (Diamond 1975) that predicts the probability of a species' occurrence based on the area of the habitat and the distance of the habitat patch to other patches. In these models, all patches are assumed to be inhabitable past a threshold size or minimum isolation (Hanski 1994). For species whose natural history is relatively unknown, an important aspect of a metapopulation occupancy model is determining if unoccupied habitats are vacant due to lack of suitable habitat or from past extinctions unrelated to the habitat patch quality.



Metapopulation Structure

Before studying habitat use, metapopulation structure should be studied to determine how the subpopulations interact with each other. In a "classical


92




Full Text
15
The majority of transition zone marshes are subject to predictable flooding by
spring lunar high tides (Williams 1991). Transition zone marshes often occur
between the mangrove community and the upland hardwood or pine
hammocks. The transition zone can be further divided into two components:
an open saltmarsh and at slightly higher elevations, a more forested area
dominated by buttonwood (Conocarpus erectus). Several species of mammals,
including the Lower Keys marsh rabbit inhabit this area.
On land where there are well-developed fresh water lenses, the pineland
community can occur. These areas are rarely inundated by salt water and are
maintained through periodic bums. Slash pine (Pinus elliottii) is the dominant
tree species, although various species of palms and ferns are also abundant
(Snyder et al. 1990). At the highest elevations (>3 m) the diverse hardwood
hammock occurs (Carlson et al. 1992). This is the climax community in the
Keys and occurs on nearly every large key. Nearly 10% of all tree species
found in this area are endemic (Long and Lakela 1971).
Although less common, fresh water wetlands dominated by sawgrass
(Cladium iamaicensisl do occur on the few keys that have a fresh water lens.
A few keys also have limited amounts of beach and dune habitat, but most of
the keys have exposed limestone rock on their coasts.
All of the native terrestrial mammals are derived from populations of the
continental United States (Layne 1974). Currently, only 5 species of native


229
Gotelli, N. J. 1991. Metapopulation models: The rescue effect, the propagule rain,
and the core-satellite hypothesis. American Naturalist 138:768-776.
Grant, W. E. 1986. Systems analysis and simulation in wildlife and fisheries
science. New York, Wiley. 388 pp.
Greenwood, P. J. 1980. Mating systems, philopatry, and dispersal in birds and
mammals. Animal Behaviour 28:1140-1162.
Greenwood, P. J. and P. H. Harvey. (1982). The natal and breeding dispersal of
birds. Annual Review of Ecology and Systematics 13:1-21.
Gyllenberg, M. and I. Hanski. 1991 Single species metapopulation dynamics: a
structured model. Theoretical Population Biology 42:35-61.
Haig, S. M., Ballou, J. D., and S. R. Derrickson. 1990. Management options for
preserving genetic diversity: Reintroduction of Guam rails to the wild.
Conservation Biology 4:290-300.
Hall, E. R. 1981. The mammals of North America. Vol. 1, John Wiley and Sons,
New York, 690 pp.
Hambright, T. L. 1991. Military history of the Florida Keys. Pages 102-105 in The
Monroe County environmental story (J. Gato, ed.). The Monroe County
Environmental Education Task Force, Big Pine Key, Florida.
Hanski, I. 1982. Dynamics of regional distribution: the core and satellite species
hypothesis. Oikos 38:210-221.
Hanski, I. 1985. Single-species spatial-dynamics may contribute to long-term rarity
and commonness. Ecology 66:335-343.
Hanski, I. 1989. Metapopulation dynamics: does it help to have more of the same?
Trends in Ecology and Evolution 4:113-114.
Hanski, I 1991. Single-species metapopulation dynamics: concepts, models, and
observations. Pages 17-38 in Metapopulation dynamics: Empirical and
theoretical investigations (M. E. Gilpin and I. Hanski, eds.), Academic
Press, London.
Hanski, I. 1994. Patch-occupancy dynamics in fragmented landscapes. Trends in
Ecology and Evolution 9:131-135.


163
model validation requires that data be available to test the predictions of the
model. Population abundance and metapopulation occupancy data are
available for 2 years, but this is not long enough to perform a rigorous test of
the model. In lieu of a validation Grant (1986) has proposed alternatives for
evaluation a PVA model without actual data. He suggests that the scenarios
used in the model be examined to determine if they accurately deal with the
management issues to be addressed and if the results are expected. Then the
parameters should be examined to determine their sensitivity to see if slight
miscalculations could have severely biased the model. Sensitivity analysis
found that none of the variables are extremely sensitive to the 25% and 50%
increases.
The major assumption of the Lower Keys marsh rabbit PVA model is
that population growth is not currently density-dependent. The data from
which this assumption was made were collected at only 6 habitat patches on 2
keys. It is possible that population growth differs at other patches and at other
keys. However, the fact that over two thirds of the populations on the Keys all
appear to be far below carrying capacity indicates that a process other than
density-dependence is in operation. It is possible that at higher survival rates,
density dependence in terms of limiting fecundity or survival at densities above
K might occur. The predicted population size of the 50% increase in marsh
rabbit survival seen in Figure 6.4 indicates that the population would increase


37
highly seasonal (Chapman et al. 1982), reproduction in marsh rabbits in
southern Florida (Holler and Conaway 1979), southern Texas (Bothnia and
Teer 1977), and southern California (Ingles 1941) exhibits only year-round
slight seasonal fluctuations.
Time between litters ranged from 1 to 5 months (X = 2.45, SE = 0.30).
Combining the results from the 11 rabbits, an average of 3.7 litters per year
was produced. Although their research methods differed from those reported
here, Holler and Conaway (1979) measured a higher fecundity rate (5.7 litters
per year) for marsh rabbits (S^ jr. paludicola) living in southern Florida.
Mortality
Fiftythree rabbit kittens observed in the nests, but only 15 young rabbits
were trapped on the grids. The survival rates (from one developmental stage to
the next) at the 5 main study sites ranged from 13 to 50%. Forty-one
individuals (26 male, 15 female) were trapped and collared at the 5 sites on
Boca Chica Key. Twenty seven (19 male, 8 female) died during the study.
Because of these small sample sizes in some of the demographic groups
statistical tests were not used. Percent mortality was highest for the second
year adults (X = 90%) and lowest for the juveniles (X = 25%). Male mortality
(%) was higher than female mortality for each age class (Figure 3.5). None of
the females in the study died during their subadult period, but nearly half of the


68
There did not appear to be any significant changes in home-range size or
shifts between the dry and wet seasons (Table 4.4, Figure 4.5). Results of a
Mann-Whitney U test, used because the variances of the home range size were
not equal, failed to find a significant result comparing either the wet and dry
95% harmonic home range (T = 251, P = 0.45) or core areas (T=245.5, P
=0.60). None of the shifts in home range between the wet and dry seasons
exceeded the radius the marsh rabbits home range. However, ANOVAs
testing the effect of time on home-range size were significant for both the 95%
harmonic mean (F = 6.25, df = 22, P = 0.02) and the core area (F = 5.55, df =
22, P = 0.03). Animals increased their home-range size with time (and number
of locations), irrespective of season.
Spacing Behavior
Figures 4.6 and 4.7 show the duration of each rabbits radio-telemetry
records and which rabbits were contemporaneous. The average amount of
overlap between same-sexed individuals occupying a patch at the same time
was significantly less than same-sexed groups occupying a site at different
times (Table 4.5). This relationship held true using both the 95% harmonic
mean home-range estimate (t_= -2.17, df = 12, P = 0.05) and the core area
(t= -3.29, df = 12, P_= 0.006). When a rabbit died, same-sexed individuals
living in adjent areas expanded their home ranges (Figure 4.8). Similarly, the


78
Table 4.2Habitat types and possible physical barriers encompassed by Lower
Keys marsh rabbit (Svlvilagus palustris hefneri) dispersal movements.
Feature
Mode of measurement
Unimproved road (dirt, gravel)
meters of length
2-land road
meters of length
4-land highway
meters of length
Mosquito drainage ditch (water < lm)
meters of length
Canal (1 10 m)
meters of length
Major body of water (water > 10m)
meters of length
Disturbed or barren habitat
ha
Mangrove
ha
Hardwood hammock
ha
Pineland
ha


57
In addition to home-range size and spacing, it is also important to
determine which habitat features a rabbit incorporates into its home range and
which features act as barriers. Rabbits generally move while under vegetative
cover (Chapman 1971, Trent and Rongstad 1974), probably to avoid predation.
Large open areas are generally seen as barriers or at least deterrents to rabbit
movements. Although the response of cottontail rabbits to roads and vehicular
traffic has not been directly studied, field studies have found that improved
roads (2-lane and larger) can be a barrier to small mammal and hare
movements (Oxley et al. 1974, Mader 1984). Other possible barriers to rabbit
movements include disturbed or developed areas with no natural vegetation,
vegetated areas with insufficient cover, and open bodies of water. The marsh
rabbit and the swamp rabbit are both known to be good swimmers (Svihla
1929, Tomkins 1935), although the frequency of swimming is not known.
While most rabbits stay within their home range and make few long
distance movements, young, subadult male rabbits have been documented
traveling longer distances (Shields 1960, Chapman and Trethewey 1972). It is
believed that these males are natal dispersers. Natal dispersal is male-biased in
most polygynous mammals (Dobson and Jones 1985). Determining which sex
disperses in a species and what factors motivate these movements are essential
to understand a species spatial organization.


109
function based on the 4 variables classified 93% of the sites correctly (3 empty
sites were classified as being occupied). The jackknife procedure classified
91%, indicating a fairly accurate model. One empty site was misclassified as
being occupied; 3 occupied sites were misclassified as being empty.
Four variables were included in the stepwise DFA comparing variably
occupied sites to the consistently occupied sites (Clump, DPopulation,
MaxHgt, and Borrichia; Table 5.6b). Consistently occupied sites had more
clump grasses and significantly higher ground vegetation (Table 5.7). Ninety-
seven percent of the sites were correctly classified by the model; 1 consistently
occupied site was misclassified as being variable. The jackknife classification
rate was also high 95% of the sites were classified correctly.
Potential Reintroduction Sites
Seven of the 17 currently empty sites were classified as being
potentially consistently occupied habitat sites using the discriminant function
and the adjusted inter-population distances (assuming multiple reintroductions
on a key). Three of these sites that were classified as being consistently
occupied occurred on North Sugarloaf, and 1 each occurred on Cudjoe Key,
Middle Torch Key, Big Torch Key and Noname Key.


79
Table 4.3--Home-range data on Lower Keys marsh rabbits (Svlvilagus palustris
hefherij observed for >1 month for juveniles, >30 locations for adults.
Rabbit
Classification
Number of
locations
Distance
between
consecutive
locations (m)
95%
harmonic
mean home
range (ha)
Core
area
(ha)
J55M
Juvenile
30
69.58
13.83
4.44
J99M
Juvenile
18
47.40
2.49
0.64
J101F
Juvenile
16
22.99
0.10
J102F
Juvenile
21
25.81
1.11
0.38
J103M
Juvenile
14
22.39
0.30
J199M
Juvenile
14
40.60
0.81
X
38.13
3.11
1.82
SE
(7.55)
(2.17)
(1.31)
A51M
Adult male
122
48.34
11.94
3.15
A55M
Adult male
70
30.81
2.73
0.89
A59M
Adult male
79
28.71
2.13
0.93
A63M
Adult male
217
25.33
5.81
1.58
A64M
Adult male
61
17.87
0.36
0.02
A67M
Adult male
99
42.10
3.47
1.10
A68M
Adult male
142
35.30
3.63
1.24
A69M
Adult male
138
50.88
7.61
2.60
A70M
Adult male
88
36.85
2.99
0.86
A71M
Adult male
67
37.47
3.18
1.22
A73M
Adult male
60
38.24
4.23
1.20
A75M
Adult male
74
28.07
4.35
1.18
A85M
Adult male
54
52.36
2.81
1.03
X
36.33
4.25
1.31
SE
(2.85)
(0.80)
(0.22)
A50F
Adult female
252
28.14
6.52
2.00
A52F
Adult female
127
34.31
10.32
3.17
A53F
Adult female
201
59.76
8.60
2.09
A57F
Adult female
220
21.90
1.67
0.53
A58F
Adult female
90
20.90
0.72
0.24
A72F
Adult female
107
26.15
2.49
1.02
A74F
Adult female
75
27.89
1.27
0.46
A76F
Adult female
81
22.12
1.53
0.53
A86F
Adult female
44
32.68
1.72
0.53
A176F
Adult female
45
30.52
0.98
0.25
X
30.44
3.58
1.08
SE
(3.56)
(1.12)
(0.32)


165
included in the simulation, the life table for these areas might be dramatically
different. This possibility suggests that a search for unoccupied habitat on the
outer islands should be a high priority for future research (see Chapter 7).
The lack of an impact from a mild disease is highly predictable if one
takes into account the rabbits sociality and the spatial structure of the
transition-zone habitat. Marsh rabbits are mainly solitary in nature (Chapman
et al. 1982), unlike their European relatives (Oryctolagus cuniculus) which are
highly susceptible to disease. Marsh rabbits have little contact with any other
rabbits from the time they are weaned to the time they mate. If the disease was
spread through physical contact it only would affect a portion of the
population. The disease would probably not spread as fast as it did in the
simulation, unless it had another vector (e.g., ticks, mosquitoes).
The scenarios involving hurricanes might be less realistic. Each year in
the Lower Keys there is a 10% chance of a hurricane (winds >74mph) making
landfall (Chen and Gerber 1990). The chance that a hurricane hits the middle
of the Lower Keys is probably <10%.
The results of the sixth scenario, decreasing migration, also should be
treated with caution. It is difficult to predict how marsh rabbits will react if
corridor habitat is altered. In the simulation, migration was reduced by 25%
and 50%, indicating that rabbits without a good corridor habitat would not
leave their patch. This may not be true; marsh rabbits might try and disperse


5
4
c=
3
ro
O 2
_co
1
0
Mainland-island
o 1
2 3 4
Area
Source-sink
o 1
2 3
Area
Classical
2 3
Area
K>
On
Figure 5.1 The 3 types of metapopulation structure: mainland-island, source-sink, and the stepping stone model a form of the classical
metapopulation. The black circles are consistently occupied sites. Grey are variably occupied sites and the white circles are empty
sites.


236
Miller, W. R., and R. E. Egler. 1950. Vegetation of the Wequetequock-Pawcatuck
tidal marshes, Connecticut. Ecological Monographs 20:143-172.
Morrison, D. F. 1976. Multivariate statistical methods. Second edition. McGraw-
Hill, New York, New York, 415 pp.
Moore, J. and R. Ali. 1984. Are dispersal and inbreeding avoidance related?
Animal Behavior 32:94-112.
Mueller, E. and G. O. Winston. 1991. Geologic history of the Florida Keys. Pages
4-11, in The Monroe County environmental story (J. Gato, ed.). The Monroe
County Environmental Education Task Force, Big Pine Key, Florida.
Murphy, D. D., Freas, K. E. and S. B. Weiss. 1990. An environment-
metapopulation approach to population viability analysis for a threatened
vertebrate. Conservation Biology 4:41-51.
Murphy, R. W., Sites, J. W. Jr., Buth, D. G. and C. H. Haufler. 1990. Proteins I:
iso2yme electrophoresis. Pages 45-126, in Molecular systematics (D. M.
Hillis and C. Moritz, eds.), Sinaeur Associates, Sunderland, Massachusetts.
Nagy, J. G. and J. B. Haufler. 1980. Wildlife nutrition. Pages 129-142 in Wildlife
management techniques, fourth edition (S. D. Schemnitz, ed.), The Wildlife
Society, Washington, D.C.
Nee, S. and R. M. May. 1992. Dynamics of metapopulations: habitat destruction
and competitive coexistence. Journal of Animal Ecology 61:37-40.
Neu, C. W., C. R. Byers and J. M. Peek. 1974. A technique for analysis of
utilization-availability data. Journal of Wildlife Management 38:541-545.
Nisbet, R. M. and W. S. C. Gurney. 1982. Modeling fluctuating populations,
Wiley, Chiester, New York, 379 pp.
NOAA. 1993. Climatological data, Florida. Volume 94:97. National Oceanic and
Atmospheric Administration, Asheville, North Carolina.
Nunney, L. and K. A. Campbell. 1993. Assessing minimum viable population size:
demography meets population genetics. Trends in Evolution and Ecology,
8:234-239.


164
far beyond the maximum carrying capacity for the habitat. It is likely that the
Lower Keys marsh rabbit is currently at the low end of the sigmoidal curve that
the logistic equation follows, and that increase survivorship would push the
population size up the curve.
Assuming density-dependence was properly addressed in the model,
there may also be questions about the realism of the scenarios. If it were
possible to reduce the amount of cat predation in the Keys, it does not appear
that predation from other predators would increase greatly. The only
confirmed raccoon (Procyon lotor) predation occurred in the crook of a fence,
where the raccoon cornered the rabbit, and this was also in an area with
extremely high raccoon density due to the many open dumpsters in the area.
Rattlesnakes (Crotalus adamanteus), the other predator in the system, are
thought to exist at low densities in the Keys (Lazell 1989) and are often killed
while crossing roads or when they enter peoples yards.
Scenario #2, stopping mortality due to vehicles, would be difficult to
achieve, but the incorporation of this mortality into the model was
straightforward. Scenario #3, recolonization, was accurately depicted in the
model if it is assumed that marsh rabbits would only be introduced to
unoccupied sites on the major keys that are connected by US-1. Other empty
areas of transition-zone habitat may exist on smaller islands off the major keys,
and these islands are likely to be devoid of feral cats. If these keys were


CHAPTER 4
SPATIAL ORGANIZATION (INTER-PATCH MOVEMENTS)
Introduction
To apply knowledge about the Lower Keys marsh rabbits population
biology in a PVA, it is necessary to define the unit of individuals that
comprises a local population. A local population can be loosely defined as a
set of individuals that has a high probability of interaction (Hanski and Gilpin
1991). Local populations may be part of a greater metapopulation where
interaction occurs via individuals moving among populations (Levins 1970).
The unit of the local population can be determined by studying the
spatial structure and movements of individuals at different spatial scales. An
individuals home range, spacing behavior, and movements (including
dispersal) will determine the spatial scale of interaction with others. These
behaviors may vary among individuals of different sexes and ages.
The home range of an individual is the area used for foraging, mating
and caring for young (Burt 1943, White and Garrott 1990). It does not include
areas used in migrations, or other occasional round-trip excursions. Home
range is calculated using a variety of methods that take periodic readings on an
individuals location.
55


241
Taylor, B. 1991. Investigating species incidence over habitat fragments of different
areas a look at error estimation. Pages 177-191 jn Metapopulation
dynamics: Empirical and theoretical investigations (M. E. Gilpin and I.
Hanski, eds.), Academic Press, London.
Teft, B. C. and J. A. Chapman. 1987. Social behavior of the New England
cottontail, Sylvilagus transitionalis (Bangs) with a review of social behavior
in New World rabbits (Mammalia: Leporidae). Review of Ecologie 42:235-
276.
Terborgh, J. 1974. The preservation of natural diversity: the problem of extinction
prone species. Bioscience 24:715-722.
Thiollay, J. M. and B. U. Meyburg. 1988. Forest fragmentation and the
conservation of raptors: a survey on the island of Java. Biological
Conservation, 44:229-250.
Thomas, C. D. 1994. Extinction, colonization, and metapopulations:
environmental tracking by rare species. Conservation Biology 8:373-378
Thomas, C. D. and T. M. Jones. 1993. Partial recovery of a skipper butterfly
(Hesparia comma! from population refuges: lessons for conservation in a
fragmented landscape. Journal of Animal Ecology 62:472-481.
Tomkins, I. R. 1935. The marsh rabbit: an incomplete life history. Journal of
Mammalogy 16:201-205.
Trent, T. T. and O. J. Rongstad. 1974. Home range and survival of cottontail
rabbits in southwestern Wisconsin. Journal of Wildlife Management 38:459-
472.
U.S. Fish and Wildlife Service. 1990. Endangered and threatened wildlife and
plants; endangered status for the Lower Keys rabbit and threatened status for
the Squirrel Chimney cave shrimp. Federal Register 55(120):25588-25591.
Vaughan, T. A. 1986. Mammalogy. Saunders College Publishing, Philadelphia,
Pennsylvania, 576 pp.
Verboom, J., Lankester, K. and J. A. J. Metz. 1991. Linking local and regional
dynamics in stochastic metapopulation models. Pages 39-55 in
Metapopulation dynamics: Empirical and theoretical investigations (M. E.
Gilpin and I. Hanski, eds.), Academic Press, London.


2
Three subspecies of marsh rabbits are recognized, Sk paludicola, S,
palustris (Chapman and Willner 1981), and 5L p^ hefneri (Figure 1.1). The
marsh rabbit is found in lowlands from the Dismal Swamp of Virginia into the
Florida Keys. Little is known about the biology of the Lower Keys marsh
rabbit (Chapman and Flux 1990, Wolfe 1992). Compared to other species of
cottontails, the biology and ecology of marsh rabbits in general (Sylvilagus
palustris spp.) is poorly understood (Chapman et al. 1982).
Tomkins (1935) and Carr (1939) commented on marsh rabbit (Sf p^
paludicola) behavior in North Carolina and north-central Florida marshes.
Blair (1935, 1936) studied the diet and habits of marsh rabbits near
Gainesville, Florida. The reproduction of the marsh rabbit (S^ p^ palustris) was
studied by Holler and Conaway (1979) in Belle Glade, Florida.
Despite these studies, large gaps in the knowledge of the population
ecology, habitat use, and risk of extinction for S. p. hefneri remain.
Extrapolation from these studies to the Lower Keys marsh rabbit may not be
accurate. The Lower Keys marsh rabbit inhabits a unique island ecosystem
and is subject to different pressures and resources.
Population Viability Analysis
Population viability analysis (PVA) is a comprehensive examination of
the interacting factors that put a population (or species) at risk of extinction


202
Other future research should be aimed at decevising means of
decreasing the mortality rate of the marsh rabbits. With all research projects
small experimental treatments should be applied first, their effects incorporated
into the PVA model, and then successful management techniques should be
used throughout the Lower Keys.


Year


154
same sex and these populations eventually went extinct. Demographic
stochasticity was incorporated into the model by varying survivorship and
reproduction (Ak?akaya 1994). The number of survivors for each stage was
drawn from a binomial distribution determined by the survival rate and the
smaple size. The number of young produced by each stage was drawn from a
Poisoon distriubtion.
A careful examination of the life-history parameters in each population
revealed that density-dependent growth does not appear to be occurring at the 5
populations from which the data was collected. Populations appeared to be
low due to the high mortality, and mortality was high at all of the patches,
regardless of rabbit density. However, these were collected only at Boca Chica
Key and Saddlebunch Key and could differ at the other keys. Therefore the 2
basic simulations were run, 1 with density dependence and 1 without.
Carrying capacity calculations (by site) for the density dependent
population growth ranged from a low of 3 rabbits (all stages) at the smallest
site to a high of 596 rabbits at the largest. Adding all the carrying capacities
together from all of the habitat patches, a total of 2,917 marsh rabbit (nestling
through adult) could occupy the Keys.


27
unbaited collapsible National live traps (80 x 30 x 30 cm), placed in a 6 x 6
array, spaced approximately 25 m apart. Each trapping session consisted of 5
nights where the traps were open, 2 nights with the traps closed and another 5
nights with traps open. Traps were checked twice daily, once in the morning
and once in the evening and were covered in burlap for shade.
All rabbits caught were sexed, weighed, and tagged (Monel no. 3,
National Band and Tag, Newport, KY). Measurements were made of the right
rear foot, ear length from notch, and total body length from nose to (and
including) tail. Marsh rabbits are relatively easy to sex (Nagy and Haufler
1980), but an aging method has not been determined. An attempt to correlate
weight, total length, ear length, and rear right foot length with age was made.
Rabbits were classified as juveniles (not sexually mature), subadults (entering
sexual maturity), or adults (fully sexually mature).
Genetic Analysis
Blood samples were obtained from all animals caught on the 6 trapping
grids between June 1992 and May 1993. In addition, blood was taken from
animals caught during preliminary trapping on Sugarloaf and Big Pine Keys.
Blood samples were obtained by lancing the ear with a hypodermic needle and
collected using heparinized capillary tubes. Blood was stored on wet ice for <1
hour and was centrifuged into plasma and hemolysate at 5,000 ipm for 10 min


74
like the marsh rabbit. Some circumstantial evidence of physical defense was
apparent in the males; 76% of the subadult and adult males had scratch marks
and scars on their face, ears and back of their heads. Boxing, scratching, and
paw-raking with the feet has been observed in other species of cottontails
(Marsden and Holler 1964) and may be responsible for the scratch marks.
Generally these behaviors are related to competition for mates (Eisenberg
1981). Female marsh rabbits showed no signs of scratch marks. It is not
readily apparent why it would be advantageous for the females to be territorial,
however there is some evidence that nesting sites may be limited (see Natality -
chapter 3). Fecal pellet marking is well-developed in other rabbit species (Teft
and Chapman 1987) and may be used by male and female marsh rabbits to
mark the boundaries of their home ranges.
Both males and females increased their home ranges following the death
of a conspecific. Over the lifetime of a rabbit this was seen as an increase in
home range with age. This phenomenon provides additional support that some
spacing behavior is partially determining home-range size. This spacing
behavior may have an impact on population density and total abundance of
rabbits in a habitat patch.
Most of these home ranges consisted almost exclusively of transition-
zone habitat. Physical features such as canals, roads, and runways were rarely
crossed by any of the adult marsh rabbits, despite the proximity of the features.


Session 6
52
Figure 5.8 (continued).
u>
.1


227
Dixon, K. R., Chapman, J. A., Rongstad, O. J. and K. M. Orhelein. 1981. A
comparison of home range size in Sylvilagus floridanus and S. bachmani.
Pages 541-548 in Proceedings of the world lagomorph conference (E. Myers
and C. D. Machines, eds.), Guelph Univ. Press, Guelph, Ontario.
Dobson, F. S. 1982. Competition for mates and predominant juvenile dispersal in
mammals. Animal Behaviour 30:184-200.
Dobson, F. S. and W. T. Jones 1985. Multiple causes of dispersal. American
Naturalist 126:855-858.
Eberhardt, L. L. 1987. Population projections from simple models. Journal of
Applied Ecology 24:103-188.
Edwards, W. R., Havera, S. P., Labisky, R. F., Ellis, J. A. and R. E. Warner. 1981.
The abundance of cottontails in relation to agricultural land use in Illinois
(U.S.A.) 1956-1978, with comments on mechanisms of regulation. Pages
761-798 in Proceedings of the world lagomorph conference (K. Meyers and
C. D. Macinnes, eds), University of Guelph, Ontario.
Eisenberg, J. F. 1981. The mammalian radiations: An analysis of trends in
evolution, adaptation and behavior. University of Chicago Press, Chicago,
Illinois. 510 pp.
Fahrig, L. and G. Merriam. 1985. Habitat patch connectivity and population
survival. Ecology 66:1762-1768.
Falconer, D. S. 1989. Introduction to quantitative genetics. 2d ed. Longman Press,
New York. 438 pp.
Festa-Bianchet, M. and W. J. King. 1984. Behaviour and dispersal of yearling
Colombian ground squirrels. Canadian Journal of Zoology 62:161-167.
Fitzgerald, B. M. and B. J. Karl. 1979. Foods of feral house cats (Telis catus L.) in
forest of the Orongorongo Valley, Wellington. New Zealand Journal of
Zoology 6:107-126.
Frank, P. A. 1993. Anastasia Island beach mouse at home at Fort Matanzas
National Monument. Park Science 13:30-31.


172
Table 6.4Number and proportion of habitat patches that existed above and
below carrying capacity (K). The total estimated number of Lower Keys marsh
rabbits (Sylvilagus palustris hefneri) was calculated for the 39 sites where
density estimates were conducted and was averaged over the 8 sessions. This
average was compared to K.
% larger or smaller
0-19% larger
20-39% larger
40-59% larger
60-79% larger
80-100% larger
Total larger
0-19% smaller
20-39% smaller
40-59% smaller
60-79% smaller
80-100% smaller
Number of sites in this Proportion of sites
category
4
10.0%
3
7.5%
2
5.0%
2
5.0%
0
0.0%
11
27.5%
4
10.0%
0
0.0%
1
2.5%
7
17.5%
15
37.5%
28
67.5%
Total smaller


72
zone significantly more than expected and used disturbed areas significantly
less.
Discussion
The data suggest that members of S. p. hefneri spend most of their lives
in 1 patch, but can move to other patches. Marsh rabbits are bom in a patch of
transition-zone habitat and remain there until they reach sexual maturity. At
sexual maturity most rabbits make a relatively long, one-way movement. Male
marsh rabbits may move a great distance away from their natal range; females
are more likely to remain in the same patches where they were bom. If a patch
is relatively small, this movement may mean leaving the habitat patch. When
these subadult marsh rabbits leave their natal ranges, they establish adult home
ranges that they maintain for their lifetime. These adult home ranges do not
incorporate any roads (2-lane and larger) bodies of water, or other types of
habitat. All mating appears to occur within the same habitat patch. These
results are consistent with the predictions of the second hypothesis, that S. p.
hefneri exists in a metapopulation (Levins 1970, Hanski 1991).
The Lower Keys marsh rabbits ability to exist as a metapopulation and
disperse over relatively long distances may be a product of inhabiting a
naturally patchy environment (Opdam 1991, Thomas 1994). Species that exist
in temporally or spatially patchy environments must be able to deal with


197
education program may help by increasing the number of pets that are
sterilized, decreasing the number of cats that are abandoned, and decreasing the
number of pet cats that are allowed to roam outdoors. Changing the publics
opinion on their cats outdoor-time through education has not generally been
successful in other areas (Proulx 1988).
Trapping the cats in marsh rabbit habitat and bringing cats to the Animal
Control Pound, might further reinforce cat owners to keep their cats indoors.
Other actions might include experimenting with cat bells to determine if they
work in reducing the cats hunting ability. Cat-proof fencing is not feasible in
most of the areas because most of the habitat exists in small patches and the
larger patches are shared with a suite of native species that would be negatively
impacted by the fences.
As a prerequisite to reduce overall marsh rabbit mortality and to prevent
further habitat loss, marsh rabbit habitat must be purchased and managed. A
cat trapping program will be easier to implement if the land is owned by the
state or federal government or an environmental non-government agency. As
the results of scenario #7 (habitat loss) of the PVA model demonstrated, even
small habitat patches are important to the overall persistence of S. p. hefneri.
A complete listing of land parcels for acquisition is given in the Lower Keys
marsh rabbit Recovery Plan (United States Fish and Wildlife Service 1993).
The largest parcel is covered by a Conservation and Recreative Lands (CARL)


TABLE OF CONTENTS
page
ACKNOWLEDGMENTS ii
ABSTRACT vii
CHAPTERS
1 INTRODUCTION 1
Population Viability Analysis 2
Spatial Structure 4
Metapopulations 6
Metapopulation Persistence 7
Dissertation Structure 9
2 THE LOWER KEYS OF FLORIDA 12
Geologic Histoiy 12
Climate 13
Flora and Fauna 14
Development 16
Study Sites 18
3 POPULATION BIOLOGY (INTRA-PATCH DYNAMICS) 22
Introduction 22
Methods 26
Trapping Girds 26
Genetic Analyses 27
Radio-telemetry 28
Natality 29
Mortality 30
Results 32
Growth and Morphology 32
Genetics 34
Demographics 35
Natality 36
Mortality 37
Discussion 39
iii


108
comparisons indicated that habitat selection occurred at least once at all of the
sites, and the pattern of selectivity did not appear to vary strongly among
seasons. In all 3 seasons, rabbits significantly used mid-marsh vegetation
(Borrichia sp. and Sporobolus virginicus) and high marsh vegetation (mainly
clump grasses) more than expected and used hammocks and low marshes less.
The over all rank of habitat use based on the fecal pellet distribution was: mid
marsh > high marsh > hammock > low marsh.
Habitat Model
Ten variables were used in the 2 DFAs (Table 5.5). All of the habitat
variables were either normal without a transformation or became normal after
the transformation (Table 5.5). None of the variables were significantly inter-
correlated. The test for homogeneity of within-group covariance matrices was
not significant (X^ = 22.75, df = 46, P > 0.36), indicating that a pooled
covariance matrix and a linear discriminant function analysis should be used
(Morrison 1976).
Four variables (DPopulation, Area, DResidence and MaxHgt) (Table
5.6a) were included in the stepwise DFA comparing empty and occupied sites.
Only 2 of the variables, DPopulation and DResidence, differed significantly
between empty and occupied sties (Table 5.7). Empty sites were more isolated
from other populations but closer to human dwellings. The discriminant


16
terrestrial mammals and perhaps 1 bat are found in the Lower Keys. All of the
terrestrial mammals have been cited as being endemic in the literature at either
the species or subspecies level (see Lazell 1989), although the accuracy of
these taxonomic claims has been debated (see Humphrey 1994). The paucity
of mammals in the Lower Keys may be related to the current isolation and
small area of the Keys, combined with a lack of fresh water (Layne 1974).
Herpetofaunal diversity is greater than the mammals, although a large
number of reptiles and amphibians in the Lower Keys are exotics (Wilson and
Porras 1983). Of importance to the Lower Keys marsh rabbit, the eastern
diamondback rattlesnake (Crotalus adamanteus) is common throughout the
Lower Keys and the alligator (Alligator mississippiensis). although rare, does
occur. Avian diversity, both breeding and over-wintering is fairly high in the
Lower Keys although lower than mainland Florida (Robertson and Kushlan
1974). Special references to birds and reptiles as predators will be made in
Chapter 3.
Development
By the 1890s Key West was the largest city in Florida and one of the
largest in the United States. The remainder of the Keys were unpopulated until
the completion of the overseas highway in 1938 and the first water pipeline in
1942 (Gallagher 1991). Today, over 78,000 people live in Monroe county,


204
Key
Site #
Area
Ownership*
Density
Saddlebunch
28
3.50
USFWS
Yes
Saddlebunch
29
2.50
USN
Yes
Saddlebunch
30
0.78
USN
Yes
Sugarloaf
31
1.00
P
Yes
Sugarloaf
32
4.00
P
Yes
Sugarloaf
33
31.50
P
Yes
Sugarloaf
34
2.30
P
Yes
Sugarloaf
35
0.86
P
No
Sugarloaf
36
0.57
P
No
N. Saddlebunch
37
0.58
Monroe
No
N. Saddlebunch
38
0.50
P
No
N. Saddlebunch
39
0.55
P
No
N. Saddlebunch
40
0.69
P
No
Cudjoe
41
2.60
P
Yes
Cudjoe
42
3.90
Air
Yes
Cudjoe
43
5.50
P
Yes
Summerland
44
2.34
P
Yes
Ramrod
45
3.90
P
Yes
Middle Torch
46
1.00
USFWS
No
Middle Torch
47
1.73
USFWS
No
Big Torch
48
0.50
P
No
Big Torch
49
0.88
USFWS
No
Big Torch
50
1.44
P
Yes
Little Torch
51
3.00
TNC
Yes
Big Pine
52
4.45
USFWS
No
Big Pine
53
19.50
USFWS
Yes
Big Pine
54
43.70
SFWMD
Yes
Big Pine
55
27.60
SFWMD
Yes
Big Pine
56
1.10
P
Yes
Big Pine
57
1.00
P
Yes
Big Pine
58
2.93
USFWS
Yes
Noname
59
1.50
USFWS
Yes
*USN United States Navy, P = privately owned, USFWS = United States Fish and
Wildlife Service, Monroe = Monroe County, Air = United States Airforce, TNC = The
Nature Conservancy.


Number of Occupied Patches Total Estimated Abundance
o 10 20 30 40 50
Year
w


95
spatial arrangement, size of the patches, and patterns of occupancy will be
examined to determine if the metapopulation is classical, mainland-island, or
source-sink in structure.
The second objective of this chapter is to elucidate factors influencing
habitat use by the Lower Keys marsh rabbit at 3 inter-dependent spatio-
temporal scales. This habitat analysis will be used to determine if currently
unoccupied habitat patches are suitable and empty because of their isolation
from other patches or if the habitat is unsuitable.
Methods
Fecal pellets were used to study diet, habitat use, and the presence/
absence of marsh rabbits. Presence of fecal pellets can be an accurate and
efficient method of studying habitat use and patch occupancy when
information about pellet persistence is estimated (Simonetti and Fuentes 1982,
Pietz and Tester 1983). It is a particularly useful method for species that are
secretive, crepuscular, at low densities, and not easily trappable (Wood 1988).
Pellet Grids
Permanent fecal pellet-sampling grids were established at each of 59
patches of transition-zone habitat in the Lower Keys. The grid was designed to
fit into the smallest patch and consisted of a square of 7 x 7 stations at 15-m


Table C.l--Continued.
C
-C
JD
9
PC
-2
o
H
cS
Xj
X
9
Pi
o
H
cd
a
pc
OP rS
13
Pi
0)
B
M VO in ^
m
cc
oo
Tf
Tf
Q
Q
VO
r-
NO
in
o
o
vq
q
q
vq
(N
H
q
q
o
in
in
vq
NO
cn
in
q
d
d
(N
Ov
d
d
d
d
^H
d
d
d
d
d
00*
CN
in
in
in
ON
o
Q
On
(N
NO
cn
(N
cc
O
q
cc
q
CN
ON
o
O
On
q
vq
oo
cn
ON
d
d
d
d
d
d
d
d
d
O
d
C'i
d
d
d
d
8888888888£P;S
00
vo
H
00
Q
cn
ON
in
oo
CN
00
-h
e'
q
Tf
o
m
in
*h
in
o
On
in
q
ON
ON
O
tj*
q
q
q
On
q
q
en
o
q
q
(N
q
00
cn
q
q
cn
d
d
d
d
d
d
in
d
On
in
d
d
d
vo
d
d
d
d
d
d
d
O
d
d
cn
o
o
NO
VO
o
VO
(N
m
*fr
o
VO
o
o
o
o
(N
tJ-
VO
in
(N
in
cc
q
in
q
q
q
q
q
in
00
q
q
o
oo
vo
oo
iH
cn
q
q
,-H
T-H
d
d
d
,H
d
d
d
d
d
d
d
d
d
d
d
d
d
d
d
d
cc
Q
ON
Q
o
ON
ON
in
n
cn
o
cn
m
o
m
r-
vo
o
VO
cn
q
o
q
q
o
in
q
q
q
q
q
in
q
q
o
q
in
tj-
o
00
in
d
d
d
d
d
d
d
d
d

in
d
in
d
d
d
d
d
d
d
d
d
VO
d
*'^Q'OOO^TfaNTfoom-HinQooommaNr^r^^mr^ooOQQQQ^ooNOO
cnooqqqqqvoqqqqqqqqinqqqqc>o(N^,o
dddddddocNddddoddddoddddddddddddddddd
00
00
(N
oo
n
c-

,-H

o
o
8
o
(N
o
oo
00
o
q
q
vo
q
q
q
cn
q
00
cn
o
VO
q
vo
q
cn
q
q
d
d
d
d
On
d
d
d
o
in
d
d
d
d
d
C'i
d
d
d
d
VO
oooooocscnvooo
(N
Tf
r-
ON
n
cn
ON
cn
o
m
00
8
8

oo
o
c-
8
oo
in

in
cn
q
q
q
T"
q
q
re
oo
q
cn
q
On
q
in
n
ih
o
q
cc
q
d
d
d
d
d
d
d
d
d
d
O
d
d
O
d
d
d
d
d
d
d
d
n
d
(Nmrj-invor^oooN
fSn^iniOhioooo\OH(S(f|TtH(snTtioOHfniohooo
218


CHAPTER 6
POPULATION VIABILITY ANALYSIS
Introduction
The process of conducting a population viability analysis (PVA; Soul
1987) involves building a model that makes probabilistic predictions about a
species future (Ginzburg et al. 1982, Shaffer 1990). Most models use
population simulation or analysis based on existing data to make these
predictions. The challenge in developing a PVA is creating a model that
captures all of the important aspects of a species ecology while using only
realistically measurable parameters (Eberhardt 1987, Boyce 1992).
This process is further complicated when the population to be modeled
exists in discrete subpopulations. If these subpopulations are related through
inter-subpopulation movements that have the potential to augment or
recolonize small or extinct subpopulations, then the PVA must include this
metapopulation structure (LaHaye et al. 1994).
Multiple Levels of Population Viability Analysis
A PVA that incoiporates metapopulation structure can involve processes
that occur at 3 increasing spatial levels: subpopulation (patch), between
135


66
(Figure 4.3). Aerial photographs were digitized (ARC/INFO 1990) to
determine the amount of each habitat type and length of all roads, runways, and
linear bodies of water was determined within this circle. The Johnson method
(Johnson 1980) was used to determine if habitat use significantly deviated from
the amount of each type of habitat available. A Waller-Duncan multiple
comparison procedure (Waller and Duncan 1969) was used to determine
differentially selected habitats. The Johnson method was chosen because it
looked for overall differences in habitat use, not at individual habitat-use
differences (Alldredge and Ratti 1992).
Results
A total of 54 Lower Keys marsh rabbits were caught and examined
during this study. Forty-three were caught on the 6 main grid sites (41 on Boca
Chica, 2 on Saddlebunch), 5 rabbits were caught while trying to recapture
collared dispersed rabbits, and 6 were caught on other sites in the Lower Keys.
Rabbits were only examined once per trapping period. Data from the 54
rabbits was recorded 130 times during eight trapping sessions (May 1991 -
May 1993).
Forty-three (28 male, 15 female) of the rabbits caught were fitted with
radio-collars, and 7 of these were juveniles. More juveniles were caught but at
the time of their capture the break-away collars were not available. Forty-one


220
Table D. 1Presence of adult and juvenile Lower Keys marsh rabbits (Svlvilagus palustris hefneri) during the last 6 sessions.
Site
code
Adult3
Juv.3
Adult4
Juv.4
Adult4
Juv.j5
Adult6
Juv.6
Adult7
Juv.7
Adult8
Juv.8
1
1
1
0
1
0
1
0
1
0
1
0
1
1
2
2
1
0
1
0
1
1
1
0
1
0
1
1
3
1
0
0
1
0
1
0
0
0
1
1
1
0
4
2
1
0
1
0
1
0
1
1
1
0
1
1
5
1
1
0
1
1
1
0
0
0
0
0
0
0
6
1
0
0
1
0
1
0
0
0
1
0
1
0
7
2
1
1
1
1
1
0
1
0
1
0
1
1
8
2
1
1
1
1
1
0
1
1
1
1
1
1
9
2
1
0
1
1
1
0
1
0
1
0
1
0
10
2
1
0
1
0
1
0
1
0
1
0
1
1
11
2
1
0
1
0
1
0
1
0
1
0
1
0
12
2
1
1
1
0
1
0
1
0
1
1
1
1
13
2
1
0
1
0
1
0
1
1
1
0
1
1
14
2
1
0
1
0
1
0
1
0
1
0
1
1
15
1
1
0
1
0
1
0
1
0
1
0
1
0
16
1
1
0
1
1
1
0
1
0
1
1
0
0
17
1
1
1
1
0
1
0
1
0
0
0
0
0
18
1
0
0
1
0
0
0
1
0
1
1
1
0
19
1
0
0
0
0
0
0
0
0
1
0
1
1
20
1
0
0
0
0
0
0
0
0
1
1
1
0
21
1
0
0
0
0
0
0
1
0
0
0
1
0
22
1
1
0
0
0
0
0
0
0
1
1
0
0
23
1
1
1
1
0
1
0
0
0
0
0
0
0
24
1
0
0
0
0
0
0
1
0
1
1
1
0
25
1
1
0
1
0
1
0
1
0
1
0
1
0
26
1
0
0
1
0
1
1
1
1
1
0
1
1
27
1
0
0
0
0
0
0
1
0
0
0
0
0
28
1
0
0
1
0
1
1
1
0
1
1
1
0
29
1
0
0
1
0
1
1
1
1
1
0
1
1
30
2
1
1
1
0
1
0
1
0
1
0
1
1


Table 6.1 --Parameters used by stage (1 = nestling, 2 = juvenile, 3 = subadult, 4 = adultl, 5 = adult2) for RAMAS simulations involving
marsh rabbit populations under different scenarios. Reproductive rates were held constant for all scenarios.
Scenario
1
Mortality
2 3
4
5
1
Catastrophe
2 3
4
5
1
Migration
2 3
4
5
Basic (exponential)
0.28
0.75
0.67
0.52
0.10
0
0
0
0
0
0
0
0.75
0
0
#la. Predator control
leads to 25% less
mortality.
0.42
0.81
0.75
0.64
0.33
0
0
0
0
0
0
0
0.75
0
0
#lb. 50% less mortality.
0.60
0.87
0.83
0.76
0.53
0
0
0
0
0
0
0
0.75
0
0
#2. All vehicular deaths
are avoided.
0.28
0.75
0.75
0.60
0.10
0
0
0
0
0
0
0
0.75
0
0
#3. Rabbits are 0.28 0.75 0.67 0.52 0.10 000 0 0 0 0 0.75
reintroduced to all vacant
patches.
All of the following scenarios are conducted assuming that mortality for all stages has been reduced by 25%.
0
0
#4a. A disease causing
0.42
0.81
0.75
0.64
0.33
0.25
0
0
0.25
0.25
0
0
0.75
0
0
some fatalities is spread
through the keys. The
effect is local, but each
patch has a 95% chance
of getting infected each
year.
#4b. More severe 0.42 0.81 0.75 0.64 0.33 0.95 0 0 0.95 0.95 0 0 0.75 0 0
disease.


96
intervals, marked with permanent flags. The grids were surveyed 3 times per
year: March (late dry season), July (mid-wet season) and November (transition
between wet and dry) from March 1991 to July 1993. Each survey consisted
of a pellet removal within a radius of 0.5 m at each station followed by a pellet
census 1 month later (a marker had pellets if 1 pellet fell within 0.5 m). A grid
was considered to be occupied (marsh rabbit present) if at least 1 of the stations
had a marsh rabbit pellet.
To ensure that fecal pellets did not begin to degrade in less than a
month, during each survey 100 of the pellets from captured rabbits were placed
on a transition-zone grid. The pellets were separated into 4 groups and placed
on rocks, on mud, in grass, and under trees. The pellets were counted weekly
to determine the rate of decomposition.
Pellet Size
To determine if the pellets produced on a grid were from juvenile or
adult marsh rabbits, a linear regression was used to determine if body mass
accurately determines pellet size. Body mass was obtained by trapping
individual marsh rabbits at 6 sites in the Lower Keys, including the 5 main
sites on Boca Chica (Figure 3.1) and 1 site on Saddlebunch. Trapping occurred
twice during the wet season (June November), and twice during the dry
season (December May). For the 5 sites on Boca Chica Key, trapping


82
Table 4.6Data on radio-collared marsh rabbits that made permanent, one-way
movements. Rabbits were caught on Boca Chica, Geiger, and Saddlebunch
Keys between June 1991 to May 1993.
Rabbit
Classification
Distance between natal
site and last location (m)
Body mass at beginning
of movment (g)
A48M
Subadult male
550
1,000
A49M
Subadult male
550
1,050
A54M
Subadult male
920
1,050
A55M
Subadult male
1,100
800
A56M
Subadult male
1,800
1,000
A60M
Subadult male
510
1,000
A66M
Subadult male
60
800
A68M
Subaudit male
180
1,000
A69M
Subadult male
980
800
A84M
Subadult male
2,050
1,050
A197M
Subadult male
400
900
X
827
950
SE
191
32
A3F
Subadult female
60
850
A65F
Subadult female
40
900
A74F
Subadult female
70
950
A76F
Subadult female
80
900
A86F
Subadult female
90
1,000
A176F
Subadult female
150
900
X
82
917
SE
15
21


238
Ralls, K., Ballou, J. D. and A. Templeton. 1988. Estimates of lethal equivalents
and the cost of inbreeding in mammals. Conservation Biology 2:185-193.
Ray, C. M, Gilpin, M. E. and A. T. Smith. 1991. The effect of conspecific
attraction on metapopulation dynamics. Biological Journal of the Linnean
Society, 42:123-134.
Ribble, D. O. 1992. Dispersal in a monogamous rodent. Ecology 73:859-866.
Richter-Dyn, N. and N. S. Goel. 1972. On the extinction of a colonizing species.
Theoretical Population Biology 3:406-433.
Robertson, W. B., Jr. and J. A. Kushlan. 1974. The southern Florida avifauna.
Pages 414-452 in Environments of south Florida: present and past (P. J.
Gleason, ed.), Miami Geological Society, Florida.
Rogers, J. S 1972. Measures of genetic similarity and genetic distance. Studies in
genetics, VII. The University of Texas Publication, 7213:145-153.
Rolstad, J. 1991. Consequences of forest fragmentation for the dynamics of bird
populations: Conceptual issues and the evidence. Pages 149-163 in
Metapopulation dynamics: Empirical and theoretical investigations (M. E.
Gilpin and I. Hanski, eds.), Academic Press, London.
Rose, G. B. 1977. Mortality rates of tagged adult cottontail rabbits. Journal of
Wildlife Management, 41:511-514.
Ross, M. S., OBrien, J. J. and L. d. S. Sternberg. 1994. Sea-level rise and the
reduction of pine forests in the Florida Keys. Ecological Applications 4:144-
156.
Salwasser, H., Mealey, S. P, and K. Johnson. Wildlife population viability.
Transactions of the North American Wildlife and Natural Resources
Conference 49:421-439.
SAS Institute, Inc. 1984. SAS/ETS users guide, version 5, SAS Institute, Cary
North Carolina, 734 pp.
SAS Institute, Inc. 1987. SAS/STAT user's guide, version 6, SAS Institute, Cary,
North Carolina, 1,028 pp.


APPENDIX C
MARSH RABBIT DENSITY ESTIMATES


10
conclusions drawn from chapters 3-5 into a PVA to test the hypothesis that the
metapopulation was in equilibrium. Predictions about the future for the Lower
Keys marsh rabbit were made under a number of different scenarios.
In the final chapter (Chapter 7), conclusions and recommendations are
made for the management and conservation of the Lower Keys marsh rabbit
and other endangered species inhabiting patch/fragmented landscapes.
Predator control, species reintroductions, the impact of more habitat loss,
diseases, and the impact of hurricanes, are examined in light of the new
research on metapopulation dynamics.


34
(females larger) and the other half exhibit no dimorphism (Champman et al.
1982). The average measurements and standard deviations of 29 marsh rabbits
were as follows: mass 1224.1 g. (80.9), total length 339.3 mm (24.9), ear from
notch 52.7 mm (3.4), and right hind foot 73.6 mm (3.7).
Genetics
Two (EST-1 and PGM-1) of the 11 loci sampled from the 19 rabbits
were polymorphic, indicating that there is genetic variation in the Lower Keys
marsh rabbit (Table 3.4). This proportion of variable loci is only slightly less
than the proportions seen in studies of cottontail rabbits in Texas (Scribner and
Warren 1986). Deviations from Hardy-Weinberg expectations were not
observed (P > 0.05) at either of the polymorphic loci, but this test may not be
valid on a so small sample. Differences in allelic proportions for each locus
appeared to vary among keys, but sample sizes were too small for statistical
comparison. Heterozygous loci and variation among individuals occurred at
Boca Chica, Geiger, and Sugarloaf Keys. The 2 rabbits sampled on the Navy
land on Saddlebunch Key were homozygous and monomorphic at all loci.
Only 1 individual was sampled from Big Pine Key; this individual was
heterozygous only at the EST-1 loci and contained an allele at the EST-1 locus
that was not seen at any of the other keys.


206
Figure A.2--A map of Saddlebuch and Lower Sugarloaf Keys. The habitat patches used
the study are numbered.


137
interact via feed-back loops leading to potential extinction via extinction
vortices (Gilpin and Soul 1986).
Interactions between patches also can be an important component of the
PVA. Dispersal rates, the mechanisms behind dispersal, and the impact of
habitat on dispersal can effect the amount of interchange between populations.
Individuals moving between patches may increase an the persistence time of an
occupied patch by augmenting the population and decreasing the chance for
stochastic extinction (rescue effect; Brown and Kodric-Brown 1977). The
dispersing individuals may also recolonize empty patches, if both males and
females or pregnant females move (in a sexual species). Incorporating inter
patch dynamics requires using spatially explicit metapopulation structure.
The third spatial level, metapopulation dynamics, considers the inter
relation between patch extinction and recolonization. For a metapopulation to
persist, the patch recolonization rates must exceed patch extinction rates
(Levins 1970, Hanski and Gilpin 1991). The inter-patch movements described
above will determine recolonization rates, but metapopulation persistence is
largely dependent on a lack of correlation among populations in extinction
rates (Murphy et al. 1990, Thomas and Jones 1993). Environmental variation
and catastrophes can cause synchronous variations in population densities
(Quinn and Hastings 1987, Gilpin 1988, Hanski 1989, Stacey and Taper 1992).
If all patches are subjected to the same adverse environmental conditions at the


189
Figure 6.9Results of Scenario #6 of the Lower Keys marsh rabbit (Sylvilagus
palustris hefneri) PVA model. Corridor habitat is destroyed, causing a 25% and 50%
reduction in dispersing individuals.


122
Table 5.5 (continued).
DResidence Distance from edge aerial photograph 1-6000 m L
of site to nearest
occupied domicile
DWater Distance from edge aerial photograph 1-1000 m L
of site to body of
water
* Transformations applied to data to achieve normality: S = square-root, L = logio> N
= no transformation needed, data already normal.


47
Table 3.4~Allele frequencies for all variable locia among 5 marsh rabbit
populations (Sylvilagus palustris hefneri) from the Lower Keys of Florida, June
1992 May 1993. Individuals were trapped on sites 7, 8 and 9 from Boca
Chica, sites 10 and 13 from Geiger Key (the southeastern portion of Boca
Chica), site 30 on Saddlebunch, site 33 on Sugarloaf, and site 53 on Big Pine.
Site
INI
Locus
Allele
Boca Chica Geiger
Saddlebunch Sugarloaf
Big Pine
(9) (5)
(2) (2)
(1)
EST-1
A
0.389
0.200
1.000
0.250
B
0.611
0.800
0.750
C
1.000
PGM-1
A
0.778
0.500
1.000
0.500
1.000
B
0.222
0.500
0.500
aThe following loci were assayed but were monomorphic: GOT, GPI, Hb, ICD-1,
LDH-1, MDH-1, MPI, PEP-1, SOD-1.


242
Wallage-Drees, J. M., Immink, H. J., De Bruyn, G. and P. A. Slim. 1986. The use
of fragment-identification to demonstrate short-term changes in the diet of
rabbits. Acta Theriologica 31:293-301.
Waller, R. A. and D. B. Duncan. 1969. A Bayes rule for the symmetric multiple
comparisons problem. Journal of the American Statistical Association
64:1484-1503.
Walter, H., E. 1985. Vegetation of the earth and ecological systems of the geo
biosphere. Springer-Verlag, New York, 318 pp.
Walters, C. J. 1986. Adaptive management of renewable resources. Macmillan,
New York, 374 pp.
Warren, R. S., and W. A. Niering. 1993. Vegetation change on a northeast tidal
marsh: interaction of sea-level rise and marsh accretion. Ecology 74:96-
103.
Webb, D. S. 1990. Historical Biogeography. Pages 70-99 in Ecosystems of Florida
(R. L. Myers and J. J. Ewel, eds.), University of Central Florida Press,
Orlando, Florida.
Westoby, M., G. R. Rost, and J. A. Weis. 1976. Problems with estimating
herbivore diets by microscopically identifying plant fragments from
stomachs. Journal of Mammalogy 57:167-172.
White, G. C., and R. A. Garrott. 1990. Analysis of Wildlife Radio-tracking Data.
Academic Press, Inc., San Diego, California, 289 pp.
Wilcox, B. A. 1986. Extinction models and conservation. Trends in Ecology and
Evolution 1:46-48.
Wilcox, B. A. and D. D. Murphy. 1985. Conservation strategy: the effects of
fragmentation on extinction. American Naturalist, 125:879-887.
Williams, A. 1991. Transitional wetlands. Pages 32-33 in The Monroe County
environmental story (J. Gato, ed.). The Monroe County Environmental
Education Task Force, Big Pine Key, Florida.
Willis, E. O. 1974. Population and local extinctions of birds on Barro Colorado
Island, Panama. Ecological Monographs 44:153-169.


215
Table B. 1-Continued.
site
code
%gcover
%ccover
H'
hgt
%bf
%clump
area
dwat
dpeop
isol
31
2
70
4
1.6
6.5
7
30
1.0
1
300
383
32
2
56
5
1.38
17.0
10
39
4.0
100
1000
383
33
2
98
5
1.12
62.0
8
87
31.50
1
2000
1800
34
1
37
13
1.57
15.0
4
10
2.30
1
1
832
35
1
60.4
12
1.84
15.0
3
11
0.86
1
700
927
36
0
74.4
15
1.47
7.6
1
1
0.57
1
624
1289
37
0
42
4
1.09
28.3
0
29
0.58
216
1
2603
38
0
59
2
0.71
40.0
0
52
0.50
200
1
2274
39
0
53
4
0.55
49.2
0
44
0.55
180
1
2511
40
0
59
18
0.69
30.5
0
0
0.69
80
1
2497
41
0
65
5
0.84
55.6
0
56
2.60
1000
1
4273
42
0
50
8
1.13
8.0
2
2
3.90
1
3000
5272
43
0
30
29
1.57
7.0
3
16
5.50
1
1
4707
44
0
96
4
1.17
60.2
0
45
2.34
125
300
6066
45
0
57
17
1.57
14.0
4
8
3.90
1
1
4746
46
0
87
3
0.91
45.2
0
63
1.00
120
480
3011
47
0
87
16
1.67
30.5
60
5
1.73
1
1
2026
48
0
72
22
2.07
20.0
15
15
0.50
72
120
3498
49
0
76
12
1.43
40.8
0
50
0.88
48
1920
4244
50
0
73
28
1.46
16.5
20
10
1.44
72
1
4749
51
0
47
81
1.64
19.3
1
6
3.00
1
100
3442
52
1
61
23
2.04
26.0
13
13
4.45
1
1
765
53
2
70
14
1.78
15.5
0
26
19.5
375
750
1772
54
1
55
30
1.16
15.0
0
30
43.7
96
1
1250
55
1
44
34
1.38
12.5
1
6
27.6
1
1
1467
56
1
38
18
1.35
25.0
3
27
1.10
50
100
1507
57
2
63
28
1.91
19.5
5
45
1.00
400
1
1811
58
1
72
35
1.46
17.4
8
5
2.93
313
780
757
59
0
87
47
1.76
15.6
6
60
1.50
1
500
1542


162
deterministic threat, and until it is removed no other management techniques
will be successful.
Based on simulation, once the 25% increase in survivorship was
implemented, only large-scale habitat destruction and a veiy severe hurricane
were able to cause the rabbits extinction. House cats have probably increased
in the Lower Keys with increased human population during the past few
decades. Extinction (or severe population declines) have been often attributed
to exotic predators (e.g., domestic cat, Felis catus), although direct evidence is
difficult to obtain.
Australia (Jones and Coman 1981), New Zealand (Fitzgerald and Karl
1979), and England (Churcher and Lawton 1989) have witnessed the decline in
many species of small mammal, bird, and herpetofauna. It is believed that cats
had a large role in the decline of these species. During recent attempts to
reintroduce the ring-tailed possum (Pseudocheims forbesi). cats killed nearly
75% of the animals (Anderson 1992). In Florida, cat density was negatively
correlated with density of the endangered Anastasia Island Beach mouse (Frank
1993).
Model Validation
The conclusions made in this chapter are based on the assumption that
the model accurately predicts the future persistence of S. p. hefneri. Proper


214
Table B.lThe variables used in the descirminant function analysis.
site
code
%gcover
%ccover
H'
hgt
%bf
%clump
area
dwat
dpeop
isol
1
2
71
30
2.05
42.1
3
46
1.05
1
270
40
2
2
64
28
2.17
34.1
2
37
1.22
108
90
40
3
1
44
19
1.24
20.2
0
4
1.62
90
540
376
4
2
77
25
1.44
15.0
15
20
2.92
1
2700
166
5
1
35
5
0.93
10.0
2
5
2.43
18
660
153
6
1
70
35
1.76
15.0
7
2
2.75
1
2880
100
7
2
72
19
1.14
18.4
1
27
3.89
135
1170
213
8
2
57
8
2.02
22.9
20
19
3.9
135
1710
124
9
2
78
20
1.25
19.4
2
30
4.86
180
1476
163
10
2
83
6
1.04
41.5
3
57
1.27
90
900
333
11
2
58
9
1.21
42.5
20
18
5.18
90
1
51
12
2
45
4
1.27
20.4
3
25
2.33
135
1710
124
13
2
55
8
1.33
40.2
15
32
3.40
98
1
51
14
2
60
6
1.31
28.5
3
35
5.83
136
1836
190
15
1
32
20
0.67
19.0
4
13
1.15
1
720
318
16
1
47
8
1.28
15.9
29
3
0.69
1
1680
153
17
1
35
14
1.49
12.1
7
2
1.04
1
2160
628
18
1
70
10
1.55
20.5
33
4
0.55
60
600
168
19
1
62
3
0.96
18.9
2
15
1.50
110
550
280
20
1
51
13
1.15
17.8
5
10
1.00
1
575
259
21
2
52
12
1.45
24.6
7
22
3.00
1
470
232
22
1
51
5
1.32
22.0
7
15
0.30
90
480
211
23
1
41
3
1.27
21.0
5
9
0.50
130
870
226
24
1
46
15
1.7
26.2
4
7
0.50
250
540
251
25
2
61
19
1.81
13.9
0
52
1.56
1
1500
353
26
2
68
25
1.2
36.6
12
36
0.45
150
1100
405
27
1
55
19
0.91
29.6
2
16
0.30
10
1560
483
28
1
93
18
1.56
9.1
10
11
3.50
110
6000
2124
29
1
50
9
1.42
20.3
1
17
0.78
500
2000
403
30
2
98
12
1.85
7.2
2
46
2.50
100
3000
472


89
Site 9
A cop
a a a a a a
MOOr
A60M -

A64M -

A66M -

A68M -

A76F -

A176F -

i i i i i i t r~
1 2 3 4 5 6 7 8
Site 13
A57F -
A58F -

A59M -

A55M -

A75M -

A86F -

A186M -

T
1 2 3 4 5 6 7 8
Trap Session
Figure 4.7 The duration that radio-collared Lower Keys marsh rabbits
(Sylvilagus palustris hemeril were followed at sites #9 and #13.


23
populations, both stochastic and environmental variation can substantially alter
sex and age ratios.
Genetic variability, at both the individual and population level, may also
be important. Highly fragmented, isolated populations are at risk of loosing
both types of variation. When a population is small and isolated, the chance of
a mating between close relatives increases. These matings can produce young
with a higher proportion of homozygous loci, thus potentially decreasing the
number of heterozygous loci and increasing the number of deleterious recessive
genes (Packer 1979). Certain heterozygous loci have been correlated with
greater tolerance to environmental variations in some species (Lemer 1954).
This increase in homozygous deleterious genes, combined with a
decrease of heterozygote loci, is believed to cause greater mortality, and
reduced fecundity, creating a phenomena called inbreeding depression (see
Lacy 1993). Loss of genetic variability among individuals or between
populations may contribute to inbreeding depression, and may decrease the
overall chance that the population can evolve to meet new environmental
conditions. Although loss of genetic variability in insular populations is well
documented (Kilpatrick 1981, Berry 1986) its overall effect on population
persistence is less known.
Inbreeding depression was documented early in domesticated animals
(Wright 1977, Falconer 1981), laboratory animals (Strong 1978) and more


119
Table 5.4Goodness-of-fit test for habitat use by Lower Keys marsh rabbits on Boca
Chica/Geiger Key. A + denotes that a habitat was used significantly more than
expected, a refers to a habitat that was used significantly less, and a 0 means the
habitat was neutral with respect to use according to the Bonferonni confidence
intervals.
March
Site
X2
Low
marsh
Mid
marsh
High
marsh
Hammock
1
6.73
0
0
0
0
4
8.14
0
0
0
-
8
23.02
0
+
0
0
9
3.95
0
0
0
0
10
14.72
-
0
+
0
11
15.13
-
+
0
0
12
31.42
-
0
+
0
13
49.38
+
0
0
Total
Mr)
3(+)
2(+)
l(-)
July
Site
X2
Low
marsh
Mid
marsh
High
marsh
Hammock
1
5.91
0
0
0
0
4
10.33
0
0
0
0
8
81.07
-
0
0
0
9
1.03
0
0
0
0
10
9.75
-
0
+
0
11
11.05
-
+
0
0
12
30.17
-
0
+
0
13
41.28
0
+
0
-
Total
4(-)
2(+)
2(+)
l(-)


Ill
Harrison (1991, 1994) and Thomas (1994) suggested that species that
inhabit early successional habitats that tend to be temporally patchy are more
likely to exist in true metapopulations; Murphy et al. (1990) extended this to all
species that display strong habitat specificity to a habitat that is fragmented.
The marsh rabbits high marsh transition zone habitat may differ from stricter
definitions of early successional habitat because it does not readily succeed to
other habitats. Frequent disturbances from tidal inundation and storms,
maintains the habitat over time. However, the Lower Keys marsh rabbit is
highly habitat specific, and high marshes are highly fragmented in the Lower
Keys. This fragmentation is mainly due to development.
Further examination of the consistently and variably occupied patches
indicates that not all patches appear to be of equal value in terms of
reproductive potential. Some of the variable patches may be sinks supported
by nearby source populations. However, not all of the variable patches were
sinks and not all of the consistently occupied sites were sources, indicating that
habitat quality is not the only factor elucidating the reproductive potential of a
patch. Local extinctions that occurred at the variable sites may have been
partly random, and perhaps all sites are variable if a long enough study was
conducted.


157
Simulation Results
The simulations using the exponential growth curve and the logistic
growth curves predicted nearly opposite outcomes for S. p. hefneri (Figure
6.3). The logistic growth curve simulation predicted that the marsh rabbit
metapopulation was stable and would vary around 1,200 rabbits (168 adults).
There were approximately 35 patches occupied during any time. The
exponential growth curve simulation estimated that S. p. hefneri would go
extinct within the next 20-30 years if all the parameters remained the same.
Because the total estimated abundances at the patches were predominantly
below carrying capacities at all of the sites and the mortality rate estimated
from the 5 main sites was high, it was determined that the exponential growth
curve was more biologically defensible. Because the exponential growth curve
more accurately mimicked population growth and was more conservative, it
was used in all of the scenarios.
Since extinction was imminent without increasing mortality or
decreasing reproduction, the sensitivity analysis only looked at increases in
survivorship and increases in reproduction. Each parameter was varied by
increasing it 25% and 50%. None of the variations appeared to stabilize the
population around a single point, however, 5 of the manipulations slowed the
extinction time to beyond 50 years (Table 6.5). Changes in the nestling
survivorship rate and the adult survival rates prolonged persistence the most.


183
Figure 6.6Results of Scenario #3 of the Lower Keys marsh rabbit (Sylvilagus
palustris hefneri) PVA model, where marsh rabbits were reintroduced to all vacant
patches.


161
if the results of the simulation model are correct, the Lower Keys marsh rabbit
is doomed to extinction in the next 20-30 years.
Migration reduction, hurricanes, mild diseases, and habitat loss, will not
significantly speed the decline of the Lower Keys marsh rabbit. A severe
disease could cause the extinction of the entire metapopulation within 3 years.
Harrison (1991, 1994) suggested that most populations that appear to
exist as stable metapopulations are actually on their way to extinction because
most metapopulations do not exist in the classic metapopulation sense with
patch recolonization exceeding patch extinction over long periods of time.
Most of these doomed metapopulations are actually mainland-island
metapopulations (where a large population sustains smaller populations) or
non-equilibrium metapopulations (where there are insufficient movements
between patches). The Lower Keys marsh rabbit differs because although it
does appear to exist in the classic metapopulations structure in many ways, it is
not lack of movements between populations that threatens its persistence.
Using the simulated management technique of decreasing cat predation
(scenario #1), the model predicted that the population would fluctuate around a
stable population size and would not experience extinction during the next 50
years. Other simulated management techniques such as recolonization and
reducing vehicular mortalities were not successful in preventing extinction.
This result indicates that the cat predation on the Lower Keys marsh rabbit is a


ACKNOWLEDGMENTS
I wish to thank Steve Humphrey for his advise and untiring support
during all of the phases of my Ph.D. work. Discussions with Buzz Holling and
John Eisenberg gave me new perspectives and insights about the science of
conservation biology. George Tanner and Lyn Branch provided constructive
comments on both my proposal and this manuscript.
This research would not have been possible without the funding from
the U. S. Navy (USN). Don Wood of the Florida Game and Fresh Water Fish
Commission (FGFWFC) effectively and efficiently administered the grant.
Additional financial support was provided by the U.S. Fish and Wildlife
Service (USFWS) and the Nature Conservancy (TNC). Many individuals
provided additional logistic support in the Lower Florida Keys: Annie
Simpkins and Amim Sheutz (USN), Tom Wilmers, Stuart Marcus, John
Andrew, Mark Yanno, and Jane Tutton (USFWS), Randy Tate (TNC), Phil
Frank (FGFWFC), Bill Keogh (wildlife photographer), and Lee Irby (long
suffering husband). My friends in both Gainesville and the Keys remained
faithful throughout my multiple moves across Florida. Many fellow graduate
students helped during the planning, analysis, and writing of my dissertation.
n


209
Figure A. 5A map of Summerland and Ramrod Keys. The habitat patches used in the
study are numbered.


103
several sites on 1 key. The distance to the nearest occupied patch was
calculated using this assumption.
Statistical analyses were done using procedures STEPDISC and
DISCRIM (SAS Institute, Inc. 1985). First, a stepwise discriminant function
analysis was used to select a subset of variables that produced a good
discrimination model. A value of P < 0.15 was used a criterion for model
inclusion. These variables were then entered to determine a discriminant
function for site classification. The model was validated using a jackknife
procedure (Capen et al. 1986). The jackknife procedure classifies each sample
using the discriminate equation derived form all samples except the 1 currently
being classified (Lachenbruch 1975).
Results
Pellet Degradation Rate and Pellet Size
During the first survey, pellets began to disappear off the grid after 7
weeks (Figure 5.2), indicating that during the dry season any sampling period
less than 7 weeks should suffice for density estimation based on pellet counts.
To compensate for a reduced persistence time anticipated for the wet season, a
sampling period of 1 month (T = 30 days) was chosen for pellet accumulation
before making counts. During the following 7 surveys, this duration was found
to be adequate for all seasons. In each session, pellets under trees in the litter


99
Fecal pellets were dried and crumbled to prepare microscope slides
following standard procedures (Johnson et al. 1983). One hundred microscope
fields (20/slide) were examined for each sample. Plant species were identified
by comparing epidermal structures from the samples with a reference collection
of plant tissues collected in the area. Average relative densities were
calculated for each food item per pellet-group sample.
Relative densities of vegetation in the pellets may not be directly
comparable to the amount of each vegetation ingested. Variation in the
digestion and retention rate of each type of vegetation can bias the results
(Wallage-Drees et al. 1986, Westoby et al. 1976, Batzli and Pitelka 1971).
These biases can be quantified by comparison of fecal and stomach samples,
but this is impractical in many studies (Johnson and Pearson 1981), especially
those involving endangered species.
To avoid these problems, and still obtain data on the differences
between available vegetation and vegetation in the rabbit's diet, we compared
the species available and the species used. We assumed that although different
types of vegetation vary in their digestibility, most vegetations would still be
present in the fecal pellet sample if they had been ingested. Available
vegetation (ground cover <1.5 m) at the 5 sites was estimated using the line
intercept method (Canfield 1941). Ten 5-m long transects were randomly


44
Table 3.1-Number of Lower Keys marsh rabbits (Svlvilagus palustris hefneri)
trapped at each site in the Lower Keys of Florida from June 1991 to May 1993.
Site
Ownership
Male
Female
Total
3
Navy
1
0
1
4
Navy
2
1
3
7a
Navy
4
3
7
8a
Navy
8
4
12
9a
Navy
6
3
9
10a
Navy
4
1
5
13a
Navy
4
4
8
16
Navy
1
0
1
19b
Navy
2
1
2
22
Private
1
2
3
34
USFWSC
1
2
3
a One of the 5 main study sites.
b The Navy-owned Saddlebunch site.
c United States Fish and Wildlife Service.


234
Lazell, J. D., Jr. 1989. Wildlife of the Florida Keys: a natural history. Island
Press, Washington, D. C. 250 pp.
Lazell, J. D., Jr. 1984. A new marsh rabbit (Svlvilagus palustris) from Florida's
Lower Keys. Journal of Mammalogy 65:424-432.
Leek, C. R. 1979. Avian extinctions in an isolated tropical wet-forest preserve,
Ecuador. The Auk 96:343-352.
Lee, K. 1993. Compass and gyroscope: Integrating science and politics for the
environment. Island Press, Washington D. C., 243 pp.
Lemer, I. M., 1954. Genetic homeostatis. John Wiley and Sons, New York, 134 pp.
Leslie, P. H. 1945. On the use of matrices in certain population mathematics.
Biometrika 33:183-212.
Levins, R. 1969. Some demographic and genetic consequences of environmental
heterogeneity for biological control. Bulletin of the Entomological Society
of America 15:237-240.
Levins, R. 1970. Extinction, Pages 70-107 in Some mathematical questions in
Biology (M. Gerstenhaber, ed.), American Mathematical Society,
Providence, Rhode Island.
Liberg, O. 1985. Food habits and prey impact by feral and house-based domestic
cats in a rural area in southern Sweden. Journal of Mammalogy 65:424-432.
Linkkila, T. E. 1971. Influence of habitat upon changes within the interspecific
Connecticut cottontail populations. Agricultural Experiment Station,
University of Connecticut, 70 pp.
Lipsey, M. W. 1990. Design sensitivity: statistical power for experimental
research. Sage Publications, Newbury Park, California, 207 pp.
Loiselle, B. A. and W. G. Hoppes. 1983. Nest predation in insular and mainland
lowland forest in Panama. Condor 85:93-95.
Long, R. W. 1974. Origin of the vascular flora of southern Florida. Pages 118-126
in Environments of South Florida: Present and past (P. J. Gleason, ed.),
Miami Geological Society, Florida.


Log Isolation (m)
131
10000
1000
100
0.1 1.0 10.0 100.0
Log Area (ha)
o
<&>
o
o
o
Q
O o U
o
o
o
Q#^§)
#
1 I I I I I I I I
O Empty
O Variable
Occupied
n 1ii i ii i
Figure 5.6 Fifty-nine sites plotted with respect to Lower Keys marsh rabbit
(Sylvilagus palustris hefneri) occupancy by logio area (ha) and logio isolation (m).
Empty patches never had marsh rabbits present, variable patches had marsh rabbits
during at least 1 session, and occupied sites had rabbits during all of the sessions.


114
Kodric-Brown 1977) in the context of classic island biogeography, or internal
rescue effect as it is applied to local patches within a metapopulation (Hanski
1982, Gotelli 1991, Holt 1993). If population numbers have been lowered at a
certain patch, new emigrants may reduce population variability, increasing the
persistence time of the local population.
Any plans to reintroduce the Lower Keys marsh rabbit should include
inoculating near by, multiple patches simultaneously. Not all of the patches
used in the reintroduction effort have to be high quality because even the
variably occupied patches were capable of supporting marsh rabbits that
produced young. Saddlebunch, Big Torch, and Middle Torch Keys all contain
multiple habitat patches. Each Key contains at least 1 patch that was classified
as being a consistently occupied patch. These Keys should be given the
highest priority for proposed reintroduction sites.
In conclusion, for S. p. hefneri, habitat quality and incidence-function
variables (area and isolation) interact to determine patch occupancy. Not all
occupied patches are of the highest habitat quality, and not all of the variable
or empty patches are empty because of low habitat quality. Several of the
empty habitat patches are sufficiently good habitat to support marsh rabbits,
but are too isolated from currently occupied habitat patches to sustain
populations. Overall, the Lower Keys marsh rabbit metapopulation did appear


62
Home Range
All juvenile rabbits that were followed for more than 1 month (12
locations) were used for analysis of home ranges. The 12-location criterion
was chosen because few juveniles were followed for more than 1 month.
Attempts to determine a minimum number of locations for the adult home-
range analysis by fitting the data to a negative exponential equation failed.
None of the individuals had home-range sizes that reached an asymptote,
despite the fact that some rabbits were followed their entire lives (Figures 4.1
and 4.2). Therefore, a minimum number of 30 locations (2.5 months) was
chosen based on other studies of similar cottontail rabbits (Dixon and Chapman
1980; Kjolhaug and Woolf 1988). Computer program HOME RANGE
(Ackerman et al. 1990) was used to determine sizes of minimum home ranges
and to plot movements.
Because of the nature of Lower Keys topography and patterns of home
ranges of marsh rabbits, several common methods of home-range estimation
were excluded. The minimum convex polygon method (Hayne 1949) was
rejected because it assumes that all areas within the perimeter of the outermost
locations are used by the animal (White and Garrott 1990). Most of the marsh
rabbits' home ranges contain unused areas, such as bodies of water, roads, or
other man-made structures. One of the most statistically rigorous methods of
calculating home-range size is the 95% probability ellipse (Jennrich and Turner


35
Demographics
The number of marsh rabbits caught on the 5 main sites varied from a
high of 18 rabbits during November 1991 and February 1992 to a low of 9
during November 1992 (Figure 3.3). The small number of rabbits at each site
precluded the use of capture/recapture statistics (Pollock et al. 1990);
population number at each site was estimated using the minimum number
known alive (MNA; Hilbom et al. 1976).
When all of the male and female individuals caught over the 8 trapping
sessions at the 5 Boca Chica sites were counted, there appeared to be more
male than female rabbits on the 5 sites during the study (Table 3.3). The sex
ratio significantly male-biased in only the subadults; nearly equal numbers of
male and female juveniles and adults were captured on the 5 grids.
Trappability (the probability that a rabbit was captured more than once
during the 10-day trapping session) differed among the demographic groups
(Table 3.5). Male and female trappability was high and not statistically
different. Juvenile trappability was lower, and females were more likely to be
retrapped during a trap session than were males. Subadult trappability was
high for the males, and very low for females. Sample sizes for the trappability
comparisons were small.


185
Figure 6.7Results of Scenario #4 of the Lower Keys marsh rabbit (Svlvilagus
palustris hefnerO PVA model. In the mild disease simulation, 25% of all nestling and
adult rabbits died at infected patches and each patch had a 75% chance of getting
infected each year. In the severe disease simulation, 75% of all nestling and adult
rabbits died.


116
Table 5.1--Relative density (%) each plant species in fecal pellet group samples (N =
40) of Svlvilagus palustris hefneri.
Plant species and family
X
SE
SDorobolus vireinicusPoaceae
35.74
5.21
Soartina spartinaePoaceae
17.33
3.87
Laeuncularia racemosaCombretaceae
10.25
2.49
Borrichia frutescensAsteraceae
8.40
1.51
Rhizophora maneleRhizophoraceae
6.81
1.97
Andropoeon elomeratus-Poaceae
6.14
2.04
Eleocharis cellulosa--Cvperaceae
3.75
1.63
Muhlenbereia filipesPoaceae
2.27
1.32
Tvpha latifolia-Tvphaceae
2.10
0.77
Coccoloba uviferaPolvgonacea
1.35
0.81
Fimbristvlis castaneaCvperaceae
0.76
0.38
Jacquinia kevensisTheophrastueae
0.75
0.30
Salicornia virginicaCheropodiaceae
0.63
0.31
Sesuvium maritimumAizoaceae
0.63
0.10
Avicennia eerminansAvicennianceae
0.38
0.23
Baccharis halimifoliaAsteraceae
0.31
0.18
Erithalis fruticosaRubiaceae
0.24
0.17
Fimbristvlis spathaceaCvperaceae
0.23
0.23
Mavtenus Dhvllanthoides-Celastraceae
0.17
0.17
Unidentified material
1.64
0.50


102
transects. The Shannon-Weaver diversity index (H) was chosen because
species composition and abundance were taken from transect data (Shannon
and Weaver 1949). In addition, the percent cover of individual species found
to be important to marsh rabbit diet and microhabitat were also included.
Results from the most recent vegetation survey (July 1993) were used in the
analysis. Landscape characteristics (area, distances to water, occupied
domiciles, and other marsh rabbit populations) were measured from USGS
topographic maps (1:24,000).
Variables were screened for normality and significant inter-correlation
before analysis. Non-normal variables were subjected to logarithmic and
square-root transformations. If a significant correlation was found between
variables, the variable with the most meaningful biological interpretation was
retained.
Two DFAs were performed on the data. The first compared the empty
sites to those that were occupied by rabbits during some portion of the study
(variable and consistently occupied sites). The second DFA compared variable
sites to the consistently occupied sites. Results of the second discriminant
function were used to reclassify the empty sites, to determine (based on habitat
features alone) if they are suitable for marsh rabbit recolonization. It was
assumed that if reintroduction was to occur, rabbits would be released at


224
Brown, J. H. 1971. Mammals on mountain tops: nonequilibrium insular
biogeography. American Naturalist 105:467-478.
Brown, J. H. and A. Kodric-Brown. 1977. Turnover rates in insular biogeography:
effects of immigration on extinction. Ecology 58:445-449.
Buechner, M. 1987. Conservation in insular parks: simulation models of factors
affecting the movement of animals across park boundaries. Biological
Conservation 41:57-76.
Burgman, M. A., Ferson, S. and H. R. Akfakaya. 1993 Risk assessment in
conservation biology. Chapman and Hall, London. 314 pp.
Burt, W. H. 1943. Territoriality and home range concepts as applied to mammals.
Journal of Mammalogy, 24:346-352.
Byers, C. R., Steinhorst, R. K. and P. R. Krausman. 1984. Clarification of a
technique for analysis of utilization-availability data. Journal of Wildlife
Management 48:1050-1053.
Canfield, R. H. 1941. Application of the line intercept method in sampling range
vegetation. Journal of Forestry 39:388-394.
Capen, D. E., Fenwich, J. W., Inkley, D. B. and A. C. Boynton. 1986. Multivariate
models of songbird habitat in New England forests. Pages 171-176 in
Wildlife 2000: Modeling habitat relationships of terrestrial vertebrates (J.
Vemer, M. L. Morrison, and C. J. Ralph, eds.), University of Wisconsin
Press, Wisconsin.
Carlson, P. C., Tanner, G. W., Wood, J. M., and S. R. Humphrey. 1993. Fire in
Key deer habitat improved browse, prevents succession, and preserves
endemic herbs. Journal of Wildlife Management 57:914-928.
Carr, A. F., Jr. 1939. Notes on the escape behavior in the Florida marsh rabbit.
Journal of Mammalogy 20:322-325.
Caughley, G. 1994. Directions in conservation biology. Journal of Animal Ecology
63:215-244.
Cervantes, F. A. and J. Martinez. 1992. Food habits of the rabbit Romerolagus
diazi (Leporidae) in Central Mexico. Journal of Mammalogy 73:830-834.


239
Schoener, T. W. and D. A. Spiller. 1987. High population persistence in a system
with high turnover. Nature 330:474-477.
Scribner, K. T., Chesser, R. K. and R. J. Warren. 1983. Spatial and temporal
genetic variability of the Eastern cottontail on west Texas playa basins.
Journal of Mammalogy 64:287-294.
Schribner, K. R. and R. J. Warren. 1986. Electrophoretic and morphologic
comparisons of Svlvilagus floridanus and S. audubonii in Texas.
Southwestern Naturalist 31:65-71.
Scribner, K. T., and R. J. Warren. 1990. Seasonal demography and movements of
cottontail rabbits on isolated playa basins. Journal of Wildlife Management
54:403-409.
Seal, U. S. and R. C. Lacy. 1989. Florida panther viability analysis. Report to the
U. S. Fish and Wildlife Service. Apple Valley, Minnesota: Captive
Breeding Specialist Group, Species Survival Commission, IUCN.
Shaffer, M. L. 1981. Minimum population sizes for species conservation.
Bioscience, 31:131-134.
Shaffer, M. L. 1990. Population viability analysis. Conservation Biology 4:39-40.
Shermyen, A. H. 1993. Florida statistical abstract, 27th edition. University Press of
Florida, Gainesville, Florida, 410 pp.
Shields, D. W. 1960. Movement patterns of brush rabbits in North Western
California. Journal of Wildlife Management, 24:381-386.
Shinn, E. A. 1988. The Geology of the Florida Keys. Oceanus 31:47-53.
Simberloff, D. S. and L. G. Abele. 1976. Island biogeography theory and
conservation practice. Science 191:285-286.
Sjogren, P. 1991. Extinction and isolation gradients in metapopulations: the case
of the pool frog (Rana lessone). Biological Journal of the Linnean Society,
42:135-147.
Smith, A. R. and M. M. Peacock. 1990. Conspecific attraction and the
determination of metapopulation colonization rates. Conservation Biology,
4:320-323.


Discussion 110
Metapopulation Structure 110
Habitat Use 112
6 POPULATION VIABILITY ANALYSIS 135
Introduction 135
Multiple Levels of Population Viability Analysis 135
Population Viability Analysis Models 138
Costs and Benefits of Using a PVA Model 139
Methods 141
Population-level Parameters 141
Inter-population Level Parameters 147
Metapopulation Parameters 148
The Model 148
Scenarios 149
Scenario #1 Decrease Predation 149
Scenario #2 Decrease Road-kills 150
Scenario #3 Reintroduce Rabbits 150
Scenario #4 Disease 150
Scenario #5 Hurricanes 151
Scenario #6 Corridor Destruction 152
Scenario #7 Habitat Destruction 152
Results 153
Population Parameter Estimations 153
Density Estimation 155
Between Population and Metapopulation Parameters 156
Simulation Results 157
Discussion 160
Model Validation 162
Minimum Viable Population, Area, and Number of Patches 167
7 CONCLUSIONS AND MANAGEMENT
RECOMMENDATIONS 193
Population Viability Analysis (PVA) and Metapopulation
Dynamics 193
Management for the Recovery of the Lower Keys Marsh Rabbit.... 196
Primary Recovery Actions 196
Secondary Recovery Actions 198
Future Research 201


75
The definition of a habitat patch put forth in the methods section appears to be
valid.
Subadult marsh rabbits were more likely to cross these barriers and use
alternate types of habitat. However, their movements did appear to be
influenced by the surrounding habitat matrix. Concordant with other studies
on rabbit movements (Chapman 1971, Trent and Rongstad 1974), marsh
rabbits were more likely to cross the more densely vegetated native habitats
(transition zone, hardwood hammock, and mangrove) than the more open,
disturbed areas. Presence of these habitats around the natal patch appeared to
facilitate movement between patches of transition zone habitat, acting as
corridors in the highly fragmented landscape (Wilson and Willis 1975). These
results support the patch geometry models (Buechner 1987, Stamps et al. 1987)
rather than the random walk theory (Berg 1983).
Dispersal has not been well studied in other species of cottontails.
Results from the telemetry indicate that most long-distance dispersers were
young males, similar to most polygynous mammalian species (Dobson 1982).
The small patch size and low density of rabbits in this study provided a unique
opportunity to investigate the cause of dispersal. Because some of the males
dispersed despite the lack of other adult males at their natal patch, the
competition-for-mates hypothesis cannot entirely explain all of the movements.
Similarly, lack of potential fathers at most of the natal patches excludes the


Table 5.8A comparison of habitat measurements between good and marginal marsh rabbit habitat in the Lower Keys of Florida.
Habitat variable (Unit)
Constitent habitat (n=22)
Variable habitat (n=20)
Test statistic*
P
X(SE)
Untransformed
X(SE)
Untransformed
GCover(%)
8.2 (0.2)
67.2
7.2 (0.2)
51.8
-3.65 (T)
0.008
CCover(%)
3.6 (1.3)
13.0
3.9 (1.2)
15.2
1.04 (U)
0.31
Borrichia(%)
2.3 (0.3)
5.2
2.2 (0.3)
4.8
-0.17 (U)
0.86
Clump(%)
5.9 (0.3)
34.8
3.1 (0.2)
9.6
-7.73 (T)
0.001
MaxHgt(cm)
26.4 (3.2)
26.4
18.3 (1.1)
18.3
-2.47 (U)
0.02
H(#)
1.5 (0.1)
1.5
1.7 (0.1)
1.7
0.99 (T)
0.32
Area(ha)
1.8 (0.5)
5.8
1.7 (0.8)
5.5
-0.14 (U)
0.89
DPopulation(m)
5.5 (0.3)
244.7
6.1 (0.2)
445.9
1.93 (U)
0.06
DResidence(m)
5.8 (0.6)
330.3
5.5 (0.6)
244.7
-0.34 (T)
0.73
DWater(m)
3.5 (1.5)
33.1
2.7 (0.5)
14.9
-1.08 (T)
0.28
to
C/1
* For all normally distributed variables (or transformed variables) a two-sample t-test (T) was used, a Mann-Whitney U-test using normal approximation (U)
was used for all non-normally distributed variables.


xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID EK6F7WZOM_Q0NUE0 INGEST_TIME 2015-03-27T19:31:48Z PACKAGE AA00029816_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES


210
Figure A.6--A map of Big Torch and Middle Torch Keys. The habitat patches used in the
study are numbered.


METAPOPULATIONS OF MARSH RABBITS: A POPULATION
VIABILITY ANALYSIS OF THE LOWER KEYS MARSH RABBIT
(SYLVILAGUS PALUSTRIS HEFNERT)
By
ELIZABETH A. FORYS
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE
UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE
REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
1995


147
abundance, the density of adult rabbits was multiplied by the area of the patch.
If juvenile pellets were found at that grid during the census, then a stable age
distribution was assumed and the number of subadults, juveniles and nestlings
was included. If no juvenile pellets were found, then it was assumed that the
rabbits at that patch had not produced any young and only the number of adult
rabbits was used.
For patches where density estimates were not possible to obtain, the
carrying capacity of the patch was used as the initial abundance. Although this
may have been higher than the actual number of rabbits, these patches were all
small and less likely to have a major impact on the estimate of total abundance.
Inter-population Level Parameters
To determine the amount of movement between populations, 6
populations of marsh rabbits on Boca Chica Key and Saddlebunch keys were
trapped. Radio-collars were placed on all rabbits >300 g. This ensured that
rabbits of all developmental stages (except for the altricial nestlings) were
represented. Radio locations were ascertained 3 times per week, or daily if it
appeared the rabbit was making a long-distance movement. The distance and
type of habitat of each long-distance movement was determined (see Chapter
4), including the crossing of major barriers (e.g., bodies of water, roads,
runways).


Month
Figure 3.2Growth curves for 3 (1 male and 2 female) Lower Keys marsh rabbits (Svlvilagus palustris hefneri).


228
Fuerst, P. A. and T. Maruyama. 1986. Considerations on the conservation of
alleles and genic heterozygosity in small managed populations. Zoo Biology
5:171-180.
Gaines, M. S., Foster, J., Diffendorfer, J. E., Sera, W. E., Holt, R. D. and G. R.
Robinson. 1992. Population processes and biological diversity.
Transactions of the North American Wildlife and Natural Resources
Conference 57:252-262.
Gallagher, D. 1991. Introduction and time line to the history of the Florida Keys.
Pages 49-53 in The Monroe County environmental story (J. Gato, ed.). The
Monroe County Environmental Education Task Force, Big Pine Key,
Florida.
Gaston, K. J. and J. H. Lawton. 1987. A test of statistical techniques for detecting
density dependence in sequential census of animal populations. Oecologia
74:404-410.
Gilpin, M. E. 1980. The role of stepping-stone islands. Theoretical Population
Biology 17:247-253.
Gilpin, M. E. 1988. Extinction of finite metapopulations in correlated
environments. Pages 177-186 in Living in a patchy environment (B.
Shorrocks and I. R. Swinglands, eds.), Oxford Scientific, Oxford, England.
Gilpin, M. E. and M. E. Soul 1986. Minimum viable populations: processes of
species extinction. Pages 19-34 in Conservation biology: The science of
scarcity and diversity (M. E. Soul, ed.), Sinauer Assoc. Inc., Sunderland,
Massachusetts.
Ginzburg, L. R. Slobodkin, L. B., Johnson, K., and A. G. Bindman. (1982).
Quasiextinction probabilities as a measure of impact on population growth.
Risk Analysis 2:171-181.
Gleason, P. J. 1984. Saving the wild places a necessity for growth. Pages 1-34 in
P. J. Gleason ed.), Environments of South Florida: present and past.
Miami Geological Society, Florida.
Goodman, D. 1987. The demography of chance extinction. Pages 11-34 in Viable
populations for conservation (M. E. Soul, ed.), Cambridge University Press,
Cambridge.


170
Table 6.2Mean number of pellets produced by captive marsh rabbits
(Sylvilagus palustris hefneri) during a 24-hour period. Equal numbers of adult
males and females were used each session.
Session
Date
Number of
rabbits
Mean number
of pellets
SE
1
Feb 1991
4
171.0
7.43
2
Jul 1991
4
120.0
10.61
3
Nov 1991
4
132.0
13.37
4
Mar 1992
4
134.0
17.20
5
Jul 1992
4
104.8
10.50
6
Nov 1992
4
120.8
14.22
7
Mar 1993
4
107.3
10.72
8
Jul 1993
4
103.8
7.13


Core Area (ha)
84
Figure 4.2The cumulative core area measurements of 6 radio-collared female
Lower Keys marsh rabbits (Sylvilagus palustris hefneri).


8
obviously important for patch colonization and establishment. These
movements also may stabilize local population variability of occupied patches
and increasing the persistence time for the local population through the rescue
effect (Brown and Kodric-Brown 1977). Patch configuration and the type of
habitat between the patches (including developed land) may ultimately have a
large impact on the individuals movements.
At the scale of the entire metapopulation, the correlation in
environmental variation among patches and the frequency of large-scale
environmental catastrophes, is vitally important (Quinn and Hastings 1987,
Gilpin 1988, Hanski 1989, Stacey and Taper 1992). If all patches are subjected
to the same adverse environmental conditions at the same time, then patch
extinctions may be correlated, leaving few or no source populations to colonize
the extinct patches.
Additionally, the cause of local extinctions may be important to the
overall metapopulation dynamics. It has been suggested (Thomas 1994) that
most local extinctions are due to deterministic changes in the environment of a
patch that render a patch unsuitable. If deterministic, these conditions are
likely to persist following the local extinction leaving the patch empty, but
unsuitable. Therefore, most empty patches are currently unsuitable for
occupancy and few suitable patches become empty. Some populations, either
because of their small size or inherently high variability, may be subject to


33
but only 3 rabbits (1 male, 2 females) were caught during this stage and
survived to adulthood. Marsh rabbit age was correlated with external
measurements using these 3 individuals.
Both body mass and total body length were found to significantly (P
<0.05) predict marsh rabbit age for all 3 rabbits (Table 3.2). Ear length and the
length of the right hind foot were not significant predictors in one of the female
rabbits. Body mass has been used as an indicator of age in cottontails (Lord
1963), but the length of the right rear foot is used more often (Bothma et al.
1972). Because measuring the length of a non-sedated marsh rabbit is difficult,
the marsh rabbit body mass was used as an indicator of age in conjunction with
appearance of external sexual organs. Body mass can be influenced by
nutritional and physiological variations but was the most objective and accurate
age indicator available. Body mass of the marsh rabbits increased almost
linearly with age for the 3 rabbits and plateaued at approximately 1,100 grams
(Figure 3.2) as the rabbits approached 1 year.
External measurements collected from 13 non-pregnant adult female
rabbits and 19 adult male rabbits were compared for differences using a
Wilcoxon test. In an attempt to use only measurements on fully grown rabbits,
only individuals that had been caught more than once as an adult were used.
No significant differences were found in any of the measurements between
sexes. Approximately half of all cottontial species are sexually dimorphic


11
Figure 1.1--Distribution of the 3 subspecies of marsh rabbit (Sylvilagus
palustris).


98
common log of area and isolation (distance to nearest occupied patch), and the
result was compared to theoretical patterns for classic, mainland-island and
source-sink (Figure 5.1).
Occupied and variable patches were further compared to determine if
they were sources and sinks. Liberal definitions of source and sink were used;
a source was any patch where reproduction occurred when occupied, and a
sink was any patch where only adult rabbits were present. Presence of adults
and juveniles was determined by measuring the size of the pellets found on the
grid. Because it was previously determined (in Chapter 4) that juveniles
remain in their natal patch until they become subadults, it was assumed that at
least 1 juvenile pellet present on the grid indicated presence of reproduction
occurring on that grid. For each habitat patch the number of surveys where
juvenile pellets were found was compared to the number of surveys occupied.
Dietary Analysis
Pellets for dietary analysis were collected from the pans beneath the
rabbit traps from the 5 sites on Boca Chica. To avoid experimental bias,
pellets were collected only from adult animals 1 time. At each of the 5 Boca
Chica sites, pellet samples were collected from eight rabbits, 2 from each sex
during both the wet and dry seasons, totaling 40 pellet-group samples.


43
Although the other hypotheses cannot be eliminated, it is highly
probable that some females were without mating opportunities during part of
the study. The low population density, combined with stochastic variation in
the sex ratio of the adults at a habitat patch, left at least 2 females (sites 8 and
9, trapping period one) without mates. These females did not produce a litter
during this period. The ability of males and females to travel between habitat
patches will be examined in Chapter 4. If these habitat patches are separate
populations, it appears that stochastic demographic variation can have a large
impact on the intrinsic rate of growth of the population (r).
This lower fecundity was matched with a mortality rate higher than most
non-hunted populations of wild rabbits (Trent and Rongstad 1974, Chapman et
al. 1982). The majority of the mortality was from anthropogenic sources
(vehicles, house cats), that are not generally density-dependent.


97
occurred from June 1991 to May 1993 (8 trapping sessions). On Saddlebunch,
trapping was conducted from June 1992 to May 1993 (4 trapping sessions).
Trapping grids were placed on each site, using unbaited collapsible National
live traps (80 x 30 x 30 cm), placed in a 6 x 6 array, spaced approximately 25
m apart. Each trapping session consisted of 5 nights where the traps were
open, 2 nights with the traps closed and another 5 nights with traps open.
Traps were checked twice daily, once in the morning and once in the evening
and were covered in burlap for shade.
All rabbits caught were sexed, weighed, and tagged (Monel no. 3,
National Band and Tag, Newport, KY). For the purposes of this part of the
study, adult marsh rabbits were those that weighed >1000 g, smaller rabbits
were classified as juveniles. Ten fecal pellets were collected in a pan
underneath each trap and the width and length of each pellet was recorded.
The average area (width x length) was used in the regression. Each rabbit was
used only once in the analysis.
Metapopulation Structure
Sites were classified based on site occupancy; if a site was never
occupied during the past 2 years it was classified as being empty, if it was
occupied at least once it was called variable and if it was consistently
occupied it was referred to as an occupied site. Sites were graphed using the


113
spartinae in the rabbit's diet is probably due to new growth during the wet
season.
Although marsh rabbits eat vegetation in proportion to its abundance,
they spend a disproportionate amount of time in the mid- and high marsh. This
may indicate that marsh rabbits use the mid- and high marsh for cover while
using all areas for foraging. Both the mid- and high marsh contain thick
ground cover (Borrichia in the mid-marsh and Spartina and Fimbrvstvlis in the
high marsh). In addition, all nests were found in the high marsh (see Chapter
3).
On a larger scale, the habitat model agreed with the habitat use/
availability results. The model indicated that thick ground cover (especially
clump grasses and B. frutescens) and distance between populations were
important factors in determining how consistently a habitat patch was
occupied. The thick grasses and forb provide the necessary cover for marsh
rabbits to nest and escape from predation. Severe predation may reduce
population size and increase the chance of stochastic demographic extinction
(Gilpin and Soul 1986). For small-to-medium size herbivores like the marsh
rabbit, predation may be the most important factor in determining local
abundance (Chapman et al. 1982).
Additionally, patches within close proximity of other occupied patches
benefit from emigration of individuals via the rescue effect (Brown and


Table 4.4A comparison between the wet and dry season home ranges (ha.) of Lower Keys marsh rabbits (Svlvilagus palustris hefheri).
The distance between the centers of each home range is compared to the radius of the 95% harmonic mean and core area estimates.
Wet season Dry season
ID
^locals
95%
harmonic
mean
Core
area
#locals
95%
harmonic
mean
Core
area
95%
harmonic
radius
Core
radius
Meters
moved
Core radius -
meters moved
A70M
36
0.81
0.24
52
2.79
0.83
50.79
27.65
15.52
12.13
A71M
36
2.99
0.90
31
1.31
0.54
97.58
53.54
24.50
29.04
A68M
64
3.79
0.90
78
1.37
0.47
109.86
53.54
38.11
15.43
A67M
45
4.03
0.96
54
0.56
0.14
113.29
55.29
37.00
18.29
A63M
124
2.28
0.57
93
5.74
1.60
85.21
42.61
14.14
28.47
A69M
60
6.31
2.04
78
3.61
1.42
141.76
80.60
13.01
67.59
A59M
43
1.95
0.87
36
1.48
0.46
78.80
52.64
45.45
7.19
A55M
36
2.16
0.80
34
0.91
0.40
82.94
50.48
24.02
26.46
A51M
73
0.53
0.33
49
0.45
0.28
41.08
32.42
5.00
27.42
A72F
36
1.15
0.37
71
3.26
1.23
60.52
34.33
39.11
-4.78
A53F
81
10.18
2.51
120
6.50
1.72
180.06
89.41
45.17
44.24
A50F
132
3.49
1.04
120
5.60
1.75
105.43
57.55
12.17
45.38
A58F
48
1.46
0.47
42
0.43
0.20
68.19
38.69
28.07
10.62
A57F
106
1.38
0.45
114
0.78
0.27
66.29
37.86
9.84
28.02
A52F
78
6.21
2.21
49
7.15
2.31
140.63
83.89
4.24
79.65


31
with special attention to the liver for signs of tularemia (Francisella tularensis).
Site description, carcass condition, and position was recorded for each death.
All rabbit carcasses thought to have been preyed upon were examined
for signs of trauma. The head, throat, and neck were examined for puncture
wounds. Hemorrhaging, particularly the presence of blood in the mouth, nares,
trachea or neck region indicated the rabbit was alive at the time of attack
(Hawthorne 1980). Feeding pattern on the carcass was also examined:
hindquarter feeding probably indicated scavenging, whereas feeding on the
shoulders and neck indicated possible predation (Hawthorne 1980).
When predation was suspected, the site was examined for sign on and
around the carcass. Tracks and scats were identified for potential information
on the predator or scavengers. Potential predators found in the Lower Keys
included Bald Eagles (Haliaeetus leucocephalus). Red-shouldered Hawks
(Buteo lineatus), eastern diamondback rattlesnakes (Crotalus adamanteus),
feral cats (Telis catus), raccoons (Procyon lotor) and possibly Black Vultures
(Corgyps atratus) and domestic house-based dogs (Canis familiaris).
Birds of prey tend to capture rabbits in the middle of the back, and will
kill using deep puncture wounds to the back and head (Hawthorne 1980).
They may take their prey back to a nesting area. Feral cat predation may be
assumed if the carcass has been dragged, eviscerated, or covered in dirt. Often
cats will leave tooth marks on every exposed bone of their prey (Anderson


59
Two general theories have been proposed to predict the direct an individual
will leave a patch. The random walk theory (Berg 1983) predicts that
individuals will disperse across random locations along the edge of a patch,
irrespective of habitat features. Patch geometry models (Stamps et al. 1987,
Buechner 1989) predict that individuals will exit patches at more permeable
areas. A landscape corridor may increase the patch edge permeability by
extending patch habitat (La Polla and Barrett 1993), and allowing individuals
to move from one habitat patch to another. The geometric and habitat features
that constitute a corridor from the animals perspective must be determined.
The objective of this part of the study was to use the information about
the spatial organization of S._g. hefneri and movements to determine how they
use the fragmented habitat in the Keys. Evidence supporting the hypothesized
patterns of spatial use will be evaluated. Three possible scenarios of habitat
use will be considered:
1. The Lower Keys marsh rabbit is confined to one habitat patch and is
incapable of moving between patches. All home ranges occur strictly within
the patch. Dispersal is within the patch, does not occur, or is unsuccessful
(rabbit leaves a patch but does not reach another patch and dies or fails to
reproduce). This will be called a relictual (Berry 1986) population.
2. Lower Keys marsh rabbits spend most of their lifetimes in a patch
but are capable of moving between patches. All home ranges occur strictly


198
proposal (#910130-44-1, Hammocks of the Lower Keys). This CARL proposal
is currently ranked at number 2 in the State of Florida and is available for
funding.
Because it will be difficult to lower the mortality from cats and to
acquire marsh rabbit habitat, additional subpopulations of marsh rabbits should
be founded in areas not occupied by domestic cats. A thorough survey of the
federally-owned back-country islands of the National Key Deer Refuge should
be conducted and marsh rabbits should be reintroduced to suitable islands.
Reintroduction to areas on Keys connected by US-1 should occur after the
number of cats has been reduced.
Secondary Recovery Actions
Even protected areas require habitat management to maintain the
maximum marsh rabbit carrying capacity of the habitat. The Lower Keys
marsh rabbit appears to rely on dense ground cover, particularly in the mid-
and high marsh (Chapter 5). Potential threats to this vegetation included exotic
vegetation, off-road vehicle (ORV) use, mowing, and trash dumping.
Australian pine (Casurina equisetifolia) and Brazilian pepper (Shinus
terebinthefolius) are rapid non-native colonizers in the transition zone that
inhibit or deter understoiy growth. These trees are particularly abundant on
Boca Chica Key, but are present on all of the Keys. Girdling trees with


CHAPTER 1
INTRODUCTION
The primary objective of this dissertation is to develop a population
viability analysis (PVA) for the endangered Lower Keys marsh rabbit
(Svlvilagus palustris hefneri). The Lower Keys marsh rabbit was first
described as a distinct subspecies by Lazell (1984). Lower Keys marsh rabbits
have a shorter molariform tooth row, a higher and more convex frontonasal
profile, a broader cranium, and a longer dentary symphysis than mainland and
Upper Keys rabbits.
Historically, the range of the Lower Keys marsh rabbit extended from
Big Pine Key to the southernmost of the Florida Keys, Key West (dePourtales
1877). During the 1970s and 1980s, a decline in Lower Keys marsh rabbits
was reported (J. Lazell, pers. comm.) and a study of the rabbit's status was
commissioned by the Florida Game and Fresh Water Fish Commission (Howe
1988). Howe (1988) recorded marsh rabbits present at 13 of J. Lazell's
(unpubl. data) original 17 sites and absent from 4 of the sites. The Lower Keys
marsh rabbit was listed as endangered by the Florida Fresh Water Fish
Commission in 1989 (F.A.C. 39-27) and by the U.S. Fish and Wildlife Service
in 1990 (U.S. Fish and Wildlife Service 1990).
l


13
Today, the Florida Keys are composed of two formations of limestone.
The Upper Keys (Soldier Key to the southeastern comer of Big Pine Key) are
composed of the Key Largo limestone and the Lower Keys (the majority of Big
Pine Key to Key West) are from the Miami limestone. The Miami limestone is
composed of small ovoid pellets of calcium carbonate (ooids; Hoffmeister
1974) that hardened when sea level fell and the land surface was exposed to
air.
Climate
Although several degrees north of the Tropic of Cancer, the close
proximity of the Gulf Stream and maritime influences produce a subtropical
climate in the Lower Keys (Chen and Gerber 1990). In 50 years of data
collection, the temperature in the Lower Keys has never fallen below freezing,
and the record low temperature is 41 F (5 C). Winter cold fronts are buffered
by the warm ocean water before proceeding to the Keys. Monthly average
temperatures differ only 15 F (9 C) from January to July (Figure 2.2). Daily
temperatures rarely vary more than 10 F (12 C) between the high and low.
There are distinct dry and wet seasons in the Lower Keys, although this
area is generally slightly drier overall than the Upper Keys and mainland
Florida (Chen and Gerber 1990). The dry season spans November through
April and accounts for less than a third of the annual precipitation. The


139
VORTEX (Lacy 1993) and RAMAS/metapop (Ak9akaya 1994) both provide
estimations about the persistence of a metapopulation using a Monte Carlo
simulation algorithm to generate stochasticity. VORTEX is slightly less
flexible in its treatment of population growth and is limited in the number of
subpopulations that can be modeled at once.
Costs and Benefits of Using a PVA Model
Although most scientists and managers consider a detailed PVA to be a
useful tool in making management decisions (Salwasser et al. 1984, Marcot
and Holthhausen 1987, Shaffer 1990), these models have recently come under
criticism. For many species, data are not currently available to conduct an in-
depth PVA, and by the time it is collected the species may be extinct (Maguire
1991, Boyce 1992). Some population dynamics such as density dependence
are poorly understood, regardless of the parameters measured in the field
(Hassell 1986). This lack of knowledge can severely detract from the most
well-developed PVA model. For other species, conducting a PVA may
estimate how long the species will persist but not indicate why the species is
declining (Caughley 1994). Managers, lawyers and the public often consider
the predictions of PVAs as facts, and do not give credence to the variance and
limitations of the model (Barthouse et al. 1984). Finally, few PVA models are


Number of Occupied Patches Total Estimated Abundance
1300
1200
1100
1000
900
800
0 10 20 30 40 50
Year
38
36
Populations
Severe disease
Mild disease
24
20 30
0
10
Year
40
50


120
Table 5.4 (continued).
November
Site
X2
Low
marsh
Mid
marsh
High
marsh
Hammock
1
30.42
-
+
0
-
4
4.02
0
0
0
0
8
17.05
0
+
0
-
9
3.95
0
+
0
0
10
8.97
-
0
+
0
11
11.46
-
0
0
-
12
31.42
-
0
+
0
13
5.43
0
0
0
-
Total
Mr)
3(+)
2(+)
Mr)
Grand Total
12(-)
8(+)
6(+)
Mr)


4
Spatial Structure
Most endangered species live in habitats that have become fragmented
(checkerspot butterfly: Murphy et al. 1990; forest-dwelling raptors: Thiollay
and Meyburg 1988, spotted owl: see Lamberson et al. 1994). For these
species, the spatial structure of the habitat patches may have large effects on
the species persistence. A PVA for a species existing in a fragmented habitat
must look at several spatial and temporal scales of resolution (Gaines et al.
1992, Lacy 1993).
Populations occurring in habitat patches may be locally subject to higher
risks of extinction than in continuous habitat due to factors (nest predation,
increased exotics, microclimate changes) that occur more frequently in smaller
areas that have high perimeter-to-area ratios (Lovejoy et al. 1986, Loiselle and
Hoppes 1983). These "edge effects" may be intensified in smaller habitat
patches, where nearly all of the patch is edge and little is interior (Wilcox and
Murphy 1985). Small patches will support fewer individuals, increasing the
rate of extinction caused by stochastic events (Gilpin and Soul 1986).
For populations of species with poor dispersal ability relative to the
distance between patches, the time to extinction will be equal to the time that
the last local population goes extinct (Hanski and Gilpin 1991). Each patch of
habitat represents a population of these species; surviving populations may be
said to be relictual (Berry 1986). Harrison (1991, 1994) uses the term non-


148
Metapopulation Parameters
To determine if the variation in population densities was correlated
among populations, a multivariate, repeated-step analysis of variance
(MANOVA) was performed on the density estimations collected during the 8
sampling sessions (PROC AUTOREG, SAS/ETS, Sas Institute 1984).
Estimated abundances of each of the populations over the 8 sessions was
compared to the other populations an the same key. Sites that never supported
any marsh rabbits were removed from the analysis.
The Model
Using the parameters described above, the simulation was repeated
1,000 times for each scenario (described below) and run for 50 years. Most
PVA models are run for 100 years however, the predicted increase in sea level
rise for south Florida (Ross et al. 1994) implies habitat changes and precludes
such a long-term outlook. To test the sensitivity of the input parameters, all of
the elements in the Leslie matrix were varied. This includes survivorship in
each of the 5 stages and reproduction in the adults.


I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope
and quality, as a dissertation for the degree of Doctor of Philosophy.
Stephen R. Humphrey, Chairman
Professor of Forest Resources 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 fully adequate, in scope
and quality, as a dissertation for the degree of Doctor of Philosophy.
Lyn C. Branch
Assistant Professor of Forest
Resources 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 fully adequate, in scope
and quality, as a dissertation for the degree of Doctor of Philosophy.
CULT^lJL
Eisenberg
Katharine Ordway^Prefessor of/
Ecosystem Conservation


159
50% extended the number of years until the metapopulation went extinct from
28 years (for the basic simulation) to 30 years. A severe disease had the largest
impact on the metapopulation persistence time; in this scenario the number of
years to extinction was reduced to 3. A mild disease only decreased the
persistence time by 8 years, producing a 20-year metapopulation persistence
time. None of the hurricanes produced long-term changes in the persistence
time of the metapopulation. A class 5 hurricane reduced the persistence time to
22 years and the class 4 and class 3 hurricanes reduced the time to 23 and 25
years, respectfully. None of the habitat loss scenarios (7a-c) had an effect on
the overall persistence time of the metapopulation.
When the mortality in all of the stages was reduced by 25%, the
scenarios had different effects on the persistence of the metapopulation.
Scenario #4 examined the effect of a mild and a more severe disease had on S.
p. hefneri (Figure 6.7). Although population numbers and the mean number of
populations occupied was slightly lower than in 1(a), neither a mild or a severe
disease appeared to have a large effect on population persistence.
When all of the Keys were simultaneously subjected to a hurricane
(scenario #5), the effects on persistence were greater and persistence decreased
with increasing hurricane impact (Figure 6.8). In all of the hurricane scenarios,
population size and metapopulation occupancy decreased slightly over the 50
years.


233
Kindvall, 0. and I. Ahln. 1992. Geometrical factors and metapopulation
dynamics of the Bush Cricket, Metrioptera bicolor Philippi (Orthoptera:
Tettigoniidae). Conservation Biology 6:520-529.
Kjolhaug, M. S. and A. Woolf. 1988. Home range of the swamp rabbit in southern
Illinois. Journal of Mammalogy 69:194-197.
Lachenbruch, P.A. 1975. Discriminant analysis. Hafner, New York. 128 pp.
Lack, D. 1933. Habitat selection in birds. Journal of Animal Ecology 2:239-262.
Lacy, R. 1993. VORTEX: A computer simulation model for population viability
analysis. Wildlife Research 20:45-65.
Lacy, R. C. and T. W. Clark. 1990. Population viability assessment of the eastern
barred bandicoot in Victoria. Pages 131-145 in The management and
conservation of small populations, (T. W. Clark and J. H. Seebeck), Chicago
Zoological Society, Brookfield, Illinois.
LaHaye, W. S., Gutirrez and H. R. Akfakaya. 1994. Spotted owl metapopulation
dynamics in Southern California. Journal of Animal Ecology 63:775-785.
Lamberson, R. H., McKelvey, K., Noon, B. R. and C. Voss. 1992. A dynamic
analysis of the northern spotted owl viability in a fragmented forest
landscape. Conservation Biology 6:505-512.
Lande, R. and G. F. Barrowclough. 1987. Effective population size, genetic
variation, and their use in population management. Pages 87-123, in Viable
populations for conservation (M. E Soul, ed.), Cambridge University Press,
Cambridge.
La Polla, V. N. and G. W. Barrett. 1993. Effects of corridor width and presence on
the population dynamics of the meadow vole (Microtus pennsylvanicus).
Landscape Ecology 8:25-37.
Laurence, W. F. 1982. Ecological impacts of tropical forest fragmentation on
nonflying mammals and their habitats. Ph.D. dissertation, University of
California, Berkeley, 259 pp.
Layne, J. N. 1974. The land mammals of South Florida. Pages 386-413 in
Environments of south Florida: Present and past (P. J. Gleason, ed.), Miami
Geological Society, Florida.


230
Hanski, L, Kuussaari, M. and M. Nieminen. 1994. Metapopulation structure and
migration in the butterfly Melitaea cinxia. Ecology 75:747-762.
Hanski, I. and M. Gilpin. 1991. Metapopulation dynamics: Brief history and
conceptual domain. Biological Journal of the Linnean Society. 42:3-16.
Hanski, I. and M. Gyllenberg. 1993. Two general metapopulation models and the
core-satellite species hypothesis. American Naturalist 142:17-41.
Harris, R. B. and R. W. Allendorf. 1989. Genetically effective population size of
large mammals: an assessment of estimators. Conservation Biology 3:181-
191.
Harris, H. and D. A Hopkinson. 1976. Handbook of enzyme electrophoresis. North
Holland Press, Amsterdam. 240 pp.
Harris, S., W. J. Cresswell, P. G. Forde, W. J. Trewhella, T. Woollard, and S.
Wray. 1990. Home-range analysis using radio-tracking data a review of
problems and techniques particularly as applied to the study of mammals.
Mammalian Review 20:97-121.
Harrison, S, 1991. Local extinction in a metapopulation context: an empirical
evaluation. Pages 89-103 in Metapopulation dynamics: empirical and
theoretical investigations (M. E. Gilpin and I. Hanski, eds.), Academic Press,
London.
Harrison, S. 1994. Metapopulations and conservation. Pages 111-128 in Large-
scale ecology and conservation biology (P. J. Edwards, R. M. May, and N.
R. Webb, eds.), Blackwell Scientific Publications, Oxford, England.
Hassell, M. P. 1986. Detecting density dependence. Trends in Ecology and
Evolution 1:90-93.
Hastings, A. 1991. Structured models of metapopulation dynamics. Pages 57-71 in
Metapopulation dynamics: empirical and theoretical investigations (M. E.
Gilpin and I. Hanski, eds.), Academic Press, London, England.
Hastings, A. and Wolin 1989. Within patch dynamics in a metapopulation.
Ecology 70:1261-1266.
Haugen, A. O. 1942. Home range of the cottontail rabbit. Ecology 23:351-367.


73
isolation at a more regional scale than those species that occupy more
permanent, interconnected habitats. Marsh rabbits throughout their range
occupy upper marshes along the coasts and interior wetlands, much of which is
patchily distributed. Before European colonization in the Keys, the transition
zone was probably more contiguous. Most of the habitat types in the Keys
exist in contiguous, concentric rings; habitat type is largely determined by
elevation.
Spatial Organization
Marsh rabbit home-range size was well within the range for cottontail
species (Chapman et al. 1982). The 95% harmonic mean estimates represent a
conservative estimate of the amount of habitat used; the core area estimate
delineated the area needed for more intense use. Despite strong seasonally of
the rainfall in the Keys, home ranges were consistent throughout the year.
Similar to the eastern cottontail and the swamp rabbit, there is little
overlap in home ranges of members of the same sex. Marsh rabbits may be
territorial within their sex. Most current definitions of territoriality specify that
an area must be used exclusively (at least within the sex) and that it must be
actively defended (Eisenberg 1981, Begon et al. 1990). To be truly territorial,
marsh rabbits would have to defend their home ranges. Territorial defense is
difficult to observe in the field, especially in a secretive, crepuscular species


REFERENCES
Ackerman, B. B., F. Leban, M. D. Samuel, and E. O. Garton. 1990. User's Manual
for Program Home Range, 2nd edition. Department of Fisheries and Wildlife
Resources, University of Idaho, Moscow. Forest, Wildlife and Range
Experiment Station, Technical Report 15, 79 pp.
Adler, G. H. and M. L. Wilson. 1987. Demography of a habitat generalist, the
white-footed mouse, in a heterogeneous environment. Ecology 68:1785-
1796.
Aguire, A. A. and E. A. Starkey. 1994. Wildlife diseases in U.S. National Parks:
Historical and coevolutionary perspectives. Conservation Biology 8:654-
661.
Akfakaya, H. R. 1994. RAMAS/metapop: Viability analysis for stage-structured
metapopulations. Applied Biomathematics, Setauket, New York. 185 pp.
Alldredge, J. R. and J. T. Ratti. 1986. Comparison of some statistical techniques
for analysis of resource selection. Journal of Wildlife Management 50:157-
165.
Anderson, I. 1992. Alien predators devastate Australian wildlife. New Scientist
135:9.
Anderson, T. E. 1969. Identifying, evaluating, and controlling wildlife damage.
Pages 497-520 in Wildlife management techniques, 3rd edition (R. H. Giles,
ed.), The Wildlife Society, Washington, D.C.
Anderson, P. K. 1989. Dispersal in rodents: A resident fitness hypothesis.
American Society of Mammalogists, Special Publication Number 9. 141 pp.
ARC/INFO, Environmental Systems Research Institute, 1990. ARC/info Users
Guide, Environmental Systems Research Institute, Redlands, California. 124
pp.
222


Figure 5.8 Patch occupancy by Lower Keys marsh rabbits (Svlvilagus palustris hefneri) during the last 6 pellet sampling sessions. A
solid circle indicates an occupied patch, a dashed circle represents a patch occupied by adults only, and an open circle indicates a vacant
patch. Dashed lines between patches indicate patches within dispersal distance from each other. The solid lines are roads.


176
Year
Figure 6.2-Departures in precipitation and temperature during the past 10 years, using
the past 30 years as a base.


208
Figure A.4A map of Cudjoe Key. The habitat patches used in the study are numbered.


112
Habitat Use
Three different temporal and spatial scales provided complementary
information on marsh rabbit habitat use. Marsh rabbits forage, nest, and hide
in the dense ground vegetation of the transition zone. With some exceptions,
S. p. hefheri appeared to feed on the plant species most abundant in its habitat.
The major plant species in the rabbit's diet did not appear to vaiy among sites,
between the wet and dry seasons (with one exception), or between males and
females. These results differ from the majority of leporid dietary studies.
Few studies have investigated the effect of site on diet. Wallage-Drees
et al. (1986) found that the diets of Orvctolagus cuniculus differed among sites.
These sites varied greatly in their vegetative composition; a portion of the sites
were in woodlands, while others were in heathlands and abandoned arable
fields. In contrast, hefneri is very specialized in its habitat use. Most
Lower Keys marsh rabbits inhabit the transition zone. Vegetation in transition-
zone areas may vary in proportion between sites but is similar in composition.
The dietary requirements of the marsh rabbit may only be met in the transition-
zone areas in the Lower Keys.
Seasonal changes in leporid diets have been extensively documented
(Cervantes and Martinez 1992, Chapman et al. 1982). In most of these studies,
the climate and vegetation changed extensively between seasons. The weather
in the Lower Keys is relatively invariant. The variation in the proportion of S.


46
Table 3.3-The number of sexually mature Lower Keys marsh rabbits
(Sylvilagus palustris hefneri) at the 5 main study sites on Boca Chica Key.
Session Site
7 8 9 10 13 Total
1 Male 0
Female 0
2 Male 0
Female 0
3 Male 2
Female 2
4 Male 2
Female 2
5 Male 2
Female 1
6 Male 1
Female 1
7 Male 1
Female 0
8 Male 1
Female 1
0 0 112
1112 5
10 113
1112 5
2 1117
1112 7
11116
2 1117
13 12 9
110 14
2 1116
2 2 0 1 6
11115
110 13
0 1114
0 2 0 2 5


Juv. Sub. Adult-1
Age Class
Adult-2
Figure 3.5-Percent mortality of Lower Keys marsh rabbits (Svlvilaeus
palustris hefneri) by sex and age class. Sample size of each age/sex cohort
indicated above each bar.


Table 3.2Correlation coefficients between body mass, body, ear, and right foot length and rabbit age for three Lower Keys
marsh rabbits (Svlvilagus palustris hefneri). In parenthesis are the number of times each rabbit was examined.
Trait Female #1 (N=4) Female #2 (N=5) Male#l(N=5)
R2
F
P
R2
F
P
R2
F
P
Body mass
0.97
70.1
0.01
0.97
84.1
0.01
0.90
18.6
0.05
Body length
0.93
27.0
0.04
0.94
49.7
0.01
0.94
29.5
0.03
Ear length
0.80
8.2
0.10
0.96
71.8
0.01
0.91
19.4
0.05
Right foot length
0.82
8.9
0.10
0.93
41.1
0.01
0.92
22.6
0.04


160
Decreasing the amount of migration (scenario #6) did not alter the mean
population size, but it did increase the amount of variance in population size
(Figure 6.9). The number of occupied populations actually increased when
there was only a 25% reduction in migration. When there was a 50%
reduction, the number of occupied metapopulations was slightly below the
results for unaltered migration rate results.
In scenario #7, patches were destroyed and removed from the simulation
(Figure 6.10). When only the large site was removed (site #33), the population
decreased to approximately 400 individuals (54 adults) and fluctuated slightly
around this value. When 5 of the largest patches (including site #33) were
removed, the population size declined nearly to extinction. Removal of the 10
smallest sites produced a similar effect, only the rate of the decline was faster.
Discussion
With its small body size, short life span, high reproductive output, and
high habitat specificity, the Lower Keys marsh rabbit typifies the good
candidate for a species that might exhibit classic metapopulation dynamics
(Murphy et al. 1990, Thomas 1994). Over the time scale of this research (3
years), S. p. hefneri did appear to fit the pattern of a species that inhabited a
patchy landscape yet was able to persist due to its dispersal ability. However,


156
density of rabbits at the largest site (site 33; Sugarloaf) had a large effect in
overall population size. These large fluctuations in the total population size did
not appear to be seasonal. The weather during the 2.5 years of the study was
normal with respect to temperature and precipitation (Figure 6.2).
Between-population and Metapopulation Parameters
Only subadults made long, one-way movements (see Chapter 4), and
75% of these successfully reached a new patch. Movement distance differed
greatly between individuals; males generally moved farther than females.
Because in RAMAS/metapop male and female dispersal can not be
distinguished, male and female dispersal was pooled. Based on the pooled
dispersal data (Chapter 4), it was assumed that if a patch was 50 m or less
away from another patch, then it could be reached by 94% of the dispersers. If
it was between 50 and 100 m, then 63% could reach it; between 100-500, then
50%. Patches 500 1000 m away could only be reached by 19% of the
dispersers and patches between 1000-2000 m away could be reached by 6%.
For patches >3000 m distant, a migration rate of <0.01 was included, to allow
for a rabbit that had superior dispersal ability.
No evidence of inter-population correlation in population size was
found using the repeated measures MANOVA. None of the correlations were
significant (P > 0.05), and the highest correlation was r = 0.65.


Middle Torch
Figure 5.4 The current distribution of the Lower Keys marsh rabbit (Svlvilagus palustris hefneri) by Key. The stipled keys are
occupied; white keys are unoccupied.


118
Table 5.3Results of univariate analysis of variance on arcsine-transformed percentage
relative density of Sporobolus virginicus (Sv), Spartina spartinae (Sp), Laguncularia
racemosa (Lr) and Borrichia ffutescnes (Bf) in the fecal pellets of SL p, hefneri in the
Lower Keys of Florida.
Source
Sv Sp
df F P F P
Lr
F P
Bf
F P
Site 5
Season 1
Sex 1
Season by sex 1
0.87
ns
1.38
ns
2.39
ns
11.61
**
0.21
ns
1.26
ns
15.71
**
6.24
*
1.15
ns
0.97
ns
2.27
ns
1.26
ns
0.04
ns
0.01
ns
2.19
ns
0.00
ns
(* 0.05 > P >0.01; ** 0.01 > P > 0.001; ns = P > 0.05)


123
Table 5.6aResults from a stepwise discriminant function analysis of 10 habitat
variables measured on "occupied and "empty" marsh rabbit habitat in the Lower Keys
ofFlorida.
Habitat Variable
Step entered
Partial R^
F
P
DPopulation
1
0.56
73.26
0.0001
Area
2
0.12
7.63
0.0007
DResidence
3
0.08
5.01
0.01
GCover
4
0.05
2.97
0.09
Table 5.6bResults from a stepwise discriminant function analysis of 10 habitat
variables measured on "variably" and "consistently" occupied marsh rabbit habitat in
the Lower Keys ofFlorida.
Habitat Variable
Step entered
Partial R^
F
P
Clump
1
0.60
60.48
0.0001
DPopulation
2
0.27
14.29
0.0005
MaxHgt
3
0.08
3.24
0.05
Borrichia
4
0.09
3.78
0.04


Genetic Threats to Persistence 42
Demographic Threats to Persistence 42
4 SPATIAL ORGANIZATION (INTER-PATCH MOVEMENTS) 55
Introduction 55
Methods 60
Radio-telemetry 61
Home Range 62
Spacing Behavior 63
Dispersal 64
Corridor Use 65
Results 66
Home Range 67
Spacing Behavior 68
Home-range Features 69
Dispersal 70
Corridor Use 71
Discussion 72
Spatial Organization 73
Conclusions 76
5 METAPOPULATION DYNAMICS: PATCH OCCUPANCY
AND HABITAT QUALITY 92
Introduction 92
Metapopulation Structure 92
Habitat Use 94
Methods 95
Pellet Grids 95
Pellet Size 96
Metapopulation Structure 97
Dietary Analysis 98
Microhabitat Use 100
Macrohabitat 101
Results 103
Pellet Degradation Rate and Pellet Size 103
Metapopulation Structure 104
Dietary Analysis 106
Resource Use and Availability 107
Habitat Model 108
Potential Reintroduction Sites 109
IV


86
^'A70M
Figure 4.4Home ranges using the 95% harmonic mean (outer boundary) and
core areas (inner boundary) of a sympatric male and female Lower Keys marsh
rabbit (Sylvilagus palustris hefneri) at site #7 on Boca Chica Key.


17
nearly half of them in the Lower Keys (Shermyen 1993). The populations has
increased nearly 65% since 1960. Tourism, which was also high during the
late 1800s, has also resurged in the Lower Keys. At the 2 National Park
beaches in the Lower Keys, over 500,000 tourists were recorded in 1993. The
majority of these visitors came to the Lower Keys and Key West via the
Overseas highway (Shermyen 1993).
Human impact on the Keys' wildlife was noticed as early as 1908, when
the Key West National Wildlife Refuge was established to provide habitat for
migratory birds. The Great White Heron National Wildlife Refuge was
established in 1938, encompassing a collection of small islands north of the
main chain of keys. In 1957, the National Key Deer Refuge was established;
its main mission was to protect the Key Deer (Odocoileus virginianus clavium)
and other wildlife. The largest tracts of land for this refuge are on Big Pine
Key and Noname Key, but other areas on the Torch Keys are currently being
added (John Andrew, National Key Deer Refuge Manager, pers. comm., 1993).
The United States military presence began in the Keys during the Civil
War, and waxed and waned throughout the past century (Hambright 1991).
Due to the Lower Keys' proximity to Cuba, the Caribbean, and open water
ideal for training aviators, the U.S. Navy currently maintains a base in the
Keys. The Navy owns a large amount of land on Key West, Boca Chica Key,


223
Barthouse, L. W., Boreman, J., Christensen, S. W., Goodyear, C. P. VanWinkle,
W., Vaughan, D. S. 1984. Population biology in the courtroom: The
Hudson River controversy. Bioscience 34:14-19.
Batzli, G. O. and F. A. Pitelka. 1971. Condition and diet of cycling populations of
the Californian vole. Journal of Mammalogy 52:141-163.
Begon, M., Harper, J. L. and C. R. Townsend. 1990. Ecology: Individuals,
populations, and communities. Sinauer Associates, Inc., Sunderland,
Massachusetts. 876 pp.
Belovsky, G. E. 1987. Extinction models and mammalian persistence. Pages 35-
59 in Viable populations for conservation, (M. E. Soul ed.), Cambridge
University Press, Cambridge, Great Britain.
Berg, H. C. 1983. Random walks in biology. Princeton University Press,
Princeton, New Jersey 142 pp.
Bergin, T. M. 1992. Habitat selection by the Western Kingbird in Western
Nebraska: a hierarchical analysis. Condor 94:903-911.
Berry, R. J. 1986. Genetics of insular populations of mammals, with particular
reference to differentiation and founder effects in British small mammals.
Biological Journal of the Linnean Society 28:205-230.
Blair, F. W. 1935. An ecological investigation of the marsh rabbit of the region of
Gainesville, Florida. M. S. thesis, University of Florida, Gainesville. 92 pp.
Blair, F. W. 1936. The Florida marsh rabbit. Journal of Mammalogy 17:197-207.
Blake, J. G. and J. R. Karr. 1987. Breeding birds of isolated woodlots: Area and
habitat relationships. Ecology 68:1724-1734.
Bothnia, J. du P. and J. G. Teer. 1977. Reproduction and productivity in South
Texas cottontail rabbits. Mammalia 41:254-281.
Bothnia, J. du P., J. G. Teer, and C. E. Gates. 1972. Growth and age determination
of the cottontail in South Texas. Journal of Wildlife Management 36:1209-
1210.
Boyce, M. S. 1992. Population viability analysis. Annual Review of Ecology and
Systematics 23:481-506.


225
Chapman, J. A. 1971. Orientation and homing of the brush rabbit. Journal of
Mammalogy 52:686-699.
Chapman, J. A. and G. Ceballos. 1990. The cottontails. Pages 95-110 in Rabbits,
hares and pikas: Status survey and conservation action plan (J. A. Chapman
and J. E. C. Flux, eds.). IUCN, Gland, Switzerland.
Chapman, J. A., Cramer, K. L., Dippenaar, N. J. and T. J. Robinson. 1992.
Systematics and biogeography of the New England cottontail Svlvilagus
transitionalis (Bangs, 1895), with the description of a new species from the
Appalachian mountains. Proceedings of the Biological Society of
Washington 105:841-866.
Chapman, J. A. and J. E. C. Flux. 1990. Introduction and overview of the order
Lagomorpha. Pages 1-6 in Rabbits, hares and pikas: Status survey and
conservation action plan (J. A. Chapman and J. E. C. Flux, eds.). IUCN,
Gland, Switzerland.
Chapman, J. A., Hannan, A. L. and D. E. Samuel. 1977. Reproductive and
physiological cycles in the cottontail complex in western Maryland and
nearby West Virginia. Wildlife Monograph 56:1-73.
Chapman, J. A., Hockman, J. G. and M. M. Ojeda. 1980. Svlvilagus floridanus.
Mammalian Species 136:1-8.
Chapman, J. A., Hockman, J. G. and W. R. Edwards. 1982. Cottontails. Pages 83-
123 in Wild mammals of North America: Biology, management, economics
(J.A. Chapman and G.A. Feldhamer, eds.) The Johns Hopkins University
Press Baltimore, Maryland.
Chapman, J. A. and R. P. Morgan. 1973. Systematic status of the cottontail
complex in western Maryland and nearby West Virginia. Wildlife
Monographs 36:1-54.
Chapman, J. A. and J. R. Stauffer. 1981. The status and distribution of the New
England cottontail. Pages 973-983 in Proceeding of the world lagomorph
conference (K. Myers and C. D. Maclnnes, eds), University of Guelph,
Guelph, Ontario, Canada.


155
Density Estimation
Average defecation rate (R) varied between seasons, ranging from an
average of 103.75 pellets/day in July 1993 to 171.0 pellet/day in March 1991
(Table 6.2). Although the data are limited in its temporal scope, marsh rabbits
appear to produce fewer pellets in the warmer months and more in the cooler
months. Similar results were found in pellet studies of other species of rabbit
(Lord 1963).
Using the 30-day sampling period, the defecation rate from the first
survey (March 1991), and the 0.5-m-radius sample, 8 of the 13 sites on Boca
Chica had a random distribution of pellets (Table 6.3). Using the same
constants and the 1-m-radius, only 2 sites had a random distribution of pellets.
Therefore, pellet counts from the 0.5-m circle were used for estimating rabbit
densities in this and all subsequent surveys.
Of the 39 sites (Appendix C) used in the density estimation, 9 did not
appear to have been used or inhabited during the study. A comparison of
carrying capacities with estimated total abundances at the patches revealed that
over two thirds of the patches were under carrying capacity (Table 6.4). The
estimated total adult marsh rabbit population size fluctuated widely between a
low of approximately 100 individuals to a high of nearly 300 (Figure 6.1).
This estimate includes all of the sites used in the density estimate (Appendix
A), but it does not include the sites that could not be censused. Changes in the


Year
Number of Occupied Patches
* N5 CO cn
o o o o o o
Year
Total Estimated Abundance
hO-t^COOOOI'O-^OOO
ooooooooo
oooooooooo
2000


Number of Occupied Patches Total Estimated Number of Individuals
Year


63
1969). It assumes that there is always only one center of activity (Harris et al.
1990), an assumption that was frequently violated in this study.
Two related methods that have worked well for analysis of rabbit home
ranges are the harmonic-mean and core-area methods (Dixon and Chapman
1980). The harmonic mean is a nonparametric method based on a volume
under a fitted 3-dimensional use distribution. It relates well to the actual
distribution of locations. The core area (as calculated by program HOME
RANGE), is the maximum area where the observed use distribution exceeds a
uniform use distribution. Thus, it shows areas of particularly high home-range
use and is relatively unaffected by outliers and sample size (Harris et al. 1990).
Seasonal differences in the home range size (95% harmonic mean and
core area) were compared for each individual that was located at least 30 times
in each season. The dry season locations occurred between November and
April; the wet season was from May to October. Shifts in home range between
season were also examined. Arithmetic centers of the dry and wet season
home range of each individual were determined and the distance these centers
shifted between seasons was calculated.
Spacing Behavior
The amount of overlap between the 95% harmonic mean and core areas
of each individual was compared to other individuals living in the same habitat


243
Willis, E. O. 1980. Species reduction in remanescent woodlots in Southern Brazil.
Proceedings of the 19th International Ornithological Congress: 783-786.
Wilson, L. D. and L. Porras. 1983. The ecological impact of man on the south
Florida herpetofauna. University of Kansas Museum of Natural History
Special Publication, 9.
Wilson, E. O. and E. O. Willis. 1975. Applied biogeography, Pages 522-536 in
Ecology and Evolution of Communities (M. L. Cody and J. M. Diamond ,
eds.), Harvard University Press, Cambridge, Massachusetts.
Wobeser, G. A. and T. R. Spraker. Post-mortem examination. Pages 89-98 in (S.
D. Schemnitz ed.). Wildlife management techniques, 4th edition. The
Wildlife Society, Washington, D.C.
Wolfe, J. L. 1992. Lower Keys marsh rabbit. Pages 71-75 in Rare and endangered
biota of Florida, Volume I: Mammals, second edition (S. R. Humphrey,
ed.),. University Presses of Florida, Gainesville.
Wolff, J. O., Lidicker, W. Z., Chesser, R. K. and M. H. Smith. 1991. Review of:
Dispersal in Rodents: a resident fitness hypothesis. Journal of Mammalogy
72:218-220.
Wood, D. H. 1988. Estimating rabbit density by counting dung pellets. Australian
Wildlife Research 15:665-671.
Woolf, A., Shoemaker, D. R., and M. Cooper. 1993. Evidence of tularemia
regulating a semi-isolated cottontail rabbit population. Journal of Wildlife
Management 57:144-157.
Wright, S. 1977. Evolution and the genetics of populations. Vol. 3. Experimental
results and evolutionary deductions. University of Chicago Press, Chicago.
456 pp.


Pellet Area (crrr)
128
0 200 400 600 800 1000 1200 1400 1600
Body Mass (g)
Figure 5.3 Average pellet size (in area cm2) in relation to body mass (g) of 53 Lower
Keys marsh rabbits (Sylvilagus palustris hefneri) trapped on Boca Chica and
Saddlebunch Keys.


117
Table 5.2~Ground cover for the 19 most abundant species measured in five transition
zone habitat patches in the Lower Keys of Florida. An asterisk following the plant
species indicates a species not found in the rabbit's diet.
Plant species and family
X
SE
SDartina SDartinaePoaceae
17.40
8.49
SDorobolus vireinicusPoaceae
14.60
4.31
Borrichia frutescensAsteraceae
8.40
3.80
Monanthochloe littoralis*--Poaceae
4.40
1.94
Andropoeon elomeratus-Poaceae
4.20
2.01
Fimbristvlis castanea-Cvperaceae
4.20
1.91
Conocarpus erecta*--Combretaceae
2.80
0.74
Muhlenbereia filipesPoaceae
2.60
1.19
Pithecellobium euadelupense*Fabaceae
2.00
2.00
Fimbristvlis spathaceaCvperaceae
1.60
1.17
Salicomia vireinicaCheropodiaceae
1.40
0.93
Laeuncularia racemosaCombretaceae
1.20
0.20
Avicennia eerminansAvicennianceae
1.00
0.32
Coccoloba uvifera--Polvgonacea
1.00
0.63
Rhizophora maneleRhizophoraceae
1.00
0.32
Tvpha latifoliaTvphaceae
1.00
1.00
Eleocharis cellulosaCvperaceae
0.80
0.49
Sesuvium maritimum-Aizoaceae
0.80
0.80
Baccharis halimifolia--Asteraceae
0.40
0.40


145
at all of the grids. The SU consisted of concentric 0.5-m- and 1.0-m-radius
circles around each permanent marker. Inside these circles, pellets were
removed, and then new pellets were counted after a specified amount of time.
To determine which SU size was optimal, the distribution pattern of pellets was
fitted to a Poisson distribution model. The Poisson distribution furnishes
values expected for a random dispersion pattern. When a chi-square goodness-
of-fit test is used to compare the 2 distributions, a significant result indicates
that the sample is non-random. The optimal sampling unit will be most likely
to produce a random distribution of pellets.
Once the preferred SU was selected, the following equation was used to
estimate rabbit density:
D = (10,000 m2/ha) X.
T*R* A
where: D = density of rabbits (rabbits per ha),
X = mean number of pellets per SU,
T = time between pellet removal and pellet counting,
R = defecation rate (number of pellets dropped per rabbit per 24-
hr day)
A = surface of each SU (m2).


58
Currently there is debate over the causes of natal dispersal (Dobson and
Jones 1985, Pusey 1987, Wolff et al. 1991). Most research has been directed
on determining the costs and benefits of dispersal to the dispersing individual.
Three main hypothesis have been proposed: inbreeding avoidance, competition
for mates, and competition for environmental resources (Packer 1979,
Greenwood 1980, Dobson 1982, Greenwood and Harvey 1982). Different
predictions for each hypothesis have been made, but research on a diverse
sample of mammalian species has not been able to falsify any of the
hypotheses. Two of the hypotheses, inbreeding avoidance and competition for
mates are not mutually exclusive in their predictions. It has been proposed that
the causal agents for dispersal may differ among species (Dobson and Jones
1985).
Recently, an alternative hypothesis has been proposed. Anderson
(1989) has suggested that the focus should be on the fitness the resident (non
dispersing) adults receive by forcing the young to leave. In most polygynous
mammals generally the father would receive the greatest benefit by coercing
his son to leave. Although some evidence of dominants evicting subordinates
exists in the literature (Moore and Ali 1984), other hypotheses can not be
eliminated (Pusey 1987, Wolff et al. 1991).
Determining what direction and what type of habitat dispersing
individuals will traverse are especially important in fragmented environments.


90
A. Both males alive (T = 3-6).
Figure 4.8The home ranges using core area, of 2 sympatric males telemetered
between session 3 and 6 (A.) and the expansion of A63Ms core area after the
death of A67M.


CHAPTER 3
POPULATION BIOLOGY (INTRA-PATCH DYNAMICS)
Introduction
The first step in developing a population viability analysis (PVA) is
determining the demographic and genetic composition of a population, how
these compositions vary, and how large a role stochastic and environmental
variability play (Salwasser et al. 1984, Gilpin and Soul 1986). Demographic
and genetic compositions of a population can influence population growth,
variability in growth, and the ultimate population size, all of which can in turn
influence the populations chance for persistence (Goodman 1987).
The demographic composition of a populations is determined by
variation in the birth and death rates and in the sex ratio of the new recruits.
Some variation in these parameters will be purely random, other sources of
variation may be from external sources (i.e., differential mortality by sex or age
group, biased sex ratios at birth ), and therefore this variability may be
correlated throughout the population. In a large, continuous population,
random variations will contribute little to the overall population variability but
environmental variability may be important; in very small, fragmented
22


69
overlap between these contemporary same-sex individuals was significantly
less than male/female overlap for both the 95% harmonic mean (T = 79.5, P =
0.05) and core area (T = 45, P = 0.02). In most cases there was little overlap
between adults of the same sex (Figure 4.9).
Most males home ranges overlapped 1 female at a time; 3 males
overlapped 2 females simultaneously. During a portion of their lives, some
males did not overlap with a female. Only 1 female overlapped with more than
1 male; this occurred for less than 1 month. Changes in overlap generally
occurred when a rabbit died and another rabbit expanded his/her home range.
Home-range Features
All 43 rabbits, including the 7 juveniles, were used in the road- and
water-crossing index. Only individuals that were currently in their natal or
adult home ranges were used. Data from individual making long-distance
movements were analyzed separately.
Home ranges of most of the marsh rabbits did not incorporate roads or
large bodies of water. No water or road crossings were recorded for any of the
juvenile rabbits. None of the rabbits crossed the major highway in the Keys
(US-1). Only 1 adult female and 1 adult male crossed paved roads during the
study. Rabbit A53F crossed a 2-lane several times. This same rabbit was the
only adult female to cross a body of water; she regularly swam in shallow


18
and North Saddlebunch Key. The U.S. Air Force manages the northernmost
portion of Cudjoe Key.
Study Sites
This study attempted to encompass all possible marsh rabbit habitat
throughout the Lower Keys, including saltwater transition-zone habitat and
fresh water marsh areas. Habitat areas were located using information from
Howe (1988) aerial photographs and ground survey. Initially, a patch of
habitat was considered isolated from another patch if the 2 areas were divided
by a major road, airplane runway, or large body of water. As a general rule,
patches of habitat <0.5 ha were not sampled, although some exceptions were
made. A previous study of marsh rabbit home ranges (Payne 1975) found that
no adult home range was <0.5 ha.
Fifty-nine habitat patches were identified throughout the Lower Keys
(Appendix A); no habitat was found on Key West, the most densely human-
populated Key. Twenty-seven habitat patches were found on Boca Chica, 25
of which were owned by the U.S. Navy. Four patches were examined at the
Saddlebunch Keys; 2 of which were owned by the U.S. Navy. On Sugarloaf
Key, 9 patches of transition zone habitat were found; 1 of which is owned by
the U.S. Fish and Wildlife Service (USFWS), and other is owned by Monroe
County. Two privately owned patches were found on Cudjoe, and 1 site


Table 4.1--Hierarchical tests and predictions of the proximate cause of dispersal.
Prediction
Cause of Dispersal
I. An equal number of males and females disperse
Competition for resources
II. Only one sex disperses.
A. The dispersers are all female.
Inbreeding avoidance
B. The dispersers are all male.
1. A male disperses only if another adult male is present
at the habitat patch.
Competition for mates
2. A male disperses only if other males are present and
could potentially be their father.
Resident fitness hypothesis
3. All males disperse regardless of the presence or
absence of other males.
Inbreeding avoidance


Table 6.1 --(Continued.)
Scenario
Mortality
Catastrophe
Migration
1
2
3
4
5
1
2
3
4
5
1
2
3
4
5
#5a. A class 3
hurricane has a 10%
0.42
0.81
0.75
0.64
0.33
0.75
0.50
0.25
0.25
0.25
0
0
0.75
0
chance of hitting the
keys.
#5b. Class 4.
0.42
0.81
0.75
0.64
0.33
0.95
0.75
0.50
0.50
0.50
0
0
0.75
0
0
#5c. Class 5.
0.42
0.81
0.75
0.64
0.33
0.99
0.90
0.75
0.75
0.75
0
0
0.75
0
0
#6a. Corridor habitat is
0.42
0.81
0.75
0.64
0.33
0
0
0
0
0
0
0
0.56
0
0
destroyed and decreases
migration by 25%.
#6b. migration
decreases by 50%.
0.42
0.81
0.75
0.64
0.33
0
0
0
0
0
0
0
0.38
0
0
#7. The largest patch
(site #33) is destroyed.
#7b. The five largest
patches are destroyed.
0.42
0.81
0.75
0.64
0.33
1
1
1
1
1
0
0
0.75
0
0
#7c. The 10 smallest
patches are destroyed.
0.42
0.81
0.75
0.64
0.33
1
1
1
1
1
0
0
0.75
0
0


42
marsh rabbit survival. Predation by cats has not previously been reported for
marsh rabbits, but cats were a major predator of adult and juvenile rabbits of
other species in several other studies (Fitzgerald and Karl 1979, Jones and
Coman 1981, Liberg 1985). The cat population of the Lower Keys is large and
includes feral and house-based cats.
Genetic Threats to Persistence
The Lower Keys marsh rabbit had a level of variation similar to large
populations of cottontails (Scribner and Warren 1986). Inbreeding depression
does not appear to be a threat for most of the Lower Keys marsh rabbits.
Although past levels of genetic variation are unknown, it does not appear that
these marsh rabbits are currently deficient in genetic variation. Future threats
may be dependent on the population size and structure.
Demographic Threats to Persistence
Fecundity of the Lower Keys marsh rabbit is lower than other
subspecies of marsh rabbits and other species of cottontails. Possible
explanations for this lower fecundity include: reabsorption of fetuses (Holler
and Conaway 1979), insufficient food or nutrients for reproduction (Cheeke
1987), adaptive pressures that have resulted in longer periods between
pregnancies, and lack of suitable mates at the habitat patch.


Table C.lEstimates of rabbit densities and numbers based on pellet counts in the Lower
Keys. (Session 1 = March 1991, Session 2 = July 1991, Session 3 = November 1991,
Session 4 = March 1992, Session 5 = July 1992, Session 6 = November 1992, Session 7 =
March 1993, Session 8 = July 1992).
Session 1
Session 2
Session 3
Session 4
Site
Rab/ha Total
Rab/ha Total
Rab/ha Total
Rab/ha Total
1
2.02
3.05
1.03
1.08
0.61
0.64
3.59
3.77
2
0.22
0.27
1.01
1.23
0.18
0.22
1.32
1.61
3
0.08
0.13
0.28
0.45
0.00
0.00
0.07
0.11
4
1.86
5.43
1.4
4.23
0.74
2.14
1.83
5.34
5
0.00
0.00
0.01
0.02
0.15
0.37
0.66
1.61
6
0.67
1.85
2.23
6.13
0.00
0.00
0.05
0.14
7
0.74
2.87
0.60
2.33
0.53
2.07
2.16
8.40
8
7.63
29.66
6.21
24.16
3.44
13.39
2.85
11.10
9
1.49
7.22
0.62
3.03
0.58
2.83
2.75
13.34
10
3.97
9.01
2.70
6.13
2.28
5.17
2.24
5.08
11
4.65
24.09
2.54
13.16
0.38
1.97
0.77
4.00
12
0.74
1.71
1.91
4.45
3.01
7.02
3.87
9.01
13
8.99
30.58
4.37
14.86
2.28
7.74
1.58
5.36
14
0.31
1.83
0.85
4.96
2.02
11.80
0.56
3.26
15
1.81
2.08
0.13
0.14
16
5.64
3.89
2.24
1.55
17
2.53
2.63
0.17
0.18
26
0.20
0.31
0.14
0.22
0.41
0.63
0.35
0.54
28
0.00
0.00
0.00
0.00
0.00
0.00
0.05
0.18
29
0.00
0.00
0.00
0.00
0.05
0.13
30
1.20
0.94
0.37
0.29
1.27
0.99
0.20
0.16
31
1.14
1.14
32
0.25
1.00
33
5.20
163.80
3.90
122.85
4.05
127.58
0.50
15.75
34
0.25
0.58
1.37
3.15
41
0.00
0.00
0.00
0.00
0.00
0.00
42
0.00
0.00
0.00
0.00
0.00
0.00
43
0.00
0.00
0.00
0.00
0.00
0.00
44
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
45
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
50
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
51
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
53
0.17
3.32
0.13
2.54
0.76
14.80
0.91
17.82
56
0.28
0.31
0.51
0.56
0.97
1.07
57
0.13
0.10
0.15
0.12
0.51
0.51
58
0.67
1.96
0.40
1.17
0.25
0.74
0.00
0.00
59
0.00
0.00
0.00
0.00
0.00
0.00
217


179
Figure 6.4Results of Scenario #1 (Predator control) of the Lower Keys marsh rabbit
(Sylvilagus palustris hefneri) PVA model, where survivorship was increased in all
stages by 25% and 50%.


94
Danielson 1991). Individuals from source populations migrate to sink
populations, giving the false appearance that these sink habitats are suitable for
the species. Habitat sources and sinks should be analyzed separately for
information on habitat requirements.
Habitat Use
An animal selects a habitat based on proximate cues from the
environment, but the decision will ultimately affect the individuals fitness in
terms of survival and reproduction (Lack 1933, Levins 1969, Cody 1985). For
many species, habitat selection is difficult to determine and habitat use is the
best available information. Assessment of habitat use is based on correlation
of the actual distribution of individuals with specific habitat features. It may
be inferred that habitat use is the outcome of habitat selection (Bergin 1992).
Habitat use may be studied at several spatial and temporal scales. For a
small-to-medium sized herbivore, the smallest scale may involve the leaf
browsed (diet). At the next level habitat use includes the areas used for
foraging and nesting within a habitat patch (microhabitat). At the largest
scales, the features that influence the presence or absence of a species over
time at a particular habitat may be studied (macrohabitat).
The first objective of this chapter is to determine the relationship among
habitat patches within the Lower Keys marsh rabbit metapopulation. The


variability, growth and morphology, natality, mortality, home range size, and
dispersal were estimated from this portion of the research. To study
metapopulation dynamics, all patches of marsh rabbit habitat in the Keys were
located and sampled with fecal-pellet grids during the 2.5 years to determine
presence/absence and density of marsh rabbits at each patch.
Fifty-three marsh rabbits were live-trapped during the study and 43 of
were fitted with radio-collars. Compared to other species of rabbits, the Lower
Keys marsh rabbits were found to have average genetic variation, lower
natality, and higher mortality. Subadult males were the primaiy dispersers; all
but one male left its natal patch at the onset of sexual maturity and moved 180
- 2050 m. Due to the amount of movement between habitat patches, the marsh
rabbits appear to exist in a metapopulation.
Fifty-nine patches of transi on-zone habitat were located throughout the
Lower Keys. Twenty of these patches had pellets present during all of the
surveys, 22 had pellets present during at least one of the surveys, and 17 never
had any pellets present. Habitat patches that were close to other patches and
that had dense vegetation were most likely to be inhabited. When all 59 habitat
patches and the information about the population biology of S. p. hefneri were
combined in a popularion viability model, the model predicted that the rabbit
would go extinct in the next 20-30 years. Reducing mortality in each life stage
had the greatest effect on overall metapopulation persistence.
viii


Table 4.5~The amount of overlap between same and opposite sexed individuals occuping a site during the same time and
between same-sexed individuals that occupied the same site during different times.
Same sex-same time
Pair 95% Core area
Same sex-different time
Pair 95% har. Core area
Opposite sex-same time
Pair 95% har. Core area
A59M/A55M
34%
12%
A51M/A69M
100%
43%
A51M/A52F
100%
67%
A63M/A67M
30%
3%
A55M/A75M
82%
44%
A59M/A57F
68%
40%
A70M/A71M
85%
42%
A59M/A55M
45%
20%
A59M/A58F
45%
22%
A50F/A74F
9%
2%
A59M/A75M
91%
72%
A60M/A53F
44%
22%
A53F/A76F
100%
21%
A53F/A76F
100%
100%
A63M/A50F
54%
32%
A57F/A58F
16%
0%
A57F/A86F
71%
74%
A64M/A53F
8%
0%
A72F/A65F
0%
0%
A58F/A86F
45%
20%
A67M/A50F
100%
100%
A67M/A74F
85%
21%
A68M/A76F
100%
90%
A70M/A65F
8%
2%
A70M/A72F
82%
48%
A71M/A72F
90%
84%
A75M/A57F
100%
100%
A75M/A86F
100%
86%
Average
39%
11%
76%
53%
70%
51%


196
management techniques are necessary to allow the endangered species to
persist as a metapopulation.
Management for Recovery of the Lower Keys Marsh Rabbit
A detailed recovery plan has been written for S. p. hefneri (United
States Fish and Wildlife Service 1993). Recommendations made in the
recovery plan are incorporated with the results of the PVA model presented in
this manuscript.
Primary Recovery Actions
If current mortality rates persist, it is likely that the Lower Keys marsh
rabbit will go extinct during the next 20-30 years. The first goal in the
recovery efforts should be increasing survivorship in all stages, but especially
in the nestling and adult stages.
The main source of juvenile to adult mortality, and probably nestling
mortality, is domesticated cats (Felis catus) (Chapter 3). Reducing mortality
by cats is particularly difficult because domestic cats can be difficult to catch,
poisoning or shooting cats in suburban areas could incite public outrage, and
because many of the cats in the rabbits habitat are free-ranging house cats.
The best method of cat control may be a combination of long-term
public education and trapping in marsh rabbit habitat. A long-term public


70
water between mangrove prop roots while being radio-tracked. Male A51M
also crossed a 2-lane road four times. Neither rabbit crossed the road to visit
another patch. Female A53F crossed the road to swim in the water adjacent
to her home patch; male A51M crossed the road when construction was
being completed in the comer of his home patch.
Dispersal
Seventeen rabbits (11 male, 6 female, all subadults) made permanent
one-way movements (Table 4.6). The minimum dispersal distance was
calculated using the diameter of the combined adult core-area size. Assuming
a circular home-range shape, the average diameter was 124 m.
Eleven of the subadults made movements >124 m. Ten of the dispersers
were male and one was female. Only 1 of the males failed to meet the criterion
for dispersal, and whereas 5 of the females did not exceed the distance. In
general, the males made long-distance movements far in excess of the criterion,
including a 3-day movement that placed a male over 2 km from his natal range.
Most of the females settled near their natal ranges, including the female that
was classified as a disperser.
The sex ratio of the dispersers was significantly male biased (binomial P
<0.001). Six of the males that dispersed left patches where there was another
adult male present, but 4 left patches where there were no adult males. Five of


Number of Occupied Patches Total Estimated Number of
Year


207
Figure A.3A map of Sugarloaf Key. The habitat patches used in the study are numbered.


211
Figure A. 7A map of Big Pine Key. The habitat patches used in the study are numbered.


201
These fire ants are attracted to the mucous found on young rabbits and were a
major source of mortality in Eastern cottontail young (Sylvilagus floridanus;
Hill 1969, 1972).
Future Research
Annual population monitoring will be important in validating the PVA
model and determining the effectiveness of management techniques. The
results of the PVA simulation (Chapter 6) showed that the number of occupied
patches was similar to the total number of rabbits. Censusing areas for
presence or absence of marsh rabbit pellets is a fast and efficient method of
determining patch occupancy. Presence/absence data can be obtained for a site
in less than one hour. Using the methods reported in this study, pellet counting
for abundances uses 4 hours per site and live-trapping takes 20 hours for each
site.
Further research on the possibility of reintroducing Lower Keys marsh
rabbits also should be explored. Potential reintroduction sites first need to be
evaluated for habitat suitability and then the ability of rabbits to interact with
the greater metapopulation should be measured. Routine translocations of
marsh rabbits might be necessary to provide genetic interchange with the
mainland Keys populations.




CHAPTER 2
THE LOWER KEYS OF FLORIDA
Geologic History
The Lower Keys of Florida are the terminal portion of an archipelago of
islands extending westward from the mainland of Florida (Figure 2.1). For the
purposes of this dissertation, the Lower Keys includes all of the islands from
Key West to Noname Key (inclusive). The Lower Keys are separated from the
Upper Keys by nearly 20 km of water (from Noname Key to Vaca Key).
The Keys are derived from coral reefs that formed during the Sangamon
interglacial period (340,000-100,000 BP). During the subsequent Wisconsin
glacial period, the Lower Keys were covered with oolitic sands and were
gradually exposed as sea level began to fall. At the height of the Wisconsin
glaciation (20,000 15,000 BP) the Keys were points of high relief amid the
large land mass of Florida (Shinn 1988, Mueller and Winston 1991). At this
point in time, Florida had nearly twice the exposed land as today (Webb 1990).
Sea level remained low until nearly 15,000 BP, when the climate began to
warm. Sea level rose rapidly from 15,000 10,000 BP, reaching a maximum
of 7 feet/year, and then the rate slowed. During the past 2,000 years sea level
has risen only 6 feet.
12


88
A70M
A72F
A71M
A65F
A182M
A185F
1 2 3 4 5 6 7 8
A50F
A63M
A67M
A74F
A52F
A51M
A69M
Site 8
A
A
a
a a
A
a
A
A
A
~T~
1
1
2
r~
3
l I
4 5
~r~
6
7
~T~
8
Site 10












I
1
~T~
2
i
3
i i
4 5
I
6
l
7
I
8
Trap Session
Site 7



i i i i r
Figure 4.6The duration that radio-collared Lower Keys marsh rabbits
(Sylvilagus palustris hemeri) were followed at sites #7, #8, and #10.


19
partially on U.S. Air Force land. One privately owned site was found on each
of Summerland and Ramrod Keys. Big Torch Key had 2 privately owned sites
and one site owned by the USFWS. Middle Torch had 2 sites, both of which
were owned by the USFWS. Little Torch had 1 habitat patch and this patch
was owned by the Nature Conservancy (TNC). Big Pine had 7 habitat patches:
3 owned by the USFWS, 2 privately owned, and 2 owned by the South Florida
Water Management District (SFWMD). Noname Key had 1 USFWS-owned
habitat site.
Additionally, habitat was found on several of the smaller islands that are
part of the Lower Keys chain but that are only accessible by boats. Habitat
patches were found on Annette and Porpoise Keys, both of which are small
islets northeast of Big Pine Key. These islets are part of the Great White
Heron National Wildlife Refuge. Transition-zone patches were observed on
Big Munson Key and Saddlehill Key. Big Munson Key is south of Big Pine
and is owned by the Boy Scouts of America, Inc. Saddlehill is southeast of
Boca Chica Key and is privately owned. The search for habitat on other
isolated islets was not complete; the existence of small patches of habitat on
more distant keys is a possibility.


I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope
and quality, as a dissertation for the degree of Doctor of Philosophy.
Professor of Forest Resources 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 fully adequate, in scope
and quality, as a dissertation for the degree of Doctor of Philosophy.
Crawford S. Holling
Arthur R. Marshall, Jr., Professor of
Ecological Sciences
This dissertation was submitted to the Graduate Faculty of the College
of Agriculture and to the Graduate School and was accepted as partial
fulfillment of the requirements for the degree of Doctor of Philosophy.
May 1995
\cuJ\ df. 3*/
Dean, college of Agriculture
Dean, Graduate School


195
Using spatially explicit models, the extinction and recolonization of
individual patches (metapopulation dynamics) can be simulated.
Metapopulations are affected by the same threats as single individual
populations, in addition to threats that occur at the metapopulation level. PVA
models can assess the chance of local extinctions and the chance of
metapopulation extinction.
Harrison (1991) has postulated that most endangered species that exist
in metapopulations are actually fragmented populations on the decline, as is the
case of the spotted owl (Strix occidentalis occidentalis; LaHaye et al. 1994)
and the Lower Keys marsh rabbit (Sylvilauus palustris hefneri). The transition-
zone marsh habitat of the Lower Keys marsh rabbit once existed as a
concentric ring located on each key between the mangroves and the upland
hardwood hammocks. Development has broken the ring, creating a patchy
distribution of transition zone habitat.
Few metapopulations of endangered species persist in the structure
predicted by Levins (1969, 1970) and Hanski and Gilpin (1991), where the
number of recolonized patches is greater or equal to the number of patches that
go extinct. Most endangered species occur in a fragmented landscape because
an intact landscape is not available, and creation of one is not possible.
Metapopulation dynamics and PVA can be used to determine what


Days Until Degredation
127
March 1991
March 1992
Session
March 1993
Figure 5.2 Number of days until the first Lower Keys marsh rabbit (Sylvilagus
palustris hefneri) pellet decayed. One group of 25 pellets was placed in each of 4
settings (in tree litter, on rocks, in grass, and in mud) and observed daily.


136
populations, and metapopulation. The subpopulation level (or patch) is the
traditional domain of PVAs that involve a single population.
Most single-population PVAs are conducted on small populations and
require that both stochastic and deterministic processes be examined (Gilpin
and Soul 1986, Wilcox 1986, Lacy and Clark 1990). Small populations are
more likely to be at risk of extinction due to stochastic variation than are large
populations. Stochastic perturbations in small populations can affect the
genetic composition (loss of genetic variability, inbreeding depression) or the
demographic composition (sex, age ratios) of a population (Shaffer 1981).
Variation in the environment (climatic variation, natural catastrophes), is also
random but affects small populations only slightly more than larger ones.
Deterministic causes of extinction include Diamonds evil quartet:
overkill, habitat destruction, invasion by introduced predators or competitors,
and food-web collapse (Diamond 1984, 1989). In addition, climatic change
(Nunney and Campell 1993) may be deterministic and disease may be
stochastic or deterministic depending on its extent and the impact it has on the
population (Aguirre and Starkey 1994). Deterministic extinction can occur
regardless of population size, if all of the population is effected. For most
species, deterministic processes play a larger role in the original decline of the
species, while stochastic process are responsible for the extinction after the
decline (Caughley 1994). Both stochastic and deterministic phenomena may


41
few degrees north of the Tropic of Cancer, the climate can be considered
tropical (Walter et al. 1985). The temperature in the Lower Keys has not fallen
below freezing in historic times. Rainfall is seasonal, but the Keys do receive
one-third of their rainfall during the "dry" season. Slight variations in marsh
rabbit reproduction might be related to inter-year variation in precipitation.
Marsh rabbit mortality was high for nearly every age class and sex. The
estimate of nestling mortality, however, may be negatively biased. The
probability of catching a rabbit during the first 3 months of its life is unknown
and it is possible that some young rabbits survived the first 3 months but were
killed later as juveniles. It seems unlikely that the young rabbits left the habitat
patch. Cottontail rabbits are extremely altricial, and very young rabbits appear
to be only capable of moving short distances, especially without cover.
Subadult males appeared to have a particularly high mortality rate,
mainly due to the number killed by vehicles. Second-year adults also have a
high mortality rate, indicating that adult marsh rabbits may become more
vulnerable to predation as they age. In general, fewer females were killed than
males. This difference in mortality data may be due to the male-skewed sex
ratio of the radio-collared subadult rabbits.
Vehicles killed nearly a quarter of radio-collared marsh rabbits in this
study, including the subadult dispersing males. Because marsh rabbits are most
active from dusk until dawn, night-time road traffic appears to be a threat to


24
recently in captively breed animals from wild stock (Ralls and Ballou 1983,
Ralls et al. 1988). Evidence for inbreeding depression in natural populations is
less clear (Lacy 1993), and in need of further study. Most examples of
possible instances for inbreeding depression in the wild come from populations
of large, long-lived species such as the grizzly bear (Ursus arctos; Harris and
Allendorf 1989) and the Florida panther (Felis concolor coryi; Seal and Lacy
1989). These species have existed at extremely small numbers in the wild for
several decades and have largely overlapping generations.
For smaller, shorter-lived species, it is currently hypothesized that
before inbreeding depression can have an effect in the wild, habitat loss and
demographic problems may have already caused the demise of the wild
populations (Lacy 1993). Management for these problems may also help in
preventing inbreeding depression and overall loss of genetic variability in the
population, although more intense management may be needed for populations
that have recently undergone severe population bottlenecks (Fuerst and
Maruyama 1986). Inbreeding depression for these species may still be an
important consideration in PVA, if captive breeding and reintroductions are
available as management options (Haig et al. 1990).
The sensitivities of demographic and genetic compositions to random
and environmental variation are dependent on the life history and population
dynamics of the species in question. Further, they may be affected by


144
standard deviation. Assuming a one-to-one ratio between males and females,
the number of males was doubled to reach the carrying capacity of adults.
Assuming a stable age distribution, the number of subadult, juveniles and
nestlings was estimated and added to the adult carrying capacity.
To determine the initial abundance of rabbits at each site, pellet counts
were used throughout the study. With knowledge about the distribution of
pellets, the number of pellets produced by a rabbit a day and the pellet
persistence rate (calculated as 30 days in Chapter 5), fairly accurate
abundances can be obtained (Wood 1988). For the purposes of this study, only
adult pellets were counted (those pellets with an area >0.4 cm see Chapter 5).
Presence or absence of juvenile pellets was recorded, but these pellets were not
included in the density estimates.
To determine the number of pellets each rabbit produced in 1 day, 4
trapped adult rabbits (2 male, 2 female), were given a collection of natural
vegetation and placed for 24 hrs in a rabbit cage with a removable pan beneath
the mesh floor. This procedure was repeated 3 times a year (March, July, and
November) to document the effect of climate and rabbit age during late dry
season, mid-wet season, and the transition time between wet and dry seasons.
Reliability of pellet counts may be affected by the patchy distribution of
pellets, which can bias the counting technique. To identify a reliable method,
during the first pellet-sampling period sampling units (SU) of 2 sizes were used


Figure 3.1--Five main habitat patches on Boca Chica used for trapping and radio-collaring of Lower Keys marsh rabbits
(Sylvilagus palustris hefneri). The solid lines are runways and taxiways. The dashed line are roads.
A


226
Chapman, J. A. and D. E. C. Trethewey, 1972. Movements within a population of
introduced eastern cottontail rabbits. Journal of Wildlife Management
36:155-158.
Chapman, J. A. and G. R. Willner. 1981. Svlvilagus palustris. Mammalian Species
153:1-3.
Cheeke, R. 1987. Rabbit feeding and nutrition. Academic Press, Inc., Orlando,
Florida. 376 pp.
Chen, E. and J. F. Gerber. 1990. Climate. Pages 11-34 in Ecosystems of Florida,
(R. L. Myers and J. J. Ewel, eds.), University of Central Florida Press,
Orlando, Florida.
Churcher, P. B. and J. H. Lawton. 1989. Beware of well-fed felines. Natural
History 7:40-46.
Cody, M. L. 1985. Habitat selection in birds. Academic Press, London. 558 pp.
dePourtales, L. F. 1877. Hints on the origin of the flora and fauna of the Florida
Keys. American Naturalist 11:137-144.
Diamond, J. M. 1975. The island dilemma: lessons of modem biogeographic
studies for the design of nature reserves. Biological Conservation 7:129-146.
Diamond, J. M. 1984. "Normal" extinctions of isolated populations. Pages 191-
246 in Extinctions (M. N. Nicety, ed.), University of Chicago Press,
Chicago.
Diamond, J. M. 1989. Overview of recent extinctions. Pages 37-41 in
Conservation for the Twenty-first Century (D. Western and M. Pearl, eds.),
Oxford University Press, New York.
Diamond, J. M. and A. C. Marshall. 1977. Niche shifts in New Hebridean birds.
Emu 77:61-72.
Diamond, J. M. and R. M. May. 1977. Island biogeography and the design of
nature reserves. Pages 228-252 in Theoretical ecology (R. M. May, ed.).
Sinauer Associates, Sunderland, Massachusetts.
Dixon, K. R., and J. A. Chapman. 1980. Harmonic mean measure of animal
activity areas. Ecology 61:1040-1044.


146
The defecation rate (R) was determined as described in the previous
section, "pellet production". The number of days between pellet removal and
counting (T), was determined to be more than 30 days in Chapter 5. Thirty
days was used as the sampling interval.
Density estimation was attempted at the permanent fecal pellet-sampling
grids that were established at each of 59 patches of transition-zone habitat in
the Lower Keys. The grid was designed to fit into the smallest patch and
consisted of a square of 7 x 7 stations at 15-m intervals, marked with
permanent flags. The grids were surveyed 3 times per year: March (late dry
season), July (mid-wet season), and November (transition between wet and
dry) from March 1991 to July 1993. Each survey consisted of pellet removal
within a radius of 0.5 m at each station followed by a pellet count 1 month
later. Because some of the grids were occasionally disturbed by vehicles, these
grids were examined only for pellet presence or absence. At several privately-
owned sites and sites that were deemed sensitive by the Navy, access was
limited and the time necessary to thoroughly count the pellets was not
available. These sites also were only examined for presence or absence of
pellets.
To determine the initial abundances for the model, all of the density
estimations collected during session 3 were used. Session 3 was chosen
because the most sites were censused during that time. To calculate initial


121
Table 5.5Variables used to measure characteristics of marsh rabbit habitat in the
Lower Keys of Florida.
Variable
abbreviation
Variable
Mode of
measurement
GCover
Ground Cover
proportion of 10
transects occluded
below 1.5 m
CCover
Canopy Cover
proportion of 10
transects occluded
above 1.5 m
Borrichia
Borrichia frutescens
cover
proportion of 10
transects occluded
at ground level
Clump
Clump grass cover
proportion of 10
transects occluded
bv Spartina spartinae.
Fimbrvstilis sp. and
Cladium iamaicensis
MaxHgt
Average maximum
height
average maximum
height of ground
vegetation measured in
0.25-m intervals in the
1st, 3rd, and 5th meters
of each transect
H
Plant diversity
Shannon-Weaver
diversity index based on
ground and canopy
species present in 10
transects
Area
Area of habitat patch
aerial photograph
DPopulation
Distance from edge
aerial photograph
of site to nearest
rabbit population
Range Trans
formation*
30-98% S
2-81% S
0-60% S
0-87% S
6.5-62.0 cm N
0.55-2.17 N
0.3-43.7 ha L
40-6066 m L


METAPOPULATIONS OF MARSH RABBITS: A POPULATION
VIABILITY ANALYSIS OF THE LOWER KEYS MARSH RABBIT
(SYLVILAGUS PALUSTRIS HEFNERT)
By
ELIZABETH A. FORYS
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE
UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE
REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
1995

ACKNOWLEDGMENTS
I wish to thank Steve Humphrey for his advise and untiring support
during all of the phases of my Ph.D. work. Discussions with Buzz Holling and
John Eisenberg gave me new perspectives and insights about the science of
conservation biology. George Tanner and Lyn Branch provided constructive
comments on both my proposal and this manuscript.
This research would not have been possible without the funding from
the U. S. Navy (USN). Don Wood of the Florida Game and Fresh Water Fish
Commission (FGFWFC) effectively and efficiently administered the grant.
Additional financial support was provided by the U.S. Fish and Wildlife
Service (USFWS) and the Nature Conservancy (TNC). Many individuals
provided additional logistic support in the Lower Florida Keys: Annie
Simpkins and Amim Sheutz (USN), Tom Wilmers, Stuart Marcus, John
Andrew, Mark Yanno, and Jane Tutton (USFWS), Randy Tate (TNC), Phil
Frank (FGFWFC), Bill Keogh (wildlife photographer), and Lee Irby (long
suffering husband). My friends in both Gainesville and the Keys remained
faithful throughout my multiple moves across Florida. Many fellow graduate
students helped during the planning, analysis, and writing of my dissertation.
n

TABLE OF CONTENTS
page
ACKNOWLEDGMENTS ii
ABSTRACT vii
CHAPTERS
1 INTRODUCTION 1
Population Viability Analysis 2
Spatial Structure 4
Metapopulations 6
Metapopulation Persistence 7
Dissertation Structure 9
2 THE LOWER KEYS OF FLORIDA 12
Geologic Histoiy 12
Climate 13
Flora and Fauna 14
Development 16
Study Sites 18
3 POPULATION BIOLOGY (INTRA-PATCH DYNAMICS) 22
Introduction 22
Methods 26
Trapping Girds 26
Genetic Analyses 27
Radio-telemetry 28
Natality 29
Mortality 30
Results 32
Growth and Morphology 32
Genetics 34
Demographics 35
Natality 36
Mortality 37
Discussion 39
iii

Genetic Threats to Persistence 42
Demographic Threats to Persistence 42
4 SPATIAL ORGANIZATION (INTER-PATCH MOVEMENTS) 55
Introduction 55
Methods 60
Radio-telemetry 61
Home Range 62
Spacing Behavior 63
Dispersal 64
Corridor Use 65
Results 66
Home Range 67
Spacing Behavior 68
Home-range Features 69
Dispersal 70
Corridor Use 71
Discussion 72
Spatial Organization 73
Conclusions 76
5 METAPOPULATION DYNAMICS: PATCH OCCUPANCY
AND HABITAT QUALITY 92
Introduction 92
Metapopulation Structure 92
Habitat Use 94
Methods 95
Pellet Grids 95
Pellet Size 96
Metapopulation Structure 97
Dietary Analysis 98
Microhabitat Use 100
Macrohabitat 101
Results 103
Pellet Degradation Rate and Pellet Size 103
Metapopulation Structure 104
Dietary Analysis 106
Resource Use and Availability 107
Habitat Model 108
Potential Reintroduction Sites 109
IV

Discussion 110
Metapopulation Structure 110
Habitat Use 112
6 POPULATION VIABILITY ANALYSIS 135
Introduction 135
Multiple Levels of Population Viability Analysis 135
Population Viability Analysis Models 138
Costs and Benefits of Using a PVA Model 139
Methods 141
Population-level Parameters 141
Inter-population Level Parameters 147
Metapopulation Parameters 148
The Model 148
Scenarios 149
Scenario #1 Decrease Predation 149
Scenario #2 Decrease Road-kills 150
Scenario #3 Reintroduce Rabbits 150
Scenario #4 Disease 150
Scenario #5 Hurricanes 151
Scenario #6 Corridor Destruction 152
Scenario #7 Habitat Destruction 152
Results 153
Population Parameter Estimations 153
Density Estimation 155
Between Population and Metapopulation Parameters 156
Simulation Results 157
Discussion 160
Model Validation 162
Minimum Viable Population, Area, and Number of Patches 167
7 CONCLUSIONS AND MANAGEMENT
RECOMMENDATIONS 193
Population Viability Analysis (PVA) and Metapopulation
Dynamics 193
Management for the Recovery of the Lower Keys Marsh Rabbit.... 196
Primary Recovery Actions 196
Secondary Recovery Actions 198
Future Research 201

APPENDICES
A DESCRIPTIVE INFORMATION ABOUT THE
HABITAT PATCHES AND MAPS 203
B VARIABLES USED IN THE DISCRIMINANT FUNCTION
ANALYSIS 213
C MARSH RABBIT DENSITY ESTIMATES 216
D MARSH RABBIT PATCH OCCUPANCY 219
REFERENCES 222
BIOGRAPHICAL SKETCH 244
vi

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
METAPOPULATIONS OF MARSH RABBITS: A POPULATION
VIABILITY ANALYSIS OF THE LOWER KEYS MARSH RABBIT
SYLVILAGUS PALUSTRIS HEFNERH
By
Elizabeth A. Forys
May, 1995
Chairman: Stephen R. Humphrey
Major Department: Wildlife Ecology and Conservation (Forest Resources and
Conservation)
The Lower Keys marsh rabbit (Svlvilagus palustris hefneri) is a state
and federally endangered subspecies that historically ranged from Big Pine
Key to the southernmost of the Florida Keys, Key West. Lower Keys marsh
rabbits inhabit the marsh transition zone, an area that is currently highly
fragmented due to development. This dissertation incorporates data collected
during a 2.5-year study of the population biology, habitat requirements, and
spatial structure of the populations of the Lower Keys marsh rabbit.
Within and between patch dynamics were studied by live-trapping
marsh rabbits and fitting them with radio-collars at 6 habitat patches. Genetic
Vll

variability, growth and morphology, natality, mortality, home range size, and
dispersal were estimated from this portion of the research. To study
metapopulation dynamics, all patches of marsh rabbit habitat in the Keys were
located and sampled with fecal-pellet grids during the 2.5 years to determine
presence/absence and density of marsh rabbits at each patch.
Fifty-three marsh rabbits were live-trapped during the study and 43 of
were fitted with radio-collars. Compared to other species of rabbits, the Lower
Keys marsh rabbits were found to have average genetic variation, lower
natality, and higher mortality. Subadult males were the primaiy dispersers; all
but one male left its natal patch at the onset of sexual maturity and moved 180
- 2050 m. Due to the amount of movement between habitat patches, the marsh
rabbits appear to exist in a metapopulation.
Fifty-nine patches of transi on-zone habitat were located throughout the
Lower Keys. Twenty of these patches had pellets present during all of the
surveys, 22 had pellets present during at least one of the surveys, and 17 never
had any pellets present. Habitat patches that were close to other patches and
that had dense vegetation were most likely to be inhabited. When all 59 habitat
patches and the information about the population biology of S. p. hefneri were
combined in a popularion viability model, the model predicted that the rabbit
would go extinct in the next 20-30 years. Reducing mortality in each life stage
had the greatest effect on overall metapopulation persistence.
viii

CHAPTER 1
INTRODUCTION
The primary objective of this dissertation is to develop a population
viability analysis (PVA) for the endangered Lower Keys marsh rabbit
(Svlvilagus palustris hefneri). The Lower Keys marsh rabbit was first
described as a distinct subspecies by Lazell (1984). Lower Keys marsh rabbits
have a shorter molariform tooth row, a higher and more convex frontonasal
profile, a broader cranium, and a longer dentary symphysis than mainland and
Upper Keys rabbits.
Historically, the range of the Lower Keys marsh rabbit extended from
Big Pine Key to the southernmost of the Florida Keys, Key West (dePourtales
1877). During the 1970s and 1980s, a decline in Lower Keys marsh rabbits
was reported (J. Lazell, pers. comm.) and a study of the rabbit's status was
commissioned by the Florida Game and Fresh Water Fish Commission (Howe
1988). Howe (1988) recorded marsh rabbits present at 13 of J. Lazell's
(unpubl. data) original 17 sites and absent from 4 of the sites. The Lower Keys
marsh rabbit was listed as endangered by the Florida Fresh Water Fish
Commission in 1989 (F.A.C. 39-27) and by the U.S. Fish and Wildlife Service
in 1990 (U.S. Fish and Wildlife Service 1990).
l

2
Three subspecies of marsh rabbits are recognized, Sk paludicola, S,
palustris (Chapman and Willner 1981), and 5L p^ hefneri (Figure 1.1). The
marsh rabbit is found in lowlands from the Dismal Swamp of Virginia into the
Florida Keys. Little is known about the biology of the Lower Keys marsh
rabbit (Chapman and Flux 1990, Wolfe 1992). Compared to other species of
cottontails, the biology and ecology of marsh rabbits in general (Sylvilagus
palustris spp.) is poorly understood (Chapman et al. 1982).
Tomkins (1935) and Carr (1939) commented on marsh rabbit (Sf p^
paludicola) behavior in North Carolina and north-central Florida marshes.
Blair (1935, 1936) studied the diet and habits of marsh rabbits near
Gainesville, Florida. The reproduction of the marsh rabbit (S^ p^ palustris) was
studied by Holler and Conaway (1979) in Belle Glade, Florida.
Despite these studies, large gaps in the knowledge of the population
ecology, habitat use, and risk of extinction for S. p. hefneri remain.
Extrapolation from these studies to the Lower Keys marsh rabbit may not be
accurate. The Lower Keys marsh rabbit inhabits a unique island ecosystem
and is subject to different pressures and resources.
Population Viability Analysis
Population viability analysis (PVA) is a comprehensive examination of
the interacting factors that put a population (or species) at risk of extinction

3
(Gilpin and Soul 1986). In small populations, both stochastic and
deterministic phenomena can affect the persistence time of a population.
Perturbations that are stochastic include variation in the environment (climatic
variation, natural catastrophes), genetic composition (loss of genetic variability,
inbreeding depression), or the demographic composition (sex, age ratios) of a
population (Shaffer 1981). Demographic and genetic stochasticity are most
important in small populations. Environmental stochasticity is important for all
populations, and its importance decreases only slightly with increasing
population size.
Deterministic causes of extinction include habitat destruction, invasion
by introduced predators, disease, and climatic change (Nunney and Campbell
1993). Deterministic extinction can occur regardless of population size, if all
of the population is affected. Both stochastic and deterministic phenomena
may interact via feed back loops leading to potential extinction via "extinction
vortices" (Gilpin and Soul 1986), and differ in magnitude and importance for
species with different life-history attributes. Because knowledge about
intrinsic population dynamics of a species is often limited and the occurrence
and impact of extrinsic factors is uncertain, PVA can only make probabilistic
predictions about a species' future (Ginzburg et al. 1982, Shaffer 1990).

4
Spatial Structure
Most endangered species live in habitats that have become fragmented
(checkerspot butterfly: Murphy et al. 1990; forest-dwelling raptors: Thiollay
and Meyburg 1988, spotted owl: see Lamberson et al. 1994). For these
species, the spatial structure of the habitat patches may have large effects on
the species persistence. A PVA for a species existing in a fragmented habitat
must look at several spatial and temporal scales of resolution (Gaines et al.
1992, Lacy 1993).
Populations occurring in habitat patches may be locally subject to higher
risks of extinction than in continuous habitat due to factors (nest predation,
increased exotics, microclimate changes) that occur more frequently in smaller
areas that have high perimeter-to-area ratios (Lovejoy et al. 1986, Loiselle and
Hoppes 1983). These "edge effects" may be intensified in smaller habitat
patches, where nearly all of the patch is edge and little is interior (Wilcox and
Murphy 1985). Small patches will support fewer individuals, increasing the
rate of extinction caused by stochastic events (Gilpin and Soul 1986).
For populations of species with poor dispersal ability relative to the
distance between patches, the time to extinction will be equal to the time that
the last local population goes extinct (Hanski and Gilpin 1991). Each patch of
habitat represents a population of these species; surviving populations may be
said to be relictual (Berry 1986). Harrison (1991, 1994) uses the term non-

5
equilibrium to describe metapopulations where movements between patches
are not great enough to increase persistence time. This population
configuration has been seen in species isolated by climatic change (Brown
1971), clear-cutting of forests (Leek 1979, Laurence 1982), and man-created
islands (Willis 1974, Karr 1982). These relictual species may lack the
locomotive ability, evolutionary history, and/or behavioral plasticity to allow
them to move between patches, or the configuration of patches may prohibit
movement. Some species may exhibit "conspecific attraction" and
preferentially colonize patches where conspecifics are present (Smith and
Peacock 1990, Ray et al. 1991). These species may be less likely to
(re)colonize empty habitat patches than other species.
If individuals of the species are exceptionally mobile, relative to the
distance between habitat patches, then the individuals within the patches may
frequently interact and form one large demographic entity inhabiting a patchy
environment (Harrison 1991). Many highly mobile species of birds (Blake and
Karr 1987, Rolstad 1991) use forest patches, wetlands, continental shelf, and
oceanic islands in this manner. Other species that are adapted to highly
variable environments such as disturbance or "r-selected" species also inhabit
temporally and spatially variable habitats but remain a single population
(Edwards et al. 1981, Adler and Wilson 1987).

6
Between these 2 extremes are species that spend at least a portion of
their life in 1 habitat patch but are capable of moving to other habitat patches
during their lifetime. For populations of these species, local extinctions may
be counteracted by colonizations from nearby patches, and the overall
persistence time of the species may be collectively longer than the persistence
time of any one patch (Levins 1969). Populations of species that exhibit this
dynamic are said to exist together as a metapopulation (Levins 1970).
Metapopulations
Recently the term "true or classic metapopulation" has been used to
distinguish between this type of population configuration and others that
outwardly appear to be metapopulations such as mainland-island and source-
sink configurations (Harrison 1991). In the mainland-island configuration,
most movement occurs from a larger patch to smaller patches. These
metapopulations do not function in the "true" reciprocal colonization pattern
unless the mainland population goes extinct (Thomas and Jones 1993) and is
recolonized by the smaller "islands. In a source-sink metapopulation, the
higher quality source habitat patch produces a surplus of descendants (rate of
population growth or A. > 0), whereas the low quality patches may be sinks that
produce a deficit (A < 0) (Pulliam 1988, Pulliam and Danielson 1991).

7
Debate currently exists over how many examples of true
metapopulations exist in nature (Harrison 1994). This may be due in part to
the lack of studies that embody the spatial and temporal scales necessary for
studying metapopulations.
Metapopulation Persistence
For a metapopulation to persist, the rate of patches being colonized and
established must exceed the rate of patches going extinct (Levins 1970, Hanski
1989). A metapopulation is said to be at equilibrium when these rates are
equal. Departures from equilibrium towards increasing extinction rate may
ultimately result in a regional extinction. There may be a threshold number
and configuration of occupied patches, below which the species is not likely to
rebound (Hanski 1991), similar to the concept of a minimum viable population
of individuals (MVP; Shaffer 1981).
Persistence of individual patches is determined by dynamics that occur
both within the patch and by movements between patches. Population
dynamics, internal to patch dynamics, such as natality, mortality, sex ratio, and
age structure, interacting with patch size, habitat quality, and predator density
determine the size of the local population and its variability.
Between-patch dynamics are shaped by a species' home range, spacing
and movement patterns, especially dispersal. Movements between patches are

8
obviously important for patch colonization and establishment. These
movements also may stabilize local population variability of occupied patches
and increasing the persistence time for the local population through the rescue
effect (Brown and Kodric-Brown 1977). Patch configuration and the type of
habitat between the patches (including developed land) may ultimately have a
large impact on the individuals movements.
At the scale of the entire metapopulation, the correlation in
environmental variation among patches and the frequency of large-scale
environmental catastrophes, is vitally important (Quinn and Hastings 1987,
Gilpin 1988, Hanski 1989, Stacey and Taper 1992). If all patches are subjected
to the same adverse environmental conditions at the same time, then patch
extinctions may be correlated, leaving few or no source populations to colonize
the extinct patches.
Additionally, the cause of local extinctions may be important to the
overall metapopulation dynamics. It has been suggested (Thomas 1994) that
most local extinctions are due to deterministic changes in the environment of a
patch that render a patch unsuitable. If deterministic, these conditions are
likely to persist following the local extinction leaving the patch empty, but
unsuitable. Therefore, most empty patches are currently unsuitable for
occupancy and few suitable patches become empty. Some populations, either
because of their small size or inherently high variability, may be subject to

9
stochastic as well as deterministic forces of extinction. In these "turnover-
prone" species (Schoener and Spiller 1987; Harrison 1991), empty patches of
suitable habitat may be relatively common.
Dissertation Structure
This dissertation incorporates data on the Lower Keys marsh rabbit at 3
spatial levels of observation: within-patch population dynamics, inter-patch
between population dynamics, and metapopulation dynamics. Using
simulation models that incorporate these three spatial scales, predictions are
made for a range of temporal scales.
Chapter 3 investigates the dynamics of 5 populations of Lower Keys
marsh rabbits living in habitat patches. Estimates of natality, mortality,
demographic structure, genetic and morphological variation were made at this
scale. In Chapter 4, the home range size, movement patterns, and dispersal
ability of the marsh rabbit were measured. This information about marsh
rabbit inter-patch use provided the basis for testing the hypothesis that the
marsh rabbits are existing in a true metapopulation configuration.
Chapter 5 explores patch occupancy, extinction, and (re)colonization at
the metapopulation scale. Marsh rabbit diet, microhabitat, and macrohabitat
selection were combined with the physical attributes of the patch to determine
if these parameters influenced patch occupancy. Chapter 6 incorporates the

10
conclusions drawn from chapters 3-5 into a PVA to test the hypothesis that the
metapopulation was in equilibrium. Predictions about the future for the Lower
Keys marsh rabbit were made under a number of different scenarios.
In the final chapter (Chapter 7), conclusions and recommendations are
made for the management and conservation of the Lower Keys marsh rabbit
and other endangered species inhabiting patch/fragmented landscapes.
Predator control, species reintroductions, the impact of more habitat loss,
diseases, and the impact of hurricanes, are examined in light of the new
research on metapopulation dynamics.

11
Figure 1.1--Distribution of the 3 subspecies of marsh rabbit (Sylvilagus
palustris).

CHAPTER 2
THE LOWER KEYS OF FLORIDA
Geologic History
The Lower Keys of Florida are the terminal portion of an archipelago of
islands extending westward from the mainland of Florida (Figure 2.1). For the
purposes of this dissertation, the Lower Keys includes all of the islands from
Key West to Noname Key (inclusive). The Lower Keys are separated from the
Upper Keys by nearly 20 km of water (from Noname Key to Vaca Key).
The Keys are derived from coral reefs that formed during the Sangamon
interglacial period (340,000-100,000 BP). During the subsequent Wisconsin
glacial period, the Lower Keys were covered with oolitic sands and were
gradually exposed as sea level began to fall. At the height of the Wisconsin
glaciation (20,000 15,000 BP) the Keys were points of high relief amid the
large land mass of Florida (Shinn 1988, Mueller and Winston 1991). At this
point in time, Florida had nearly twice the exposed land as today (Webb 1990).
Sea level remained low until nearly 15,000 BP, when the climate began to
warm. Sea level rose rapidly from 15,000 10,000 BP, reaching a maximum
of 7 feet/year, and then the rate slowed. During the past 2,000 years sea level
has risen only 6 feet.
12

13
Today, the Florida Keys are composed of two formations of limestone.
The Upper Keys (Soldier Key to the southeastern comer of Big Pine Key) are
composed of the Key Largo limestone and the Lower Keys (the majority of Big
Pine Key to Key West) are from the Miami limestone. The Miami limestone is
composed of small ovoid pellets of calcium carbonate (ooids; Hoffmeister
1974) that hardened when sea level fell and the land surface was exposed to
air.
Climate
Although several degrees north of the Tropic of Cancer, the close
proximity of the Gulf Stream and maritime influences produce a subtropical
climate in the Lower Keys (Chen and Gerber 1990). In 50 years of data
collection, the temperature in the Lower Keys has never fallen below freezing,
and the record low temperature is 41 F (5 C). Winter cold fronts are buffered
by the warm ocean water before proceeding to the Keys. Monthly average
temperatures differ only 15 F (9 C) from January to July (Figure 2.2). Daily
temperatures rarely vary more than 10 F (12 C) between the high and low.
There are distinct dry and wet seasons in the Lower Keys, although this
area is generally slightly drier overall than the Upper Keys and mainland
Florida (Chen and Gerber 1990). The dry season spans November through
April and accounts for less than a third of the annual precipitation. The

14
remaining two-thirds of the rainfall occurs during the wet season which begins
in May and ends in October (Figure 2.2). Relative humidity remains high
throughout the year, averaging nearly 75% (NOAA 1993).
Hurricane season occurs mainly during the wet season, from June until
November. The chance that a hurricane (winds >74 mph) will hit the Lower
Keys is slightly less than 10% each year. Historically, most hurricanes have
occurred during the months of September and October. Tropical storms (i.e.,
winds 39-74) occur more frequently, but do not produce the large storm surges
and destruction that hurricanes are capable of producing.
Flora and Fauna
The flora of the Lower Keys is derived from 4 sources: the Caribbean,
the eastern U.S. coastal plain, endemic taxa evolved in place, and exotic taxa
from cosmopolitan sources (Long 1974). Four plant associations predominate:
mangrove, transitional salt-marsh, pineland, and hardwood hammock.
Mangrove community is dominated by 3 saline-tolerant, unrelated
species: red mangrove (Rhizophora mangle), black mangrove (Avicennia
germinans), and white mangrove (Laguncularia racemosa). These trees occur
in areas that are either continually submerged or tidally inundated.
Transitional salt-marsh (also called the transition zone) is a grassy,
nearly treeless marsh area that generally occurs from 1 to 3 m above sea level.

15
The majority of transition zone marshes are subject to predictable flooding by
spring lunar high tides (Williams 1991). Transition zone marshes often occur
between the mangrove community and the upland hardwood or pine
hammocks. The transition zone can be further divided into two components:
an open saltmarsh and at slightly higher elevations, a more forested area
dominated by buttonwood (Conocarpus erectus). Several species of mammals,
including the Lower Keys marsh rabbit inhabit this area.
On land where there are well-developed fresh water lenses, the pineland
community can occur. These areas are rarely inundated by salt water and are
maintained through periodic bums. Slash pine (Pinus elliottii) is the dominant
tree species, although various species of palms and ferns are also abundant
(Snyder et al. 1990). At the highest elevations (>3 m) the diverse hardwood
hammock occurs (Carlson et al. 1992). This is the climax community in the
Keys and occurs on nearly every large key. Nearly 10% of all tree species
found in this area are endemic (Long and Lakela 1971).
Although less common, fresh water wetlands dominated by sawgrass
(Cladium iamaicensisl do occur on the few keys that have a fresh water lens.
A few keys also have limited amounts of beach and dune habitat, but most of
the keys have exposed limestone rock on their coasts.
All of the native terrestrial mammals are derived from populations of the
continental United States (Layne 1974). Currently, only 5 species of native

16
terrestrial mammals and perhaps 1 bat are found in the Lower Keys. All of the
terrestrial mammals have been cited as being endemic in the literature at either
the species or subspecies level (see Lazell 1989), although the accuracy of
these taxonomic claims has been debated (see Humphrey 1994). The paucity
of mammals in the Lower Keys may be related to the current isolation and
small area of the Keys, combined with a lack of fresh water (Layne 1974).
Herpetofaunal diversity is greater than the mammals, although a large
number of reptiles and amphibians in the Lower Keys are exotics (Wilson and
Porras 1983). Of importance to the Lower Keys marsh rabbit, the eastern
diamondback rattlesnake (Crotalus adamanteus) is common throughout the
Lower Keys and the alligator (Alligator mississippiensis). although rare, does
occur. Avian diversity, both breeding and over-wintering is fairly high in the
Lower Keys although lower than mainland Florida (Robertson and Kushlan
1974). Special references to birds and reptiles as predators will be made in
Chapter 3.
Development
By the 1890s Key West was the largest city in Florida and one of the
largest in the United States. The remainder of the Keys were unpopulated until
the completion of the overseas highway in 1938 and the first water pipeline in
1942 (Gallagher 1991). Today, over 78,000 people live in Monroe county,

17
nearly half of them in the Lower Keys (Shermyen 1993). The populations has
increased nearly 65% since 1960. Tourism, which was also high during the
late 1800s, has also resurged in the Lower Keys. At the 2 National Park
beaches in the Lower Keys, over 500,000 tourists were recorded in 1993. The
majority of these visitors came to the Lower Keys and Key West via the
Overseas highway (Shermyen 1993).
Human impact on the Keys' wildlife was noticed as early as 1908, when
the Key West National Wildlife Refuge was established to provide habitat for
migratory birds. The Great White Heron National Wildlife Refuge was
established in 1938, encompassing a collection of small islands north of the
main chain of keys. In 1957, the National Key Deer Refuge was established;
its main mission was to protect the Key Deer (Odocoileus virginianus clavium)
and other wildlife. The largest tracts of land for this refuge are on Big Pine
Key and Noname Key, but other areas on the Torch Keys are currently being
added (John Andrew, National Key Deer Refuge Manager, pers. comm., 1993).
The United States military presence began in the Keys during the Civil
War, and waxed and waned throughout the past century (Hambright 1991).
Due to the Lower Keys' proximity to Cuba, the Caribbean, and open water
ideal for training aviators, the U.S. Navy currently maintains a base in the
Keys. The Navy owns a large amount of land on Key West, Boca Chica Key,

18
and North Saddlebunch Key. The U.S. Air Force manages the northernmost
portion of Cudjoe Key.
Study Sites
This study attempted to encompass all possible marsh rabbit habitat
throughout the Lower Keys, including saltwater transition-zone habitat and
fresh water marsh areas. Habitat areas were located using information from
Howe (1988) aerial photographs and ground survey. Initially, a patch of
habitat was considered isolated from another patch if the 2 areas were divided
by a major road, airplane runway, or large body of water. As a general rule,
patches of habitat <0.5 ha were not sampled, although some exceptions were
made. A previous study of marsh rabbit home ranges (Payne 1975) found that
no adult home range was <0.5 ha.
Fifty-nine habitat patches were identified throughout the Lower Keys
(Appendix A); no habitat was found on Key West, the most densely human-
populated Key. Twenty-seven habitat patches were found on Boca Chica, 25
of which were owned by the U.S. Navy. Four patches were examined at the
Saddlebunch Keys; 2 of which were owned by the U.S. Navy. On Sugarloaf
Key, 9 patches of transition zone habitat were found; 1 of which is owned by
the U.S. Fish and Wildlife Service (USFWS), and other is owned by Monroe
County. Two privately owned patches were found on Cudjoe, and 1 site

19
partially on U.S. Air Force land. One privately owned site was found on each
of Summerland and Ramrod Keys. Big Torch Key had 2 privately owned sites
and one site owned by the USFWS. Middle Torch had 2 sites, both of which
were owned by the USFWS. Little Torch had 1 habitat patch and this patch
was owned by the Nature Conservancy (TNC). Big Pine had 7 habitat patches:
3 owned by the USFWS, 2 privately owned, and 2 owned by the South Florida
Water Management District (SFWMD). Noname Key had 1 USFWS-owned
habitat site.
Additionally, habitat was found on several of the smaller islands that are
part of the Lower Keys chain but that are only accessible by boats. Habitat
patches were found on Annette and Porpoise Keys, both of which are small
islets northeast of Big Pine Key. These islets are part of the Great White
Heron National Wildlife Refuge. Transition-zone patches were observed on
Big Munson Key and Saddlehill Key. Big Munson Key is south of Big Pine
and is owned by the Boy Scouts of America, Inc. Saddlehill is southeast of
Boca Chica Key and is privately owned. The search for habitat on other
isolated islets was not complete; the existence of small patches of habitat on
more distant keys is a possibility.

Middle Torch
\
Big Torch
Big Pine
Cudjoe
N. Saddlebunch\
Saddlebunch \
1J Qjk
jmmerland Parr1
e.
Little Torch
O
Sugarloaf
10
Key West
Boca Chica
Figure 2.1'The Lower Keys of Florida. The solid line connecting the Keys is highway US-1.

Average Temperature ( F)
21
1 2 3 4 5 6 7 8 9 10 11 12
Month
10
9
8
7
6
5
4
3
2
1
0
Figure 2.2-The average temperature (line, left axis) and average precipitation
(bars, right axis) during the past 25(1965-1990) years measured at Key West
airport, Key West, Florida.
Average Precipitation (in.)

CHAPTER 3
POPULATION BIOLOGY (INTRA-PATCH DYNAMICS)
Introduction
The first step in developing a population viability analysis (PVA) is
determining the demographic and genetic composition of a population, how
these compositions vary, and how large a role stochastic and environmental
variability play (Salwasser et al. 1984, Gilpin and Soul 1986). Demographic
and genetic compositions of a population can influence population growth,
variability in growth, and the ultimate population size, all of which can in turn
influence the populations chance for persistence (Goodman 1987).
The demographic composition of a populations is determined by
variation in the birth and death rates and in the sex ratio of the new recruits.
Some variation in these parameters will be purely random, other sources of
variation may be from external sources (i.e., differential mortality by sex or age
group, biased sex ratios at birth ), and therefore this variability may be
correlated throughout the population. In a large, continuous population,
random variations will contribute little to the overall population variability but
environmental variability may be important; in very small, fragmented
22

23
populations, both stochastic and environmental variation can substantially alter
sex and age ratios.
Genetic variability, at both the individual and population level, may also
be important. Highly fragmented, isolated populations are at risk of loosing
both types of variation. When a population is small and isolated, the chance of
a mating between close relatives increases. These matings can produce young
with a higher proportion of homozygous loci, thus potentially decreasing the
number of heterozygous loci and increasing the number of deleterious recessive
genes (Packer 1979). Certain heterozygous loci have been correlated with
greater tolerance to environmental variations in some species (Lemer 1954).
This increase in homozygous deleterious genes, combined with a
decrease of heterozygote loci, is believed to cause greater mortality, and
reduced fecundity, creating a phenomena called inbreeding depression (see
Lacy 1993). Loss of genetic variability among individuals or between
populations may contribute to inbreeding depression, and may decrease the
overall chance that the population can evolve to meet new environmental
conditions. Although loss of genetic variability in insular populations is well
documented (Kilpatrick 1981, Berry 1986) its overall effect on population
persistence is less known.
Inbreeding depression was documented early in domesticated animals
(Wright 1977, Falconer 1981), laboratory animals (Strong 1978) and more

24
recently in captively breed animals from wild stock (Ralls and Ballou 1983,
Ralls et al. 1988). Evidence for inbreeding depression in natural populations is
less clear (Lacy 1993), and in need of further study. Most examples of
possible instances for inbreeding depression in the wild come from populations
of large, long-lived species such as the grizzly bear (Ursus arctos; Harris and
Allendorf 1989) and the Florida panther (Felis concolor coryi; Seal and Lacy
1989). These species have existed at extremely small numbers in the wild for
several decades and have largely overlapping generations.
For smaller, shorter-lived species, it is currently hypothesized that
before inbreeding depression can have an effect in the wild, habitat loss and
demographic problems may have already caused the demise of the wild
populations (Lacy 1993). Management for these problems may also help in
preventing inbreeding depression and overall loss of genetic variability in the
population, although more intense management may be needed for populations
that have recently undergone severe population bottlenecks (Fuerst and
Maruyama 1986). Inbreeding depression for these species may still be an
important consideration in PVA, if captive breeding and reintroductions are
available as management options (Haig et al. 1990).
The sensitivities of demographic and genetic compositions to random
and environmental variation are dependent on the life history and population
dynamics of the species in question. Further, they may be affected by

25
idiosyncratic factors such as the habitat available and climatic variation, and
therefore require in-depth study.
Lagomorph population biology is characterized by high reproductive
rates and high rates of mortality. All lagomorphs are entirely herbivorous,
although they feed on a diversity of vegetation. Due to their high abundances
and intermediate size, lagomorphs are the base of many predator-prey systems
involving small to medium-sized predators (Chapman and Flux 1990).
Most research on rabbits has been on the most abundant species (e.g.,
eastern cottontail, Sylvilagus floridanus), primarily in relation to hunting.
There is little information on the management of rare or endangered lagomorph
species. Additionally, most research on endangered species (in general) has
historically concentrated on larger species that are long-lived and have longer
generation-times (Murphy et al. 1990). Threats to the persistence of smaller
species with higher reproductive rates, but shorter life-spans (r-selected) will
differ. These "r" selected species may be more habitat specific than "K"
species and may experience higher variability in population numbers (Pimm
1991).
The objective of this portion of the research was to determine the
demographic and genetic compositions of the Lower Keys marsh rabbit.
Special emphasis was placed on parameters that affect the intrinsic rate of
growth of the marsh rabbit populations (natality, mortality, sex and age ratios),

26
the variability of these parameters, and the extrinsic factors that influence this
variability. This information about the population biology was used in the final
model predicting the future of the Lower Keys marsh rabbit presented in
Chapter 6.
Methods
Two methods were employed to study the population biology of the
Lower Keys marsh rabbit: live-trapping and radio-telemetry. Marsh rabbits
were studied at a subsample of 6 habitat patches. All of the patches chosen
were on Navy-owned land; a diversity of patch size and shape was sampled.
Five habitat patches were selected on Boca Chica and 1 on Saddlebunch Key.
Additional sites throughout the Lower Keys were used for the portion of the
study that examined the genetic composition of the marsh rabbits.
Trapping Grids
Individual marsh rabbits were examined by trapping the 6 main sites.
Trapping occurred twice during the wet season (June November), and twice
during the dry season (December May). For the 5 sites on Boca Chica Key
(Figure 3.1), trapping occurred from June 1991 to May 1993 (8 trapping
sessions). On Saddlebunch, trapping was conducted from June 1992 to May
1993 (4 trapping sessions). Trapping grids were placed on each site, using

27
unbaited collapsible National live traps (80 x 30 x 30 cm), placed in a 6 x 6
array, spaced approximately 25 m apart. Each trapping session consisted of 5
nights where the traps were open, 2 nights with the traps closed and another 5
nights with traps open. Traps were checked twice daily, once in the morning
and once in the evening and were covered in burlap for shade.
All rabbits caught were sexed, weighed, and tagged (Monel no. 3,
National Band and Tag, Newport, KY). Measurements were made of the right
rear foot, ear length from notch, and total body length from nose to (and
including) tail. Marsh rabbits are relatively easy to sex (Nagy and Haufler
1980), but an aging method has not been determined. An attempt to correlate
weight, total length, ear length, and rear right foot length with age was made.
Rabbits were classified as juveniles (not sexually mature), subadults (entering
sexual maturity), or adults (fully sexually mature).
Genetic Analysis
Blood samples were obtained from all animals caught on the 6 trapping
grids between June 1992 and May 1993. In addition, blood was taken from
animals caught during preliminary trapping on Sugarloaf and Big Pine Keys.
Blood samples were obtained by lancing the ear with a hypodermic needle and
collected using heparinized capillary tubes. Blood was stored on wet ice for <1
hour and was centrifuged into plasma and hemolysate at 5,000 ipm for 10 min

28
(Scribner et al. 1983). The plasma and hemolysate were stored in liquid
nitrogen and were analyzed used starch-gel electrophoresis within 1 year.
Eleven presumptive genetic loci were scored for the samples. Standard
procedures for starch-gel electrophoresis were used (Harris and Hopkinson
1976). Locus nomenclature followed McAlpine et al. (1985) for mapped
human genes. Two buffer systems were used. Tris-citrate, pH 6.7 was used
for the following enzymes: esterase, E.C. 3.1.1.1 (EST-1); glucose phosphate
isomerase, E.C. 5.3.1.9 (GPI); isocitrate dehydrogenase, E.C. 1.1.1.42 (ICD-1);
lactate dehydrogenase, E.C. 1.1.1.27 (LDH-1), malate dehydrogenase, E.C.
1.1.1.37 (MDH-1); mannose phosphate isomerase, E.C. (MPI); and superoxide
dismutase, E.C. 1.15.1.1 (SOD-1). Tris citrate, pH 8.0 was used for the
following enzymes: beta hemoglobin (PHb); glutamate oxaloacetic
transaminase, E.C. 2.6.1.1 (GOT-1); peptidase, E.C. 3.4.13 (PEP-1); and
phosphoglucomutase, E.C. 2.7.5.1 (PGM-1).
Radio-telemetry
Each rabbit weighing >1,000 grams trapped at the 5 Boca Chica sites
and the Saddlebunch site was fitted with a radio-collar (weight <3 grams) and a
transmitter with an estimated 10-month operational life (Telonics, Inc., Mesa,
AZ). Smaller rabbits (300-1,000 g) were fitted with a similar radio-collar, but
with a velcro break-away device added to allow the animals to lose the

29
equipment as their bodies grew larger. Collars were replaced as the animals
aged. Previous studies on cottontail rabbits (S. floridanus) found that mortality
rates were not statistically different for radio-telemetered rabbits and rabbits
marked using other methods (Trent and Rongstad 1974, Rose 1977).
Collared rabbits were located on separate days three times a week, once
in the early morning (7-9 a.m.), once at mid-day (11 a.m. 1 p.m.), and once in
the evening (4-6 p.m.). Signals were followed until the animal could be seen
or the exact location was found. All locations were made >24 hours apart to
ensure independence of observations. Locations were plotted on 1:2400 aerial
maps with 10-m2 grid overlays. Because the 5 sites on Boca Chica were
studied for 2 years and the Saddlebunch site was only studied during the final
year, natality and mortality portions of the study only used data from the 5 sites
on Boca Chica Key (Figure 3.1).
Natality
Natality was studied by examining females during each trap session for
pregnancy (by palpating the uterus) and lactation. Nesting was determined by
following collared female rabbits until they centered their activities around one
small area. Nest confirmation was made by observing the young through a
tunnel in the grass or by finding deposits of juvenile pellets near the nest area.
Where possible, the number of young observed in or fleeing the main chamber

30
of the nest was recorded. Although this method is approximate, it provided an
estimate of the number of individuals surviving at the time of nest discovery.
Number of litters for each female was calculated for both the wet and dry
seasons and compared using a likelihood-ratio Chi-square (G^). The nest area
was mapped, and the dominant vegetation was identified.
Mortality
Percent mortality was calculated for 5 age classes: nestling, juvenile,
subadult, first-year adult, and second-year adult. Nestling mortality (0-3
months) was calculated by comparing the number of young observed in nests to
the number of juveniles caught on the trapping grid. Juvenile mortality (4-7
months) was estimated using telemetry for individuals >300 g. Subadult (8-10
months), first-year and second-year adult mortality was determined from the
telemetry data alone. First-year adults were rabbits that had been followed as
an adult for 1 year; most had been collared as juvenile or subadults. Second-
year adults were rabbits that had been followed as adults for 2 years. First- and
second-year adult classes may have contained individuals older than 2 and 3
years old.
All mortalities were located within 2 days of death. When possible,
field necropsies were performed as outlined by Wobeser and Spraker (1980),

31
with special attention to the liver for signs of tularemia (Francisella tularensis).
Site description, carcass condition, and position was recorded for each death.
All rabbit carcasses thought to have been preyed upon were examined
for signs of trauma. The head, throat, and neck were examined for puncture
wounds. Hemorrhaging, particularly the presence of blood in the mouth, nares,
trachea or neck region indicated the rabbit was alive at the time of attack
(Hawthorne 1980). Feeding pattern on the carcass was also examined:
hindquarter feeding probably indicated scavenging, whereas feeding on the
shoulders and neck indicated possible predation (Hawthorne 1980).
When predation was suspected, the site was examined for sign on and
around the carcass. Tracks and scats were identified for potential information
on the predator or scavengers. Potential predators found in the Lower Keys
included Bald Eagles (Haliaeetus leucocephalus). Red-shouldered Hawks
(Buteo lineatus), eastern diamondback rattlesnakes (Crotalus adamanteus),
feral cats (Telis catus), raccoons (Procyon lotor) and possibly Black Vultures
(Corgyps atratus) and domestic house-based dogs (Canis familiaris).
Birds of prey tend to capture rabbits in the middle of the back, and will
kill using deep puncture wounds to the back and head (Hawthorne 1980).
They may take their prey back to a nesting area. Feral cat predation may be
assumed if the carcass has been dragged, eviscerated, or covered in dirt. Often
cats will leave tooth marks on every exposed bone of their prey (Anderson

32
1969). Raccoons are more likely than cats to eat the breast, crop, and entrails
of their prey. They may also carry portions of the prey to water (Anderson
1969).
Results
A total of 54 Lower Keys marsh rabbits was caught and examined
during this study (Table 3.1). Forty-three were caught on the 6 main grid sites
(41 on Boca Chica, 2 on Saddlebunch), 5 rabbits were caught while trying to
recapture collared dispersed rabbits, and 6 were caught on other sites in the
Lower Keys. Rabbits were only examined once per trapping period. Data
from the 54 rabbits was recorded 130 times during 8 trapping sessions (June
1991 May 1993).
Forty-three (28 male, 15 female) of the rabbits caught were fitted with
radio-collars, and 7 of these were juveniles. More juveniles were caught
before the break-away collar technology was complete. Forty-one of the radio-
collared rabbits were on Boca Chica Key and two were on Saddlebunch Key.
Growth and Morphology
Mass of the 54 marsh rabbits ranged from 100 to 1400 grams (including
pregnant females). It was judged that the 100-gram rabbits had just left the
nest and were approximately 1 month old. Several young rabbits were causght,

33
but only 3 rabbits (1 male, 2 females) were caught during this stage and
survived to adulthood. Marsh rabbit age was correlated with external
measurements using these 3 individuals.
Both body mass and total body length were found to significantly (P
<0.05) predict marsh rabbit age for all 3 rabbits (Table 3.2). Ear length and the
length of the right hind foot were not significant predictors in one of the female
rabbits. Body mass has been used as an indicator of age in cottontails (Lord
1963), but the length of the right rear foot is used more often (Bothma et al.
1972). Because measuring the length of a non-sedated marsh rabbit is difficult,
the marsh rabbit body mass was used as an indicator of age in conjunction with
appearance of external sexual organs. Body mass can be influenced by
nutritional and physiological variations but was the most objective and accurate
age indicator available. Body mass of the marsh rabbits increased almost
linearly with age for the 3 rabbits and plateaued at approximately 1,100 grams
(Figure 3.2) as the rabbits approached 1 year.
External measurements collected from 13 non-pregnant adult female
rabbits and 19 adult male rabbits were compared for differences using a
Wilcoxon test. In an attempt to use only measurements on fully grown rabbits,
only individuals that had been caught more than once as an adult were used.
No significant differences were found in any of the measurements between
sexes. Approximately half of all cottontial species are sexually dimorphic

34
(females larger) and the other half exhibit no dimorphism (Champman et al.
1982). The average measurements and standard deviations of 29 marsh rabbits
were as follows: mass 1224.1 g. (80.9), total length 339.3 mm (24.9), ear from
notch 52.7 mm (3.4), and right hind foot 73.6 mm (3.7).
Genetics
Two (EST-1 and PGM-1) of the 11 loci sampled from the 19 rabbits
were polymorphic, indicating that there is genetic variation in the Lower Keys
marsh rabbit (Table 3.4). This proportion of variable loci is only slightly less
than the proportions seen in studies of cottontail rabbits in Texas (Scribner and
Warren 1986). Deviations from Hardy-Weinberg expectations were not
observed (P > 0.05) at either of the polymorphic loci, but this test may not be
valid on a so small sample. Differences in allelic proportions for each locus
appeared to vary among keys, but sample sizes were too small for statistical
comparison. Heterozygous loci and variation among individuals occurred at
Boca Chica, Geiger, and Sugarloaf Keys. The 2 rabbits sampled on the Navy
land on Saddlebunch Key were homozygous and monomorphic at all loci.
Only 1 individual was sampled from Big Pine Key; this individual was
heterozygous only at the EST-1 loci and contained an allele at the EST-1 locus
that was not seen at any of the other keys.

35
Demographics
The number of marsh rabbits caught on the 5 main sites varied from a
high of 18 rabbits during November 1991 and February 1992 to a low of 9
during November 1992 (Figure 3.3). The small number of rabbits at each site
precluded the use of capture/recapture statistics (Pollock et al. 1990);
population number at each site was estimated using the minimum number
known alive (MNA; Hilbom et al. 1976).
When all of the male and female individuals caught over the 8 trapping
sessions at the 5 Boca Chica sites were counted, there appeared to be more
male than female rabbits on the 5 sites during the study (Table 3.3). The sex
ratio significantly male-biased in only the subadults; nearly equal numbers of
male and female juveniles and adults were captured on the 5 grids.
Trappability (the probability that a rabbit was captured more than once
during the 10-day trapping session) differed among the demographic groups
(Table 3.5). Male and female trappability was high and not statistically
different. Juvenile trappability was lower, and females were more likely to be
retrapped during a trap session than were males. Subadult trappability was
high for the males, and very low for females. Sample sizes for the trappability
comparisons were small.

36
Natality
Eleven adult female marsh rabbits were radio-collared, followed for >1
month, and used in this portion of the analysis. Length of time followed
ranged from 2 to 22 months (X = 9.09 months, SE = 1.94 ) for a total of >100
months. Thirty-one nesting events were recorded; all females were observed to
nest and produce a litter at least once. The number of young observed per nest
ranged from 1 to 3 with an average of 1.77 kittens per nest (SE = 0.09).
All nests were made in clump grasses (thick grasses and sedges), with
22 of the nests predominately in Spartina spartinae and the other 9 in
Fimbrystvlis castanae. In general, the nests consisted of a main chamber with
several smaller chambers and exit/entry routes. None of the nests were
obviously lined with fur as reported in the northern subspecies (Tomkins
1935). Only 2 of the females used the same nesting area more than once, and
none of the female rabbits used the nest of another rabbit during this study.
There was little apparent seasonal pattern in the reproduction of the
marsh rabbit (Figure 3.4). Combining data from the 2 years of the study, the
proportion of rabbits that produced litters each month ranged from 0-56%. The
highest proportion of females with litters was seen in March and September;
the lowest proportion was seen in April and December. The average number of
litters produced during the wet and diy seasons did not differ significantly (G^
= 0.15, df = 1, P > 0.05). Although reproduction in most cottontail species is

37
highly seasonal (Chapman et al. 1982), reproduction in marsh rabbits in
southern Florida (Holler and Conaway 1979), southern Texas (Bothnia and
Teer 1977), and southern California (Ingles 1941) exhibits only year-round
slight seasonal fluctuations.
Time between litters ranged from 1 to 5 months (X = 2.45, SE = 0.30).
Combining the results from the 11 rabbits, an average of 3.7 litters per year
was produced. Although their research methods differed from those reported
here, Holler and Conaway (1979) measured a higher fecundity rate (5.7 litters
per year) for marsh rabbits (S^ jr. paludicola) living in southern Florida.
Mortality
Fiftythree rabbit kittens observed in the nests, but only 15 young rabbits
were trapped on the grids. The survival rates (from one developmental stage to
the next) at the 5 main study sites ranged from 13 to 50%. Forty-one
individuals (26 male, 15 female) were trapped and collared at the 5 sites on
Boca Chica Key. Twenty seven (19 male, 8 female) died during the study.
Because of these small sample sizes in some of the demographic groups
statistical tests were not used. Percent mortality was highest for the second
year adults (X = 90%) and lowest for the juveniles (X = 25%). Male mortality
(%) was higher than female mortality for each age class (Figure 3.5). None of
the females in the study died during their subadult period, but nearly half of the

38
subadult males were killed. One female was caught as a subadult and then
followed until the study ended, establishing a maximum longevity of 3 years
for this study.
Cause of mortality was determined for 24 rabbits (Figure 3.6). None of
the juvenile mortalities were included because their bodies were never found.
Presence of blood and fur on the break-away collar and subsequent
disappearance from the trapping grid were assumed to mean the animals had
been killed. It was possible to determine the cause of mortality for all of the
subadults (all male) and 19 of the adults. Only 1 of the adults disappeared
during the study.
Cause of mortality was organized into 6 classes: cat/raccoon, vehicle,
rattlesnake, raccoon, cat and poaching. The cat/raccoon class was used for
mortalities where it was not possible to distinguish between predation by a
feral cat or raccoon. Sign from both predators was abundant at the kill sites.
In nearly all of the kills the rabbit was dragged and partially buried, indicating
cat predation. In 3 cases the cat or raccoon was seen with the rabbit, and it was
assumed that the predator had killed the animal. In addition, Amim Sheutz (the
Natural Resource Manager for the Navy on Boca Chica Key, pers. comm.)
reported an observation oft a Bald Eagle grasping a medium-sized marsh
rabbit.

39
Full necropsies were only possible on 2 of the rabbits. The remainder of
the carcasses were entirely eaten, eviscerated, or decomposed. Neither of the
necropsies revealed any sign of disease.
The most subadult and adult mortalities were caused by vehicles (Figure
3.6). It was followed closely by the number of individuals killed by feral cats
and raccoons. Three rabbits were eaten by rattlesnakes (the radio-collar
transmissions were traced to the rattlesnake) and one rabbit was shot. Both
sexes appeared to be susceptible to predation by cats and raccoons, but more
males were killed by vehicles. Seven males were killed by vehicles on the
road, 4 of them subadult males. The only female road-kill was actually killed
in the marsh adjacent to the road. The vehicle had apparently been driven
through the habitat. Mortality on and off the base on Boca Chica Key was
similar. Road-kills, feral cat predation, and raccoon predation occurred on
both on and off the base on Boca Chica Key. All of the predation by
rattlesnakes and the poaching occurred on Navy land off the Boca Chica base.
Discussion
The high recapture rate of adult marsh rabbits indicates that most adult
rabbits living on a grid were probably captured during the study. The lower
trappability percentages for juveniles and female subadults suggests that some
of these individuals might have been missed. Because of this difference in

40
trappability, most notably in the female subadults, the apparent male bias in the
sex ratio may be due to differences in the trappability. Both male and female
biased sex ratios are common in the literature (Chapman et al. 1982).
Speculation about the role of biased sex ratios of litter and of adult rabbits has
been made in relation to mating systems (Chapman et al. 1977), population
density, precipitation, and success of hunters (Edwards et al. 1981).
Average productivity for female Lower Keys marsh rabbit (6.36
young/year) was slightly lower than for marsh rabbits in southern Florida
(Holler and Conaway 1979). However, the productivity measure in this study
only accounts for young seen in the nest during a 2-week period after birth. It
is possible more young were bom and not seen in the nest or that some young
died before, during, or shortly after birth. Productivity estimates are
substantially lower than for eastern cottontail rabbits (Chapman et al. 1982).
Lord (1960, 1963) noted that a decrease in the litter size of cottontails was
correlated with a decrease in latitude. As the size of the litter decreases, the
potential length of the breeding season increases. The breeding season and
litter size of the Lower Keys marsh rabbit are consistent with this general
observation.
Breeding was year-round in the Lower Keys. Cottontail breeding and
reproduction are generally tied to temperature and precipitation. In the Keys,
these climatic changes may be more subtle. Although the Florida Keys lie a

41
few degrees north of the Tropic of Cancer, the climate can be considered
tropical (Walter et al. 1985). The temperature in the Lower Keys has not fallen
below freezing in historic times. Rainfall is seasonal, but the Keys do receive
one-third of their rainfall during the "dry" season. Slight variations in marsh
rabbit reproduction might be related to inter-year variation in precipitation.
Marsh rabbit mortality was high for nearly every age class and sex. The
estimate of nestling mortality, however, may be negatively biased. The
probability of catching a rabbit during the first 3 months of its life is unknown
and it is possible that some young rabbits survived the first 3 months but were
killed later as juveniles. It seems unlikely that the young rabbits left the habitat
patch. Cottontail rabbits are extremely altricial, and very young rabbits appear
to be only capable of moving short distances, especially without cover.
Subadult males appeared to have a particularly high mortality rate,
mainly due to the number killed by vehicles. Second-year adults also have a
high mortality rate, indicating that adult marsh rabbits may become more
vulnerable to predation as they age. In general, fewer females were killed than
males. This difference in mortality data may be due to the male-skewed sex
ratio of the radio-collared subadult rabbits.
Vehicles killed nearly a quarter of radio-collared marsh rabbits in this
study, including the subadult dispersing males. Because marsh rabbits are most
active from dusk until dawn, night-time road traffic appears to be a threat to

42
marsh rabbit survival. Predation by cats has not previously been reported for
marsh rabbits, but cats were a major predator of adult and juvenile rabbits of
other species in several other studies (Fitzgerald and Karl 1979, Jones and
Coman 1981, Liberg 1985). The cat population of the Lower Keys is large and
includes feral and house-based cats.
Genetic Threats to Persistence
The Lower Keys marsh rabbit had a level of variation similar to large
populations of cottontails (Scribner and Warren 1986). Inbreeding depression
does not appear to be a threat for most of the Lower Keys marsh rabbits.
Although past levels of genetic variation are unknown, it does not appear that
these marsh rabbits are currently deficient in genetic variation. Future threats
may be dependent on the population size and structure.
Demographic Threats to Persistence
Fecundity of the Lower Keys marsh rabbit is lower than other
subspecies of marsh rabbits and other species of cottontails. Possible
explanations for this lower fecundity include: reabsorption of fetuses (Holler
and Conaway 1979), insufficient food or nutrients for reproduction (Cheeke
1987), adaptive pressures that have resulted in longer periods between
pregnancies, and lack of suitable mates at the habitat patch.

43
Although the other hypotheses cannot be eliminated, it is highly
probable that some females were without mating opportunities during part of
the study. The low population density, combined with stochastic variation in
the sex ratio of the adults at a habitat patch, left at least 2 females (sites 8 and
9, trapping period one) without mates. These females did not produce a litter
during this period. The ability of males and females to travel between habitat
patches will be examined in Chapter 4. If these habitat patches are separate
populations, it appears that stochastic demographic variation can have a large
impact on the intrinsic rate of growth of the population (r).
This lower fecundity was matched with a mortality rate higher than most
non-hunted populations of wild rabbits (Trent and Rongstad 1974, Chapman et
al. 1982). The majority of the mortality was from anthropogenic sources
(vehicles, house cats), that are not generally density-dependent.

44
Table 3.1-Number of Lower Keys marsh rabbits (Svlvilagus palustris hefneri)
trapped at each site in the Lower Keys of Florida from June 1991 to May 1993.
Site
Ownership
Male
Female
Total
3
Navy
1
0
1
4
Navy
2
1
3
7a
Navy
4
3
7
8a
Navy
8
4
12
9a
Navy
6
3
9
10a
Navy
4
1
5
13a
Navy
4
4
8
16
Navy
1
0
1
19b
Navy
2
1
2
22
Private
1
2
3
34
USFWSC
1
2
3
a One of the 5 main study sites.
b The Navy-owned Saddlebunch site.
c United States Fish and Wildlife Service.

Table 3.2Correlation coefficients between body mass, body, ear, and right foot length and rabbit age for three Lower Keys
marsh rabbits (Svlvilagus palustris hefneri). In parenthesis are the number of times each rabbit was examined.
Trait Female #1 (N=4) Female #2 (N=5) Male#l(N=5)
R2
F
P
R2
F
P
R2
F
P
Body mass
0.97
70.1
0.01
0.97
84.1
0.01
0.90
18.6
0.05
Body length
0.93
27.0
0.04
0.94
49.7
0.01
0.94
29.5
0.03
Ear length
0.80
8.2
0.10
0.96
71.8
0.01
0.91
19.4
0.05
Right foot length
0.82
8.9
0.10
0.93
41.1
0.01
0.92
22.6
0.04

46
Table 3.3-The number of sexually mature Lower Keys marsh rabbits
(Sylvilagus palustris hefneri) at the 5 main study sites on Boca Chica Key.
Session Site
7 8 9 10 13 Total
1 Male 0
Female 0
2 Male 0
Female 0
3 Male 2
Female 2
4 Male 2
Female 2
5 Male 2
Female 1
6 Male 1
Female 1
7 Male 1
Female 0
8 Male 1
Female 1
0 0 112
1112 5
10 113
1112 5
2 1117
1112 7
11116
2 1117
13 12 9
110 14
2 1116
2 2 0 1 6
11115
110 13
0 1114
0 2 0 2 5

47
Table 3.4~Allele frequencies for all variable locia among 5 marsh rabbit
populations (Sylvilagus palustris hefneri) from the Lower Keys of Florida, June
1992 May 1993. Individuals were trapped on sites 7, 8 and 9 from Boca
Chica, sites 10 and 13 from Geiger Key (the southeastern portion of Boca
Chica), site 30 on Saddlebunch, site 33 on Sugarloaf, and site 53 on Big Pine.
Site
INI
Locus
Allele
Boca Chica Geiger
Saddlebunch Sugarloaf
Big Pine
(9) (5)
(2) (2)
(1)
EST-1
A
0.389
0.200
1.000
0.250
B
0.611
0.800
0.750
C
1.000
PGM-1
A
0.778
0.500
1.000
0.500
1.000
B
0.222
0.500
0.500
aThe following loci were assayed but were monomorphic: GOT, GPI, Hb, ICD-1,
LDH-1, MDH-1, MPI, PEP-1, SOD-1.

Table 3.5--The sum of all marsh rabbits (Sylvilagus palustris hefneri) trapped at the 5 main sites on Boca Chica during the 8
trapping sessions (June 1991 to May 1993), and the proportion of these rabbits that were trapped twice during a 10-day
trapping session. The G2 for the number indicates a test to determine if the sex ratio of the trapped rabbits differed from
1:1. TheG2 for the trappability indiciates a test to determine if the trappability differed between sexes.
Age
Male
Female
Male:Female
Number
Trappability
Number
Trappability
Number G^
Trappability G^
Juvenile
8
50%
6
67%
1.0
11.6*
OO
Subadult
14
71%
4
25%
14.3*
59.6**
Adult
42
83%
42
88%
0.0
0.6
* P < 0.05
** P <0.001

Figure 3.1--Five main habitat patches on Boca Chica used for trapping and radio-collaring of Lower Keys marsh rabbits
(Sylvilagus palustris hefneri). The solid lines are runways and taxiways. The dashed line are roads.
A

Month
Figure 3.2Growth curves for 3 (1 male and 2 female) Lower Keys marsh rabbits (Svlvilagus palustris hefneri).

Number of Rabbits Caught
51
May 1991 May 1992
Session
Figure 3.3-The number of Lower Keys marsh rabbits Svlvilagus palustris
hefheri) caught during each trap session at the 5 main sites.

52
Month
Figure 3.4-Proportion (bar) and number (above bar) of Lower Keys marsh
rabbits (Sylvilagus palustris hefneri) that produced a litter during each month.
Data from 2 years were combined and averaged. (January = month 1,
December month 12).

Juv. Sub. Adult-1
Age Class
Adult-2
Figure 3.5-Percent mortality of Lower Keys marsh rabbits (Svlvilaeus
palustris hefneri) by sex and age class. Sample size of each age/sex cohort
indicated above each bar.

54
(/)
0
ro
n
o
0
_Q
E
3
Vehicle Cat Cat/Rac Snake Rac Poach
Cause of Mortality
Figure 3.6Cause of death for 24 Lower Keys marsh rabbits (Sylvilagus
palustris hefneri). (Vehicle car, truck, or plane; Cat = domestic cat; Cat/Rac
= unable to distinguish if predator was a cat or a raccoon; Snake = eastern
diamond back rattlesnake; Rac = raccoon; Poach = death by gun)

CHAPTER 4
SPATIAL ORGANIZATION (INTER-PATCH MOVEMENTS)
Introduction
To apply knowledge about the Lower Keys marsh rabbits population
biology in a PVA, it is necessary to define the unit of individuals that
comprises a local population. A local population can be loosely defined as a
set of individuals that has a high probability of interaction (Hanski and Gilpin
1991). Local populations may be part of a greater metapopulation where
interaction occurs via individuals moving among populations (Levins 1970).
The unit of the local population can be determined by studying the
spatial structure and movements of individuals at different spatial scales. An
individuals home range, spacing behavior, and movements (including
dispersal) will determine the spatial scale of interaction with others. These
behaviors may vary among individuals of different sexes and ages.
The home range of an individual is the area used for foraging, mating
and caring for young (Burt 1943, White and Garrott 1990). It does not include
areas used in migrations, or other occasional round-trip excursions. Home
range is calculated using a variety of methods that take periodic readings on an
individuals location.
55

56
Home-range size varies greatly among individuals, populations and
species of the genus Svlvilagus (see Chapman et al. 1982); Only 1 study has
attempted to estimate home-range size for the marsh rabbit (S. palustris). Blair
(1936) estimated a linear home range based on a maximum home range
width of 183 m from trapping data.
It is generally assumed that most cottontails are not territorial (Chapman
et al. 1982). This assumption is based largely on studies of the eastern
cottontail (S. floridanus; Haugen 1942; Chapman and Trethewey 1972, Dixon
et al. 1981) and the brush rabbit (S. bachmani; Shields 1960, Chapman 1971).
Yet, Trent and Rongstad (1974) reported that female eastern cottontails did not
have overlapping home ranges during the breeding season. Jurewicz et al.
(1981) found small amounts of overlap in the nocturnal home ranges of
breeding female eastern cottontails, and suggested that a spacing mechanism
was in operation. Male swamp rabbits (S. aquaticus) have exhibited linear
dominance hierarchies involving non-overlapping, defended home ranges in
several studies (Marsden and Holler 1964, Sorenson et al. 1968, Holler and
Sorenson 1969). Additionally, Kjolhaug and Woolf (1988) found no overlap
between individuals of the same sex in both male and female swamp rabbits.
Information on the spacing behavior of the other 10 species of cottontails
(genus Svlvilagus) is lacking.

57
In addition to home-range size and spacing, it is also important to
determine which habitat features a rabbit incorporates into its home range and
which features act as barriers. Rabbits generally move while under vegetative
cover (Chapman 1971, Trent and Rongstad 1974), probably to avoid predation.
Large open areas are generally seen as barriers or at least deterrents to rabbit
movements. Although the response of cottontail rabbits to roads and vehicular
traffic has not been directly studied, field studies have found that improved
roads (2-lane and larger) can be a barrier to small mammal and hare
movements (Oxley et al. 1974, Mader 1984). Other possible barriers to rabbit
movements include disturbed or developed areas with no natural vegetation,
vegetated areas with insufficient cover, and open bodies of water. The marsh
rabbit and the swamp rabbit are both known to be good swimmers (Svihla
1929, Tomkins 1935), although the frequency of swimming is not known.
While most rabbits stay within their home range and make few long
distance movements, young, subadult male rabbits have been documented
traveling longer distances (Shields 1960, Chapman and Trethewey 1972). It is
believed that these males are natal dispersers. Natal dispersal is male-biased in
most polygynous mammals (Dobson and Jones 1985). Determining which sex
disperses in a species and what factors motivate these movements are essential
to understand a species spatial organization.

58
Currently there is debate over the causes of natal dispersal (Dobson and
Jones 1985, Pusey 1987, Wolff et al. 1991). Most research has been directed
on determining the costs and benefits of dispersal to the dispersing individual.
Three main hypothesis have been proposed: inbreeding avoidance, competition
for mates, and competition for environmental resources (Packer 1979,
Greenwood 1980, Dobson 1982, Greenwood and Harvey 1982). Different
predictions for each hypothesis have been made, but research on a diverse
sample of mammalian species has not been able to falsify any of the
hypotheses. Two of the hypotheses, inbreeding avoidance and competition for
mates are not mutually exclusive in their predictions. It has been proposed that
the causal agents for dispersal may differ among species (Dobson and Jones
1985).
Recently, an alternative hypothesis has been proposed. Anderson
(1989) has suggested that the focus should be on the fitness the resident (non
dispersing) adults receive by forcing the young to leave. In most polygynous
mammals generally the father would receive the greatest benefit by coercing
his son to leave. Although some evidence of dominants evicting subordinates
exists in the literature (Moore and Ali 1984), other hypotheses can not be
eliminated (Pusey 1987, Wolff et al. 1991).
Determining what direction and what type of habitat dispersing
individuals will traverse are especially important in fragmented environments.

59
Two general theories have been proposed to predict the direct an individual
will leave a patch. The random walk theory (Berg 1983) predicts that
individuals will disperse across random locations along the edge of a patch,
irrespective of habitat features. Patch geometry models (Stamps et al. 1987,
Buechner 1989) predict that individuals will exit patches at more permeable
areas. A landscape corridor may increase the patch edge permeability by
extending patch habitat (La Polla and Barrett 1993), and allowing individuals
to move from one habitat patch to another. The geometric and habitat features
that constitute a corridor from the animals perspective must be determined.
The objective of this part of the study was to use the information about
the spatial organization of S._g. hefneri and movements to determine how they
use the fragmented habitat in the Keys. Evidence supporting the hypothesized
patterns of spatial use will be evaluated. Three possible scenarios of habitat
use will be considered:
1. The Lower Keys marsh rabbit is confined to one habitat patch and is
incapable of moving between patches. All home ranges occur strictly within
the patch. Dispersal is within the patch, does not occur, or is unsuccessful
(rabbit leaves a patch but does not reach another patch and dies or fails to
reproduce). This will be called a relictual (Berry 1986) population.
2. Lower Keys marsh rabbits spend most of their lifetimes in a patch
but are capable of moving between patches. All home ranges occur strictly

60
within the patch and individuals only interact with others living at the same
patch. There may be occasional movements between patches and/or some of
the dispersing individuals will successfully move to new patches. The areas
surrounding the patches act as a barrier to most movements and is not included
in the home range of the individuals. These patches contain sub or local
populations that exist within a greater metapopulation (Levins 1970). These
patches are not necessarily genetically distinct but are demographically
isolated. Matings between individuals at different patches does not occur.
3. Lower Keys marsh rabbits move regularly between patches and may
use several patches at once. The home ranges encompass several patches at
one period of time and individuals interact with others from other patches
regularly. All of the patches are part of one large population, which inhabits a
highly fragmented environment.
Methods
Marsh rabbits were trapped at the 5 main sites on Boca Chica. Trapping
occurred twice during the wet season (May October), and twice during the
dry season (November April) from May 1991 to May 1993 (8 trapping
sessions). Trapping grids were placed on each site, using unbaited collapsible
National live traps (80 x 30 x 30 cm), placed in a 6 x 6 array, spaced
approximately 25 m apart. Each trapping session consisted of 5 nights where

61
the traps were open, 2 nights with the traps closed and another 5 nights with
traps open. Traps were checked twice daily, once in the morning and once in
the evening and were covered in burlap for shade. All rabbits caught were
sexed, weighed, and tagged (Monel no. 3, National Band and Tag, Newport,
KY).
Radio-telemetry
Each rabbit weighing >1,000 grams was fitted with a radio-collar and a
transmitter with an estimated 10-month operational life (Telonics, Inc., Mesa,
AZ). Smaller rabbits (300-1,000 g) were fitted with a similar radio-collar, but
with a velcro break-away device added to allow the animals to lose the
equipment as their bodies grew larger. Collars were replaced as the animals
aged.
Collared rabbits were located on separate days 3 times a week, once in
the early morning (7-9 a.m.), once at mid-day (11 a.m. 1 p.m.), and once in
the evening (4-6 p.m.). Signals were followed until the animal could be seen
or the exact location was found. All locations were made >24 hours apart to
ensure independence of observations. Locations were plotted on 1:2400 aerial
maps with 10-m2 grid overlays. All road and water crossings were recorded
and the distance of the water crossing was measured.

62
Home Range
All juvenile rabbits that were followed for more than 1 month (12
locations) were used for analysis of home ranges. The 12-location criterion
was chosen because few juveniles were followed for more than 1 month.
Attempts to determine a minimum number of locations for the adult home-
range analysis by fitting the data to a negative exponential equation failed.
None of the individuals had home-range sizes that reached an asymptote,
despite the fact that some rabbits were followed their entire lives (Figures 4.1
and 4.2). Therefore, a minimum number of 30 locations (2.5 months) was
chosen based on other studies of similar cottontail rabbits (Dixon and Chapman
1980; Kjolhaug and Woolf 1988). Computer program HOME RANGE
(Ackerman et al. 1990) was used to determine sizes of minimum home ranges
and to plot movements.
Because of the nature of Lower Keys topography and patterns of home
ranges of marsh rabbits, several common methods of home-range estimation
were excluded. The minimum convex polygon method (Hayne 1949) was
rejected because it assumes that all areas within the perimeter of the outermost
locations are used by the animal (White and Garrott 1990). Most of the marsh
rabbits' home ranges contain unused areas, such as bodies of water, roads, or
other man-made structures. One of the most statistically rigorous methods of
calculating home-range size is the 95% probability ellipse (Jennrich and Turner

63
1969). It assumes that there is always only one center of activity (Harris et al.
1990), an assumption that was frequently violated in this study.
Two related methods that have worked well for analysis of rabbit home
ranges are the harmonic-mean and core-area methods (Dixon and Chapman
1980). The harmonic mean is a nonparametric method based on a volume
under a fitted 3-dimensional use distribution. It relates well to the actual
distribution of locations. The core area (as calculated by program HOME
RANGE), is the maximum area where the observed use distribution exceeds a
uniform use distribution. Thus, it shows areas of particularly high home-range
use and is relatively unaffected by outliers and sample size (Harris et al. 1990).
Seasonal differences in the home range size (95% harmonic mean and
core area) were compared for each individual that was located at least 30 times
in each season. The dry season locations occurred between November and
April; the wet season was from May to October. Shifts in home range between
season were also examined. Arithmetic centers of the dry and wet season
home range of each individual were determined and the distance these centers
shifted between seasons was calculated.
Spacing Behavior
The amount of overlap between the 95% harmonic mean and core areas
of each individual was compared to other individuals living in the same habitat

64
patch. The first set of comparisons looked at all of the individuals of the same
sex present at a patch during the same period of time. Percent overlap was
determined by plotting all the home range contours together using Program
HOME RANGE, and overlapping a 10 m grid. For each pair of individuals the
amount of overlap was calculated as a percentage of the smaller home range.
This percent of overlap was compared to overlaps of same-sexed
individuals living at the same patch but at different times, and pairs of opposite
sexed individuals. A t-test was used for normally distributed data and when
the variances between the 2 groups were equal. A Mann-Whitney U test was
used when these conditions were not met.
Dispersal
If an animal appeared to be making a long-distance movement, it was
located at least once every day. These locations were not used in any home
range calculations. A rabbit that made a one-way long-distance movement was
said to have dispersed. The criterion for dispersal as a movement was the
diameter of the average marsh rabbit home range (Ribble 1992).
Dispersing animals were classified by sex and age class. Locations
from the daily radio-telemetry session were plotted and straight lines were
drawn between them to approximate the minimum distance traveled. Data on
the demography of dispersers was used to test hypotheses on the cause of

65
dispersal. A hierarchical hypothesis design was used (Table 4.1) to attempt to
narrow the potential cause of dispersal.
First, a binomial probability test was used to determine if the sex ratio
of the dispersers was significantly different from unity. If this sex-ratio
showed a significant bias, then trapping and radio-telemetry data were
examined to determine if other adult males inhabited the dispersers natal
patch. Trapping records of the adult males were further examined to determine
if these males could potentially be the father of the dispersers. Potential
fatherhood was assumed if the adult male was present at the dispersers natal
patch around the time when the dispersers were believed to have been
conceived.
Corridor Use
To determine if dispersing individuals were randomly leaving a patch or
if they were influenced by potential corridors, paths of all individuals making
one-way long distance movements were plotted. The proportion of each
movers trip (measured in meters) that covered different habitat types and the
number of roads, runways, and bodies of water they crossed was measured.
Table 4.2 lists the habitat types and features compared.
Using this distance as a radius with the arithmetic centerpoint of the
natal home range as the center, a circle was drawn for each moving individual

66
(Figure 4.3). Aerial photographs were digitized (ARC/INFO 1990) to
determine the amount of each habitat type and length of all roads, runways, and
linear bodies of water was determined within this circle. The Johnson method
(Johnson 1980) was used to determine if habitat use significantly deviated from
the amount of each type of habitat available. A Waller-Duncan multiple
comparison procedure (Waller and Duncan 1969) was used to determine
differentially selected habitats. The Johnson method was chosen because it
looked for overall differences in habitat use, not at individual habitat-use
differences (Alldredge and Ratti 1992).
Results
A total of 54 Lower Keys marsh rabbits were caught and examined
during this study. Forty-three were caught on the 6 main grid sites (41 on Boca
Chica, 2 on Saddlebunch), 5 rabbits were caught while trying to recapture
collared dispersed rabbits, and 6 were caught on other sites in the Lower Keys.
Rabbits were only examined once per trapping period. Data from the 54
rabbits was recorded 130 times during eight trapping sessions (May 1991 -
May 1993).
Forty-three (28 male, 15 female) of the rabbits caught were fitted with
radio-collars, and 7 of these were juveniles. More juveniles were caught but at
the time of their capture the break-away collars were not available. Forty-one

67
of the radio-collared rabbits were on Boca Chica Key and 2 were on
Saddlebunch Key.
Home Range
Sufficient data were available for seven juveniles, 13 adult males, and
10 adult females for the home-range analysis (Table 4.3). Adult males were
followed an average of >8 months (98 locations) and adult females were
followed an average of >10 months (124 locations). Comparisons using the
Wilcoxon test for differences between adult males and females did not show
significant differences between the average distance moved between locations
(Z = -1.71, P = 0.09), the 95% harmonic-mean home range (Z = -1.33, P =
0.18), or the core area (Z = -1.21, P =0.23). Resident adult males and females
were similar with respect to their home-range shape and size (Figure 4.4).
Both sexes appeared to establish a permanent home range shortly after maturity
(9-10 months), and both resided in that area for the duration of their lives.
Combining both sexes, the average 95% harmonic mean was 3.96 ha (SE =
0.65) and the average core area was 1.21 ha (SE = 0.87). Home-range size
estimates for juvenile rabbits were not significantly different from adults, but 3
of the juveniles did not develop core areas. The lack of core area formation
may be an attribute of juvenile home ranges or an artifact of the low number of
locations used in the analysis.

68
There did not appear to be any significant changes in home-range size or
shifts between the dry and wet seasons (Table 4.4, Figure 4.5). Results of a
Mann-Whitney U test, used because the variances of the home range size were
not equal, failed to find a significant result comparing either the wet and dry
95% harmonic home range (T = 251, P = 0.45) or core areas (T=245.5, P
=0.60). None of the shifts in home range between the wet and dry seasons
exceeded the radius the marsh rabbits home range. However, ANOVAs
testing the effect of time on home-range size were significant for both the 95%
harmonic mean (F = 6.25, df = 22, P = 0.02) and the core area (F = 5.55, df =
22, P = 0.03). Animals increased their home-range size with time (and number
of locations), irrespective of season.
Spacing Behavior
Figures 4.6 and 4.7 show the duration of each rabbits radio-telemetry
records and which rabbits were contemporaneous. The average amount of
overlap between same-sexed individuals occupying a patch at the same time
was significantly less than same-sexed groups occupying a site at different
times (Table 4.5). This relationship held true using both the 95% harmonic
mean home-range estimate (t_= -2.17, df = 12, P = 0.05) and the core area
(t= -3.29, df = 12, P_= 0.006). When a rabbit died, same-sexed individuals
living in adjent areas expanded their home ranges (Figure 4.8). Similarly, the

69
overlap between these contemporary same-sex individuals was significantly
less than male/female overlap for both the 95% harmonic mean (T = 79.5, P =
0.05) and core area (T = 45, P = 0.02). In most cases there was little overlap
between adults of the same sex (Figure 4.9).
Most males home ranges overlapped 1 female at a time; 3 males
overlapped 2 females simultaneously. During a portion of their lives, some
males did not overlap with a female. Only 1 female overlapped with more than
1 male; this occurred for less than 1 month. Changes in overlap generally
occurred when a rabbit died and another rabbit expanded his/her home range.
Home-range Features
All 43 rabbits, including the 7 juveniles, were used in the road- and
water-crossing index. Only individuals that were currently in their natal or
adult home ranges were used. Data from individual making long-distance
movements were analyzed separately.
Home ranges of most of the marsh rabbits did not incorporate roads or
large bodies of water. No water or road crossings were recorded for any of the
juvenile rabbits. None of the rabbits crossed the major highway in the Keys
(US-1). Only 1 adult female and 1 adult male crossed paved roads during the
study. Rabbit A53F crossed a 2-lane several times. This same rabbit was the
only adult female to cross a body of water; she regularly swam in shallow

70
water between mangrove prop roots while being radio-tracked. Male A51M
also crossed a 2-lane road four times. Neither rabbit crossed the road to visit
another patch. Female A53F crossed the road to swim in the water adjacent
to her home patch; male A51M crossed the road when construction was
being completed in the comer of his home patch.
Dispersal
Seventeen rabbits (11 male, 6 female, all subadults) made permanent
one-way movements (Table 4.6). The minimum dispersal distance was
calculated using the diameter of the combined adult core-area size. Assuming
a circular home-range shape, the average diameter was 124 m.
Eleven of the subadults made movements >124 m. Ten of the dispersers
were male and one was female. Only 1 of the males failed to meet the criterion
for dispersal, and whereas 5 of the females did not exceed the distance. In
general, the males made long-distance movements far in excess of the criterion,
including a 3-day movement that placed a male over 2 km from his natal range.
Most of the females settled near their natal ranges, including the female that
was classified as a disperser.
The sex ratio of the dispersers was significantly male biased (binomial P
<0.001). Six of the males that dispersed left patches where there was another
adult male present, but 4 left patches where there were no adult males. Five of

71
the males that dispersed left patches were the male present could have been
their father, but the other 5 left patches were there either were no males
present, or a male too young to be their parent was present. The 1 subadult
male that did not disperse occupied a patch where there were no other adult
males.
Corridor Use
Subadult rabbits traveled through a variety of habitats between their
natal and permanent home ranges. Three rabbits crossed a dirt road, 7 crossed
a 2-lane road and 3 rabbits were observed crossing taxiways and runways.
None of the rabbits crossed the 4-lane highway (US 1), but none of the
dispersal radii encompassed or were adjacent to the highway. Two rabbits
swam across ditches, 1 across a canal, and 1 crossed a (12 m) body of water.
In general, most of the subadult rabbits traveled through areas with
dense ground cover. These marsh rabbits were recorded traveling through
mangroves, upland hardwood hammocks, and in the vegetation between the
shoulder of the road and the water. The narrowest strip of plant cover used
(corridor) by a dispersing marsh rabbit was 3-5 m wide. The Johnson test that
compared the amount of each habitat traveled through compared to its use was
significant (JF = 6.62, df = 3,14, P = 0.005). A Waller-Duncan comparison
found that rabbits used areas of mangrove, hardwood hammock, and transition

72
zone significantly more than expected and used disturbed areas significantly
less.
Discussion
The data suggest that members of S. p. hefneri spend most of their lives
in 1 patch, but can move to other patches. Marsh rabbits are bom in a patch of
transition-zone habitat and remain there until they reach sexual maturity. At
sexual maturity most rabbits make a relatively long, one-way movement. Male
marsh rabbits may move a great distance away from their natal range; females
are more likely to remain in the same patches where they were bom. If a patch
is relatively small, this movement may mean leaving the habitat patch. When
these subadult marsh rabbits leave their natal ranges, they establish adult home
ranges that they maintain for their lifetime. These adult home ranges do not
incorporate any roads (2-lane and larger) bodies of water, or other types of
habitat. All mating appears to occur within the same habitat patch. These
results are consistent with the predictions of the second hypothesis, that S. p.
hefneri exists in a metapopulation (Levins 1970, Hanski 1991).
The Lower Keys marsh rabbits ability to exist as a metapopulation and
disperse over relatively long distances may be a product of inhabiting a
naturally patchy environment (Opdam 1991, Thomas 1994). Species that exist
in temporally or spatially patchy environments must be able to deal with

73
isolation at a more regional scale than those species that occupy more
permanent, interconnected habitats. Marsh rabbits throughout their range
occupy upper marshes along the coasts and interior wetlands, much of which is
patchily distributed. Before European colonization in the Keys, the transition
zone was probably more contiguous. Most of the habitat types in the Keys
exist in contiguous, concentric rings; habitat type is largely determined by
elevation.
Spatial Organization
Marsh rabbit home-range size was well within the range for cottontail
species (Chapman et al. 1982). The 95% harmonic mean estimates represent a
conservative estimate of the amount of habitat used; the core area estimate
delineated the area needed for more intense use. Despite strong seasonally of
the rainfall in the Keys, home ranges were consistent throughout the year.
Similar to the eastern cottontail and the swamp rabbit, there is little
overlap in home ranges of members of the same sex. Marsh rabbits may be
territorial within their sex. Most current definitions of territoriality specify that
an area must be used exclusively (at least within the sex) and that it must be
actively defended (Eisenberg 1981, Begon et al. 1990). To be truly territorial,
marsh rabbits would have to defend their home ranges. Territorial defense is
difficult to observe in the field, especially in a secretive, crepuscular species

74
like the marsh rabbit. Some circumstantial evidence of physical defense was
apparent in the males; 76% of the subadult and adult males had scratch marks
and scars on their face, ears and back of their heads. Boxing, scratching, and
paw-raking with the feet has been observed in other species of cottontails
(Marsden and Holler 1964) and may be responsible for the scratch marks.
Generally these behaviors are related to competition for mates (Eisenberg
1981). Female marsh rabbits showed no signs of scratch marks. It is not
readily apparent why it would be advantageous for the females to be territorial,
however there is some evidence that nesting sites may be limited (see Natality -
chapter 3). Fecal pellet marking is well-developed in other rabbit species (Teft
and Chapman 1987) and may be used by male and female marsh rabbits to
mark the boundaries of their home ranges.
Both males and females increased their home ranges following the death
of a conspecific. Over the lifetime of a rabbit this was seen as an increase in
home range with age. This phenomenon provides additional support that some
spacing behavior is partially determining home-range size. This spacing
behavior may have an impact on population density and total abundance of
rabbits in a habitat patch.
Most of these home ranges consisted almost exclusively of transition-
zone habitat. Physical features such as canals, roads, and runways were rarely
crossed by any of the adult marsh rabbits, despite the proximity of the features.

75
The definition of a habitat patch put forth in the methods section appears to be
valid.
Subadult marsh rabbits were more likely to cross these barriers and use
alternate types of habitat. However, their movements did appear to be
influenced by the surrounding habitat matrix. Concordant with other studies
on rabbit movements (Chapman 1971, Trent and Rongstad 1974), marsh
rabbits were more likely to cross the more densely vegetated native habitats
(transition zone, hardwood hammock, and mangrove) than the more open,
disturbed areas. Presence of these habitats around the natal patch appeared to
facilitate movement between patches of transition zone habitat, acting as
corridors in the highly fragmented landscape (Wilson and Willis 1975). These
results support the patch geometry models (Buechner 1987, Stamps et al. 1987)
rather than the random walk theory (Berg 1983).
Dispersal has not been well studied in other species of cottontails.
Results from the telemetry indicate that most long-distance dispersers were
young males, similar to most polygynous mammalian species (Dobson 1982).
The small patch size and low density of rabbits in this study provided a unique
opportunity to investigate the cause of dispersal. Because some of the males
dispersed despite the lack of other adult males at their natal patch, the
competition-for-mates hypothesis cannot entirely explain all of the movements.
Similarly, lack of potential fathers at most of the natal patches excludes the

76
resident fitness hypothesis as being the sole explanation. Only the inbreeding
avoidance hypothesis is consistent with all of the data collected. It is possible
that dispersal may be motivated by several factors (Dobson and Jones 1985)
and that there may be variance between individuals. This interpretation is
based only on current dispersal patterns. Current dispersal behavior may be a
response to population structure in evolutionary time. The marsh rabbit
population structure has probably undergone dramatic changes since the
colonization of man in the Keys.
Conclusions
These results imply that the S. p. hefneri should be managed as a
metapopulation. Each local population is socially isolated from the other
populations. Interchange generally occurs by movement of subadult males and
this movement is facilitated by the occurrence of habitat corridors. The impact
these conclusions has on the persistence of S._g. hefneri will be addressed in
chapter 6.

Table 4.1--Hierarchical tests and predictions of the proximate cause of dispersal.
Prediction
Cause of Dispersal
I. An equal number of males and females disperse
Competition for resources
II. Only one sex disperses.
A. The dispersers are all female.
Inbreeding avoidance
B. The dispersers are all male.
1. A male disperses only if another adult male is present
at the habitat patch.
Competition for mates
2. A male disperses only if other males are present and
could potentially be their father.
Resident fitness hypothesis
3. All males disperse regardless of the presence or
absence of other males.
Inbreeding avoidance

78
Table 4.2Habitat types and possible physical barriers encompassed by Lower
Keys marsh rabbit (Svlvilagus palustris hefneri) dispersal movements.
Feature
Mode of measurement
Unimproved road (dirt, gravel)
meters of length
2-land road
meters of length
4-land highway
meters of length
Mosquito drainage ditch (water < lm)
meters of length
Canal (1 10 m)
meters of length
Major body of water (water > 10m)
meters of length
Disturbed or barren habitat
ha
Mangrove
ha
Hardwood hammock
ha
Pineland
ha

79
Table 4.3--Home-range data on Lower Keys marsh rabbits (Svlvilagus palustris
hefherij observed for >1 month for juveniles, >30 locations for adults.
Rabbit
Classification
Number of
locations
Distance
between
consecutive
locations (m)
95%
harmonic
mean home
range (ha)
Core
area
(ha)
J55M
Juvenile
30
69.58
13.83
4.44
J99M
Juvenile
18
47.40
2.49
0.64
J101F
Juvenile
16
22.99
0.10
J102F
Juvenile
21
25.81
1.11
0.38
J103M
Juvenile
14
22.39
0.30
J199M
Juvenile
14
40.60
0.81
X
38.13
3.11
1.82
SE
(7.55)
(2.17)
(1.31)
A51M
Adult male
122
48.34
11.94
3.15
A55M
Adult male
70
30.81
2.73
0.89
A59M
Adult male
79
28.71
2.13
0.93
A63M
Adult male
217
25.33
5.81
1.58
A64M
Adult male
61
17.87
0.36
0.02
A67M
Adult male
99
42.10
3.47
1.10
A68M
Adult male
142
35.30
3.63
1.24
A69M
Adult male
138
50.88
7.61
2.60
A70M
Adult male
88
36.85
2.99
0.86
A71M
Adult male
67
37.47
3.18
1.22
A73M
Adult male
60
38.24
4.23
1.20
A75M
Adult male
74
28.07
4.35
1.18
A85M
Adult male
54
52.36
2.81
1.03
X
36.33
4.25
1.31
SE
(2.85)
(0.80)
(0.22)
A50F
Adult female
252
28.14
6.52
2.00
A52F
Adult female
127
34.31
10.32
3.17
A53F
Adult female
201
59.76
8.60
2.09
A57F
Adult female
220
21.90
1.67
0.53
A58F
Adult female
90
20.90
0.72
0.24
A72F
Adult female
107
26.15
2.49
1.02
A74F
Adult female
75
27.89
1.27
0.46
A76F
Adult female
81
22.12
1.53
0.53
A86F
Adult female
44
32.68
1.72
0.53
A176F
Adult female
45
30.52
0.98
0.25
X
30.44
3.58
1.08
SE
(3.56)
(1.12)
(0.32)

Table 4.4A comparison between the wet and dry season home ranges (ha.) of Lower Keys marsh rabbits (Svlvilagus palustris hefheri).
The distance between the centers of each home range is compared to the radius of the 95% harmonic mean and core area estimates.
Wet season Dry season
ID
^locals
95%
harmonic
mean
Core
area
#locals
95%
harmonic
mean
Core
area
95%
harmonic
radius
Core
radius
Meters
moved
Core radius -
meters moved
A70M
36
0.81
0.24
52
2.79
0.83
50.79
27.65
15.52
12.13
A71M
36
2.99
0.90
31
1.31
0.54
97.58
53.54
24.50
29.04
A68M
64
3.79
0.90
78
1.37
0.47
109.86
53.54
38.11
15.43
A67M
45
4.03
0.96
54
0.56
0.14
113.29
55.29
37.00
18.29
A63M
124
2.28
0.57
93
5.74
1.60
85.21
42.61
14.14
28.47
A69M
60
6.31
2.04
78
3.61
1.42
141.76
80.60
13.01
67.59
A59M
43
1.95
0.87
36
1.48
0.46
78.80
52.64
45.45
7.19
A55M
36
2.16
0.80
34
0.91
0.40
82.94
50.48
24.02
26.46
A51M
73
0.53
0.33
49
0.45
0.28
41.08
32.42
5.00
27.42
A72F
36
1.15
0.37
71
3.26
1.23
60.52
34.33
39.11
-4.78
A53F
81
10.18
2.51
120
6.50
1.72
180.06
89.41
45.17
44.24
A50F
132
3.49
1.04
120
5.60
1.75
105.43
57.55
12.17
45.38
A58F
48
1.46
0.47
42
0.43
0.20
68.19
38.69
28.07
10.62
A57F
106
1.38
0.45
114
0.78
0.27
66.29
37.86
9.84
28.02
A52F
78
6.21
2.21
49
7.15
2.31
140.63
83.89
4.24
79.65

Table 4.5~The amount of overlap between same and opposite sexed individuals occuping a site during the same time and
between same-sexed individuals that occupied the same site during different times.
Same sex-same time
Pair 95% Core area
Same sex-different time
Pair 95% har. Core area
Opposite sex-same time
Pair 95% har. Core area
A59M/A55M
34%
12%
A51M/A69M
100%
43%
A51M/A52F
100%
67%
A63M/A67M
30%
3%
A55M/A75M
82%
44%
A59M/A57F
68%
40%
A70M/A71M
85%
42%
A59M/A55M
45%
20%
A59M/A58F
45%
22%
A50F/A74F
9%
2%
A59M/A75M
91%
72%
A60M/A53F
44%
22%
A53F/A76F
100%
21%
A53F/A76F
100%
100%
A63M/A50F
54%
32%
A57F/A58F
16%
0%
A57F/A86F
71%
74%
A64M/A53F
8%
0%
A72F/A65F
0%
0%
A58F/A86F
45%
20%
A67M/A50F
100%
100%
A67M/A74F
85%
21%
A68M/A76F
100%
90%
A70M/A65F
8%
2%
A70M/A72F
82%
48%
A71M/A72F
90%
84%
A75M/A57F
100%
100%
A75M/A86F
100%
86%
Average
39%
11%
76%
53%
70%
51%

82
Table 4.6Data on radio-collared marsh rabbits that made permanent, one-way
movements. Rabbits were caught on Boca Chica, Geiger, and Saddlebunch
Keys between June 1991 to May 1993.
Rabbit
Classification
Distance between natal
site and last location (m)
Body mass at beginning
of movment (g)
A48M
Subadult male
550
1,000
A49M
Subadult male
550
1,050
A54M
Subadult male
920
1,050
A55M
Subadult male
1,100
800
A56M
Subadult male
1,800
1,000
A60M
Subadult male
510
1,000
A66M
Subadult male
60
800
A68M
Subaudit male
180
1,000
A69M
Subadult male
980
800
A84M
Subadult male
2,050
1,050
A197M
Subadult male
400
900
X
827
950
SE
191
32
A3F
Subadult female
60
850
A65F
Subadult female
40
900
A74F
Subadult female
70
950
A76F
Subadult female
80
900
A86F
Subadult female
90
1,000
A176F
Subadult female
150
900
X
82
917
SE
15
21

Core Area (ha)
83
Figure 4.1The cumulative core area measurements of 8 radio-collared male
Lower Keys marsh rabbits fSylvilagus palustris hefneriV

Core Area (ha)
84
Figure 4.2The cumulative core area measurements of 6 radio-collared female
Lower Keys marsh rabbits (Sylvilagus palustris hefneri).

oo
Figure 4.3To determine if a dispersing Lower Keys marsh rabbit (Sylvilagus palustris hefneri) used certain habitat as a
corridor, the habitat type the rabbit moved through was compared to the habitat available. Using the centerpoint of the
rabbits natal home range and the distance dispersed as a radius, a circle of available habitat was drawn. The arrow
indicates the actual patch the rabbit took.

86
^'A70M
Figure 4.4Home ranges using the 95% harmonic mean (outer boundary) and
core areas (inner boundary) of a sympatric male and female Lower Keys marsh
rabbit (Sylvilagus palustris hefneri) at site #7 on Boca Chica Key.

87
Figure 4.5--A comparison of the wet and dry season home ranges using the
95% harmonic mean (outer boundary) and the core area (inner boundary) for
an adult male at site 10.

88
A70M
A72F
A71M
A65F
A182M
A185F
1 2 3 4 5 6 7 8
A50F
A63M
A67M
A74F
A52F
A51M
A69M
Site 8
A
A
a
a a
A
a
A
A
A
~T~
1
1
2
r~
3
l I
4 5
~r~
6
7
~T~
8
Site 10












I
1
~T~
2
i
3
i i
4 5
I
6
l
7
I
8
Trap Session
Site 7



i i i i r
Figure 4.6The duration that radio-collared Lower Keys marsh rabbits
(Sylvilagus palustris hemeri) were followed at sites #7, #8, and #10.

89
Site 9
A cop
a a a a a a
MOOr
A60M -

A64M -

A66M -

A68M -

A76F -

A176F -

i i i i i i t r~
1 2 3 4 5 6 7 8
Site 13
A57F -
A58F -

A59M -

A55M -

A75M -

A86F -

A186M -

T
1 2 3 4 5 6 7 8
Trap Session
Figure 4.7 The duration that radio-collared Lower Keys marsh rabbits
(Sylvilagus palustris hemeril were followed at sites #9 and #13.

90
A. Both males alive (T = 3-6).
Figure 4.8The home ranges using core area, of 2 sympatric males telemetered
between session 3 and 6 (A.) and the expansion of A63Ms core area after the
death of A67M.

91
Figure 4.9The core areas of 4 sympatric Lower Keys marsh rabbits
(Sylvilagus palustris hefneri). Rabbits A55M and A59M are adult males,
A57F and A58F are adult females.

CHAPTER 5
METAPOPULATION DYNAMICS: PATCH OCCUPANCY AND
HABITAT QUALITY
Introduction
In Levins (1969) metapopulation model, all habitat patches were
identical in size and quality. More recent models (Taylor 1991, Thomas et al.
1992, Hanski 1994, Hanski et al. 1994), have taken an incidence-function
approach (Diamond 1975) that predicts the probability of a species occurrence
based on the area of the habitat and the distance of the habitat patch to other
patches. In these models, all patches are assumed to be inhabitable past a
threshold size or minimum isolation (Hanski 1994). For species whose natural
history is relatively unknown, an important aspect of a metapopulation
occupancy model is determining if unoccupied habitats are vacant due to lack
of suitable habitat or from past extinctions unrelated to the habitat patch
quality.
Metapopulation Structure
Before studying habitat use, metapopulation structure should be studied
to determine how the subpopulations interact with each other. In a classical
92

93
metapopulation all patches greater than a minimum area and less than a
maximum isolation are suitable for supporting a population. Local extinctions
(due to deterministic or stochastic causes) are counteracted by colonizations
from near-by patches (Hanski and Gilpin 1991). Some classic metapopulations
may resemble the stepping-stone model of occupancy (Gilpin 1980). In this
model, the probability of occupancy increases as patch size increases and
distance to the nearest occupied patch decreases as a dynamic consequence of
local extinction and colonization.
In a mainland-island metapopulation, 1 or a few large patches are
occupied continuously while surrounding small patches frequently experience
extinction but are recolonized due to their proximity to a mainland patch
(Harrison 1991, 1994). Only patch size determines which patches will be
mainlands and which patches are islands. Larger patches (and hence larger
populations) are able to maintain occupancy despite population variability and
thereby act as mainlands from which individuals emigrate to the smaller
islands.
Although similar to mainland-island metapopulations, source-sink
metapopulations differ because habitat quality explains the patterns of
occurrence. In a source-sink metapopulation the higher quality source habitat
patch produces a surplus of descendants (X > 1) whereas the low-quality
patches may be sinks that produce a deficit (X < 1) (Pulliam 1988, Pulliam and

94
Danielson 1991). Individuals from source populations migrate to sink
populations, giving the false appearance that these sink habitats are suitable for
the species. Habitat sources and sinks should be analyzed separately for
information on habitat requirements.
Habitat Use
An animal selects a habitat based on proximate cues from the
environment, but the decision will ultimately affect the individuals fitness in
terms of survival and reproduction (Lack 1933, Levins 1969, Cody 1985). For
many species, habitat selection is difficult to determine and habitat use is the
best available information. Assessment of habitat use is based on correlation
of the actual distribution of individuals with specific habitat features. It may
be inferred that habitat use is the outcome of habitat selection (Bergin 1992).
Habitat use may be studied at several spatial and temporal scales. For a
small-to-medium sized herbivore, the smallest scale may involve the leaf
browsed (diet). At the next level habitat use includes the areas used for
foraging and nesting within a habitat patch (microhabitat). At the largest
scales, the features that influence the presence or absence of a species over
time at a particular habitat may be studied (macrohabitat).
The first objective of this chapter is to determine the relationship among
habitat patches within the Lower Keys marsh rabbit metapopulation. The

95
spatial arrangement, size of the patches, and patterns of occupancy will be
examined to determine if the metapopulation is classical, mainland-island, or
source-sink in structure.
The second objective of this chapter is to elucidate factors influencing
habitat use by the Lower Keys marsh rabbit at 3 inter-dependent spatio-
temporal scales. This habitat analysis will be used to determine if currently
unoccupied habitat patches are suitable and empty because of their isolation
from other patches or if the habitat is unsuitable.
Methods
Fecal pellets were used to study diet, habitat use, and the presence/
absence of marsh rabbits. Presence of fecal pellets can be an accurate and
efficient method of studying habitat use and patch occupancy when
information about pellet persistence is estimated (Simonetti and Fuentes 1982,
Pietz and Tester 1983). It is a particularly useful method for species that are
secretive, crepuscular, at low densities, and not easily trappable (Wood 1988).
Pellet Grids
Permanent fecal pellet-sampling grids were established at each of 59
patches of transition-zone habitat in the Lower Keys. The grid was designed to
fit into the smallest patch and consisted of a square of 7 x 7 stations at 15-m

96
intervals, marked with permanent flags. The grids were surveyed 3 times per
year: March (late dry season), July (mid-wet season) and November (transition
between wet and dry) from March 1991 to July 1993. Each survey consisted
of a pellet removal within a radius of 0.5 m at each station followed by a pellet
census 1 month later (a marker had pellets if 1 pellet fell within 0.5 m). A grid
was considered to be occupied (marsh rabbit present) if at least 1 of the stations
had a marsh rabbit pellet.
To ensure that fecal pellets did not begin to degrade in less than a
month, during each survey 100 of the pellets from captured rabbits were placed
on a transition-zone grid. The pellets were separated into 4 groups and placed
on rocks, on mud, in grass, and under trees. The pellets were counted weekly
to determine the rate of decomposition.
Pellet Size
To determine if the pellets produced on a grid were from juvenile or
adult marsh rabbits, a linear regression was used to determine if body mass
accurately determines pellet size. Body mass was obtained by trapping
individual marsh rabbits at 6 sites in the Lower Keys, including the 5 main
sites on Boca Chica (Figure 3.1) and 1 site on Saddlebunch. Trapping occurred
twice during the wet season (June November), and twice during the dry
season (December May). For the 5 sites on Boca Chica Key, trapping

97
occurred from June 1991 to May 1993 (8 trapping sessions). On Saddlebunch,
trapping was conducted from June 1992 to May 1993 (4 trapping sessions).
Trapping grids were placed on each site, using unbaited collapsible National
live traps (80 x 30 x 30 cm), placed in a 6 x 6 array, spaced approximately 25
m apart. Each trapping session consisted of 5 nights where the traps were
open, 2 nights with the traps closed and another 5 nights with traps open.
Traps were checked twice daily, once in the morning and once in the evening
and were covered in burlap for shade.
All rabbits caught were sexed, weighed, and tagged (Monel no. 3,
National Band and Tag, Newport, KY). For the purposes of this part of the
study, adult marsh rabbits were those that weighed >1000 g, smaller rabbits
were classified as juveniles. Ten fecal pellets were collected in a pan
underneath each trap and the width and length of each pellet was recorded.
The average area (width x length) was used in the regression. Each rabbit was
used only once in the analysis.
Metapopulation Structure
Sites were classified based on site occupancy; if a site was never
occupied during the past 2 years it was classified as being empty, if it was
occupied at least once it was called variable and if it was consistently
occupied it was referred to as an occupied site. Sites were graphed using the

98
common log of area and isolation (distance to nearest occupied patch), and the
result was compared to theoretical patterns for classic, mainland-island and
source-sink (Figure 5.1).
Occupied and variable patches were further compared to determine if
they were sources and sinks. Liberal definitions of source and sink were used;
a source was any patch where reproduction occurred when occupied, and a
sink was any patch where only adult rabbits were present. Presence of adults
and juveniles was determined by measuring the size of the pellets found on the
grid. Because it was previously determined (in Chapter 4) that juveniles
remain in their natal patch until they become subadults, it was assumed that at
least 1 juvenile pellet present on the grid indicated presence of reproduction
occurring on that grid. For each habitat patch the number of surveys where
juvenile pellets were found was compared to the number of surveys occupied.
Dietary Analysis
Pellets for dietary analysis were collected from the pans beneath the
rabbit traps from the 5 sites on Boca Chica. To avoid experimental bias,
pellets were collected only from adult animals 1 time. At each of the 5 Boca
Chica sites, pellet samples were collected from eight rabbits, 2 from each sex
during both the wet and dry seasons, totaling 40 pellet-group samples.

99
Fecal pellets were dried and crumbled to prepare microscope slides
following standard procedures (Johnson et al. 1983). One hundred microscope
fields (20/slide) were examined for each sample. Plant species were identified
by comparing epidermal structures from the samples with a reference collection
of plant tissues collected in the area. Average relative densities were
calculated for each food item per pellet-group sample.
Relative densities of vegetation in the pellets may not be directly
comparable to the amount of each vegetation ingested. Variation in the
digestion and retention rate of each type of vegetation can bias the results
(Wallage-Drees et al. 1986, Westoby et al. 1976, Batzli and Pitelka 1971).
These biases can be quantified by comparison of fecal and stomach samples,
but this is impractical in many studies (Johnson and Pearson 1981), especially
those involving endangered species.
To avoid these problems, and still obtain data on the differences
between available vegetation and vegetation in the rabbit's diet, we compared
the species available and the species used. We assumed that although different
types of vegetation vary in their digestibility, most vegetations would still be
present in the fecal pellet sample if they had been ingested. Available
vegetation (ground cover <1.5 m) at the 5 sites was estimated using the line
intercept method (Canfield 1941). Ten 5-m long transects were randomly

100
located at each site and percent occlusion was determined for each plant
species.
The relative densities are comparable between samples, because the
digestion and retention rates are fairly constant for each plant species
(Wallage-Drees et al. 1986). Univariate analyses of variance (PROC GLM,
SAS Institute, Inc. 1988) were performed on arcsine-transformed estimates of
dietary composition (%) for the most abundant species in the samples. Data
from the 2 years were pooled to provide sufficient sample sizes to examine the
differences between pellet groups from different sites, seasons, and sexes.
Microhabitat Use
Microhabitat use was studied at each of the most accessible sites (sites
that were not high securtity) on Boca Chica Key (sites 1-17, Appendix A).
Detailed data were recorded on vegetation at each of the 49 permanent markers
(49 quadrats) on the 17 grids. Based on species composition, each site was
classified as being in 1 of 4 habitat categories: low marsh (dominated by
Monanthocloe littoralis. Sesuvium maritimum. Salicomia virginica. Batis
martima, and some small mangrove trees), mid-marsh (dominated by Borrichia
sp. and Sporobolus virginicus). high marsh (predominantly the clump grasses
Spartina spartinae, and Fimbrvstvlis sp.), and hammock (dominated by tree
species, mainly Conocarpus erectas). Four categories is an adequate number of

101
habitat groups suggested by Alldredge and Ratti (1986) for controlling Type II
statistical errors.
Presence or absence of pellets was recorded at each marker during
March, July, and November (from November 1991 July 1993). A Chi-square
Goodness-of-fit test was used to determine whether there was a significant
difference between the "expected" use of habitat classes (the proportion of
quadrats in each habitat category) and the observed frequency of habitat usage
(the proportion of quadrats with pellets). If a statistically significant overall
result was found, habitat selection was determined by using Bonferroni
confidence intervals (Neu et al. 1974, Byers and Steinhorst 1984).
Macrohabitat
Macrohabitat variables were measured at each of the 59 sites in the
Lower Keys (Appendix A). The variables included the traditional
metapopulation geometric variables, area and isolation. In addition, at each
grid vegetation characteristics were measured using a line intercept method
(Canfield 1941). Ten 5-m long transects were randomly located at each site
and percent occlusion was determined for each plant species. Two vegetation
layers were used: ground vegetation of plants <1.5 m in height and canopy
vegetation of plants >1.5 m. Ground cover, canopy cover, average maximum
ground vegetation height, and plant diversity were calculated from these

102
transects. The Shannon-Weaver diversity index (H) was chosen because
species composition and abundance were taken from transect data (Shannon
and Weaver 1949). In addition, the percent cover of individual species found
to be important to marsh rabbit diet and microhabitat were also included.
Results from the most recent vegetation survey (July 1993) were used in the
analysis. Landscape characteristics (area, distances to water, occupied
domiciles, and other marsh rabbit populations) were measured from USGS
topographic maps (1:24,000).
Variables were screened for normality and significant inter-correlation
before analysis. Non-normal variables were subjected to logarithmic and
square-root transformations. If a significant correlation was found between
variables, the variable with the most meaningful biological interpretation was
retained.
Two DFAs were performed on the data. The first compared the empty
sites to those that were occupied by rabbits during some portion of the study
(variable and consistently occupied sites). The second DFA compared variable
sites to the consistently occupied sites. Results of the second discriminant
function were used to reclassify the empty sites, to determine (based on habitat
features alone) if they are suitable for marsh rabbit recolonization. It was
assumed that if reintroduction was to occur, rabbits would be released at

103
several sites on 1 key. The distance to the nearest occupied patch was
calculated using this assumption.
Statistical analyses were done using procedures STEPDISC and
DISCRIM (SAS Institute, Inc. 1985). First, a stepwise discriminant function
analysis was used to select a subset of variables that produced a good
discrimination model. A value of P < 0.15 was used a criterion for model
inclusion. These variables were then entered to determine a discriminant
function for site classification. The model was validated using a jackknife
procedure (Capen et al. 1986). The jackknife procedure classifies each sample
using the discriminate equation derived form all samples except the 1 currently
being classified (Lachenbruch 1975).
Results
Pellet Degradation Rate and Pellet Size
During the first survey, pellets began to disappear off the grid after 7
weeks (Figure 5.2), indicating that during the dry season any sampling period
less than 7 weeks should suffice for density estimation based on pellet counts.
To compensate for a reduced persistence time anticipated for the wet season, a
sampling period of 1 month (T = 30 days) was chosen for pellet accumulation
before making counts. During the following 7 surveys, this duration was found
to be adequate for all seasons. In each session, pellets under trees in the litter

104
took the longest time to decay, whereas those in mud decayed in the shortest
time.
Fifty-three rabbits (19 females and 34 males) each produced 10 intact
pellets and were included in the regression of body mass and pellet area. The
relationship was strong and significant for both males (F = 95.8, P < 0.0001, r2
= 0.75) and females (F = 198.6, P < 0.0001, r2 = 0.92). The regression lines
were very similar, and both regression equations produced a pellet area of 0.4
cm for a body mass of 1000 g (Figure 5.3). This pellet area was used to
distinguish between adult and juvenile pellets on the pellet grids. It was
determined in Chapter 3 (Growth and Morphology) that most rabbits were
sexually mature by 1000 g and considered to be adults.
Metapopulation Structure
Marsh rabbits were found at sites on Boca Chica, Saddlebunch,
Sugarloaf, and Big Pine Keys (Figure 5.4). Between Sugarloaf and Big Pine
Key, a gap in the distribution of marsh rabbit populations occured, despite the
presence of transition-zone habitat.
Of the 59 patches of transition zone habitat used in the analysis (Figure
5.5), 20 had pellets present during all of the surveys (occupied patches), 22 had
pellets present during at least 1 survey (variable patches), and 17 never had any
pellets present (empty). When these patches were plotted by log area and

105
isolation (Figure 5.6), area did not have a large effect on whether a patch had
rabbits present (variable or occupied). There did appear to be a maximum
isolation (inter-patch distance) beyond which no rabbits occurred on patches.
This maximum isolation occurred around 2000 m (2 km) and is similar to the
maximum dispersal distance for males (Figure 5.7).
When the metapopulation structure of the S. p. hefneri was compared to
the predicted structures (Figure 5.1), it appeared that it did not fit 1 pattern
perfectly. The structure was most similar to the classical metapopulation
structure. An examination of patch extinction and recolonization on Big Pine
(Figure 5.8) demonstrates that patches of all sizes went extinct during the study
and were often recolonized. Of the 22 sites that had had pellets present for at
least 1 but not all of the surveys, 11 were observed to go extinct during the
study. Four of those local extinctions were recolonized.
The Lower Keys marsh rabbit metapopulation was least similar to the
mainland-island structure. In the mainland-island pattern, only large patches
are permanently occupied; yet in the observed pattern, patches of all size
appeared to be permanently occupied and some of the largest sites were
variably occupied.
In the source-sink model, low-quality patches that are near source
patches are often variably occupied, while low-quality patches that are further
away are empty. This appeared to be partially true in the marsh rabbit

106
metapopulation structure, yet some of the small, nearby patches were
permanently occupied as might be seen in a classic metapopulation.
Consistently occupied patches had juvenile pellets present for
significantly more sessions than variably occupied patches (t = -2.67, P <
0.01). The average proportion of sessions with juveniles was 54.2% (SD =
26.1) for the consistently occupied sites and 29.9% (SD 26.05) for the variably
occupied. Six of variably occupied sites and only 1 of the consistently
occupied sites never had juvenile pellets present. The variable sites could
produce juveniles, but they were less likely to than the consistent patches.
Further analysis of differences in habitat quality was warranted.
Dietary Analysis
Nineteen plant species representing 14 families were found in the fecal
pellets of 40 rabbits (Table 5.1). Two grasses, Sporobolus virginicus and
Spartina spartinae. composed over 50% of all diets. Sporobolus virginicus
occurs predominantly in the low marsh while S. spartinae dominates the high
marsh. Two mangrove species, Laguncularia racemosa and Rhizophora
mangle, and a small succulent shrub Borrichia frutescens. made up another
25%. The remainder of the species were grasses, sedges, shrubs, and trees.
Three species, Monanthochloe littoralis. Conocarpus erecta, and
Pithecellobium guadelupense. were abundant at the transition-zone sites but

107
absent from the marsh rabbit's diet (Table 5.2). Key grass (M. littoralis) is a
short wiry grass, and buttonwood (C. erecta) and blackbead (P. guadelupense)
are tree species. Excluding these species, marsh rabbits generally fed on the
most abundant species available.
The 4 most abundant species in the fecal pellets, S. virginicus.
spartinae. B. frutescens. and L. racemosa. were used in the univariate analysis.
Univariate analysis indicated a significant between-season difference in the
relative density of S,. spartinae in the fecal pellets, and a significant season-by
sex interactions in presence of virginicus and of spartinae (Table 5.3).
There were no significant differences found among sites or between sexes for
any of the plant species, despite the fact that the statistical power of the
univariate analyses were high (0.48 0.98).
Resource Use and Availability
Eight of the original 17 sites on Boca Chica/Geiger Key were used to
determine habitat selection (Table 5.4). Nine sites could not be included in the
analysis because either there were no pellets present during 1 of the surveys or
1 of the habitat types was missing. To simplify the analysis data, the 2 years of
data were combined and habitat selection was determined by season.
The majority of the overall chi-square goodness-of-fit statistics were
significant (P < 0.05) for all of the sites during the 3 seasons. Goodness-of-fit

108
comparisons indicated that habitat selection occurred at least once at all of the
sites, and the pattern of selectivity did not appear to vary strongly among
seasons. In all 3 seasons, rabbits significantly used mid-marsh vegetation
(Borrichia sp. and Sporobolus virginicus) and high marsh vegetation (mainly
clump grasses) more than expected and used hammocks and low marshes less.
The over all rank of habitat use based on the fecal pellet distribution was: mid
marsh > high marsh > hammock > low marsh.
Habitat Model
Ten variables were used in the 2 DFAs (Table 5.5). All of the habitat
variables were either normal without a transformation or became normal after
the transformation (Table 5.5). None of the variables were significantly inter-
correlated. The test for homogeneity of within-group covariance matrices was
not significant (X^ = 22.75, df = 46, P > 0.36), indicating that a pooled
covariance matrix and a linear discriminant function analysis should be used
(Morrison 1976).
Four variables (DPopulation, Area, DResidence and MaxHgt) (Table
5.6a) were included in the stepwise DFA comparing empty and occupied sites.
Only 2 of the variables, DPopulation and DResidence, differed significantly
between empty and occupied sties (Table 5.7). Empty sites were more isolated
from other populations but closer to human dwellings. The discriminant

109
function based on the 4 variables classified 93% of the sites correctly (3 empty
sites were classified as being occupied). The jackknife procedure classified
91%, indicating a fairly accurate model. One empty site was misclassified as
being occupied; 3 occupied sites were misclassified as being empty.
Four variables were included in the stepwise DFA comparing variably
occupied sites to the consistently occupied sites (Clump, DPopulation,
MaxHgt, and Borrichia; Table 5.6b). Consistently occupied sites had more
clump grasses and significantly higher ground vegetation (Table 5.7). Ninety-
seven percent of the sites were correctly classified by the model; 1 consistently
occupied site was misclassified as being variable. The jackknife classification
rate was also high 95% of the sites were classified correctly.
Potential Reintroduction Sites
Seven of the 17 currently empty sites were classified as being
potentially consistently occupied habitat sites using the discriminant function
and the adjusted inter-population distances (assuming multiple reintroductions
on a key). Three of these sites that were classified as being consistently
occupied occurred on North Sugarloaf, and 1 each occurred on Cudjoe Key,
Middle Torch Key, Big Torch Key and Noname Key.

110
Discussion
Metapopulation Structure
In the comparison of occupied sites (consistently and variably) and
empty patches, the metapopulation structure of S. p. hefheri conforms to the
classic definition of a metapopulation (Levins 1969, Hanski and Gilpin 1991).
Nearly all patches that are below a maximum inter-patch distance are occupied
and patches of all areas are occupied. Lack of a minimum area effect may be
due to the methods for patch identification; habitat patches were included in
the study only when they exceeded 0.5 ha.
Few good examples of classic metapopulations in nature exist in the
literature (Harrison 1991, 1994). This may be a function of the duration of
most population studies and scarcity of larger-scale research; otherwise, most
systems studied are not true metapopulations. Species that have been found to
exist in a classic metapopulation may have some unifying life-histories and
behaviors (Murphy et al. 1990), although debate exists as to what these
characteristics include (see Harrison 1994). Small-bodied, short-lived species
with high reproductive rates (r-selected species) appear to be the most likely
candidates for classic metapopulations (Murphy et al. 1990), because such
species generally are good colonists and have exceptional dispersal abilities
(Harrison 1994). The Lower Keys marsh rabbit fits this description well, as do
many species of small mammals, invertebrates, and annual plants.

Ill
Harrison (1991, 1994) and Thomas (1994) suggested that species that
inhabit early successional habitats that tend to be temporally patchy are more
likely to exist in true metapopulations; Murphy et al. (1990) extended this to all
species that display strong habitat specificity to a habitat that is fragmented.
The marsh rabbits high marsh transition zone habitat may differ from stricter
definitions of early successional habitat because it does not readily succeed to
other habitats. Frequent disturbances from tidal inundation and storms,
maintains the habitat over time. However, the Lower Keys marsh rabbit is
highly habitat specific, and high marshes are highly fragmented in the Lower
Keys. This fragmentation is mainly due to development.
Further examination of the consistently and variably occupied patches
indicates that not all patches appear to be of equal value in terms of
reproductive potential. Some of the variable patches may be sinks supported
by nearby source populations. However, not all of the variable patches were
sinks and not all of the consistently occupied sites were sources, indicating that
habitat quality is not the only factor elucidating the reproductive potential of a
patch. Local extinctions that occurred at the variable sites may have been
partly random, and perhaps all sites are variable if a long enough study was
conducted.

112
Habitat Use
Three different temporal and spatial scales provided complementary
information on marsh rabbit habitat use. Marsh rabbits forage, nest, and hide
in the dense ground vegetation of the transition zone. With some exceptions,
S. p. hefheri appeared to feed on the plant species most abundant in its habitat.
The major plant species in the rabbit's diet did not appear to vaiy among sites,
between the wet and dry seasons (with one exception), or between males and
females. These results differ from the majority of leporid dietary studies.
Few studies have investigated the effect of site on diet. Wallage-Drees
et al. (1986) found that the diets of Orvctolagus cuniculus differed among sites.
These sites varied greatly in their vegetative composition; a portion of the sites
were in woodlands, while others were in heathlands and abandoned arable
fields. In contrast, hefneri is very specialized in its habitat use. Most
Lower Keys marsh rabbits inhabit the transition zone. Vegetation in transition-
zone areas may vary in proportion between sites but is similar in composition.
The dietary requirements of the marsh rabbit may only be met in the transition-
zone areas in the Lower Keys.
Seasonal changes in leporid diets have been extensively documented
(Cervantes and Martinez 1992, Chapman et al. 1982). In most of these studies,
the climate and vegetation changed extensively between seasons. The weather
in the Lower Keys is relatively invariant. The variation in the proportion of S.

113
spartinae in the rabbit's diet is probably due to new growth during the wet
season.
Although marsh rabbits eat vegetation in proportion to its abundance,
they spend a disproportionate amount of time in the mid- and high marsh. This
may indicate that marsh rabbits use the mid- and high marsh for cover while
using all areas for foraging. Both the mid- and high marsh contain thick
ground cover (Borrichia in the mid-marsh and Spartina and Fimbrvstvlis in the
high marsh). In addition, all nests were found in the high marsh (see Chapter
3).
On a larger scale, the habitat model agreed with the habitat use/
availability results. The model indicated that thick ground cover (especially
clump grasses and B. frutescens) and distance between populations were
important factors in determining how consistently a habitat patch was
occupied. The thick grasses and forb provide the necessary cover for marsh
rabbits to nest and escape from predation. Severe predation may reduce
population size and increase the chance of stochastic demographic extinction
(Gilpin and Soul 1986). For small-to-medium size herbivores like the marsh
rabbit, predation may be the most important factor in determining local
abundance (Chapman et al. 1982).
Additionally, patches within close proximity of other occupied patches
benefit from emigration of individuals via the rescue effect (Brown and

114
Kodric-Brown 1977) in the context of classic island biogeography, or internal
rescue effect as it is applied to local patches within a metapopulation (Hanski
1982, Gotelli 1991, Holt 1993). If population numbers have been lowered at a
certain patch, new emigrants may reduce population variability, increasing the
persistence time of the local population.
Any plans to reintroduce the Lower Keys marsh rabbit should include
inoculating near by, multiple patches simultaneously. Not all of the patches
used in the reintroduction effort have to be high quality because even the
variably occupied patches were capable of supporting marsh rabbits that
produced young. Saddlebunch, Big Torch, and Middle Torch Keys all contain
multiple habitat patches. Each Key contains at least 1 patch that was classified
as being a consistently occupied patch. These Keys should be given the
highest priority for proposed reintroduction sites.
In conclusion, for S. p. hefneri, habitat quality and incidence-function
variables (area and isolation) interact to determine patch occupancy. Not all
occupied patches are of the highest habitat quality, and not all of the variable
or empty patches are empty because of low habitat quality. Several of the
empty habitat patches are sufficiently good habitat to support marsh rabbits,
but are too isolated from currently occupied habitat patches to sustain
populations. Overall, the Lower Keys marsh rabbit metapopulation did appear

115
to be closest to the classic metapopulation in structure. Implications of these
conclusions will be further examined in Chapter 7.

116
Table 5.1--Relative density (%) each plant species in fecal pellet group samples (N =
40) of Svlvilagus palustris hefneri.
Plant species and family
X
SE
SDorobolus vireinicusPoaceae
35.74
5.21
Soartina spartinaePoaceae
17.33
3.87
Laeuncularia racemosaCombretaceae
10.25
2.49
Borrichia frutescensAsteraceae
8.40
1.51
Rhizophora maneleRhizophoraceae
6.81
1.97
Andropoeon elomeratus-Poaceae
6.14
2.04
Eleocharis cellulosa--Cvperaceae
3.75
1.63
Muhlenbereia filipesPoaceae
2.27
1.32
Tvpha latifolia-Tvphaceae
2.10
0.77
Coccoloba uviferaPolvgonacea
1.35
0.81
Fimbristvlis castaneaCvperaceae
0.76
0.38
Jacquinia kevensisTheophrastueae
0.75
0.30
Salicornia virginicaCheropodiaceae
0.63
0.31
Sesuvium maritimumAizoaceae
0.63
0.10
Avicennia eerminansAvicennianceae
0.38
0.23
Baccharis halimifoliaAsteraceae
0.31
0.18
Erithalis fruticosaRubiaceae
0.24
0.17
Fimbristvlis spathaceaCvperaceae
0.23
0.23
Mavtenus Dhvllanthoides-Celastraceae
0.17
0.17
Unidentified material
1.64
0.50

117
Table 5.2~Ground cover for the 19 most abundant species measured in five transition
zone habitat patches in the Lower Keys of Florida. An asterisk following the plant
species indicates a species not found in the rabbit's diet.
Plant species and family
X
SE
SDartina SDartinaePoaceae
17.40
8.49
SDorobolus vireinicusPoaceae
14.60
4.31
Borrichia frutescensAsteraceae
8.40
3.80
Monanthochloe littoralis*--Poaceae
4.40
1.94
Andropoeon elomeratus-Poaceae
4.20
2.01
Fimbristvlis castanea-Cvperaceae
4.20
1.91
Conocarpus erecta*--Combretaceae
2.80
0.74
Muhlenbereia filipesPoaceae
2.60
1.19
Pithecellobium euadelupense*Fabaceae
2.00
2.00
Fimbristvlis spathaceaCvperaceae
1.60
1.17
Salicomia vireinicaCheropodiaceae
1.40
0.93
Laeuncularia racemosaCombretaceae
1.20
0.20
Avicennia eerminansAvicennianceae
1.00
0.32
Coccoloba uvifera--Polvgonacea
1.00
0.63
Rhizophora maneleRhizophoraceae
1.00
0.32
Tvpha latifoliaTvphaceae
1.00
1.00
Eleocharis cellulosaCvperaceae
0.80
0.49
Sesuvium maritimum-Aizoaceae
0.80
0.80
Baccharis halimifolia--Asteraceae
0.40
0.40

118
Table 5.3Results of univariate analysis of variance on arcsine-transformed percentage
relative density of Sporobolus virginicus (Sv), Spartina spartinae (Sp), Laguncularia
racemosa (Lr) and Borrichia ffutescnes (Bf) in the fecal pellets of SL p, hefneri in the
Lower Keys of Florida.
Source
Sv Sp
df F P F P
Lr
F P
Bf
F P
Site 5
Season 1
Sex 1
Season by sex 1
0.87
ns
1.38
ns
2.39
ns
11.61
**
0.21
ns
1.26
ns
15.71
**
6.24
*
1.15
ns
0.97
ns
2.27
ns
1.26
ns
0.04
ns
0.01
ns
2.19
ns
0.00
ns
(* 0.05 > P >0.01; ** 0.01 > P > 0.001; ns = P > 0.05)

119
Table 5.4Goodness-of-fit test for habitat use by Lower Keys marsh rabbits on Boca
Chica/Geiger Key. A + denotes that a habitat was used significantly more than
expected, a refers to a habitat that was used significantly less, and a 0 means the
habitat was neutral with respect to use according to the Bonferonni confidence
intervals.
March
Site
X2
Low
marsh
Mid
marsh
High
marsh
Hammock
1
6.73
0
0
0
0
4
8.14
0
0
0
-
8
23.02
0
+
0
0
9
3.95
0
0
0
0
10
14.72
-
0
+
0
11
15.13
-
+
0
0
12
31.42
-
0
+
0
13
49.38
+
0
0
Total
Mr)
3(+)
2(+)
l(-)
July
Site
X2
Low
marsh
Mid
marsh
High
marsh
Hammock
1
5.91
0
0
0
0
4
10.33
0
0
0
0
8
81.07
-
0
0
0
9
1.03
0
0
0
0
10
9.75
-
0
+
0
11
11.05
-
+
0
0
12
30.17
-
0
+
0
13
41.28
0
+
0
-
Total
4(-)
2(+)
2(+)
l(-)

120
Table 5.4 (continued).
November
Site
X2
Low
marsh
Mid
marsh
High
marsh
Hammock
1
30.42
-
+
0
-
4
4.02
0
0
0
0
8
17.05
0
+
0
-
9
3.95
0
+
0
0
10
8.97
-
0
+
0
11
11.46
-
0
0
-
12
31.42
-
0
+
0
13
5.43
0
0
0
-
Total
Mr)
3(+)
2(+)
Mr)
Grand Total
12(-)
8(+)
6(+)
Mr)

121
Table 5.5Variables used to measure characteristics of marsh rabbit habitat in the
Lower Keys of Florida.
Variable
abbreviation
Variable
Mode of
measurement
GCover
Ground Cover
proportion of 10
transects occluded
below 1.5 m
CCover
Canopy Cover
proportion of 10
transects occluded
above 1.5 m
Borrichia
Borrichia frutescens
cover
proportion of 10
transects occluded
at ground level
Clump
Clump grass cover
proportion of 10
transects occluded
bv Spartina spartinae.
Fimbrvstilis sp. and
Cladium iamaicensis
MaxHgt
Average maximum
height
average maximum
height of ground
vegetation measured in
0.25-m intervals in the
1st, 3rd, and 5th meters
of each transect
H
Plant diversity
Shannon-Weaver
diversity index based on
ground and canopy
species present in 10
transects
Area
Area of habitat patch
aerial photograph
DPopulation
Distance from edge
aerial photograph
of site to nearest
rabbit population
Range Trans
formation*
30-98% S
2-81% S
0-60% S
0-87% S
6.5-62.0 cm N
0.55-2.17 N
0.3-43.7 ha L
40-6066 m L

122
Table 5.5 (continued).
DResidence Distance from edge aerial photograph 1-6000 m L
of site to nearest
occupied domicile
DWater Distance from edge aerial photograph 1-1000 m L
of site to body of
water
* Transformations applied to data to achieve normality: S = square-root, L = logio> N
= no transformation needed, data already normal.

123
Table 5.6aResults from a stepwise discriminant function analysis of 10 habitat
variables measured on "occupied and "empty" marsh rabbit habitat in the Lower Keys
ofFlorida.
Habitat Variable
Step entered
Partial R^
F
P
DPopulation
1
0.56
73.26
0.0001
Area
2
0.12
7.63
0.0007
DResidence
3
0.08
5.01
0.01
GCover
4
0.05
2.97
0.09
Table 5.6bResults from a stepwise discriminant function analysis of 10 habitat
variables measured on "variably" and "consistently" occupied marsh rabbit habitat in
the Lower Keys ofFlorida.
Habitat Variable
Step entered
Partial R^
F
P
Clump
1
0.60
60.48
0.0001
DPopulation
2
0.27
14.29
0.0005
MaxHgt
3
0.08
3.24
0.05
Borrichia
4
0.09
3.78
0.04

Table 5.7--A comparison of habitat measurements between occupied and empty marsh rabbit habitat in the Lower Keys of Florida.
Habitat variable (Unit)
Occupied habitat (n=42)
Empty habitat (n=17)
Test statistic*
P
X(SE~)
Untransformed
X(SE)
Untransformed
GCover(%)
7.7(0.2)
59.3
8.0(0.3)
64.0
1.12 (T)
0.27
CCover(%)
3.7(0.2)
13.7
3.8(0.5)
14.4
-0.44 (U)
0.66
Borrichia(%)
2.3(0.2)
5.3
1 5(0.5)
2.3
-2.36 (U)
0.02
Clump(%)
4.4(0.3)
19.4
4.5(0.6)
20.3
0.15 (T)
0.88
MaxHgt(cm)
22.2(1.7)
22.2
28.7(4.2)
28.7
1.10(U)
0.27
H(#)
1.4(0.1)
1.4
1.3(0.1)
1.3
-1.35 (T)
0.19
Area(ha)
1.6(0.3)
4.8
0.6(0.1)
1.8
-1.15 (T)
0.25
DPopulation(m)
5.8(0.2)
330.3
8.1(0.1)
3294.5
5.78 (U)
0.001
DResidence(m)
5.7(0.4)
298.9
2.9(0.8)
18.2
-3.10 (T)
0.005
DWater(m)
3.0(0.4)
20.1
2.9(0.6)
18.2
-0.20 (U)
0.84
* For all normally distributed variables (or transformed variables) a two-sample t-test (T) was used, a Mann-Whitney U-test using normal approximation (U)
was used for all non-normally distributed variables.

Table 5.8A comparison of habitat measurements between good and marginal marsh rabbit habitat in the Lower Keys of Florida.
Habitat variable (Unit)
Constitent habitat (n=22)
Variable habitat (n=20)
Test statistic*
P
X(SE)
Untransformed
X(SE)
Untransformed
GCover(%)
8.2 (0.2)
67.2
7.2 (0.2)
51.8
-3.65 (T)
0.008
CCover(%)
3.6 (1.3)
13.0
3.9 (1.2)
15.2
1.04 (U)
0.31
Borrichia(%)
2.3 (0.3)
5.2
2.2 (0.3)
4.8
-0.17 (U)
0.86
Clump(%)
5.9 (0.3)
34.8
3.1 (0.2)
9.6
-7.73 (T)
0.001
MaxHgt(cm)
26.4 (3.2)
26.4
18.3 (1.1)
18.3
-2.47 (U)
0.02
H(#)
1.5 (0.1)
1.5
1.7 (0.1)
1.7
0.99 (T)
0.32
Area(ha)
1.8 (0.5)
5.8
1.7 (0.8)
5.5
-0.14 (U)
0.89
DPopulation(m)
5.5 (0.3)
244.7
6.1 (0.2)
445.9
1.93 (U)
0.06
DResidence(m)
5.8 (0.6)
330.3
5.5 (0.6)
244.7
-0.34 (T)
0.73
DWater(m)
3.5 (1.5)
33.1
2.7 (0.5)
14.9
-1.08 (T)
0.28
to
C/1
* For all normally distributed variables (or transformed variables) a two-sample t-test (T) was used, a Mann-Whitney U-test using normal approximation (U)
was used for all non-normally distributed variables.

5
4
c=
3
ro
O 2
_co
1
0
Mainland-island
o 1
2 3 4
Area
Source-sink
o 1
2 3
Area
Classical
2 3
Area
K>
On
Figure 5.1 The 3 types of metapopulation structure: mainland-island, source-sink, and the stepping stone model a form of the classical
metapopulation. The black circles are consistently occupied sites. Grey are variably occupied sites and the white circles are empty
sites.

Days Until Degredation
127
March 1991
March 1992
Session
March 1993
Figure 5.2 Number of days until the first Lower Keys marsh rabbit (Sylvilagus
palustris hefneri) pellet decayed. One group of 25 pellets was placed in each of 4
settings (in tree litter, on rocks, in grass, and in mud) and observed daily.

Pellet Area (crrr)
128
0 200 400 600 800 1000 1200 1400 1600
Body Mass (g)
Figure 5.3 Average pellet size (in area cm2) in relation to body mass (g) of 53 Lower
Keys marsh rabbits (Sylvilagus palustris hefneri) trapped on Boca Chica and
Saddlebunch Keys.

Middle Torch
Figure 5.4 The current distribution of the Lower Keys marsh rabbit (Svlvilagus palustris hefneri) by Key. The stipled keys are
occupied; white keys are unoccupied.

Number of Patches
130
Number of Sessions Occupied
Figure 5.5--Number of sessions (from session 3-8) occupied by the Lower Keys marsh
rabbit (Sylvilagus palustris hefneri) at each of the 59 transition-zone sites.

Log Isolation (m)
131
10000
1000
100
0.1 1.0 10.0 100.0
Log Area (ha)
o
<&>
o
o
o
Q
O o U
o
o
o
Q#^§)
#
1 I I I I I I I I
O Empty
O Variable
Occupied
n 1ii i ii i
Figure 5.6 Fifty-nine sites plotted with respect to Lower Keys marsh rabbit
(Sylvilagus palustris hefneri) occupancy by logio area (ha) and logio isolation (m).
Empty patches never had marsh rabbits present, variable patches had marsh rabbits
during at least 1 session, and occupied sites had rabbits during all of the sessions.

Proportion
132
Dispersal (m)
Figure 5.7 Average distance between a patch and the nearest occupied patch (patch
isolation) for the 59 patches, in comparison to the dispersal distances of the 6 female
Lower Keys marsh rabbits (Svlvilaeus palustris hefneri) and the 11 male.

Figure 5.8 Patch occupancy by Lower Keys marsh rabbits (Svlvilagus palustris hefneri) during the last 6 pellet sampling sessions. A
solid circle indicates an occupied patch, a dashed circle represents a patch occupied by adults only, and an open circle indicates a vacant
patch. Dashed lines between patches indicate patches within dispersal distance from each other. The solid lines are roads.

Session 6
52
Figure 5.8 (continued).
u>
.1

CHAPTER 6
POPULATION VIABILITY ANALYSIS
Introduction
The process of conducting a population viability analysis (PVA; Soul
1987) involves building a model that makes probabilistic predictions about a
species future (Ginzburg et al. 1982, Shaffer 1990). Most models use
population simulation or analysis based on existing data to make these
predictions. The challenge in developing a PVA is creating a model that
captures all of the important aspects of a species ecology while using only
realistically measurable parameters (Eberhardt 1987, Boyce 1992).
This process is further complicated when the population to be modeled
exists in discrete subpopulations. If these subpopulations are related through
inter-subpopulation movements that have the potential to augment or
recolonize small or extinct subpopulations, then the PVA must include this
metapopulation structure (LaHaye et al. 1994).
Multiple Levels of Population Viability Analysis
A PVA that incoiporates metapopulation structure can involve processes
that occur at 3 increasing spatial levels: subpopulation (patch), between
135

136
populations, and metapopulation. The subpopulation level (or patch) is the
traditional domain of PVAs that involve a single population.
Most single-population PVAs are conducted on small populations and
require that both stochastic and deterministic processes be examined (Gilpin
and Soul 1986, Wilcox 1986, Lacy and Clark 1990). Small populations are
more likely to be at risk of extinction due to stochastic variation than are large
populations. Stochastic perturbations in small populations can affect the
genetic composition (loss of genetic variability, inbreeding depression) or the
demographic composition (sex, age ratios) of a population (Shaffer 1981).
Variation in the environment (climatic variation, natural catastrophes), is also
random but affects small populations only slightly more than larger ones.
Deterministic causes of extinction include Diamonds evil quartet:
overkill, habitat destruction, invasion by introduced predators or competitors,
and food-web collapse (Diamond 1984, 1989). In addition, climatic change
(Nunney and Campell 1993) may be deterministic and disease may be
stochastic or deterministic depending on its extent and the impact it has on the
population (Aguirre and Starkey 1994). Deterministic extinction can occur
regardless of population size, if all of the population is effected. For most
species, deterministic processes play a larger role in the original decline of the
species, while stochastic process are responsible for the extinction after the
decline (Caughley 1994). Both stochastic and deterministic phenomena may

137
interact via feed-back loops leading to potential extinction via extinction
vortices (Gilpin and Soul 1986).
Interactions between patches also can be an important component of the
PVA. Dispersal rates, the mechanisms behind dispersal, and the impact of
habitat on dispersal can effect the amount of interchange between populations.
Individuals moving between patches may increase an the persistence time of an
occupied patch by augmenting the population and decreasing the chance for
stochastic extinction (rescue effect; Brown and Kodric-Brown 1977). The
dispersing individuals may also recolonize empty patches, if both males and
females or pregnant females move (in a sexual species). Incorporating inter
patch dynamics requires using spatially explicit metapopulation structure.
The third spatial level, metapopulation dynamics, considers the inter
relation between patch extinction and recolonization. For a metapopulation to
persist, the patch recolonization rates must exceed patch extinction rates
(Levins 1970, Hanski and Gilpin 1991). The inter-patch movements described
above will determine recolonization rates, but metapopulation persistence is
largely dependent on a lack of correlation among populations in extinction
rates (Murphy et al. 1990, Thomas and Jones 1993). Environmental variation
and catastrophes can cause synchronous variations in population densities
(Quinn and Hastings 1987, Gilpin 1988, Hanski 1989, Stacey and Taper 1992).
If all patches are subjected to the same adverse environmental conditions at the

138
same time, then patch extinctions may be correlated, leaving few or no source
populations to colonize the extinct patches.
Population Viability Analysis Models
Most metapopulation models (e.g., Nisbet and Gurney 1982, Hanski
1985, 1991, Hastings and Wolin 1989, Hastings 1991, Verboom et al. 1991,
Gyllenberg and Hanski 1992, Nee and May 1992) predict metapopulation
persistence based on phenomena that occur at the metapopulation level only
(Hanski 1994). Most of these models are theoretical in nature, and the
parameters can not be readily measured in the field. A few models (Fahrig and
Merriam 1985, Hanski and Gyllenburg 1993), incorporate the inter-population
level and the metapopulation level phenomena, creating spatially realistic
models that require only modest amounts of data to construct. These models
provide information on metapopulation dynamics of real populations but are
less useful in making management decisions. Models that incorporate all 3
levels require substantially more information but may be used to test
hypotheses regarding threats to the species at all 3 levels.
Several simulation models are currently available that combine
stochastic birth and death process models (Richter-Dyn and Goel 1972,
Goodman 1987), differing migration rates based on spatially explicit patch
structure and extinction and recolonization (correlated or uncorrelated).

139
VORTEX (Lacy 1993) and RAMAS/metapop (Ak9akaya 1994) both provide
estimations about the persistence of a metapopulation using a Monte Carlo
simulation algorithm to generate stochasticity. VORTEX is slightly less
flexible in its treatment of population growth and is limited in the number of
subpopulations that can be modeled at once.
Costs and Benefits of Using a PVA Model
Although most scientists and managers consider a detailed PVA to be a
useful tool in making management decisions (Salwasser et al. 1984, Marcot
and Holthhausen 1987, Shaffer 1990), these models have recently come under
criticism. For many species, data are not currently available to conduct an in-
depth PVA, and by the time it is collected the species may be extinct (Maguire
1991, Boyce 1992). Some population dynamics such as density dependence
are poorly understood, regardless of the parameters measured in the field
(Hassell 1986). This lack of knowledge can severely detract from the most
well-developed PVA model. For other species, conducting a PVA may
estimate how long the species will persist but not indicate why the species is
declining (Caughley 1994). Managers, lawyers and the public often consider
the predictions of PVAs as facts, and do not give credence to the variance and
limitations of the model (Barthouse et al. 1984). Finally, few PVA models are

140
ever validated and therefore it is difficult to determine which model approaches
are most successful (Grant 1986).
Despite these problems, PVA has been found to be a useful management
tool for a number of species (grizzly bear, Suchy 1985; northern Spotted owl,
LaHaye et al. 1994). The strongest potential of PVA is in the development of
adaptive management strategies (Holling 1978, Walters 1986, Lee 1993) to
recover endangered species. A well-developed PVA can refine multiple
hypotheses about the impact of multiple management tactics. Small-scale
management techniques can be attempted at different (sub)populations, the
(sub)populations monitored, the PVA model validated, and then the most
successful suite of management techniques applied to all of the populations.
The main objective of this chapter is to develop a PVA model for the
Lower Keys marsh rabbit (Sylvilagus palustris hefneri). This model will make
predictions about the persistence of the subspecies under a variety of potential
scenarios and management techniques. This information will be used to make
recommendations about the direction of future research and management of
this subspecies.
The Lower Keys marsh rabbit is an endangered subspecies occurring in
a classical metapopulation structure. It is believed to be declining during
recent years (Howe 1988) and was only protected 4 years ago. Since the

141
formal listing, habitat destruction has abated somewhat, although development
of the rabbits habitat is still possible.
Methods
The RAMAS/metapop (Ak9akaya 1994) simulation model was used for
the PVA because of its flexibility and ability to incorporate a large number of
populations. Processes occurring at all 3 spatial levels were incoiporated in the
model. At the level of the population, a life-history table in the form of a
Leslie matrix (Leslie 1945) was completed, the impact of demographic and
genetic stochasticity assessed, the form of the population growth curve was
estimated, and the initial abundances for each population were estimated. At
the between-population level, the number and age of the individuals moving
between populations was determined, and the spatial structure of the
metapopulation and the impact this has on inter-population movements was
estimated. At the metapopulation level, the degree of correlation between
population density was estimated.
Population-level Parameters
A stage rather than age Leslie matrix was used based on the results of
Chapter 3, which suggested that S. p. hefneri differs in sum val and
reproduction depending on the developmental stage. Five stages were used:

142
nestlings (0-3 months), juveniles (4-7), subadults (8-10), first-year adults, and
second- (and older-) year adults. Survivorship was estimated using a
combination of live-trapping and radio-telemetry, as was reproductive effect
per female (see Chapter 3). Males and females were combined when
determining all rates.
Demographic stochasticity (e.g., sex ration, demographic structure) was
studied by monitoring marsh rabbit populations at 5 patches of habitat.
Genetic variation was determined by taking blood from the ears of marsh
rabbits and performing starch-gel enzyme electrophoresis (as described in
Chapter 3).
Determining whether population growth of S. p. hefneri is dependent or
independent of density provided a challenge. Detecting density dependence in
natural populations is difficult because of the variation caused by stochastic
demographic and environmental processes (Burgman et al. 1993). Few studies
have actually documented that density-dependence occurs, despite a profusion
of studies. Several methods of detecting density dependence using sequential
censuses have been developed, but none of these methods reliably test density
dependence (Hassell 1986). An experimental manipulation of population
density over time is the best way to determine density dependence. However,
when working with endangered species, manipulations may be difficult
because they may affect the population and because the necessary time may not

143
be available to conduct such an experiment. Gaston and Lawton (1987)
suggest that a qualitative analysis of the population dynamics and life history
of a species might be used to infer if density dependence could occur.
For the purposes of this model, it was assumed that the rabbit, a short-lived,
quickly reproducing mammal, might exhibit some density dependence. The
first run of the model incorporated simple density dependence using the logistic
equation, where population growth follows the logistic equation until carrying
capacity (K) is reach. At carrying capacity the model uses the life-table to
predict population growth.
Yet, because of the severe predation pressure (Chapter 3) on the marsh
rabbits, population growth might be controlled by the predation, limiting the
ability of the Lower Keys marsh rabbit to exhibit density-dependent population
growth. The model was additionally run without density dependence, using the
exponential equation in conjunction with the life table to predict population
growth.
To determine the carrying capacity for the density-dependent growth,
the number of males that could fit into a patch was calculated by assuming
non-overlapping core areas. For areas of consistently occupied habitat
(Chapter 5), the size of the male core area was the mean core area calculated in
Chapter 4 plus one standard deviation. For areas of variably occupied
habitat, the size of the male core area was this mean core area minus one

144
standard deviation. Assuming a one-to-one ratio between males and females,
the number of males was doubled to reach the carrying capacity of adults.
Assuming a stable age distribution, the number of subadult, juveniles and
nestlings was estimated and added to the adult carrying capacity.
To determine the initial abundance of rabbits at each site, pellet counts
were used throughout the study. With knowledge about the distribution of
pellets, the number of pellets produced by a rabbit a day and the pellet
persistence rate (calculated as 30 days in Chapter 5), fairly accurate
abundances can be obtained (Wood 1988). For the purposes of this study, only
adult pellets were counted (those pellets with an area >0.4 cm see Chapter 5).
Presence or absence of juvenile pellets was recorded, but these pellets were not
included in the density estimates.
To determine the number of pellets each rabbit produced in 1 day, 4
trapped adult rabbits (2 male, 2 female), were given a collection of natural
vegetation and placed for 24 hrs in a rabbit cage with a removable pan beneath
the mesh floor. This procedure was repeated 3 times a year (March, July, and
November) to document the effect of climate and rabbit age during late dry
season, mid-wet season, and the transition time between wet and dry seasons.
Reliability of pellet counts may be affected by the patchy distribution of
pellets, which can bias the counting technique. To identify a reliable method,
during the first pellet-sampling period sampling units (SU) of 2 sizes were used

145
at all of the grids. The SU consisted of concentric 0.5-m- and 1.0-m-radius
circles around each permanent marker. Inside these circles, pellets were
removed, and then new pellets were counted after a specified amount of time.
To determine which SU size was optimal, the distribution pattern of pellets was
fitted to a Poisson distribution model. The Poisson distribution furnishes
values expected for a random dispersion pattern. When a chi-square goodness-
of-fit test is used to compare the 2 distributions, a significant result indicates
that the sample is non-random. The optimal sampling unit will be most likely
to produce a random distribution of pellets.
Once the preferred SU was selected, the following equation was used to
estimate rabbit density:
D = (10,000 m2/ha) X.
T*R* A
where: D = density of rabbits (rabbits per ha),
X = mean number of pellets per SU,
T = time between pellet removal and pellet counting,
R = defecation rate (number of pellets dropped per rabbit per 24-
hr day)
A = surface of each SU (m2).

146
The defecation rate (R) was determined as described in the previous
section, "pellet production". The number of days between pellet removal and
counting (T), was determined to be more than 30 days in Chapter 5. Thirty
days was used as the sampling interval.
Density estimation was attempted at the permanent fecal pellet-sampling
grids that were established at each of 59 patches of transition-zone habitat in
the Lower Keys. The grid was designed to fit into the smallest patch and
consisted of a square of 7 x 7 stations at 15-m intervals, marked with
permanent flags. The grids were surveyed 3 times per year: March (late dry
season), July (mid-wet season), and November (transition between wet and
dry) from March 1991 to July 1993. Each survey consisted of pellet removal
within a radius of 0.5 m at each station followed by a pellet count 1 month
later. Because some of the grids were occasionally disturbed by vehicles, these
grids were examined only for pellet presence or absence. At several privately-
owned sites and sites that were deemed sensitive by the Navy, access was
limited and the time necessary to thoroughly count the pellets was not
available. These sites also were only examined for presence or absence of
pellets.
To determine the initial abundances for the model, all of the density
estimations collected during session 3 were used. Session 3 was chosen
because the most sites were censused during that time. To calculate initial

147
abundance, the density of adult rabbits was multiplied by the area of the patch.
If juvenile pellets were found at that grid during the census, then a stable age
distribution was assumed and the number of subadults, juveniles and nestlings
was included. If no juvenile pellets were found, then it was assumed that the
rabbits at that patch had not produced any young and only the number of adult
rabbits was used.
For patches where density estimates were not possible to obtain, the
carrying capacity of the patch was used as the initial abundance. Although this
may have been higher than the actual number of rabbits, these patches were all
small and less likely to have a major impact on the estimate of total abundance.
Inter-population Level Parameters
To determine the amount of movement between populations, 6
populations of marsh rabbits on Boca Chica Key and Saddlebunch keys were
trapped. Radio-collars were placed on all rabbits >300 g. This ensured that
rabbits of all developmental stages (except for the altricial nestlings) were
represented. Radio locations were ascertained 3 times per week, or daily if it
appeared the rabbit was making a long-distance movement. The distance and
type of habitat of each long-distance movement was determined (see Chapter
4), including the crossing of major barriers (e.g., bodies of water, roads,
runways).

148
Metapopulation Parameters
To determine if the variation in population densities was correlated
among populations, a multivariate, repeated-step analysis of variance
(MANOVA) was performed on the density estimations collected during the 8
sampling sessions (PROC AUTOREG, SAS/ETS, Sas Institute 1984).
Estimated abundances of each of the populations over the 8 sessions was
compared to the other populations an the same key. Sites that never supported
any marsh rabbits were removed from the analysis.
The Model
Using the parameters described above, the simulation was repeated
1,000 times for each scenario (described below) and run for 50 years. Most
PVA models are run for 100 years however, the predicted increase in sea level
rise for south Florida (Ross et al. 1994) implies habitat changes and precludes
such a long-term outlook. To test the sensitivity of the input parameters, all of
the elements in the Leslie matrix were varied. This includes survivorship in
each of the 5 stages and reproduction in the adults.

149
Scenarios
Each scenario is described in depth below, but the specific numerical
modifications in the parameters for each scenario are listed in Table 6.1. The
first 2 simulations used the original parameters but differed in the way they
simulated population growth. The first simulation used the logistic equation
for population growth; the second used the exponential. After comparing the
results of the simulation with knowledge about the biology of S. p. hefneri, 1
simulation was chosen to be used in the scenarios.
Scenario #1- Decrease Predation.
(a). Predator control leads to a 25% decrease in mortality in each stage.
This would require a 50% reduction in the current amount of cat predation.
This would be likely to occur if the house-based and feral cats (Felis catus)
were removed and prevented from hunting in marsh rabbit habitat. Some
predation would probably occur before cats were captured in the trap but the
overall mortality rate would decrease. The radio-telemetry study conducted on
the Lower Keys marsh rabbit (see Chapter 3) found that cats were the major
predator of juvenile, subadult, and adult rabbits. It is assumed that nestling
rabbits also would be vulnerable to cat predation.
(b). Predator control leads to a 50% decrease in mortality at each stage.
This would require complete extermination of all cats in marsh rabbit habitat

150
and a reduction in mortalities attributable to raccoons. Raccoon population
reduction might occur when supplemental food sources are eliminated (i.e.,
open dumpsters, outdoor pet food).
Scenario #2 Decrease Road-kills.
All vehicular deaths are prevented. This would mean a roughly 25%
increase in the survivorship of subadults and 10% increase in the number of
adults. Implementing this management strategy would require strict speed limit
reductions, or an over(under) pass for the rabbits to use.
Scenario #3- Reintroduce Rabbits.
Rabbits are reintroduced to all vacant habitat patches. This would entail
using captive breeding to produce rabbits to be released to the 17 empty habitat
patches (see Chapter 5). The number and stage of the rabbits to be introduced
would be determined by calculating carrying capacity for each patch and
assuming a stable age distribution.
Scenario #4 Disease.
(a). A relatively mild disease (e.g. tularemia) that kills some of the
infected individuals spreads through the keys. The disease kills animals in the
stages that have social contact (i.e., the nestlings and adults). The disease

151
affects 1 patch at a time, but each patch has 75% chance of getting infected
each year. Tulemia epizootics have been observed in a number of rabbit
species (Jellison 1969) and may play a role in population regulation (Woolf et
al. 1993). The rate of spread between infected populations is not known
(Woolf et al. 1993). Infected populations generally pass the disease to most,
but not all, nearby populations.
(b). A more severe disease, perhaps similar to the effects of
myxamatosis in Australia and Europe on Oryctolagus cuniculus (Chapman and
Flux 1990), spreads through the Keys. This disease also kills only the nestlings
and adults, but its mortality rate is much higher than the disease described in
(a). This disease also affects one patch at a time and with a 75% chance each
year.
Scenario #5 Hurricanes.
(a). A class 3 hurricane directly hits the center of the Lower Keys
during the first year of the simulation. The effect is regional, and all keys
suffer some mortality. Storm surge inflicts the greatest effect. Mortality is
highest for the nestlings and lowest for the adults that can swim. The
probability that a hurricane of this strength hits the Lower Keys is <10% each
year.

152
(b). Same scenario as (a), only the hurricane is class 4 and therefore the
mortality is higher. Class 4 hurricanes occur less frequently than class 3. The
probability of this strength of a hurricane is <5%.
(c). Same scenario as (a) and (b), but the hurricane is class 5 and
therefore will cause the greatest amount of mortality. This type of hurricane
has a <1% chance of occurring each year.
Scenario #6 Corridor Destruction.
(a). Corridor habitat is destroyed and this decreases migration by 25%.
All patches are affected.
(b). Corridor habitat is destroyed and decreases migration by 50%. All
patches are affected.
Scenario #7 Habitat Destruction.
(a). The largest patch (site 33, 31.5 ha) is destroyed. This patch is
privately owned and could potentially be developed.
(b). The 5 largest patches are destroyed. This would include site #33
and 4 large areas on Big Pine. This is a less likely scenario because the 4 large
areas on Big Pine curr ently are protected.

153
(c). The 10 smallest patches are destroyed. The majority of these
patches are on Boca Chica and Saddlebunch Keys. This scenario is very likely,
the patches are all owned by the Navy or privately owned.
Results
Population Parameter Estimations
The Leslie matrix was completed using estimates of survivorship and
reproduction given in Chapter 3. Survivorship of nestlings and the standard
error was 0.28 (0.17), for juveniles it was higher at 0.75 (0.10), for subadults it
was 0.67 (0.14), for first-year adults it was 0.52 (0.09) and for second-year and
older adults it was 0.10 (0.12). Reproductive rates were assumed to be the
same for first- and second- year adults. The reproductive rate was obtained by
multiplying the average female fecundity rate by the survivorship of the stage
class and then dividing by 2 to account for males. The rate for first-year adults
was 2.19 (0.07) and 1.70 (0.10) for second-year adults.
The marsh rabbit populations did not exhibit evidence of inbreeding
depression or low genetic variation. Sylvilagus palustris hefneri had a level of
genetic variation similar to large populations of cottontail (Scribner and Warren
1986); genetic effects were not factored into the model. Demographic
stochasticity did appear in the marsh rabbit populations. Several populations
were so small that at times they occasionally only supported members of the

154
same sex and these populations eventually went extinct. Demographic
stochasticity was incorporated into the model by varying survivorship and
reproduction (Ak?akaya 1994). The number of survivors for each stage was
drawn from a binomial distribution determined by the survival rate and the
smaple size. The number of young produced by each stage was drawn from a
Poisoon distriubtion.
A careful examination of the life-history parameters in each population
revealed that density-dependent growth does not appear to be occurring at the 5
populations from which the data was collected. Populations appeared to be
low due to the high mortality, and mortality was high at all of the patches,
regardless of rabbit density. However, these were collected only at Boca Chica
Key and Saddlebunch Key and could differ at the other keys. Therefore the 2
basic simulations were run, 1 with density dependence and 1 without.
Carrying capacity calculations (by site) for the density dependent
population growth ranged from a low of 3 rabbits (all stages) at the smallest
site to a high of 596 rabbits at the largest. Adding all the carrying capacities
together from all of the habitat patches, a total of 2,917 marsh rabbit (nestling
through adult) could occupy the Keys.

155
Density Estimation
Average defecation rate (R) varied between seasons, ranging from an
average of 103.75 pellets/day in July 1993 to 171.0 pellet/day in March 1991
(Table 6.2). Although the data are limited in its temporal scope, marsh rabbits
appear to produce fewer pellets in the warmer months and more in the cooler
months. Similar results were found in pellet studies of other species of rabbit
(Lord 1963).
Using the 30-day sampling period, the defecation rate from the first
survey (March 1991), and the 0.5-m-radius sample, 8 of the 13 sites on Boca
Chica had a random distribution of pellets (Table 6.3). Using the same
constants and the 1-m-radius, only 2 sites had a random distribution of pellets.
Therefore, pellet counts from the 0.5-m circle were used for estimating rabbit
densities in this and all subsequent surveys.
Of the 39 sites (Appendix C) used in the density estimation, 9 did not
appear to have been used or inhabited during the study. A comparison of
carrying capacities with estimated total abundances at the patches revealed that
over two thirds of the patches were under carrying capacity (Table 6.4). The
estimated total adult marsh rabbit population size fluctuated widely between a
low of approximately 100 individuals to a high of nearly 300 (Figure 6.1).
This estimate includes all of the sites used in the density estimate (Appendix
A), but it does not include the sites that could not be censused. Changes in the

156
density of rabbits at the largest site (site 33; Sugarloaf) had a large effect in
overall population size. These large fluctuations in the total population size did
not appear to be seasonal. The weather during the 2.5 years of the study was
normal with respect to temperature and precipitation (Figure 6.2).
Between-population and Metapopulation Parameters
Only subadults made long, one-way movements (see Chapter 4), and
75% of these successfully reached a new patch. Movement distance differed
greatly between individuals; males generally moved farther than females.
Because in RAMAS/metapop male and female dispersal can not be
distinguished, male and female dispersal was pooled. Based on the pooled
dispersal data (Chapter 4), it was assumed that if a patch was 50 m or less
away from another patch, then it could be reached by 94% of the dispersers. If
it was between 50 and 100 m, then 63% could reach it; between 100-500, then
50%. Patches 500 1000 m away could only be reached by 19% of the
dispersers and patches between 1000-2000 m away could be reached by 6%.
For patches >3000 m distant, a migration rate of <0.01 was included, to allow
for a rabbit that had superior dispersal ability.
No evidence of inter-population correlation in population size was
found using the repeated measures MANOVA. None of the correlations were
significant (P > 0.05), and the highest correlation was r = 0.65.

157
Simulation Results
The simulations using the exponential growth curve and the logistic
growth curves predicted nearly opposite outcomes for S. p. hefneri (Figure
6.3). The logistic growth curve simulation predicted that the marsh rabbit
metapopulation was stable and would vary around 1,200 rabbits (168 adults).
There were approximately 35 patches occupied during any time. The
exponential growth curve simulation estimated that S. p. hefneri would go
extinct within the next 20-30 years if all the parameters remained the same.
Because the total estimated abundances at the patches were predominantly
below carrying capacities at all of the sites and the mortality rate estimated
from the 5 main sites was high, it was determined that the exponential growth
curve was more biologically defensible. Because the exponential growth curve
more accurately mimicked population growth and was more conservative, it
was used in all of the scenarios.
Since extinction was imminent without increasing mortality or
decreasing reproduction, the sensitivity analysis only looked at increases in
survivorship and increases in reproduction. Each parameter was varied by
increasing it 25% and 50%. None of the variations appeared to stabilize the
population around a single point, however, 5 of the manipulations slowed the
extinction time to beyond 50 years (Table 6.5). Changes in the nestling
survivorship rate and the adult survival rates prolonged persistence the most.

158
Increasing the survival rate of the juveniles and subadults and increasing
reproduction did not produce large changes in the fate of the population.
In scenario #1, the increase in survivorship in all stages by 25 %
resulted in a stabilization of the predicted population size and metapopulation
occupancy (Figure 6.4). The population varied around 1,200 total individuals
occupying approximately 30 patches. When survivorship was increased by
50%, the population exponentially grew to nearly 8,000 individuals, and all
patches were eventually occupied.
In scenario #2, elimination of vehicular mortalities only prolonged the
persistence time by approximately 10 years (Figure 6.5). Similarly, in scenario
#3, when all the currently empty patches were recolonized by a carrying
capacity level of marsh rabbits in scenario #1, the population only persisted
slightly longer than without the recolonizations (Figure 6.6). The average time
to extinction was 25 years.
Because scenarios 4-7 represented further threats to the persistence of
the marsh rabbit, these simulations were run both with the original life table
and using the life history parameters in scenario #l(a) (survivorship was
increased by 25% in all stages). When scenarios 4-7 were run using the actual
life table survivorship estimates, 1 of the scenarios slightly increased the
number of years before extinction and the other 3 scenarios hastened the
amount of time before extinction (Table 6.6). Reducing migration 25% and

159
50% extended the number of years until the metapopulation went extinct from
28 years (for the basic simulation) to 30 years. A severe disease had the largest
impact on the metapopulation persistence time; in this scenario the number of
years to extinction was reduced to 3. A mild disease only decreased the
persistence time by 8 years, producing a 20-year metapopulation persistence
time. None of the hurricanes produced long-term changes in the persistence
time of the metapopulation. A class 5 hurricane reduced the persistence time to
22 years and the class 4 and class 3 hurricanes reduced the time to 23 and 25
years, respectfully. None of the habitat loss scenarios (7a-c) had an effect on
the overall persistence time of the metapopulation.
When the mortality in all of the stages was reduced by 25%, the
scenarios had different effects on the persistence of the metapopulation.
Scenario #4 examined the effect of a mild and a more severe disease had on S.
p. hefneri (Figure 6.7). Although population numbers and the mean number of
populations occupied was slightly lower than in 1(a), neither a mild or a severe
disease appeared to have a large effect on population persistence.
When all of the Keys were simultaneously subjected to a hurricane
(scenario #5), the effects on persistence were greater and persistence decreased
with increasing hurricane impact (Figure 6.8). In all of the hurricane scenarios,
population size and metapopulation occupancy decreased slightly over the 50
years.

160
Decreasing the amount of migration (scenario #6) did not alter the mean
population size, but it did increase the amount of variance in population size
(Figure 6.9). The number of occupied populations actually increased when
there was only a 25% reduction in migration. When there was a 50%
reduction, the number of occupied metapopulations was slightly below the
results for unaltered migration rate results.
In scenario #7, patches were destroyed and removed from the simulation
(Figure 6.10). When only the large site was removed (site #33), the population
decreased to approximately 400 individuals (54 adults) and fluctuated slightly
around this value. When 5 of the largest patches (including site #33) were
removed, the population size declined nearly to extinction. Removal of the 10
smallest sites produced a similar effect, only the rate of the decline was faster.
Discussion
With its small body size, short life span, high reproductive output, and
high habitat specificity, the Lower Keys marsh rabbit typifies the good
candidate for a species that might exhibit classic metapopulation dynamics
(Murphy et al. 1990, Thomas 1994). Over the time scale of this research (3
years), S. p. hefneri did appear to fit the pattern of a species that inhabited a
patchy landscape yet was able to persist due to its dispersal ability. However,

161
if the results of the simulation model are correct, the Lower Keys marsh rabbit
is doomed to extinction in the next 20-30 years.
Migration reduction, hurricanes, mild diseases, and habitat loss, will not
significantly speed the decline of the Lower Keys marsh rabbit. A severe
disease could cause the extinction of the entire metapopulation within 3 years.
Harrison (1991, 1994) suggested that most populations that appear to
exist as stable metapopulations are actually on their way to extinction because
most metapopulations do not exist in the classic metapopulation sense with
patch recolonization exceeding patch extinction over long periods of time.
Most of these doomed metapopulations are actually mainland-island
metapopulations (where a large population sustains smaller populations) or
non-equilibrium metapopulations (where there are insufficient movements
between patches). The Lower Keys marsh rabbit differs because although it
does appear to exist in the classic metapopulations structure in many ways, it is
not lack of movements between populations that threatens its persistence.
Using the simulated management technique of decreasing cat predation
(scenario #1), the model predicted that the population would fluctuate around a
stable population size and would not experience extinction during the next 50
years. Other simulated management techniques such as recolonization and
reducing vehicular mortalities were not successful in preventing extinction.
This result indicates that the cat predation on the Lower Keys marsh rabbit is a

162
deterministic threat, and until it is removed no other management techniques
will be successful.
Based on simulation, once the 25% increase in survivorship was
implemented, only large-scale habitat destruction and a veiy severe hurricane
were able to cause the rabbits extinction. House cats have probably increased
in the Lower Keys with increased human population during the past few
decades. Extinction (or severe population declines) have been often attributed
to exotic predators (e.g., domestic cat, Felis catus), although direct evidence is
difficult to obtain.
Australia (Jones and Coman 1981), New Zealand (Fitzgerald and Karl
1979), and England (Churcher and Lawton 1989) have witnessed the decline in
many species of small mammal, bird, and herpetofauna. It is believed that cats
had a large role in the decline of these species. During recent attempts to
reintroduce the ring-tailed possum (Pseudocheims forbesi). cats killed nearly
75% of the animals (Anderson 1992). In Florida, cat density was negatively
correlated with density of the endangered Anastasia Island Beach mouse (Frank
1993).
Model Validation
The conclusions made in this chapter are based on the assumption that
the model accurately predicts the future persistence of S. p. hefneri. Proper

163
model validation requires that data be available to test the predictions of the
model. Population abundance and metapopulation occupancy data are
available for 2 years, but this is not long enough to perform a rigorous test of
the model. In lieu of a validation Grant (1986) has proposed alternatives for
evaluation a PVA model without actual data. He suggests that the scenarios
used in the model be examined to determine if they accurately deal with the
management issues to be addressed and if the results are expected. Then the
parameters should be examined to determine their sensitivity to see if slight
miscalculations could have severely biased the model. Sensitivity analysis
found that none of the variables are extremely sensitive to the 25% and 50%
increases.
The major assumption of the Lower Keys marsh rabbit PVA model is
that population growth is not currently density-dependent. The data from
which this assumption was made were collected at only 6 habitat patches on 2
keys. It is possible that population growth differs at other patches and at other
keys. However, the fact that over two thirds of the populations on the Keys all
appear to be far below carrying capacity indicates that a process other than
density-dependence is in operation. It is possible that at higher survival rates,
density dependence in terms of limiting fecundity or survival at densities above
K might occur. The predicted population size of the 50% increase in marsh
rabbit survival seen in Figure 6.4 indicates that the population would increase

164
far beyond the maximum carrying capacity for the habitat. It is likely that the
Lower Keys marsh rabbit is currently at the low end of the sigmoidal curve that
the logistic equation follows, and that increase survivorship would push the
population size up the curve.
Assuming density-dependence was properly addressed in the model,
there may also be questions about the realism of the scenarios. If it were
possible to reduce the amount of cat predation in the Keys, it does not appear
that predation from other predators would increase greatly. The only
confirmed raccoon (Procyon lotor) predation occurred in the crook of a fence,
where the raccoon cornered the rabbit, and this was also in an area with
extremely high raccoon density due to the many open dumpsters in the area.
Rattlesnakes (Crotalus adamanteus), the other predator in the system, are
thought to exist at low densities in the Keys (Lazell 1989) and are often killed
while crossing roads or when they enter peoples yards.
Scenario #2, stopping mortality due to vehicles, would be difficult to
achieve, but the incorporation of this mortality into the model was
straightforward. Scenario #3, recolonization, was accurately depicted in the
model if it is assumed that marsh rabbits would only be introduced to
unoccupied sites on the major keys that are connected by US-1. Other empty
areas of transition-zone habitat may exist on smaller islands off the major keys,
and these islands are likely to be devoid of feral cats. If these keys were

165
included in the simulation, the life table for these areas might be dramatically
different. This possibility suggests that a search for unoccupied habitat on the
outer islands should be a high priority for future research (see Chapter 7).
The lack of an impact from a mild disease is highly predictable if one
takes into account the rabbits sociality and the spatial structure of the
transition-zone habitat. Marsh rabbits are mainly solitary in nature (Chapman
et al. 1982), unlike their European relatives (Oryctolagus cuniculus) which are
highly susceptible to disease. Marsh rabbits have little contact with any other
rabbits from the time they are weaned to the time they mate. If the disease was
spread through physical contact it only would affect a portion of the
population. The disease would probably not spread as fast as it did in the
simulation, unless it had another vector (e.g., ticks, mosquitoes).
The scenarios involving hurricanes might be less realistic. Each year in
the Lower Keys there is a 10% chance of a hurricane (winds >74mph) making
landfall (Chen and Gerber 1990). The chance that a hurricane hits the middle
of the Lower Keys is probably <10%.
The results of the sixth scenario, decreasing migration, also should be
treated with caution. It is difficult to predict how marsh rabbits will react if
corridor habitat is altered. In the simulation, migration was reduced by 25%
and 50%, indicating that rabbits without a good corridor habitat would not
leave their patch. This may not be true; marsh rabbits might try and disperse

166
despite the lack of habitat and then perhaps face greater mortality without
adequate food and cover. An experimental treatment of movements of Lower
Keys marsh rabbits without corridor habitat is needed to clarify this process.
When migration was decreased by 25%, the overall abundance and
number of patches occupied actually increased. When migration was cut by
50%, the overall population size appeared to increase, but the number of
occupied patches was lower than with full migration or the 25% decline. So
between the current amount of migration and a 25% reduction is the optimal
dispersal rate for maintaining metapopulation persistence.
Results from the last scenario provide insight into the importance of
individual patches to the overall metapopulation persistence. When mortality
was reduced by 25%, the elimination of the largest site simply decreased the
overall number of rabbits and the number of patches occupied. The destruction
of 5 large and 10 small patches lead to extinction. These last 2 scenarios lead
to extinction at nearly the same rate despite the fact that the 5 largest patches
contain 128.1 ha of habitat while the 10 small contain less than 10 ha. The
importance of a patch to the overall metapopulation persistence is more than
just the amount of area is encompasses. Thus, further habitat loss of small,
individual patches is a severe threat to the subspecies persistence.

167
Minimum Viable Population, Area, and Number of Patches
Using the current Leslie matrix values, it does not seem that there is any
minimum population size (Soul 1980) that could reasonably be called viable
for even short-term persistence. With the 25% increase in survivorship, the
effective population size (Ne) varied around 167 (or 334 adults if a 1:1 sex ratio
is assumed). This is above the hypothesized minimum level of 50 to avoid
inbreeding problems, but below the level of 500 individuals, which is the level
necessary to maintain genetic diversity (see Lande and Barrowclough 1987).
Minimum viable area seems less appropriate in this situation than
minimum number of patches. The number of occupied patches was seen to be
more important than the size of the patches, although the connectivity of the
patches might also be important. Judging from the results of scenario #7
(habitat loss), this minimum number of patches for metapopulation persistence
is between 20 and 30 patches spread over at least 3 keys.
These results suggest that although S. p. hefneri is probably currently
moving toward extinction, an increase in survivorship might cause the
subspecies to persist in a classic equilibrium metapopulation. For the
Lower Keys marsh rabbit, the metapopulation paradigm can form a useful
framework from which to build recovery actions.

Table 6.1 --Parameters used by stage (1 = nestling, 2 = juvenile, 3 = subadult, 4 = adultl, 5 = adult2) for RAMAS simulations involving
marsh rabbit populations under different scenarios. Reproductive rates were held constant for all scenarios.
Scenario
1
Mortality
2 3
4
5
1
Catastrophe
2 3
4
5
1
Migration
2 3
4
5
Basic (exponential)
0.28
0.75
0.67
0.52
0.10
0
0
0
0
0
0
0
0.75
0
0
#la. Predator control
leads to 25% less
mortality.
0.42
0.81
0.75
0.64
0.33
0
0
0
0
0
0
0
0.75
0
0
#lb. 50% less mortality.
0.60
0.87
0.83
0.76
0.53
0
0
0
0
0
0
0
0.75
0
0
#2. All vehicular deaths
are avoided.
0.28
0.75
0.75
0.60
0.10
0
0
0
0
0
0
0
0.75
0
0
#3. Rabbits are 0.28 0.75 0.67 0.52 0.10 000 0 0 0 0 0.75
reintroduced to all vacant
patches.
All of the following scenarios are conducted assuming that mortality for all stages has been reduced by 25%.
0
0
#4a. A disease causing
0.42
0.81
0.75
0.64
0.33
0.25
0
0
0.25
0.25
0
0
0.75
0
0
some fatalities is spread
through the keys. The
effect is local, but each
patch has a 95% chance
of getting infected each
year.
#4b. More severe 0.42 0.81 0.75 0.64 0.33 0.95 0 0 0.95 0.95 0 0 0.75 0 0
disease.

Table 6.1 --(Continued.)
Scenario
Mortality
Catastrophe
Migration
1
2
3
4
5
1
2
3
4
5
1
2
3
4
5
#5a. A class 3
hurricane has a 10%
0.42
0.81
0.75
0.64
0.33
0.75
0.50
0.25
0.25
0.25
0
0
0.75
0
chance of hitting the
keys.
#5b. Class 4.
0.42
0.81
0.75
0.64
0.33
0.95
0.75
0.50
0.50
0.50
0
0
0.75
0
0
#5c. Class 5.
0.42
0.81
0.75
0.64
0.33
0.99
0.90
0.75
0.75
0.75
0
0
0.75
0
0
#6a. Corridor habitat is
0.42
0.81
0.75
0.64
0.33
0
0
0
0
0
0
0
0.56
0
0
destroyed and decreases
migration by 25%.
#6b. migration
decreases by 50%.
0.42
0.81
0.75
0.64
0.33
0
0
0
0
0
0
0
0.38
0
0
#7. The largest patch
(site #33) is destroyed.
#7b. The five largest
patches are destroyed.
0.42
0.81
0.75
0.64
0.33
1
1
1
1
1
0
0
0.75
0
0
#7c. The 10 smallest
patches are destroyed.
0.42
0.81
0.75
0.64
0.33
1
1
1
1
1
0
0
0.75
0
0

170
Table 6.2Mean number of pellets produced by captive marsh rabbits
(Sylvilagus palustris hefneri) during a 24-hour period. Equal numbers of adult
males and females were used each session.
Session
Date
Number of
rabbits
Mean number
of pellets
SE
1
Feb 1991
4
171.0
7.43
2
Jul 1991
4
120.0
10.61
3
Nov 1991
4
132.0
13.37
4
Mar 1992
4
134.0
17.20
5
Jul 1992
4
104.8
10.50
6
Nov 1992
4
120.8
14.22
7
Mar 1993
4
107.3
10.72
8
Jul 1993
4
103.8
7.13

171
Table 6.3 Fitting the frequency distribution of the number of pellets per
sampling unit to the Poisson model for sample units using 0.5-m and 1-m radii.
The degree of correspondence is measured using a chi-square test. A
significant result indicates that the distribution is not random.
0.5-m radius 1.0-m radius
Site
Chi-square d.f.
P < 0.05
Chi-square
d.f.
P<0.
1
12.46
6
no
86.48
6
yes
2
3.88
2
no
7.48
2
yes
3
0.07
1
no
0.11
1
no
4
16.66
9
no
36.22
10
yes
5
0
0
0
0
6
2.70
3
no
7.18
2
yes
7
3.52
9
no
3.88
6
no
8
17.73
6
yes
101.31
6
yes
9
24.70
9
yes
26.79
9
yes
10
31.76
7
yes
37.68
7
yes
11
34.71
8
yes
61.58
8
yes
12
7.47
4
no
12.35
9
yes
13
60.64
9
yes
77.73
9
yes
14
0.84
4
no
8.96
3
yes

172
Table 6.4Number and proportion of habitat patches that existed above and
below carrying capacity (K). The total estimated number of Lower Keys marsh
rabbits (Sylvilagus palustris hefneri) was calculated for the 39 sites where
density estimates were conducted and was averaged over the 8 sessions. This
average was compared to K.
% larger or smaller
0-19% larger
20-39% larger
40-59% larger
60-79% larger
80-100% larger
Total larger
0-19% smaller
20-39% smaller
40-59% smaller
60-79% smaller
80-100% smaller
Number of sites in this Proportion of sites
category
4
10.0%
3
7.5%
2
5.0%
2
5.0%
0
0.0%
11
27.5%
4
10.0%
0
0.0%
1
2.5%
7
17.5%
15
37.5%
28
67.5%
Total smaller

Table 6.5Sensitivity analysis for population parameters. Each of the parameters used in the model was increased by 25% and 50% to
see how sensitive these parameters are to changes in value or errors in estimation.
Simulation
Abunance (SE)
Number of
Populations
(SE)
Years to Extinction
- 1 SE Mean + 1SE
25% more nestlings
180.3 (25.6)
11.6(0.9)
39
>50*
>50
50% more nestlings
765.8 (17.2)
26.8 (0.5)
>50
>50
>50
25% more juveniles
105.1 (31.8)
16.7(1.3)
26
38
46
50% more juveniles
344.2 (34.2)
18.1 (1.2)
32
49
>50
25% more subadults
154.1 (37.3)
9.0 (1.4)
28
39
>50
50% more subadults
169.3 (37.9)
10.1 (1.4)
28
45
49
25% more adult 1
205.3 (28.1)
23.5 (0.8)
>50
>50
>50
50% more adult 1
239.6 (32.6)
24.5 (0.7)
>50
>50
>50
25% more adult 2
173.1 (34.7)
23.1 (0.8)
>50
>50
>50
50% more adult 2
301.5 (34.8)
23.7(0.8)
>50
>50
>50
25% more reproduction
176.2 (38.3)
10.2(1.4)
30
45
>50
50% more reproduction
275.3 (42.5)
14.9(1.3)
41
>50
>50
*>50 indicates that the population size did not reach zero during the 50 year simulation.

174
Table 6.6.The number of years until the Lower Keys marsh rabbit (Sylvilagus
palustris hefneri) goes extinct as predicted by the PVA simulation model under
4 different scenarios. The mortality rate for each stage was taken from the
original life table.
Scenario Number of years until extinction
-1 SE Average +1SE
Basic
21
28
33
4a. Mild disease
15
20
25
4b. Severe disease
2
3
4
5a. Class 3 hurricane
22
25
28
5b. Class 4 hurricane
21
23
25
5c. Class 5 hurricane
20
22
24
6a. Reduce migration (25%)
21
30
34
6b. Reduce migration (50%)
19
30
34
7a. Destroy largest patch
20
27
32
7b. Destroy 5 largest patches
21
27
33
7c. Destroy 10 smallest patches
21
28
33

Estimated Number of Adults
175
Figure 6.1-The number of adult Lower Keys marsh rabbits (Svlvilagus palustris
hefneri) estimated during the 8 pellet-counting sessions (March 1991 July 1993), by
Key group (see Appendix A).

176
Year
Figure 6.2-Departures in precipitation and temperature during the past 10 years, using
the past 30 years as a base.

177
Figure 6.3 Predictions (and error bars) of the population viability analysis model for
the Lower Keys marsh rabbit (Svlvilagus palustris hefneri) using density-dependent
population growth (logistic equation) and density-independent growth (exponential
equation).

Number of Occupied Patches

179
Figure 6.4Results of Scenario #1 (Predator control) of the Lower Keys marsh rabbit
(Sylvilagus palustris hefneri) PVA model, where survivorship was increased in all
stages by 25% and 50%.

Number of Occupied Patches Total Estimated Abundance
o 10 20 30 40 50
Year
w

181
Figure 6.5Results of Scenario #2 (all vehicular deaths are avoided) of the Lower
Keys marsh rabbit (Sylvilagus palustris hefhen) PVA model, where survivorship was
increased in the subadult stage by 25% and 10% in the adult 1 stage.

Year
Number of Occupied Populations
^ ^ hO hO CO CO
ocnocnocnocn
Year
Total Estimated Abundance
ro cd co o
o o o o o
o o o o o o
1200

183
Figure 6.6Results of Scenario #3 of the Lower Keys marsh rabbit (Sylvilagus
palustris hefneri) PVA model, where marsh rabbits were reintroduced to all vacant
patches.

Year
Number of Occupied Patches
* N5 CO cn
o o o o o o
Year
Total Estimated Abundance
hO-t^COOOOI'O-^OOO
ooooooooo
oooooooooo
2000

185
Figure 6.7Results of Scenario #4 of the Lower Keys marsh rabbit (Svlvilagus
palustris hefnerO PVA model. In the mild disease simulation, 25% of all nestling and
adult rabbits died at infected patches and each patch had a 75% chance of getting
infected each year. In the severe disease simulation, 75% of all nestling and adult
rabbits died.

Number of Occupied Patches Total Estimated Abundance
1300
1200
1100
1000
900
800
0 10 20 30 40 50
Year
38
36
Populations
Severe disease
Mild disease
24
20 30
0
10
Year
40
50

187
Figure 6.8Results of Scenario #5 of the Lower Keys marsh rabbit (Sylvilagus
palustris hefneri) PVA model. The effects of a hurricane kill rabbits in each stage,
with the youngest rabbits being the most likely to die. Class 3 produces the lowest
mortality and class 5 the highest.

Number of Occupied Patches Total Estimated Number of Individuals
Year

189
Figure 6.9Results of Scenario #6 of the Lower Keys marsh rabbit (Sylvilagus
palustris hefneri) PVA model. Corridor habitat is destroyed, causing a 25% and 50%
reduction in dispersing individuals.

Number of Occupied Patches Total Estimated Number of
Year

191
Figure 6.10Results of Scenario #7 of the Lower Keys marsh rabbit (Sylvagus
palustris hefheri) PVA model. In the first simulation the largest patch (#33) is
destroyed, in the second the 5 largest patches are destroyed, and in the last simulation
10 of the smallest patches are destroyed.

Year

CHAPTER 7
CONCLUSIONS AND MANAGEMENT RECOMMENDATIONS
Population Viability Analysis (PVA) and Metapopulation Dynamics
Recently, the political and legal battle over the spotted owl (Strix
occidentalis occidentalis) and other listed species have spurred debate about
the uses of PVA (Boyce 1992, Harrison 1994). A PVA model can have vastly
different outcomes depending on the parameters and algorithms used, and
therefore might not be legally defensible. Because of these legal problems,
Harrison (1994) suggests that we should seek sensible alternatives to PVA
such as comparing the abundances of species of concern in forests (or other
habitats) of varying degrees of fragmentation. However, it is difficult to see
how correlative studies will exceed the power of a good PVA and how this will
solve the legal problems associated with endangered species.
At its worst, conducting a PVA can demonstrate where more data is
needed, and the results can be interpreted knowing the assumptions of the
model. This will provide more information than simple estimates of minimum
viable population size (MVP) and extinction probabilities associated with MVP
(Boyce 1992).
193

194
At its best, a PVA can help understand how a species interacts with its
environment and define the processes that are most important in persistence.
Combined with the scientific method, a PVA can be used in hypothesis testing
about the cause of the species decline and how different management
techniques can halt this decline. These techniques can be applied at small
scales, and then the results extended throughout the species using a PVA
model, but appropriate validation of the model should be conducted at a variety
of locations.
Recently, several population simulation models (VORTEX, Lacy 1993,
RAMAS, Akijakaya 1994) have been extended to include spatial structure in
their analysis of population viability. Most endangered species exist in highly
fragmented habitats and the spatial structure of the PVA model should include
the interchange among habitat fragments. If habitat patches are highly
independent of each other with little exchange between patches, then it may be
best to model each patch independently. If movement between patches is so
great that the rabbits exist as a single population, then the spatial structure of
the habitat patches is not an important feature to be included in the model.
However, when the species exists in a metapopulation, the spatial structure of
the patches should be included. In VORTEX and RAMAS, inter-population
movements are incorporated into these models based on measured dispersal
information or based solely on the spatial structure of the metapopulation.

195
Using spatially explicit models, the extinction and recolonization of
individual patches (metapopulation dynamics) can be simulated.
Metapopulations are affected by the same threats as single individual
populations, in addition to threats that occur at the metapopulation level. PVA
models can assess the chance of local extinctions and the chance of
metapopulation extinction.
Harrison (1991) has postulated that most endangered species that exist
in metapopulations are actually fragmented populations on the decline, as is the
case of the spotted owl (Strix occidentalis occidentalis; LaHaye et al. 1994)
and the Lower Keys marsh rabbit (Sylvilauus palustris hefneri). The transition-
zone marsh habitat of the Lower Keys marsh rabbit once existed as a
concentric ring located on each key between the mangroves and the upland
hardwood hammocks. Development has broken the ring, creating a patchy
distribution of transition zone habitat.
Few metapopulations of endangered species persist in the structure
predicted by Levins (1969, 1970) and Hanski and Gilpin (1991), where the
number of recolonized patches is greater or equal to the number of patches that
go extinct. Most endangered species occur in a fragmented landscape because
an intact landscape is not available, and creation of one is not possible.
Metapopulation dynamics and PVA can be used to determine what

196
management techniques are necessary to allow the endangered species to
persist as a metapopulation.
Management for Recovery of the Lower Keys Marsh Rabbit
A detailed recovery plan has been written for S. p. hefneri (United
States Fish and Wildlife Service 1993). Recommendations made in the
recovery plan are incorporated with the results of the PVA model presented in
this manuscript.
Primary Recovery Actions
If current mortality rates persist, it is likely that the Lower Keys marsh
rabbit will go extinct during the next 20-30 years. The first goal in the
recovery efforts should be increasing survivorship in all stages, but especially
in the nestling and adult stages.
The main source of juvenile to adult mortality, and probably nestling
mortality, is domesticated cats (Felis catus) (Chapter 3). Reducing mortality
by cats is particularly difficult because domestic cats can be difficult to catch,
poisoning or shooting cats in suburban areas could incite public outrage, and
because many of the cats in the rabbits habitat are free-ranging house cats.
The best method of cat control may be a combination of long-term
public education and trapping in marsh rabbit habitat. A long-term public

197
education program may help by increasing the number of pets that are
sterilized, decreasing the number of cats that are abandoned, and decreasing the
number of pet cats that are allowed to roam outdoors. Changing the publics
opinion on their cats outdoor-time through education has not generally been
successful in other areas (Proulx 1988).
Trapping the cats in marsh rabbit habitat and bringing cats to the Animal
Control Pound, might further reinforce cat owners to keep their cats indoors.
Other actions might include experimenting with cat bells to determine if they
work in reducing the cats hunting ability. Cat-proof fencing is not feasible in
most of the areas because most of the habitat exists in small patches and the
larger patches are shared with a suite of native species that would be negatively
impacted by the fences.
As a prerequisite to reduce overall marsh rabbit mortality and to prevent
further habitat loss, marsh rabbit habitat must be purchased and managed. A
cat trapping program will be easier to implement if the land is owned by the
state or federal government or an environmental non-government agency. As
the results of scenario #7 (habitat loss) of the PVA model demonstrated, even
small habitat patches are important to the overall persistence of S. p. hefneri.
A complete listing of land parcels for acquisition is given in the Lower Keys
marsh rabbit Recovery Plan (United States Fish and Wildlife Service 1993).
The largest parcel is covered by a Conservation and Recreative Lands (CARL)

198
proposal (#910130-44-1, Hammocks of the Lower Keys). This CARL proposal
is currently ranked at number 2 in the State of Florida and is available for
funding.
Because it will be difficult to lower the mortality from cats and to
acquire marsh rabbit habitat, additional subpopulations of marsh rabbits should
be founded in areas not occupied by domestic cats. A thorough survey of the
federally-owned back-country islands of the National Key Deer Refuge should
be conducted and marsh rabbits should be reintroduced to suitable islands.
Reintroduction to areas on Keys connected by US-1 should occur after the
number of cats has been reduced.
Secondary Recovery Actions
Even protected areas require habitat management to maintain the
maximum marsh rabbit carrying capacity of the habitat. The Lower Keys
marsh rabbit appears to rely on dense ground cover, particularly in the mid-
and high marsh (Chapter 5). Potential threats to this vegetation included exotic
vegetation, off-road vehicle (ORV) use, mowing, and trash dumping.
Australian pine (Casurina equisetifolia) and Brazilian pepper (Shinus
terebinthefolius) are rapid non-native colonizers in the transition zone that
inhibit or deter understoiy growth. These trees are particularly abundant on
Boca Chica Key, but are present on all of the Keys. Girdling trees with

199
applications of GARLON has been found to be a fast, effective method of
killing these exotics.
Off-road vehicle traffic (ORV) compacts soil, destroys vegetation and
increases erosion. Off-road vehicle traffic through marsh rabbit habitat can be
stopped by preventing access to habitats from the road. Local education
programs involving law enforcement and military police officers have been
helpful in preventing ORV use of wetlands. Deterring ORV use of wetlands
also decreases the amount of illegal dumping.
Restoration of vegetation in areas that have been badly damaged by
ORV use, dumping, or other land uses may be needed at some areas.
Restoration, although costly, may increase the total amount of habitat available
to the marsh rabbit. A trial restoration project has been initiated at site #10 on
Boca Chica.
In addition to protecting habitat, protection of marsh rabbit dispersal
corridors is also important in preserving the subspecies. Marsh rabbits will
move through any habitat that is not disturbed (Chapter 4). Future
development plans in the Keys should be evaluated for their impact on corridor
habitat.
Although cats are the main predator of the marsh rabbit, reducing other
sources of mortality may significantly increase the marsh rabbit population.
Raccoons (Procvon lotor) are not as great of a threat to the persistence of S. p.

200
hefneri (Chapter 3) as cats, but raccoon numbers may be increased locally in
areas with open dumpsters. Providing top-closing lid dumpsters might
decrease the numbers of raccoons. Other native species such as Bald Eagles
(Haliaeetus leucocephalus) and Rattlesnakes (Crotalus adamanteus) and exotic
species (e.g. dogs, Canis familiaris) are not seen as being a threat to Lower
Keys marsh rabbits survival.
A significant number of rabbits are killed by vehicles (Chapter 3).
While it is not likely to prevent all of these mortalities, the number of deaths
might be reduced if speed limits were enforced. The most important time to
enforce speedlimits is from dusk to dawn, when the rabbits are most active.
Although undocumented, there may be additional sources of mortality in
the nestling stage. At this stage, rabbits are extremely altricial and perhaps
more susceptible to some types of predation than older rabbits. Fire, while
fairly rare in the transition zone, might have a impact on younger rabbits that
are incapable of escaping. Overall, fire probably has a positive effect on the
rabbit because of its effect on understory vegetation.
Mowing, while not having a large impact on older rabbits that can run
away, might have a large effect on nestlings. Mowing may additionally change
the composition and diversity of the plants in the marsh rabbit habitat.
Red imported fire ants (Solenopsis invicta! have recently colonized the
Keys and are abundant in some of the transition zone areas (Forys pers. obs.).

201
These fire ants are attracted to the mucous found on young rabbits and were a
major source of mortality in Eastern cottontail young (Sylvilagus floridanus;
Hill 1969, 1972).
Future Research
Annual population monitoring will be important in validating the PVA
model and determining the effectiveness of management techniques. The
results of the PVA simulation (Chapter 6) showed that the number of occupied
patches was similar to the total number of rabbits. Censusing areas for
presence or absence of marsh rabbit pellets is a fast and efficient method of
determining patch occupancy. Presence/absence data can be obtained for a site
in less than one hour. Using the methods reported in this study, pellet counting
for abundances uses 4 hours per site and live-trapping takes 20 hours for each
site.
Further research on the possibility of reintroducing Lower Keys marsh
rabbits also should be explored. Potential reintroduction sites first need to be
evaluated for habitat suitability and then the ability of rabbits to interact with
the greater metapopulation should be measured. Routine translocations of
marsh rabbits might be necessary to provide genetic interchange with the
mainland Keys populations.

202
Other future research should be aimed at decevising means of
decreasing the mortality rate of the marsh rabbits. With all research projects
small experimental treatments should be applied first, their effects incorporated
into the PVA model, and then successful management techniques should be
used throughout the Lower Keys.

APPENDIX A
DESCRIPTIVE INFORMATION ABOUT THE HABITAT PATCHES AND MAPS
Each transition zone site in the Lower Keys of Florida is listed by the Key it occurs
on, the area of the habitat patch, the ownership, and whether the patch was included in the
density estimates. Maps with each site located by number follow this appendix.
Key
Site #
Area
Ownership*
Density
Boca Chica
1
1.05
USN
Yes
Boca Chica
2
1.22
USN
Yes
Boca Chica
3
1.62
USN
Yes
Boca Chica
4
2.92
USN
Yes
Boca Chica
5
2.43
USN
Yes
Boca Chica
6
2.75
USN
Yes
Boca Chica
7
3.89
USN
Yes
Boca Chica
8
3.90
USN
Yes
Boca Chica
9
4.86
USN
Yes
Boca Chica
10
2.27
USN
Yes
Boca Chica
11
5.18
USN
Yes
Boca Chica
12
2.33
USN
Yes
Boca Chica
13
3.40
USN
Yes
Boca Chica
14
5.83
USN
Yes
Boca Chica
15
1.15
USN
Yes
Boca Chica
16
0.69
USN
Yes
Boca Chica
17
1.04
USN
Yes
Boca Chica
18
0.55
USN
No
Boca Chica
19
1.50
USN
No
Boca Chica
20
1.00
USN
No
Boca Chica
21
3.00
USN
No
Boca Chica
22
0.30
USN
No
Boca Chica
23
0.50
USN
No
Boca Chica
24
0.50
USN
No
Boca Chica
25
1.56
?
Yes
Boca Chica
26
0.45
USN
No
Boca Chica
27
0.30
?
No
203

204
Key
Site #
Area
Ownership*
Density
Saddlebunch
28
3.50
USFWS
Yes
Saddlebunch
29
2.50
USN
Yes
Saddlebunch
30
0.78
USN
Yes
Sugarloaf
31
1.00
P
Yes
Sugarloaf
32
4.00
P
Yes
Sugarloaf
33
31.50
P
Yes
Sugarloaf
34
2.30
P
Yes
Sugarloaf
35
0.86
P
No
Sugarloaf
36
0.57
P
No
N. Saddlebunch
37
0.58
Monroe
No
N. Saddlebunch
38
0.50
P
No
N. Saddlebunch
39
0.55
P
No
N. Saddlebunch
40
0.69
P
No
Cudjoe
41
2.60
P
Yes
Cudjoe
42
3.90
Air
Yes
Cudjoe
43
5.50
P
Yes
Summerland
44
2.34
P
Yes
Ramrod
45
3.90
P
Yes
Middle Torch
46
1.00
USFWS
No
Middle Torch
47
1.73
USFWS
No
Big Torch
48
0.50
P
No
Big Torch
49
0.88
USFWS
No
Big Torch
50
1.44
P
Yes
Little Torch
51
3.00
TNC
Yes
Big Pine
52
4.45
USFWS
No
Big Pine
53
19.50
USFWS
Yes
Big Pine
54
43.70
SFWMD
Yes
Big Pine
55
27.60
SFWMD
Yes
Big Pine
56
1.10
P
Yes
Big Pine
57
1.00
P
Yes
Big Pine
58
2.93
USFWS
Yes
Noname
59
1.50
USFWS
Yes
*USN United States Navy, P = privately owned, USFWS = United States Fish and
Wildlife Service, Monroe = Monroe County, Air = United States Airforce, TNC = The
Nature Conservancy.


206
Figure A.2--A map of Saddlebuch and Lower Sugarloaf Keys. The habitat patches used
the study are numbered.

207
Figure A.3A map of Sugarloaf Key. The habitat patches used in the study are numbered.

208
Figure A.4A map of Cudjoe Key. The habitat patches used in the study are numbered.

209
Figure A. 5A map of Summerland and Ramrod Keys. The habitat patches used in the
study are numbered.

210
Figure A.6--A map of Big Torch and Middle Torch Keys. The habitat patches used in the
study are numbered.

211
Figure A. 7A map of Big Pine Key. The habitat patches used in the study are numbered.

212
Figure A.8--A map of Noname Key. The habitat patches used in the study are numbered.

APPENDIX B
VARIABLES USED IN THE DISCRIMINANT FUNCTION ANALYSIS

214
Table B.lThe variables used in the descirminant function analysis.
site
code
%gcover
%ccover
H'
hgt
%bf
%clump
area
dwat
dpeop
isol
1
2
71
30
2.05
42.1
3
46
1.05
1
270
40
2
2
64
28
2.17
34.1
2
37
1.22
108
90
40
3
1
44
19
1.24
20.2
0
4
1.62
90
540
376
4
2
77
25
1.44
15.0
15
20
2.92
1
2700
166
5
1
35
5
0.93
10.0
2
5
2.43
18
660
153
6
1
70
35
1.76
15.0
7
2
2.75
1
2880
100
7
2
72
19
1.14
18.4
1
27
3.89
135
1170
213
8
2
57
8
2.02
22.9
20
19
3.9
135
1710
124
9
2
78
20
1.25
19.4
2
30
4.86
180
1476
163
10
2
83
6
1.04
41.5
3
57
1.27
90
900
333
11
2
58
9
1.21
42.5
20
18
5.18
90
1
51
12
2
45
4
1.27
20.4
3
25
2.33
135
1710
124
13
2
55
8
1.33
40.2
15
32
3.40
98
1
51
14
2
60
6
1.31
28.5
3
35
5.83
136
1836
190
15
1
32
20
0.67
19.0
4
13
1.15
1
720
318
16
1
47
8
1.28
15.9
29
3
0.69
1
1680
153
17
1
35
14
1.49
12.1
7
2
1.04
1
2160
628
18
1
70
10
1.55
20.5
33
4
0.55
60
600
168
19
1
62
3
0.96
18.9
2
15
1.50
110
550
280
20
1
51
13
1.15
17.8
5
10
1.00
1
575
259
21
2
52
12
1.45
24.6
7
22
3.00
1
470
232
22
1
51
5
1.32
22.0
7
15
0.30
90
480
211
23
1
41
3
1.27
21.0
5
9
0.50
130
870
226
24
1
46
15
1.7
26.2
4
7
0.50
250
540
251
25
2
61
19
1.81
13.9
0
52
1.56
1
1500
353
26
2
68
25
1.2
36.6
12
36
0.45
150
1100
405
27
1
55
19
0.91
29.6
2
16
0.30
10
1560
483
28
1
93
18
1.56
9.1
10
11
3.50
110
6000
2124
29
1
50
9
1.42
20.3
1
17
0.78
500
2000
403
30
2
98
12
1.85
7.2
2
46
2.50
100
3000
472

215
Table B. 1-Continued.
site
code
%gcover
%ccover
H'
hgt
%bf
%clump
area
dwat
dpeop
isol
31
2
70
4
1.6
6.5
7
30
1.0
1
300
383
32
2
56
5
1.38
17.0
10
39
4.0
100
1000
383
33
2
98
5
1.12
62.0
8
87
31.50
1
2000
1800
34
1
37
13
1.57
15.0
4
10
2.30
1
1
832
35
1
60.4
12
1.84
15.0
3
11
0.86
1
700
927
36
0
74.4
15
1.47
7.6
1
1
0.57
1
624
1289
37
0
42
4
1.09
28.3
0
29
0.58
216
1
2603
38
0
59
2
0.71
40.0
0
52
0.50
200
1
2274
39
0
53
4
0.55
49.2
0
44
0.55
180
1
2511
40
0
59
18
0.69
30.5
0
0
0.69
80
1
2497
41
0
65
5
0.84
55.6
0
56
2.60
1000
1
4273
42
0
50
8
1.13
8.0
2
2
3.90
1
3000
5272
43
0
30
29
1.57
7.0
3
16
5.50
1
1
4707
44
0
96
4
1.17
60.2
0
45
2.34
125
300
6066
45
0
57
17
1.57
14.0
4
8
3.90
1
1
4746
46
0
87
3
0.91
45.2
0
63
1.00
120
480
3011
47
0
87
16
1.67
30.5
60
5
1.73
1
1
2026
48
0
72
22
2.07
20.0
15
15
0.50
72
120
3498
49
0
76
12
1.43
40.8
0
50
0.88
48
1920
4244
50
0
73
28
1.46
16.5
20
10
1.44
72
1
4749
51
0
47
81
1.64
19.3
1
6
3.00
1
100
3442
52
1
61
23
2.04
26.0
13
13
4.45
1
1
765
53
2
70
14
1.78
15.5
0
26
19.5
375
750
1772
54
1
55
30
1.16
15.0
0
30
43.7
96
1
1250
55
1
44
34
1.38
12.5
1
6
27.6
1
1
1467
56
1
38
18
1.35
25.0
3
27
1.10
50
100
1507
57
2
63
28
1.91
19.5
5
45
1.00
400
1
1811
58
1
72
35
1.46
17.4
8
5
2.93
313
780
757
59
0
87
47
1.76
15.6
6
60
1.50
1
500
1542

APPENDIX C
MARSH RABBIT DENSITY ESTIMATES

Table C.lEstimates of rabbit densities and numbers based on pellet counts in the Lower
Keys. (Session 1 = March 1991, Session 2 = July 1991, Session 3 = November 1991,
Session 4 = March 1992, Session 5 = July 1992, Session 6 = November 1992, Session 7 =
March 1993, Session 8 = July 1992).
Session 1
Session 2
Session 3
Session 4
Site
Rab/ha Total
Rab/ha Total
Rab/ha Total
Rab/ha Total
1
2.02
3.05
1.03
1.08
0.61
0.64
3.59
3.77
2
0.22
0.27
1.01
1.23
0.18
0.22
1.32
1.61
3
0.08
0.13
0.28
0.45
0.00
0.00
0.07
0.11
4
1.86
5.43
1.4
4.23
0.74
2.14
1.83
5.34
5
0.00
0.00
0.01
0.02
0.15
0.37
0.66
1.61
6
0.67
1.85
2.23
6.13
0.00
0.00
0.05
0.14
7
0.74
2.87
0.60
2.33
0.53
2.07
2.16
8.40
8
7.63
29.66
6.21
24.16
3.44
13.39
2.85
11.10
9
1.49
7.22
0.62
3.03
0.58
2.83
2.75
13.34
10
3.97
9.01
2.70
6.13
2.28
5.17
2.24
5.08
11
4.65
24.09
2.54
13.16
0.38
1.97
0.77
4.00
12
0.74
1.71
1.91
4.45
3.01
7.02
3.87
9.01
13
8.99
30.58
4.37
14.86
2.28
7.74
1.58
5.36
14
0.31
1.83
0.85
4.96
2.02
11.80
0.56
3.26
15
1.81
2.08
0.13
0.14
16
5.64
3.89
2.24
1.55
17
2.53
2.63
0.17
0.18
26
0.20
0.31
0.14
0.22
0.41
0.63
0.35
0.54
28
0.00
0.00
0.00
0.00
0.00
0.00
0.05
0.18
29
0.00
0.00
0.00
0.00
0.05
0.13
30
1.20
0.94
0.37
0.29
1.27
0.99
0.20
0.16
31
1.14
1.14
32
0.25
1.00
33
5.20
163.80
3.90
122.85
4.05
127.58
0.50
15.75
34
0.25
0.58
1.37
3.15
41
0.00
0.00
0.00
0.00
0.00
0.00
42
0.00
0.00
0.00
0.00
0.00
0.00
43
0.00
0.00
0.00
0.00
0.00
0.00
44
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
45
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
50
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
51
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
53
0.17
3.32
0.13
2.54
0.76
14.80
0.91
17.82
56
0.28
0.31
0.51
0.56
0.97
1.07
57
0.13
0.10
0.15
0.12
0.51
0.51
58
0.67
1.96
0.40
1.17
0.25
0.74
0.00
0.00
59
0.00
0.00
0.00
0.00
0.00
0.00
217

Table C.l--Continued.
C
-C
JD
9
PC
-2
o
H
cS
Xj
X
9
Pi
o
H
cd
a
pc
OP rS
13
Pi
0)
B
M VO in ^
m
cc
oo
Tf
Tf
Q
Q
VO
r-
NO
in
o
o
vq
q
q
vq
(N
H
q
q
o
in
in
vq
NO
cn
in
q
d
d
(N
Ov
d
d
d
d
^H
d
d
d
d
d
00*
CN
in
in
in
ON
o
Q
On
(N
NO
cn
(N
cc
O
q
cc
q
CN
ON
o
O
On
q
vq
oo
cn
ON
d
d
d
d
d
d
d
d
d
O
d
C'i
d
d
d
d
8888888888£P;S
00
vo
H
00
Q
cn
ON
in
oo
CN
00
-h
e'
q
Tf
o
m
in
*h
in
o
On
in
q
ON
ON
O
tj*
q
q
q
On
q
q
en
o
q
q
(N
q
00
cn
q
q
cn
d
d
d
d
d
d
in
d
On
in
d
d
d
vo
d
d
d
d
d
d
d
O
d
d
cn
o
o
NO
VO
o
VO
(N
m
*fr
o
VO
o
o
o
o
(N
tJ-
VO
in
(N
in
cc
q
in
q
q
q
q
q
in
00
q
q
o
oo
vo
oo
iH
cn
q
q
,-H
T-H
d
d
d
,H
d
d
d
d
d
d
d
d
d
d
d
d
d
d
d
d
cc
Q
ON
Q
o
ON
ON
in
n
cn
o
cn
m
o
m
r-
vo
o
VO
cn
q
o
q
q
o
in
q
q
q
q
q
in
q
q
o
q
in
tj-
o
00
in
d
d
d
d
d
d
d
d
d

in
d
in
d
d
d
d
d
d
d
d
d
VO
d
*'^Q'OOO^TfaNTfoom-HinQooommaNr^r^^mr^ooOQQQQ^ooNOO
cnooqqqqqvoqqqqqqqqinqqqqc>o(N^,o
dddddddocNddddoddddoddddddddddddddddd
00
00
(N
oo
n
c-

,-H

o
o
8
o
(N
o
oo
00
o
q
q
vo
q
q
q
cn
q
00
cn
o
VO
q
vo
q
cn
q
q
d
d
d
d
On
d
d
d
o
in
d
d
d
d
d
C'i
d
d
d
d
VO
oooooocscnvooo
(N
Tf
r-
ON
n
cn
ON
cn
o
m
00
8
8

oo
o
c-
8
oo
in

in
cn
q
q
q
T"
q
q
re
oo
q
cn
q
On
q
in
n
ih
o
q
cc
q
d
d
d
d
d
d
d
d
d
d
O
d
d
O
d
d
d
d
d
d
d
d
n
d
(Nmrj-invor^oooN
fSn^iniOhioooo\OH(S(f|TtH(snTtioOHfniohooo
218

APPENDIX D
MARSH RABBIT PATCH OCCUPANCY

220
Table D. 1Presence of adult and juvenile Lower Keys marsh rabbits (Svlvilagus palustris hefneri) during the last 6 sessions.
Site
code
Adult3
Juv.3
Adult4
Juv.4
Adult4
Juv.j5
Adult6
Juv.6
Adult7
Juv.7
Adult8
Juv.8
1
1
1
0
1
0
1
0
1
0
1
0
1
1
2
2
1
0
1
0
1
1
1
0
1
0
1
1
3
1
0
0
1
0
1
0
0
0
1
1
1
0
4
2
1
0
1
0
1
0
1
1
1
0
1
1
5
1
1
0
1
1
1
0
0
0
0
0
0
0
6
1
0
0
1
0
1
0
0
0
1
0
1
0
7
2
1
1
1
1
1
0
1
0
1
0
1
1
8
2
1
1
1
1
1
0
1
1
1
1
1
1
9
2
1
0
1
1
1
0
1
0
1
0
1
0
10
2
1
0
1
0
1
0
1
0
1
0
1
1
11
2
1
0
1
0
1
0
1
0
1
0
1
0
12
2
1
1
1
0
1
0
1
0
1
1
1
1
13
2
1
0
1
0
1
0
1
1
1
0
1
1
14
2
1
0
1
0
1
0
1
0
1
0
1
1
15
1
1
0
1
0
1
0
1
0
1
0
1
0
16
1
1
0
1
1
1
0
1
0
1
1
0
0
17
1
1
1
1
0
1
0
1
0
0
0
0
0
18
1
0
0
1
0
0
0
1
0
1
1
1
0
19
1
0
0
0
0
0
0
0
0
1
0
1
1
20
1
0
0
0
0
0
0
0
0
1
1
1
0
21
1
0
0
0
0
0
0
1
0
0
0
1
0
22
1
1
0
0
0
0
0
0
0
1
1
0
0
23
1
1
1
1
0
1
0
0
0
0
0
0
0
24
1
0
0
0
0
0
0
1
0
1
1
1
0
25
1
1
0
1
0
1
0
1
0
1
0
1
0
26
1
0
0
1
0
1
1
1
1
1
0
1
1
27
1
0
0
0
0
0
0
1
0
0
0
0
0
28
1
0
0
1
0
1
1
1
0
1
1
1
0
29
1
0
0
1
0
1
1
1
1
1
0
1
1
30
2
1
1
1
0
1
0
1
0
1
0
1
1

Table D. 1--Continued.
Site
code
Adult3
Juv.3
Adult4
Juv.4
Adult4
Juv.j5
Adult6
Juv.6
Adult7
Juv.7
Adult8
Juv.8
31
2
1
0
1
0
1
0
1
0
1
0
1
1
32
2
1
0
1
0
1
0
1
0
1
0
1
0
33
2
1
1
1
1
1
1
1
0
1
1
1
1
34
0
1
0
0
0
0
0
0
0
0
0
0
0
35
1
0
0
0
0
0
0
0
0
0
0
1
0
36
0
0
0
0
0
0
0
0
0
0
0
0
0
37
0
0
0
0
0
0
0
0
0
0
0
0
0
38
0
0
0
0
0
0
0
0
0
0
0
0
0
39
0
0
0
0
0
0
0
0
0
0
0
0
0
40
0
0
0
0
0
0
0
0
0
0
0
0
0
41
0
0
0
0
0
0
0
0
0
0
0
0
0
42
0
0
0
0
0
0
0
0
0
0
0
0
0
43
0
0
0
0
0
0
0
0
0
0
0
0
0
44
0
0
0
0
0
0
0
0
0
0
0
0
0
45
0
0
0
0
0
0
0
0
0
0
0
0
0
46
0
0
0
0
0
0
0
0
0
0
0
0
0
47
0
0
0
0
0
0
0
0
0
0
0
0
0
48
0
0
0
0
0
0
0
0
0
0
0
0
0
49
0
0
0
0
0
0
0
0
0
0
0
0
0
50
0
0
0
0
0
0
0
0
0
0
0
0
0
51
0
0
0
0
0
0
0
0
0
0
0
0
0
52
1
0
0
0
0
0
0
0
0
0
0
1
0
53
2
1
1
1
1
1
0
1
0
1
1
1
0
54
1
1
0
1
1
1
0
0
0
0
0
0
0
55
1
1
0
1
0
0
0
0
0
0
0
0
0
56
1
1
0
1
1
1
0
0
0
0
0
0
0
57
2
1
1
1
1
1
0
1
1
1
0
1
1
58
1
1
0
0
0
0
0
0
0
1
0
1
1
59
0
0
0
0
0
0
0
0
0
0
0
0
0

REFERENCES
Ackerman, B. B., F. Leban, M. D. Samuel, and E. O. Garton. 1990. User's Manual
for Program Home Range, 2nd edition. Department of Fisheries and Wildlife
Resources, University of Idaho, Moscow. Forest, Wildlife and Range
Experiment Station, Technical Report 15, 79 pp.
Adler, G. H. and M. L. Wilson. 1987. Demography of a habitat generalist, the
white-footed mouse, in a heterogeneous environment. Ecology 68:1785-
1796.
Aguire, A. A. and E. A. Starkey. 1994. Wildlife diseases in U.S. National Parks:
Historical and coevolutionary perspectives. Conservation Biology 8:654-
661.
Akfakaya, H. R. 1994. RAMAS/metapop: Viability analysis for stage-structured
metapopulations. Applied Biomathematics, Setauket, New York. 185 pp.
Alldredge, J. R. and J. T. Ratti. 1986. Comparison of some statistical techniques
for analysis of resource selection. Journal of Wildlife Management 50:157-
165.
Anderson, I. 1992. Alien predators devastate Australian wildlife. New Scientist
135:9.
Anderson, T. E. 1969. Identifying, evaluating, and controlling wildlife damage.
Pages 497-520 in Wildlife management techniques, 3rd edition (R. H. Giles,
ed.), The Wildlife Society, Washington, D.C.
Anderson, P. K. 1989. Dispersal in rodents: A resident fitness hypothesis.
American Society of Mammalogists, Special Publication Number 9. 141 pp.
ARC/INFO, Environmental Systems Research Institute, 1990. ARC/info Users
Guide, Environmental Systems Research Institute, Redlands, California. 124
pp.
222

223
Barthouse, L. W., Boreman, J., Christensen, S. W., Goodyear, C. P. VanWinkle,
W., Vaughan, D. S. 1984. Population biology in the courtroom: The
Hudson River controversy. Bioscience 34:14-19.
Batzli, G. O. and F. A. Pitelka. 1971. Condition and diet of cycling populations of
the Californian vole. Journal of Mammalogy 52:141-163.
Begon, M., Harper, J. L. and C. R. Townsend. 1990. Ecology: Individuals,
populations, and communities. Sinauer Associates, Inc., Sunderland,
Massachusetts. 876 pp.
Belovsky, G. E. 1987. Extinction models and mammalian persistence. Pages 35-
59 in Viable populations for conservation, (M. E. Soul ed.), Cambridge
University Press, Cambridge, Great Britain.
Berg, H. C. 1983. Random walks in biology. Princeton University Press,
Princeton, New Jersey 142 pp.
Bergin, T. M. 1992. Habitat selection by the Western Kingbird in Western
Nebraska: a hierarchical analysis. Condor 94:903-911.
Berry, R. J. 1986. Genetics of insular populations of mammals, with particular
reference to differentiation and founder effects in British small mammals.
Biological Journal of the Linnean Society 28:205-230.
Blair, F. W. 1935. An ecological investigation of the marsh rabbit of the region of
Gainesville, Florida. M. S. thesis, University of Florida, Gainesville. 92 pp.
Blair, F. W. 1936. The Florida marsh rabbit. Journal of Mammalogy 17:197-207.
Blake, J. G. and J. R. Karr. 1987. Breeding birds of isolated woodlots: Area and
habitat relationships. Ecology 68:1724-1734.
Bothnia, J. du P. and J. G. Teer. 1977. Reproduction and productivity in South
Texas cottontail rabbits. Mammalia 41:254-281.
Bothnia, J. du P., J. G. Teer, and C. E. Gates. 1972. Growth and age determination
of the cottontail in South Texas. Journal of Wildlife Management 36:1209-
1210.
Boyce, M. S. 1992. Population viability analysis. Annual Review of Ecology and
Systematics 23:481-506.

224
Brown, J. H. 1971. Mammals on mountain tops: nonequilibrium insular
biogeography. American Naturalist 105:467-478.
Brown, J. H. and A. Kodric-Brown. 1977. Turnover rates in insular biogeography:
effects of immigration on extinction. Ecology 58:445-449.
Buechner, M. 1987. Conservation in insular parks: simulation models of factors
affecting the movement of animals across park boundaries. Biological
Conservation 41:57-76.
Burgman, M. A., Ferson, S. and H. R. Akfakaya. 1993 Risk assessment in
conservation biology. Chapman and Hall, London. 314 pp.
Burt, W. H. 1943. Territoriality and home range concepts as applied to mammals.
Journal of Mammalogy, 24:346-352.
Byers, C. R., Steinhorst, R. K. and P. R. Krausman. 1984. Clarification of a
technique for analysis of utilization-availability data. Journal of Wildlife
Management 48:1050-1053.
Canfield, R. H. 1941. Application of the line intercept method in sampling range
vegetation. Journal of Forestry 39:388-394.
Capen, D. E., Fenwich, J. W., Inkley, D. B. and A. C. Boynton. 1986. Multivariate
models of songbird habitat in New England forests. Pages 171-176 in
Wildlife 2000: Modeling habitat relationships of terrestrial vertebrates (J.
Vemer, M. L. Morrison, and C. J. Ralph, eds.), University of Wisconsin
Press, Wisconsin.
Carlson, P. C., Tanner, G. W., Wood, J. M., and S. R. Humphrey. 1993. Fire in
Key deer habitat improved browse, prevents succession, and preserves
endemic herbs. Journal of Wildlife Management 57:914-928.
Carr, A. F., Jr. 1939. Notes on the escape behavior in the Florida marsh rabbit.
Journal of Mammalogy 20:322-325.
Caughley, G. 1994. Directions in conservation biology. Journal of Animal Ecology
63:215-244.
Cervantes, F. A. and J. Martinez. 1992. Food habits of the rabbit Romerolagus
diazi (Leporidae) in Central Mexico. Journal of Mammalogy 73:830-834.

225
Chapman, J. A. 1971. Orientation and homing of the brush rabbit. Journal of
Mammalogy 52:686-699.
Chapman, J. A. and G. Ceballos. 1990. The cottontails. Pages 95-110 in Rabbits,
hares and pikas: Status survey and conservation action plan (J. A. Chapman
and J. E. C. Flux, eds.). IUCN, Gland, Switzerland.
Chapman, J. A., Cramer, K. L., Dippenaar, N. J. and T. J. Robinson. 1992.
Systematics and biogeography of the New England cottontail Svlvilagus
transitionalis (Bangs, 1895), with the description of a new species from the
Appalachian mountains. Proceedings of the Biological Society of
Washington 105:841-866.
Chapman, J. A. and J. E. C. Flux. 1990. Introduction and overview of the order
Lagomorpha. Pages 1-6 in Rabbits, hares and pikas: Status survey and
conservation action plan (J. A. Chapman and J. E. C. Flux, eds.). IUCN,
Gland, Switzerland.
Chapman, J. A., Hannan, A. L. and D. E. Samuel. 1977. Reproductive and
physiological cycles in the cottontail complex in western Maryland and
nearby West Virginia. Wildlife Monograph 56:1-73.
Chapman, J. A., Hockman, J. G. and M. M. Ojeda. 1980. Svlvilagus floridanus.
Mammalian Species 136:1-8.
Chapman, J. A., Hockman, J. G. and W. R. Edwards. 1982. Cottontails. Pages 83-
123 in Wild mammals of North America: Biology, management, economics
(J.A. Chapman and G.A. Feldhamer, eds.) The Johns Hopkins University
Press Baltimore, Maryland.
Chapman, J. A. and R. P. Morgan. 1973. Systematic status of the cottontail
complex in western Maryland and nearby West Virginia. Wildlife
Monographs 36:1-54.
Chapman, J. A. and J. R. Stauffer. 1981. The status and distribution of the New
England cottontail. Pages 973-983 in Proceeding of the world lagomorph
conference (K. Myers and C. D. Maclnnes, eds), University of Guelph,
Guelph, Ontario, Canada.

226
Chapman, J. A. and D. E. C. Trethewey, 1972. Movements within a population of
introduced eastern cottontail rabbits. Journal of Wildlife Management
36:155-158.
Chapman, J. A. and G. R. Willner. 1981. Svlvilagus palustris. Mammalian Species
153:1-3.
Cheeke, R. 1987. Rabbit feeding and nutrition. Academic Press, Inc., Orlando,
Florida. 376 pp.
Chen, E. and J. F. Gerber. 1990. Climate. Pages 11-34 in Ecosystems of Florida,
(R. L. Myers and J. J. Ewel, eds.), University of Central Florida Press,
Orlando, Florida.
Churcher, P. B. and J. H. Lawton. 1989. Beware of well-fed felines. Natural
History 7:40-46.
Cody, M. L. 1985. Habitat selection in birds. Academic Press, London. 558 pp.
dePourtales, L. F. 1877. Hints on the origin of the flora and fauna of the Florida
Keys. American Naturalist 11:137-144.
Diamond, J. M. 1975. The island dilemma: lessons of modem biogeographic
studies for the design of nature reserves. Biological Conservation 7:129-146.
Diamond, J. M. 1984. "Normal" extinctions of isolated populations. Pages 191-
246 in Extinctions (M. N. Nicety, ed.), University of Chicago Press,
Chicago.
Diamond, J. M. 1989. Overview of recent extinctions. Pages 37-41 in
Conservation for the Twenty-first Century (D. Western and M. Pearl, eds.),
Oxford University Press, New York.
Diamond, J. M. and A. C. Marshall. 1977. Niche shifts in New Hebridean birds.
Emu 77:61-72.
Diamond, J. M. and R. M. May. 1977. Island biogeography and the design of
nature reserves. Pages 228-252 in Theoretical ecology (R. M. May, ed.).
Sinauer Associates, Sunderland, Massachusetts.
Dixon, K. R., and J. A. Chapman. 1980. Harmonic mean measure of animal
activity areas. Ecology 61:1040-1044.

227
Dixon, K. R., Chapman, J. A., Rongstad, O. J. and K. M. Orhelein. 1981. A
comparison of home range size in Sylvilagus floridanus and S. bachmani.
Pages 541-548 in Proceedings of the world lagomorph conference (E. Myers
and C. D. Machines, eds.), Guelph Univ. Press, Guelph, Ontario.
Dobson, F. S. 1982. Competition for mates and predominant juvenile dispersal in
mammals. Animal Behaviour 30:184-200.
Dobson, F. S. and W. T. Jones 1985. Multiple causes of dispersal. American
Naturalist 126:855-858.
Eberhardt, L. L. 1987. Population projections from simple models. Journal of
Applied Ecology 24:103-188.
Edwards, W. R., Havera, S. P., Labisky, R. F., Ellis, J. A. and R. E. Warner. 1981.
The abundance of cottontails in relation to agricultural land use in Illinois
(U.S.A.) 1956-1978, with comments on mechanisms of regulation. Pages
761-798 in Proceedings of the world lagomorph conference (K. Meyers and
C. D. Macinnes, eds), University of Guelph, Ontario.
Eisenberg, J. F. 1981. The mammalian radiations: An analysis of trends in
evolution, adaptation and behavior. University of Chicago Press, Chicago,
Illinois. 510 pp.
Fahrig, L. and G. Merriam. 1985. Habitat patch connectivity and population
survival. Ecology 66:1762-1768.
Falconer, D. S. 1989. Introduction to quantitative genetics. 2d ed. Longman Press,
New York. 438 pp.
Festa-Bianchet, M. and W. J. King. 1984. Behaviour and dispersal of yearling
Colombian ground squirrels. Canadian Journal of Zoology 62:161-167.
Fitzgerald, B. M. and B. J. Karl. 1979. Foods of feral house cats (Telis catus L.) in
forest of the Orongorongo Valley, Wellington. New Zealand Journal of
Zoology 6:107-126.
Frank, P. A. 1993. Anastasia Island beach mouse at home at Fort Matanzas
National Monument. Park Science 13:30-31.

228
Fuerst, P. A. and T. Maruyama. 1986. Considerations on the conservation of
alleles and genic heterozygosity in small managed populations. Zoo Biology
5:171-180.
Gaines, M. S., Foster, J., Diffendorfer, J. E., Sera, W. E., Holt, R. D. and G. R.
Robinson. 1992. Population processes and biological diversity.
Transactions of the North American Wildlife and Natural Resources
Conference 57:252-262.
Gallagher, D. 1991. Introduction and time line to the history of the Florida Keys.
Pages 49-53 in The Monroe County environmental story (J. Gato, ed.). The
Monroe County Environmental Education Task Force, Big Pine Key,
Florida.
Gaston, K. J. and J. H. Lawton. 1987. A test of statistical techniques for detecting
density dependence in sequential census of animal populations. Oecologia
74:404-410.
Gilpin, M. E. 1980. The role of stepping-stone islands. Theoretical Population
Biology 17:247-253.
Gilpin, M. E. 1988. Extinction of finite metapopulations in correlated
environments. Pages 177-186 in Living in a patchy environment (B.
Shorrocks and I. R. Swinglands, eds.), Oxford Scientific, Oxford, England.
Gilpin, M. E. and M. E. Soul 1986. Minimum viable populations: processes of
species extinction. Pages 19-34 in Conservation biology: The science of
scarcity and diversity (M. E. Soul, ed.), Sinauer Assoc. Inc., Sunderland,
Massachusetts.
Ginzburg, L. R. Slobodkin, L. B., Johnson, K., and A. G. Bindman. (1982).
Quasiextinction probabilities as a measure of impact on population growth.
Risk Analysis 2:171-181.
Gleason, P. J. 1984. Saving the wild places a necessity for growth. Pages 1-34 in
P. J. Gleason ed.), Environments of South Florida: present and past.
Miami Geological Society, Florida.
Goodman, D. 1987. The demography of chance extinction. Pages 11-34 in Viable
populations for conservation (M. E. Soul, ed.), Cambridge University Press,
Cambridge.

229
Gotelli, N. J. 1991. Metapopulation models: The rescue effect, the propagule rain,
and the core-satellite hypothesis. American Naturalist 138:768-776.
Grant, W. E. 1986. Systems analysis and simulation in wildlife and fisheries
science. New York, Wiley. 388 pp.
Greenwood, P. J. 1980. Mating systems, philopatry, and dispersal in birds and
mammals. Animal Behaviour 28:1140-1162.
Greenwood, P. J. and P. H. Harvey. (1982). The natal and breeding dispersal of
birds. Annual Review of Ecology and Systematics 13:1-21.
Gyllenberg, M. and I. Hanski. 1991 Single species metapopulation dynamics: a
structured model. Theoretical Population Biology 42:35-61.
Haig, S. M., Ballou, J. D., and S. R. Derrickson. 1990. Management options for
preserving genetic diversity: Reintroduction of Guam rails to the wild.
Conservation Biology 4:290-300.
Hall, E. R. 1981. The mammals of North America. Vol. 1, John Wiley and Sons,
New York, 690 pp.
Hambright, T. L. 1991. Military history of the Florida Keys. Pages 102-105 in The
Monroe County environmental story (J. Gato, ed.). The Monroe County
Environmental Education Task Force, Big Pine Key, Florida.
Hanski, I. 1982. Dynamics of regional distribution: the core and satellite species
hypothesis. Oikos 38:210-221.
Hanski, I. 1985. Single-species spatial-dynamics may contribute to long-term rarity
and commonness. Ecology 66:335-343.
Hanski, I. 1989. Metapopulation dynamics: does it help to have more of the same?
Trends in Ecology and Evolution 4:113-114.
Hanski, I 1991. Single-species metapopulation dynamics: concepts, models, and
observations. Pages 17-38 in Metapopulation dynamics: Empirical and
theoretical investigations (M. E. Gilpin and I. Hanski, eds.), Academic
Press, London.
Hanski, I. 1994. Patch-occupancy dynamics in fragmented landscapes. Trends in
Ecology and Evolution 9:131-135.

230
Hanski, L, Kuussaari, M. and M. Nieminen. 1994. Metapopulation structure and
migration in the butterfly Melitaea cinxia. Ecology 75:747-762.
Hanski, I. and M. Gilpin. 1991. Metapopulation dynamics: Brief history and
conceptual domain. Biological Journal of the Linnean Society. 42:3-16.
Hanski, I. and M. Gyllenberg. 1993. Two general metapopulation models and the
core-satellite species hypothesis. American Naturalist 142:17-41.
Harris, R. B. and R. W. Allendorf. 1989. Genetically effective population size of
large mammals: an assessment of estimators. Conservation Biology 3:181-
191.
Harris, H. and D. A Hopkinson. 1976. Handbook of enzyme electrophoresis. North
Holland Press, Amsterdam. 240 pp.
Harris, S., W. J. Cresswell, P. G. Forde, W. J. Trewhella, T. Woollard, and S.
Wray. 1990. Home-range analysis using radio-tracking data a review of
problems and techniques particularly as applied to the study of mammals.
Mammalian Review 20:97-121.
Harrison, S, 1991. Local extinction in a metapopulation context: an empirical
evaluation. Pages 89-103 in Metapopulation dynamics: empirical and
theoretical investigations (M. E. Gilpin and I. Hanski, eds.), Academic Press,
London.
Harrison, S. 1994. Metapopulations and conservation. Pages 111-128 in Large-
scale ecology and conservation biology (P. J. Edwards, R. M. May, and N.
R. Webb, eds.), Blackwell Scientific Publications, Oxford, England.
Hassell, M. P. 1986. Detecting density dependence. Trends in Ecology and
Evolution 1:90-93.
Hastings, A. 1991. Structured models of metapopulation dynamics. Pages 57-71 in
Metapopulation dynamics: empirical and theoretical investigations (M. E.
Gilpin and I. Hanski, eds.), Academic Press, London, England.
Hastings, A. and Wolin 1989. Within patch dynamics in a metapopulation.
Ecology 70:1261-1266.
Haugen, A. O. 1942. Home range of the cottontail rabbit. Ecology 23:351-367.

231
Hawthorne, D. W. 1980. Wildlife damage and control techniques. Pages 411-440
in Wildlife management techniques, 4th edition (S. D. Schemnitz, ed.), The
Wildlife Society, Washington, D.C.
Hayne, D. W. 1949. Calculation of size of home range. Journal of Mammalogy
39:190-206.
Hilbom, R., Redfield, J. A. and C. J. Krebs. 1976. On the reliability of
enumberation for mark and recapture census of voles. Canadian Journal of
Zoology 54:1019-1024.
Hill, E. P. 1969. Observations of imported fire ant predation on nesting cottontails.
Proceedings of the Annual Conference of the Southeastern Association of
Game and Fish Commission 23:171-181.
Hill, E. P. 1972. The cottontail rabbit in Alabama. Bulletin 440. Agricultural
Experimentation Station, Auburn University, Auburn, Alabama, 103 pp.
Hoffmeister, J. E. 1974. Land from the sea, the geologic story of South Florida.
University of Miami Press, Coral Gable, Florida.
Holler, N. R. and M. F. Sorenson. 1969. Changes in behavior of a male swamp
rabbit. Journal of Mammalogy 50:832-833.
Holler, N. R., and C. H. Conaway. 1979. Reproduction of the marsh rabbit
(Sylvilagus palustris) in south Florida. Journal of Mammalogy 60:768-777.
Holling, C. S. 1978. Adaptive environmental assessment and management, Wiley,
New York, 377 pp.
Holt, R. D. 1993. The influence of regional processes on local communities. Pages
77-88 in Ecological communities: historical and geologic perspectives (R. E.
Ricklefs and D. Schluter, eds.), University of Chicago Press, Chicago.
Howe, S. E. 1988. Lower Keys marsh rabbit status survey. Report to U.S. Fish
and Wildlife Service, 3100 University Blvd. South, Suite 120, Jacksonville.
8 pp.
Humphrey, S. R. 1992. Rare and endangered biota of Florida: Volume 1.
Mammals. University Press of Florida, Gainesville, Florida.

232
Humphrey, S. R. 1994. Endangered mammals of Florida: is Mammalogy credible?
Journal of Mammalogy 75:551-552.
Humphrey, S. R. and D. B. Barbour. 1981. Status and habitat of three subspecies
of Peromysus polionotus in Florida. Journal of Mammalogy 62:840-844.
Ingles, L. G. 1941. Natural history observations on the Audubon cottontail. Journal
of Mammalogy 22:227-250.
Jellison, W. L., Owen, C. R., Bell, J. F., and G. M. Kohls. 1961. Tularemia and
animal populations: ecology and epizootiology. Wildlife Diseases 17:1-22.
Jennrich, R. I., and F. B. Turner. 1969. Measurements of noncircular home range.
Journal of Theoretical Biology 22:227-237.
Johnson, D. H. 1980. The comparison of usage and availability measurements for
evaluating resource preference. Ecology 61:65-71.
Johnson, M. K. and H. A. Pearson. 1981. Esophageal, fecal and exclosure
estimates of cattle diets on a longleaf pine-bluestem range. Journal of Range
Management 34:232-234.
Johnson, M. K., Wofford, H. and H. A. Pearson. 1983. Microhistological
techniques for food habits analyses. Research paper SO-199, U.S.
Department of Agriculture, Southern Forest Experimental Station, New
Orleans, Louisiana, 40 pp.
Jones, E., and B. J. Coman. 1981. Ecology of the feral cat, Felis catus (L.), in
south-eastern Australia. Australian Wildlife Research. 8:537-547.
Jurewicz, R. L., Cary, J. R and O. J. Ronstad. 1981. Spatial relationships of
breeding female cottontial rabbits in southwestern Wisconsin. Pages 295-
309 in Proceedings of the world lagomorph conference (E. Myers, and C. D.
Machines, eds.), Guelph University Press, Guelph, Ontario.
Karr, J. R. 1982. Avian based extinction of Barro Colorado Island, Panama: a
reassessment. American Naturalist, 119:220-239.
Kilpatrick, C. W. 1981. Genetic structure of insular populations. Pages 28-59 in
Mammalian population genetics (M. H. Smith and J. Joule, eds.), University
of Georgia Press, Athens, Georgia.

233
Kindvall, 0. and I. Ahln. 1992. Geometrical factors and metapopulation
dynamics of the Bush Cricket, Metrioptera bicolor Philippi (Orthoptera:
Tettigoniidae). Conservation Biology 6:520-529.
Kjolhaug, M. S. and A. Woolf. 1988. Home range of the swamp rabbit in southern
Illinois. Journal of Mammalogy 69:194-197.
Lachenbruch, P.A. 1975. Discriminant analysis. Hafner, New York. 128 pp.
Lack, D. 1933. Habitat selection in birds. Journal of Animal Ecology 2:239-262.
Lacy, R. 1993. VORTEX: A computer simulation model for population viability
analysis. Wildlife Research 20:45-65.
Lacy, R. C. and T. W. Clark. 1990. Population viability assessment of the eastern
barred bandicoot in Victoria. Pages 131-145 in The management and
conservation of small populations, (T. W. Clark and J. H. Seebeck), Chicago
Zoological Society, Brookfield, Illinois.
LaHaye, W. S., Gutirrez and H. R. Akfakaya. 1994. Spotted owl metapopulation
dynamics in Southern California. Journal of Animal Ecology 63:775-785.
Lamberson, R. H., McKelvey, K., Noon, B. R. and C. Voss. 1992. A dynamic
analysis of the northern spotted owl viability in a fragmented forest
landscape. Conservation Biology 6:505-512.
Lande, R. and G. F. Barrowclough. 1987. Effective population size, genetic
variation, and their use in population management. Pages 87-123, in Viable
populations for conservation (M. E Soul, ed.), Cambridge University Press,
Cambridge.
La Polla, V. N. and G. W. Barrett. 1993. Effects of corridor width and presence on
the population dynamics of the meadow vole (Microtus pennsylvanicus).
Landscape Ecology 8:25-37.
Laurence, W. F. 1982. Ecological impacts of tropical forest fragmentation on
nonflying mammals and their habitats. Ph.D. dissertation, University of
California, Berkeley, 259 pp.
Layne, J. N. 1974. The land mammals of South Florida. Pages 386-413 in
Environments of south Florida: Present and past (P. J. Gleason, ed.), Miami
Geological Society, Florida.

234
Lazell, J. D., Jr. 1989. Wildlife of the Florida Keys: a natural history. Island
Press, Washington, D. C. 250 pp.
Lazell, J. D., Jr. 1984. A new marsh rabbit (Svlvilagus palustris) from Florida's
Lower Keys. Journal of Mammalogy 65:424-432.
Leek, C. R. 1979. Avian extinctions in an isolated tropical wet-forest preserve,
Ecuador. The Auk 96:343-352.
Lee, K. 1993. Compass and gyroscope: Integrating science and politics for the
environment. Island Press, Washington D. C., 243 pp.
Lemer, I. M., 1954. Genetic homeostatis. John Wiley and Sons, New York, 134 pp.
Leslie, P. H. 1945. On the use of matrices in certain population mathematics.
Biometrika 33:183-212.
Levins, R. 1969. Some demographic and genetic consequences of environmental
heterogeneity for biological control. Bulletin of the Entomological Society
of America 15:237-240.
Levins, R. 1970. Extinction, Pages 70-107 in Some mathematical questions in
Biology (M. Gerstenhaber, ed.), American Mathematical Society,
Providence, Rhode Island.
Liberg, O. 1985. Food habits and prey impact by feral and house-based domestic
cats in a rural area in southern Sweden. Journal of Mammalogy 65:424-432.
Linkkila, T. E. 1971. Influence of habitat upon changes within the interspecific
Connecticut cottontail populations. Agricultural Experiment Station,
University of Connecticut, 70 pp.
Lipsey, M. W. 1990. Design sensitivity: statistical power for experimental
research. Sage Publications, Newbury Park, California, 207 pp.
Loiselle, B. A. and W. G. Hoppes. 1983. Nest predation in insular and mainland
lowland forest in Panama. Condor 85:93-95.
Long, R. W. 1974. Origin of the vascular flora of southern Florida. Pages 118-126
in Environments of South Florida: Present and past (P. J. Gleason, ed.),
Miami Geological Society, Florida.

235
Long, R. W. and O. Lakela. 1971. A flora of tropical Florida. University of Miami
Press, Coral Gables, Florida, 962 pp.
Lord, R. D., Jr. 1963. The cottontail rabbit in Illinois. Illinois Department of
Conservation Technical Bulletin Number 3. Southern Illinois University
Press, Carbondale, 94 pp.
Lord, R. D., Jr. 1960. Litter size and latitude in North American mammals.
American Midland. Naturalist 64:488-499.
Lovejoy, T. E., Bierregaard, R. O. Jr., Rylands, A. B., Malcolm, J. R., Quinela, C.
E., Harper, L. H., Brown, K. S. Jr., Powell, A. H., Powell, B. V. N., Schubart
H. O. R. and M. B. Hays. 1986. Edge and other effects of isolation on
Amazon forest fragments. Pages. 257-285 in Conservation biology: the
science of scarcity and diversity (M. E. Soul, ed.), Sinauer Associates, Inc.,
Sunderland, Massachusetts.
MacArthur, R. H. and E. O. Wilson. 1967. The theory of island biogeography.
Princeton University Press, Princeton, New Jersey.
Mader, H. J. 1984. Animal habitat isolation by roads and agricultural fields.
Biological Conservation 29:81-96.
Maguire, L. A. 1991. Risk analysis for conservation biologists. Conservation
Biology 5:123-125.
Marcot, B. G. and R. Holthausen. 1987. Analyzing population viability of the
spotted owl in the Pacific Northwest. Transactions of the North American
Wildlife and Natural Resources Conference 52:333-347.
Marsden, H. M. and N. R. Holler, 1964. Social behavior in confined populations of
the cottontail and swamp rabbit. Wildlife Monograph 13:1-39.
McAlpine, P. J., Shows, T. B., Miller, R. L. and A. J. Pakstis. 1985. The 1985
catalog of mapped genes and report of the nomenclature committee. Human
Gene Mapping 8. Cytogenetics and Cell Genetics 40:8-66.
McCreedy, C. D. and H. P. Weeks, Jr. 1993. Growth of cottontail rabbits
(Sylvilagus floridanus) in response to ancillary sodium. Journal of
Mammalogy 74:217-223.

236
Miller, W. R., and R. E. Egler. 1950. Vegetation of the Wequetequock-Pawcatuck
tidal marshes, Connecticut. Ecological Monographs 20:143-172.
Morrison, D. F. 1976. Multivariate statistical methods. Second edition. McGraw-
Hill, New York, New York, 415 pp.
Moore, J. and R. Ali. 1984. Are dispersal and inbreeding avoidance related?
Animal Behavior 32:94-112.
Mueller, E. and G. O. Winston. 1991. Geologic history of the Florida Keys. Pages
4-11, in The Monroe County environmental story (J. Gato, ed.). The Monroe
County Environmental Education Task Force, Big Pine Key, Florida.
Murphy, D. D., Freas, K. E. and S. B. Weiss. 1990. An environment-
metapopulation approach to population viability analysis for a threatened
vertebrate. Conservation Biology 4:41-51.
Murphy, R. W., Sites, J. W. Jr., Buth, D. G. and C. H. Haufler. 1990. Proteins I:
iso2yme electrophoresis. Pages 45-126, in Molecular systematics (D. M.
Hillis and C. Moritz, eds.), Sinaeur Associates, Sunderland, Massachusetts.
Nagy, J. G. and J. B. Haufler. 1980. Wildlife nutrition. Pages 129-142 in Wildlife
management techniques, fourth edition (S. D. Schemnitz, ed.), The Wildlife
Society, Washington, D.C.
Nee, S. and R. M. May. 1992. Dynamics of metapopulations: habitat destruction
and competitive coexistence. Journal of Animal Ecology 61:37-40.
Neu, C. W., C. R. Byers and J. M. Peek. 1974. A technique for analysis of
utilization-availability data. Journal of Wildlife Management 38:541-545.
Nisbet, R. M. and W. S. C. Gurney. 1982. Modeling fluctuating populations,
Wiley, Chiester, New York, 379 pp.
NOAA. 1993. Climatological data, Florida. Volume 94:97. National Oceanic and
Atmospheric Administration, Asheville, North Carolina.
Nunney, L. and K. A. Campbell. 1993. Assessing minimum viable population size:
demography meets population genetics. Trends in Evolution and Ecology,
8:234-239.

237
Opdam, P. 1990. Metapopulation theory and habitat fragmentation: a review of
holartic breeding bird studies. Landscape Ecology 5:93-106.
Oxley, D. J., Fenton, M. B. and G. R. Carmody. 1974. The effects of roads on
small mammals. Journal of Applied Ecology, 11:51-59.
Packer, C. 1979. Inter-troup transfer and inbreeding avoidance in Papio anubis.
Animal Behaviour 27:1-36.
Payne, N. F. 1975. Range extension of the marsh rabbit in Virginia. Chesapeake
Science 16:77-78.
Pietz, P. J. and J. R. Tester. 1983. Habitat selection by snowshoe hares in North
Central Minnesota. Journal of Wildlife Management 47:686-696.
Pimm, S. L. 1991. The balance of nature? Ecological issues in the conservation of
species and communities. University of Chicago Press, Chicago, Illinois.
434 pp.
Pollock, K. H., Nichols, J. D., Brownie, C. and J. E. Hines. 1990. Statistical
inference for capture-recapture experiments. Wildlife Monographs 107:1-97.
Proulx, G. 1988. Control of urban wildlife predation by cats through public
education. Environmental Education 15:358-359.
Pulliam, H. R. 1988. Sources, sinks, and population regulation. American
Naturalist 132:652-661.
Pulliam, H. R. and B. J. Danielson. 1991. Sources, sinks and habitat selection: A
landscape perspective on population dynamics. American Naturalist
137:S50-S66.
Pusey, 1987. Sex-biased dispersal and inbreeding avoidance in birds and mammals.
Trends in Ecology and Evolution. 2:295-299.
Quinn, J. F. and A. Hastings. 1987. Extinction in subdivided habitats.
Conservation Biology 1:198-208.
Ralls, K. and J. Ballou. 1983. Extinction: lessons from zoos. Pages 164-184 in
Genetics and Conservation: a reference for managing wild animal and plant
populations (C. M. Schonewald-Cox, S. M. Chambers, B. MacBryde, and L.
Thomas, eds.), Benjamin/Cummings, Menlo Park, California.

238
Ralls, K., Ballou, J. D. and A. Templeton. 1988. Estimates of lethal equivalents
and the cost of inbreeding in mammals. Conservation Biology 2:185-193.
Ray, C. M, Gilpin, M. E. and A. T. Smith. 1991. The effect of conspecific
attraction on metapopulation dynamics. Biological Journal of the Linnean
Society, 42:123-134.
Ribble, D. O. 1992. Dispersal in a monogamous rodent. Ecology 73:859-866.
Richter-Dyn, N. and N. S. Goel. 1972. On the extinction of a colonizing species.
Theoretical Population Biology 3:406-433.
Robertson, W. B., Jr. and J. A. Kushlan. 1974. The southern Florida avifauna.
Pages 414-452 in Environments of south Florida: present and past (P. J.
Gleason, ed.), Miami Geological Society, Florida.
Rogers, J. S 1972. Measures of genetic similarity and genetic distance. Studies in
genetics, VII. The University of Texas Publication, 7213:145-153.
Rolstad, J. 1991. Consequences of forest fragmentation for the dynamics of bird
populations: Conceptual issues and the evidence. Pages 149-163 in
Metapopulation dynamics: Empirical and theoretical investigations (M. E.
Gilpin and I. Hanski, eds.), Academic Press, London.
Rose, G. B. 1977. Mortality rates of tagged adult cottontail rabbits. Journal of
Wildlife Management, 41:511-514.
Ross, M. S., OBrien, J. J. and L. d. S. Sternberg. 1994. Sea-level rise and the
reduction of pine forests in the Florida Keys. Ecological Applications 4:144-
156.
Salwasser, H., Mealey, S. P, and K. Johnson. Wildlife population viability.
Transactions of the North American Wildlife and Natural Resources
Conference 49:421-439.
SAS Institute, Inc. 1984. SAS/ETS users guide, version 5, SAS Institute, Cary
North Carolina, 734 pp.
SAS Institute, Inc. 1987. SAS/STAT user's guide, version 6, SAS Institute, Cary,
North Carolina, 1,028 pp.

239
Schoener, T. W. and D. A. Spiller. 1987. High population persistence in a system
with high turnover. Nature 330:474-477.
Scribner, K. T., Chesser, R. K. and R. J. Warren. 1983. Spatial and temporal
genetic variability of the Eastern cottontail on west Texas playa basins.
Journal of Mammalogy 64:287-294.
Schribner, K. R. and R. J. Warren. 1986. Electrophoretic and morphologic
comparisons of Svlvilagus floridanus and S. audubonii in Texas.
Southwestern Naturalist 31:65-71.
Scribner, K. T., and R. J. Warren. 1990. Seasonal demography and movements of
cottontail rabbits on isolated playa basins. Journal of Wildlife Management
54:403-409.
Seal, U. S. and R. C. Lacy. 1989. Florida panther viability analysis. Report to the
U. S. Fish and Wildlife Service. Apple Valley, Minnesota: Captive
Breeding Specialist Group, Species Survival Commission, IUCN.
Shaffer, M. L. 1981. Minimum population sizes for species conservation.
Bioscience, 31:131-134.
Shaffer, M. L. 1990. Population viability analysis. Conservation Biology 4:39-40.
Shermyen, A. H. 1993. Florida statistical abstract, 27th edition. University Press of
Florida, Gainesville, Florida, 410 pp.
Shields, D. W. 1960. Movement patterns of brush rabbits in North Western
California. Journal of Wildlife Management, 24:381-386.
Shinn, E. A. 1988. The Geology of the Florida Keys. Oceanus 31:47-53.
Simberloff, D. S. and L. G. Abele. 1976. Island biogeography theory and
conservation practice. Science 191:285-286.
Sjogren, P. 1991. Extinction and isolation gradients in metapopulations: the case
of the pool frog (Rana lessone). Biological Journal of the Linnean Society,
42:135-147.
Smith, A. R. and M. M. Peacock. 1990. Conspecific attraction and the
determination of metapopulation colonization rates. Conservation Biology,
4:320-323.

240
Smith, A. T. 1978. Comparative demography of pikas (Ochotona): effect of spatial
and temporal age-specific mortality. Ecology 59:133-139.
Smith, A. T., 1980. Temporal changes in insular populations of the pika (Ochotona
princeps!. Ecology 61:8-13.
Snyder, J. R., Herndon, I. and W. B. Robertson. 1990. South Florida rockland.
Pages 230-272 in Ecosystems of Florida (R. L. Myers and J. J. Ewel, eds.),
University of Central Florida Press, Orlando, Florida.
Sorenson, M. F., Rogers, J. P. and T. S. Baskett. 1968. Reproduction and
development in confined swamp rabbits. Journal of Wildlife Management
32:520-531.
Soul, M. E. 1981. Thresholds of survival: maintaining fitness and evolutionary
potential, Pages 151-169 in Conservation biology: An evolutionary-ecology
perspective (M. E. Soul and B. A. Wilcox eds.), Sinauer, Sunderland,
Massachusetts.
Soul, M. E. 1987. Introduction. Pages 1-11 in Viable populations for conservation
(M. E. Soul, ed.), Cambridge University Press, New York.
Stacy, P. B. and M. Taper. 1992. Environmental variation and the persistence of
small populations. Ecological Applications 2:18-29.
Stamps, J. A., Buechner, M. and V. V. Krishnan. 1987. The effects of edge
permeability and habitat geometry on emigration from patches of habitat.
American Naturalist 129:533-552.
Strong, L. C. 1978. Inbred mice in science. Pages 45-67 in Origins of Inbred Mice.
(H. C. Morse, ed.), Academic Press, New York.
Suchy, W., McDonald, L. L., Strickland, M. D., and S. H. Anderson. 1985. New
estimates of minimum viable population size for grizzly bears of the
Yellowstone ecosystem. Wildlife Society Bulletin 13:223-228.
Swofford, D. L., and R. Selander. 1981. BIOSYS-1, a computer program for the
analysis of allelic variation in genetics. Department of Genetics and
Development, University of Illinois, Urbana, 65 pp.

241
Taylor, B. 1991. Investigating species incidence over habitat fragments of different
areas a look at error estimation. Pages 177-191 jn Metapopulation
dynamics: Empirical and theoretical investigations (M. E. Gilpin and I.
Hanski, eds.), Academic Press, London.
Teft, B. C. and J. A. Chapman. 1987. Social behavior of the New England
cottontail, Sylvilagus transitionalis (Bangs) with a review of social behavior
in New World rabbits (Mammalia: Leporidae). Review of Ecologie 42:235-
276.
Terborgh, J. 1974. The preservation of natural diversity: the problem of extinction
prone species. Bioscience 24:715-722.
Thiollay, J. M. and B. U. Meyburg. 1988. Forest fragmentation and the
conservation of raptors: a survey on the island of Java. Biological
Conservation, 44:229-250.
Thomas, C. D. 1994. Extinction, colonization, and metapopulations:
environmental tracking by rare species. Conservation Biology 8:373-378
Thomas, C. D. and T. M. Jones. 1993. Partial recovery of a skipper butterfly
(Hesparia comma! from population refuges: lessons for conservation in a
fragmented landscape. Journal of Animal Ecology 62:472-481.
Tomkins, I. R. 1935. The marsh rabbit: an incomplete life history. Journal of
Mammalogy 16:201-205.
Trent, T. T. and O. J. Rongstad. 1974. Home range and survival of cottontail
rabbits in southwestern Wisconsin. Journal of Wildlife Management 38:459-
472.
U.S. Fish and Wildlife Service. 1990. Endangered and threatened wildlife and
plants; endangered status for the Lower Keys rabbit and threatened status for
the Squirrel Chimney cave shrimp. Federal Register 55(120):25588-25591.
Vaughan, T. A. 1986. Mammalogy. Saunders College Publishing, Philadelphia,
Pennsylvania, 576 pp.
Verboom, J., Lankester, K. and J. A. J. Metz. 1991. Linking local and regional
dynamics in stochastic metapopulation models. Pages 39-55 in
Metapopulation dynamics: Empirical and theoretical investigations (M. E.
Gilpin and I. Hanski, eds.), Academic Press, London.

242
Wallage-Drees, J. M., Immink, H. J., De Bruyn, G. and P. A. Slim. 1986. The use
of fragment-identification to demonstrate short-term changes in the diet of
rabbits. Acta Theriologica 31:293-301.
Waller, R. A. and D. B. Duncan. 1969. A Bayes rule for the symmetric multiple
comparisons problem. Journal of the American Statistical Association
64:1484-1503.
Walter, H., E. 1985. Vegetation of the earth and ecological systems of the geo
biosphere. Springer-Verlag, New York, 318 pp.
Walters, C. J. 1986. Adaptive management of renewable resources. Macmillan,
New York, 374 pp.
Warren, R. S., and W. A. Niering. 1993. Vegetation change on a northeast tidal
marsh: interaction of sea-level rise and marsh accretion. Ecology 74:96-
103.
Webb, D. S. 1990. Historical Biogeography. Pages 70-99 in Ecosystems of Florida
(R. L. Myers and J. J. Ewel, eds.), University of Central Florida Press,
Orlando, Florida.
Westoby, M., G. R. Rost, and J. A. Weis. 1976. Problems with estimating
herbivore diets by microscopically identifying plant fragments from
stomachs. Journal of Mammalogy 57:167-172.
White, G. C., and R. A. Garrott. 1990. Analysis of Wildlife Radio-tracking Data.
Academic Press, Inc., San Diego, California, 289 pp.
Wilcox, B. A. 1986. Extinction models and conservation. Trends in Ecology and
Evolution 1:46-48.
Wilcox, B. A. and D. D. Murphy. 1985. Conservation strategy: the effects of
fragmentation on extinction. American Naturalist, 125:879-887.
Williams, A. 1991. Transitional wetlands. Pages 32-33 in The Monroe County
environmental story (J. Gato, ed.). The Monroe County Environmental
Education Task Force, Big Pine Key, Florida.
Willis, E. O. 1974. Population and local extinctions of birds on Barro Colorado
Island, Panama. Ecological Monographs 44:153-169.

243
Willis, E. O. 1980. Species reduction in remanescent woodlots in Southern Brazil.
Proceedings of the 19th International Ornithological Congress: 783-786.
Wilson, L. D. and L. Porras. 1983. The ecological impact of man on the south
Florida herpetofauna. University of Kansas Museum of Natural History
Special Publication, 9.
Wilson, E. O. and E. O. Willis. 1975. Applied biogeography, Pages 522-536 in
Ecology and Evolution of Communities (M. L. Cody and J. M. Diamond ,
eds.), Harvard University Press, Cambridge, Massachusetts.
Wobeser, G. A. and T. R. Spraker. Post-mortem examination. Pages 89-98 in (S.
D. Schemnitz ed.). Wildlife management techniques, 4th edition. The
Wildlife Society, Washington, D.C.
Wolfe, J. L. 1992. Lower Keys marsh rabbit. Pages 71-75 in Rare and endangered
biota of Florida, Volume I: Mammals, second edition (S. R. Humphrey,
ed.),. University Presses of Florida, Gainesville.
Wolff, J. O., Lidicker, W. Z., Chesser, R. K. and M. H. Smith. 1991. Review of:
Dispersal in Rodents: a resident fitness hypothesis. Journal of Mammalogy
72:218-220.
Wood, D. H. 1988. Estimating rabbit density by counting dung pellets. Australian
Wildlife Research 15:665-671.
Woolf, A., Shoemaker, D. R., and M. Cooper. 1993. Evidence of tularemia
regulating a semi-isolated cottontail rabbit population. Journal of Wildlife
Management 57:144-157.
Wright, S. 1977. Evolution and the genetics of populations. Vol. 3. Experimental
results and evolutionary deductions. University of Chicago Press, Chicago.
456 pp.

244
BIOGRAPHICAL SKETCH
Elizabeth A. Forys was bom September 12, 1966, in Berkeley, California.
She attended the University of Virginia in Charlottesville, where she received a
B.A. in environmental sciences and biology, in 1988. In 1990, Elizabeth received
her M.S. in environmental sciences with a concentration in ecology. Her thesis title
was The effect of immigration on the demographic and genetic composition of
Oryzomvs palustris on the Virginia Barrier Islands. In August of 1990, she entered
the graduate program in the Department of Wildlife and Range Sciences (currently
the Department of Wildlife Ecology and Conservation) and will receive her Ph.D. in
May of 1995. After graduation, Beth is returning to the Lower Keys to work on the
recovery efforts of several endangered species.

I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope
and quality, as a dissertation for the degree of Doctor of Philosophy.
Stephen R. Humphrey, Chairman
Professor of Forest Resources 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 fully adequate, in scope
and quality, as a dissertation for the degree of Doctor of Philosophy.
Lyn C. Branch
Assistant Professor of Forest
Resources 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 fully adequate, in scope
and quality, as a dissertation for the degree of Doctor of Philosophy.
CULT^lJL
Eisenberg
Katharine Ordway^Prefessor of/
Ecosystem Conservation

I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope
and quality, as a dissertation for the degree of Doctor of Philosophy.
Professor of Forest Resources 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 fully adequate, in scope
and quality, as a dissertation for the degree of Doctor of Philosophy.
Crawford S. Holling
Arthur R. Marshall, Jr., Professor of
Ecological Sciences
This dissertation was submitted to the Graduate Faculty of the College
of Agriculture and to the Graduate School and was accepted as partial
fulfillment of the requirements for the degree of Doctor of Philosophy.
May 1995
\cuJ\ df. 3*/
Dean, college of Agriculture
Dean, Graduate School



101
habitat groups suggested by Alldredge and Ratti (1986) for controlling Type II
statistical errors.
Presence or absence of pellets was recorded at each marker during
March, July, and November (from November 1991 July 1993). A Chi-square
Goodness-of-fit test was used to determine whether there was a significant
difference between the "expected" use of habitat classes (the proportion of
quadrats in each habitat category) and the observed frequency of habitat usage
(the proportion of quadrats with pellets). If a statistically significant overall
result was found, habitat selection was determined by using Bonferroni
confidence intervals (Neu et al. 1974, Byers and Steinhorst 1984).
Macrohabitat
Macrohabitat variables were measured at each of the 59 sites in the
Lower Keys (Appendix A). The variables included the traditional
metapopulation geometric variables, area and isolation. In addition, at each
grid vegetation characteristics were measured using a line intercept method
(Canfield 1941). Ten 5-m long transects were randomly located at each site
and percent occlusion was determined for each plant species. Two vegetation
layers were used: ground vegetation of plants <1.5 m in height and canopy
vegetation of plants >1.5 m. Ground cover, canopy cover, average maximum
ground vegetation height, and plant diversity were calculated from these


149
Scenarios
Each scenario is described in depth below, but the specific numerical
modifications in the parameters for each scenario are listed in Table 6.1. The
first 2 simulations used the original parameters but differed in the way they
simulated population growth. The first simulation used the logistic equation
for population growth; the second used the exponential. After comparing the
results of the simulation with knowledge about the biology of S. p. hefneri, 1
simulation was chosen to be used in the scenarios.
Scenario #1- Decrease Predation.
(a). Predator control leads to a 25% decrease in mortality in each stage.
This would require a 50% reduction in the current amount of cat predation.
This would be likely to occur if the house-based and feral cats (Felis catus)
were removed and prevented from hunting in marsh rabbit habitat. Some
predation would probably occur before cats were captured in the trap but the
overall mortality rate would decrease. The radio-telemetry study conducted on
the Lower Keys marsh rabbit (see Chapter 3) found that cats were the major
predator of juvenile, subadult, and adult rabbits. It is assumed that nestling
rabbits also would be vulnerable to cat predation.
(b). Predator control leads to a 50% decrease in mortality at each stage.
This would require complete extermination of all cats in marsh rabbit habitat


Table 3.5--The sum of all marsh rabbits (Sylvilagus palustris hefneri) trapped at the 5 main sites on Boca Chica during the 8
trapping sessions (June 1991 to May 1993), and the proportion of these rabbits that were trapped twice during a 10-day
trapping session. The G2 for the number indicates a test to determine if the sex ratio of the trapped rabbits differed from
1:1. TheG2 for the trappability indiciates a test to determine if the trappability differed between sexes.
Age
Male
Female
Male:Female
Number
Trappability
Number
Trappability
Number G^
Trappability G^
Juvenile
8
50%
6
67%
1.0
11.6*
OO
Subadult
14
71%
4
25%
14.3*
59.6**
Adult
42
83%
42
88%
0.0
0.6
* P < 0.05
** P <0.001


138
same time, then patch extinctions may be correlated, leaving few or no source
populations to colonize the extinct patches.
Population Viability Analysis Models
Most metapopulation models (e.g., Nisbet and Gurney 1982, Hanski
1985, 1991, Hastings and Wolin 1989, Hastings 1991, Verboom et al. 1991,
Gyllenberg and Hanski 1992, Nee and May 1992) predict metapopulation
persistence based on phenomena that occur at the metapopulation level only
(Hanski 1994). Most of these models are theoretical in nature, and the
parameters can not be readily measured in the field. A few models (Fahrig and
Merriam 1985, Hanski and Gyllenburg 1993), incorporate the inter-population
level and the metapopulation level phenomena, creating spatially realistic
models that require only modest amounts of data to construct. These models
provide information on metapopulation dynamics of real populations but are
less useful in making management decisions. Models that incorporate all 3
levels require substantially more information but may be used to test
hypotheses regarding threats to the species at all 3 levels.
Several simulation models are currently available that combine
stochastic birth and death process models (Richter-Dyn and Goel 1972,
Goodman 1987), differing migration rates based on spatially explicit patch
structure and extinction and recolonization (correlated or uncorrelated).


106
metapopulation structure, yet some of the small, nearby patches were
permanently occupied as might be seen in a classic metapopulation.
Consistently occupied patches had juvenile pellets present for
significantly more sessions than variably occupied patches (t = -2.67, P <
0.01). The average proportion of sessions with juveniles was 54.2% (SD =
26.1) for the consistently occupied sites and 29.9% (SD 26.05) for the variably
occupied. Six of variably occupied sites and only 1 of the consistently
occupied sites never had juvenile pellets present. The variable sites could
produce juveniles, but they were less likely to than the consistent patches.
Further analysis of differences in habitat quality was warranted.
Dietary Analysis
Nineteen plant species representing 14 families were found in the fecal
pellets of 40 rabbits (Table 5.1). Two grasses, Sporobolus virginicus and
Spartina spartinae. composed over 50% of all diets. Sporobolus virginicus
occurs predominantly in the low marsh while S. spartinae dominates the high
marsh. Two mangrove species, Laguncularia racemosa and Rhizophora
mangle, and a small succulent shrub Borrichia frutescens. made up another
25%. The remainder of the species were grasses, sedges, shrubs, and trees.
Three species, Monanthochloe littoralis. Conocarpus erecta, and
Pithecellobium guadelupense. were abundant at the transition-zone sites but


28
(Scribner et al. 1983). The plasma and hemolysate were stored in liquid
nitrogen and were analyzed used starch-gel electrophoresis within 1 year.
Eleven presumptive genetic loci were scored for the samples. Standard
procedures for starch-gel electrophoresis were used (Harris and Hopkinson
1976). Locus nomenclature followed McAlpine et al. (1985) for mapped
human genes. Two buffer systems were used. Tris-citrate, pH 6.7 was used
for the following enzymes: esterase, E.C. 3.1.1.1 (EST-1); glucose phosphate
isomerase, E.C. 5.3.1.9 (GPI); isocitrate dehydrogenase, E.C. 1.1.1.42 (ICD-1);
lactate dehydrogenase, E.C. 1.1.1.27 (LDH-1), malate dehydrogenase, E.C.
1.1.1.37 (MDH-1); mannose phosphate isomerase, E.C. (MPI); and superoxide
dismutase, E.C. 1.15.1.1 (SOD-1). Tris citrate, pH 8.0 was used for the
following enzymes: beta hemoglobin (PHb); glutamate oxaloacetic
transaminase, E.C. 2.6.1.1 (GOT-1); peptidase, E.C. 3.4.13 (PEP-1); and
phosphoglucomutase, E.C. 2.7.5.1 (PGM-1).
Radio-telemetry
Each rabbit weighing >1,000 grams trapped at the 5 Boca Chica sites
and the Saddlebunch site was fitted with a radio-collar (weight <3 grams) and a
transmitter with an estimated 10-month operational life (Telonics, Inc., Mesa,
AZ). Smaller rabbits (300-1,000 g) were fitted with a similar radio-collar, but
with a velcro break-away device added to allow the animals to lose the


200
hefneri (Chapter 3) as cats, but raccoon numbers may be increased locally in
areas with open dumpsters. Providing top-closing lid dumpsters might
decrease the numbers of raccoons. Other native species such as Bald Eagles
(Haliaeetus leucocephalus) and Rattlesnakes (Crotalus adamanteus) and exotic
species (e.g. dogs, Canis familiaris) are not seen as being a threat to Lower
Keys marsh rabbits survival.
A significant number of rabbits are killed by vehicles (Chapter 3).
While it is not likely to prevent all of these mortalities, the number of deaths
might be reduced if speed limits were enforced. The most important time to
enforce speedlimits is from dusk to dawn, when the rabbits are most active.
Although undocumented, there may be additional sources of mortality in
the nestling stage. At this stage, rabbits are extremely altricial and perhaps
more susceptible to some types of predation than older rabbits. Fire, while
fairly rare in the transition zone, might have a impact on younger rabbits that
are incapable of escaping. Overall, fire probably has a positive effect on the
rabbit because of its effect on understory vegetation.
Mowing, while not having a large impact on older rabbits that can run
away, might have a large effect on nestlings. Mowing may additionally change
the composition and diversity of the plants in the marsh rabbit habitat.
Red imported fire ants (Solenopsis invicta! have recently colonized the
Keys and are abundant in some of the transition zone areas (Forys pers. obs.).


40
trappability, most notably in the female subadults, the apparent male bias in the
sex ratio may be due to differences in the trappability. Both male and female
biased sex ratios are common in the literature (Chapman et al. 1982).
Speculation about the role of biased sex ratios of litter and of adult rabbits has
been made in relation to mating systems (Chapman et al. 1977), population
density, precipitation, and success of hunters (Edwards et al. 1981).
Average productivity for female Lower Keys marsh rabbit (6.36
young/year) was slightly lower than for marsh rabbits in southern Florida
(Holler and Conaway 1979). However, the productivity measure in this study
only accounts for young seen in the nest during a 2-week period after birth. It
is possible more young were bom and not seen in the nest or that some young
died before, during, or shortly after birth. Productivity estimates are
substantially lower than for eastern cottontail rabbits (Chapman et al. 1982).
Lord (1960, 1963) noted that a decrease in the litter size of cottontails was
correlated with a decrease in latitude. As the size of the litter decreases, the
potential length of the breeding season increases. The breeding season and
litter size of the Lower Keys marsh rabbit are consistent with this general
observation.
Breeding was year-round in the Lower Keys. Cottontail breeding and
reproduction are generally tied to temperature and precipitation. In the Keys,
these climatic changes may be more subtle. Although the Florida Keys lie a


52
Month
Figure 3.4-Proportion (bar) and number (above bar) of Lower Keys marsh
rabbits (Sylvilagus palustris hefneri) that produced a litter during each month.
Data from 2 years were combined and averaged. (January = month 1,
December month 12).


5
equilibrium to describe metapopulations where movements between patches
are not great enough to increase persistence time. This population
configuration has been seen in species isolated by climatic change (Brown
1971), clear-cutting of forests (Leek 1979, Laurence 1982), and man-created
islands (Willis 1974, Karr 1982). These relictual species may lack the
locomotive ability, evolutionary history, and/or behavioral plasticity to allow
them to move between patches, or the configuration of patches may prohibit
movement. Some species may exhibit "conspecific attraction" and
preferentially colonize patches where conspecifics are present (Smith and
Peacock 1990, Ray et al. 1991). These species may be less likely to
(re)colonize empty habitat patches than other species.
If individuals of the species are exceptionally mobile, relative to the
distance between habitat patches, then the individuals within the patches may
frequently interact and form one large demographic entity inhabiting a patchy
environment (Harrison 1991). Many highly mobile species of birds (Blake and
Karr 1987, Rolstad 1991) use forest patches, wetlands, continental shelf, and
oceanic islands in this manner. Other species that are adapted to highly
variable environments such as disturbance or "r-selected" species also inhabit
temporally and spatially variable habitats but remain a single population
(Edwards et al. 1981, Adler and Wilson 1987).


240
Smith, A. T. 1978. Comparative demography of pikas (Ochotona): effect of spatial
and temporal age-specific mortality. Ecology 59:133-139.
Smith, A. T., 1980. Temporal changes in insular populations of the pika (Ochotona
princeps!. Ecology 61:8-13.
Snyder, J. R., Herndon, I. and W. B. Robertson. 1990. South Florida rockland.
Pages 230-272 in Ecosystems of Florida (R. L. Myers and J. J. Ewel, eds.),
University of Central Florida Press, Orlando, Florida.
Sorenson, M. F., Rogers, J. P. and T. S. Baskett. 1968. Reproduction and
development in confined swamp rabbits. Journal of Wildlife Management
32:520-531.
Soul, M. E. 1981. Thresholds of survival: maintaining fitness and evolutionary
potential, Pages 151-169 in Conservation biology: An evolutionary-ecology
perspective (M. E. Soul and B. A. Wilcox eds.), Sinauer, Sunderland,
Massachusetts.
Soul, M. E. 1987. Introduction. Pages 1-11 in Viable populations for conservation
(M. E. Soul, ed.), Cambridge University Press, New York.
Stacy, P. B. and M. Taper. 1992. Environmental variation and the persistence of
small populations. Ecological Applications 2:18-29.
Stamps, J. A., Buechner, M. and V. V. Krishnan. 1987. The effects of edge
permeability and habitat geometry on emigration from patches of habitat.
American Naturalist 129:533-552.
Strong, L. C. 1978. Inbred mice in science. Pages 45-67 in Origins of Inbred Mice.
(H. C. Morse, ed.), Academic Press, New York.
Suchy, W., McDonald, L. L., Strickland, M. D., and S. H. Anderson. 1985. New
estimates of minimum viable population size for grizzly bears of the
Yellowstone ecosystem. Wildlife Society Bulletin 13:223-228.
Swofford, D. L., and R. Selander. 1981. BIOSYS-1, a computer program for the
analysis of allelic variation in genetics. Department of Genetics and
Development, University of Illinois, Urbana, 65 pp.


9
stochastic as well as deterministic forces of extinction. In these "turnover-
prone" species (Schoener and Spiller 1987; Harrison 1991), empty patches of
suitable habitat may be relatively common.
Dissertation Structure
This dissertation incorporates data on the Lower Keys marsh rabbit at 3
spatial levels of observation: within-patch population dynamics, inter-patch
between population dynamics, and metapopulation dynamics. Using
simulation models that incorporate these three spatial scales, predictions are
made for a range of temporal scales.
Chapter 3 investigates the dynamics of 5 populations of Lower Keys
marsh rabbits living in habitat patches. Estimates of natality, mortality,
demographic structure, genetic and morphological variation were made at this
scale. In Chapter 4, the home range size, movement patterns, and dispersal
ability of the marsh rabbit were measured. This information about marsh
rabbit inter-patch use provided the basis for testing the hypothesis that the
marsh rabbits are existing in a true metapopulation configuration.
Chapter 5 explores patch occupancy, extinction, and (re)colonization at
the metapopulation scale. Marsh rabbit diet, microhabitat, and macrohabitat
selection were combined with the physical attributes of the patch to determine
if these parameters influenced patch occupancy. Chapter 6 incorporates the


64
patch. The first set of comparisons looked at all of the individuals of the same
sex present at a patch during the same period of time. Percent overlap was
determined by plotting all the home range contours together using Program
HOME RANGE, and overlapping a 10 m grid. For each pair of individuals the
amount of overlap was calculated as a percentage of the smaller home range.
This percent of overlap was compared to overlaps of same-sexed
individuals living at the same patch but at different times, and pairs of opposite
sexed individuals. A t-test was used for normally distributed data and when
the variances between the 2 groups were equal. A Mann-Whitney U test was
used when these conditions were not met.
Dispersal
If an animal appeared to be making a long-distance movement, it was
located at least once every day. These locations were not used in any home
range calculations. A rabbit that made a one-way long-distance movement was
said to have dispersed. The criterion for dispersal as a movement was the
diameter of the average marsh rabbit home range (Ribble 1992).
Dispersing animals were classified by sex and age class. Locations
from the daily radio-telemetry session were plotted and straight lines were
drawn between them to approximate the minimum distance traveled. Data on
the demography of dispersers was used to test hypotheses on the cause of


3
(Gilpin and Soul 1986). In small populations, both stochastic and
deterministic phenomena can affect the persistence time of a population.
Perturbations that are stochastic include variation in the environment (climatic
variation, natural catastrophes), genetic composition (loss of genetic variability,
inbreeding depression), or the demographic composition (sex, age ratios) of a
population (Shaffer 1981). Demographic and genetic stochasticity are most
important in small populations. Environmental stochasticity is important for all
populations, and its importance decreases only slightly with increasing
population size.
Deterministic causes of extinction include habitat destruction, invasion
by introduced predators, disease, and climatic change (Nunney and Campbell
1993). Deterministic extinction can occur regardless of population size, if all
of the population is affected. Both stochastic and deterministic phenomena
may interact via feed back loops leading to potential extinction via "extinction
vortices" (Gilpin and Soul 1986), and differ in magnitude and importance for
species with different life-history attributes. Because knowledge about
intrinsic population dynamics of a species is often limited and the occurrence
and impact of extrinsic factors is uncertain, PVA can only make probabilistic
predictions about a species' future (Ginzburg et al. 1982, Shaffer 1990).


166
despite the lack of habitat and then perhaps face greater mortality without
adequate food and cover. An experimental treatment of movements of Lower
Keys marsh rabbits without corridor habitat is needed to clarify this process.
When migration was decreased by 25%, the overall abundance and
number of patches occupied actually increased. When migration was cut by
50%, the overall population size appeared to increase, but the number of
occupied patches was lower than with full migration or the 25% decline. So
between the current amount of migration and a 25% reduction is the optimal
dispersal rate for maintaining metapopulation persistence.
Results from the last scenario provide insight into the importance of
individual patches to the overall metapopulation persistence. When mortality
was reduced by 25%, the elimination of the largest site simply decreased the
overall number of rabbits and the number of patches occupied. The destruction
of 5 large and 10 small patches lead to extinction. These last 2 scenarios lead
to extinction at nearly the same rate despite the fact that the 5 largest patches
contain 128.1 ha of habitat while the 10 small contain less than 10 ha. The
importance of a patch to the overall metapopulation persistence is more than
just the amount of area is encompasses. Thus, further habitat loss of small,
individual patches is a severe threat to the subspecies persistence.


38
subadult males were killed. One female was caught as a subadult and then
followed until the study ended, establishing a maximum longevity of 3 years
for this study.
Cause of mortality was determined for 24 rabbits (Figure 3.6). None of
the juvenile mortalities were included because their bodies were never found.
Presence of blood and fur on the break-away collar and subsequent
disappearance from the trapping grid were assumed to mean the animals had
been killed. It was possible to determine the cause of mortality for all of the
subadults (all male) and 19 of the adults. Only 1 of the adults disappeared
during the study.
Cause of mortality was organized into 6 classes: cat/raccoon, vehicle,
rattlesnake, raccoon, cat and poaching. The cat/raccoon class was used for
mortalities where it was not possible to distinguish between predation by a
feral cat or raccoon. Sign from both predators was abundant at the kill sites.
In nearly all of the kills the rabbit was dragged and partially buried, indicating
cat predation. In 3 cases the cat or raccoon was seen with the rabbit, and it was
assumed that the predator had killed the animal. In addition, Amim Sheutz (the
Natural Resource Manager for the Navy on Boca Chica Key, pers. comm.)
reported an observation oft a Bald Eagle grasping a medium-sized marsh
rabbit.


Number of Rabbits Caught
51
May 1991 May 1992
Session
Figure 3.3-The number of Lower Keys marsh rabbits Svlvilagus palustris
hefheri) caught during each trap session at the 5 main sites.


39
Full necropsies were only possible on 2 of the rabbits. The remainder of
the carcasses were entirely eaten, eviscerated, or decomposed. Neither of the
necropsies revealed any sign of disease.
The most subadult and adult mortalities were caused by vehicles (Figure
3.6). It was followed closely by the number of individuals killed by feral cats
and raccoons. Three rabbits were eaten by rattlesnakes (the radio-collar
transmissions were traced to the rattlesnake) and one rabbit was shot. Both
sexes appeared to be susceptible to predation by cats and raccoons, but more
males were killed by vehicles. Seven males were killed by vehicles on the
road, 4 of them subadult males. The only female road-kill was actually killed
in the marsh adjacent to the road. The vehicle had apparently been driven
through the habitat. Mortality on and off the base on Boca Chica Key was
similar. Road-kills, feral cat predation, and raccoon predation occurred on
both on and off the base on Boca Chica Key. All of the predation by
rattlesnakes and the poaching occurred on Navy land off the Boca Chica base.
Discussion
The high recapture rate of adult marsh rabbits indicates that most adult
rabbits living on a grid were probably captured during the study. The lower
trappability percentages for juveniles and female subadults suggests that some
of these individuals might have been missed. Because of this difference in


25
idiosyncratic factors such as the habitat available and climatic variation, and
therefore require in-depth study.
Lagomorph population biology is characterized by high reproductive
rates and high rates of mortality. All lagomorphs are entirely herbivorous,
although they feed on a diversity of vegetation. Due to their high abundances
and intermediate size, lagomorphs are the base of many predator-prey systems
involving small to medium-sized predators (Chapman and Flux 1990).
Most research on rabbits has been on the most abundant species (e.g.,
eastern cottontail, Sylvilagus floridanus), primarily in relation to hunting.
There is little information on the management of rare or endangered lagomorph
species. Additionally, most research on endangered species (in general) has
historically concentrated on larger species that are long-lived and have longer
generation-times (Murphy et al. 1990). Threats to the persistence of smaller
species with higher reproductive rates, but shorter life-spans (r-selected) will
differ. These "r" selected species may be more habitat specific than "K"
species and may experience higher variability in population numbers (Pimm
1991).
The objective of this portion of the research was to determine the
demographic and genetic compositions of the Lower Keys marsh rabbit.
Special emphasis was placed on parameters that affect the intrinsic rate of
growth of the marsh rabbit populations (natality, mortality, sex and age ratios),


Number of Patches
130
Number of Sessions Occupied
Figure 5.5--Number of sessions (from session 3-8) occupied by the Lower Keys marsh
rabbit (Sylvilagus palustris hefneri) at each of the 59 transition-zone sites.


93
metapopulation all patches greater than a minimum area and less than a
maximum isolation are suitable for supporting a population. Local extinctions
(due to deterministic or stochastic causes) are counteracted by colonizations
from near-by patches (Hanski and Gilpin 1991). Some classic metapopulations
may resemble the stepping-stone model of occupancy (Gilpin 1980). In this
model, the probability of occupancy increases as patch size increases and
distance to the nearest occupied patch decreases as a dynamic consequence of
local extinction and colonization.
In a mainland-island metapopulation, 1 or a few large patches are
occupied continuously while surrounding small patches frequently experience
extinction but are recolonized due to their proximity to a mainland patch
(Harrison 1991, 1994). Only patch size determines which patches will be
mainlands and which patches are islands. Larger patches (and hence larger
populations) are able to maintain occupancy despite population variability and
thereby act as mainlands from which individuals emigrate to the smaller
islands.
Although similar to mainland-island metapopulations, source-sink
metapopulations differ because habitat quality explains the patterns of
occurrence. In a source-sink metapopulation the higher quality source habitat
patch produces a surplus of descendants (X > 1) whereas the low-quality
patches may be sinks that produce a deficit (X < 1) (Pulliam 1988, Pulliam and


Table 6.5Sensitivity analysis for population parameters. Each of the parameters used in the model was increased by 25% and 50% to
see how sensitive these parameters are to changes in value or errors in estimation.
Simulation
Abunance (SE)
Number of
Populations
(SE)
Years to Extinction
- 1 SE Mean + 1SE
25% more nestlings
180.3 (25.6)
11.6(0.9)
39
>50*
>50
50% more nestlings
765.8 (17.2)
26.8 (0.5)
>50
>50
>50
25% more juveniles
105.1 (31.8)
16.7(1.3)
26
38
46
50% more juveniles
344.2 (34.2)
18.1 (1.2)
32
49
>50
25% more subadults
154.1 (37.3)
9.0 (1.4)
28
39
>50
50% more subadults
169.3 (37.9)
10.1 (1.4)
28
45
49
25% more adult 1
205.3 (28.1)
23.5 (0.8)
>50
>50
>50
50% more adult 1
239.6 (32.6)
24.5 (0.7)
>50
>50
>50
25% more adult 2
173.1 (34.7)
23.1 (0.8)
>50
>50
>50
50% more adult 2
301.5 (34.8)
23.7(0.8)
>50
>50
>50
25% more reproduction
176.2 (38.3)
10.2(1.4)
30
45
>50
50% more reproduction
275.3 (42.5)
14.9(1.3)
41
>50
>50
*>50 indicates that the population size did not reach zero during the 50 year simulation.


191
Figure 6.10Results of Scenario #7 of the Lower Keys marsh rabbit (Sylvagus
palustris hefheri) PVA model. In the first simulation the largest patch (#33) is
destroyed, in the second the 5 largest patches are destroyed, and in the last simulation
10 of the smallest patches are destroyed.


Middle Torch
\
Big Torch
Big Pine
Cudjoe
N. Saddlebunch\
Saddlebunch \
1J Qjk
jmmerland Parr1
e.
Little Torch
O
Sugarloaf
10
Key West
Boca Chica
Figure 2.1'The Lower Keys of Florida. The solid line connecting the Keys is highway US-1.


194
At its best, a PVA can help understand how a species interacts with its
environment and define the processes that are most important in persistence.
Combined with the scientific method, a PVA can be used in hypothesis testing
about the cause of the species decline and how different management
techniques can halt this decline. These techniques can be applied at small
scales, and then the results extended throughout the species using a PVA
model, but appropriate validation of the model should be conducted at a variety
of locations.
Recently, several population simulation models (VORTEX, Lacy 1993,
RAMAS, Akijakaya 1994) have been extended to include spatial structure in
their analysis of population viability. Most endangered species exist in highly
fragmented habitats and the spatial structure of the PVA model should include
the interchange among habitat fragments. If habitat patches are highly
independent of each other with little exchange between patches, then it may be
best to model each patch independently. If movement between patches is so
great that the rabbits exist as a single population, then the spatial structure of
the habitat patches is not an important feature to be included in the model.
However, when the species exists in a metapopulation, the spatial structure of
the patches should be included. In VORTEX and RAMAS, inter-population
movements are incorporated into these models based on measured dispersal
information or based solely on the spatial structure of the metapopulation.


143
be available to conduct such an experiment. Gaston and Lawton (1987)
suggest that a qualitative analysis of the population dynamics and life history
of a species might be used to infer if density dependence could occur.
For the purposes of this model, it was assumed that the rabbit, a short-lived,
quickly reproducing mammal, might exhibit some density dependence. The
first run of the model incorporated simple density dependence using the logistic
equation, where population growth follows the logistic equation until carrying
capacity (K) is reach. At carrying capacity the model uses the life-table to
predict population growth.
Yet, because of the severe predation pressure (Chapter 3) on the marsh
rabbits, population growth might be controlled by the predation, limiting the
ability of the Lower Keys marsh rabbit to exhibit density-dependent population
growth. The model was additionally run without density dependence, using the
exponential equation in conjunction with the life table to predict population
growth.
To determine the carrying capacity for the density-dependent growth,
the number of males that could fit into a patch was calculated by assuming
non-overlapping core areas. For areas of consistently occupied habitat
(Chapter 5), the size of the male core area was the mean core area calculated in
Chapter 4 plus one standard deviation. For areas of variably occupied
habitat, the size of the male core area was this mean core area minus one


100
located at each site and percent occlusion was determined for each plant
species.
The relative densities are comparable between samples, because the
digestion and retention rates are fairly constant for each plant species
(Wallage-Drees et al. 1986). Univariate analyses of variance (PROC GLM,
SAS Institute, Inc. 1988) were performed on arcsine-transformed estimates of
dietary composition (%) for the most abundant species in the samples. Data
from the 2 years were pooled to provide sufficient sample sizes to examine the
differences between pellet groups from different sites, seasons, and sexes.
Microhabitat Use
Microhabitat use was studied at each of the most accessible sites (sites
that were not high securtity) on Boca Chica Key (sites 1-17, Appendix A).
Detailed data were recorded on vegetation at each of the 49 permanent markers
(49 quadrats) on the 17 grids. Based on species composition, each site was
classified as being in 1 of 4 habitat categories: low marsh (dominated by
Monanthocloe littoralis. Sesuvium maritimum. Salicomia virginica. Batis
martima, and some small mangrove trees), mid-marsh (dominated by Borrichia
sp. and Sporobolus virginicus). high marsh (predominantly the clump grasses
Spartina spartinae, and Fimbrvstvlis sp.), and hammock (dominated by tree
species, mainly Conocarpus erectas). Four categories is an adequate number of


177
Figure 6.3 Predictions (and error bars) of the population viability analysis model for
the Lower Keys marsh rabbit (Svlvilagus palustris hefneri) using density-dependent
population growth (logistic equation) and density-independent growth (exponential
equation).


199
applications of GARLON has been found to be a fast, effective method of
killing these exotics.
Off-road vehicle traffic (ORV) compacts soil, destroys vegetation and
increases erosion. Off-road vehicle traffic through marsh rabbit habitat can be
stopped by preventing access to habitats from the road. Local education
programs involving law enforcement and military police officers have been
helpful in preventing ORV use of wetlands. Deterring ORV use of wetlands
also decreases the amount of illegal dumping.
Restoration of vegetation in areas that have been badly damaged by
ORV use, dumping, or other land uses may be needed at some areas.
Restoration, although costly, may increase the total amount of habitat available
to the marsh rabbit. A trial restoration project has been initiated at site #10 on
Boca Chica.
In addition to protecting habitat, protection of marsh rabbit dispersal
corridors is also important in preserving the subspecies. Marsh rabbits will
move through any habitat that is not disturbed (Chapter 4). Future
development plans in the Keys should be evaluated for their impact on corridor
habitat.
Although cats are the main predator of the marsh rabbit, reducing other
sources of mortality may significantly increase the marsh rabbit population.
Raccoons (Procvon lotor) are not as great of a threat to the persistence of S. p.


140
ever validated and therefore it is difficult to determine which model approaches
are most successful (Grant 1986).
Despite these problems, PVA has been found to be a useful management
tool for a number of species (grizzly bear, Suchy 1985; northern Spotted owl,
LaHaye et al. 1994). The strongest potential of PVA is in the development of
adaptive management strategies (Holling 1978, Walters 1986, Lee 1993) to
recover endangered species. A well-developed PVA can refine multiple
hypotheses about the impact of multiple management tactics. Small-scale
management techniques can be attempted at different (sub)populations, the
(sub)populations monitored, the PVA model validated, and then the most
successful suite of management techniques applied to all of the populations.
The main objective of this chapter is to develop a PVA model for the
Lower Keys marsh rabbit (Sylvilagus palustris hefneri). This model will make
predictions about the persistence of the subspecies under a variety of potential
scenarios and management techniques. This information will be used to make
recommendations about the direction of future research and management of
this subspecies.
The Lower Keys marsh rabbit is an endangered subspecies occurring in
a classical metapopulation structure. It is believed to be declining during
recent years (Howe 1988) and was only protected 4 years ago. Since the


151
affects 1 patch at a time, but each patch has 75% chance of getting infected
each year. Tulemia epizootics have been observed in a number of rabbit
species (Jellison 1969) and may play a role in population regulation (Woolf et
al. 1993). The rate of spread between infected populations is not known
(Woolf et al. 1993). Infected populations generally pass the disease to most,
but not all, nearby populations.
(b). A more severe disease, perhaps similar to the effects of
myxamatosis in Australia and Europe on Oryctolagus cuniculus (Chapman and
Flux 1990), spreads through the Keys. This disease also kills only the nestlings
and adults, but its mortality rate is much higher than the disease described in
(a). This disease also affects one patch at a time and with a 75% chance each
year.
Scenario #5 Hurricanes.
(a). A class 3 hurricane directly hits the center of the Lower Keys
during the first year of the simulation. The effect is regional, and all keys
suffer some mortality. Storm surge inflicts the greatest effect. Mortality is
highest for the nestlings and lowest for the adults that can swim. The
probability that a hurricane of this strength hits the Lower Keys is <10% each
year.


Number of Occupied Patches


187
Figure 6.8Results of Scenario #5 of the Lower Keys marsh rabbit (Sylvilagus
palustris hefneri) PVA model. The effects of a hurricane kill rabbits in each stage,
with the youngest rabbits being the most likely to die. Class 3 produces the lowest
mortality and class 5 the highest.


107
absent from the marsh rabbit's diet (Table 5.2). Key grass (M. littoralis) is a
short wiry grass, and buttonwood (C. erecta) and blackbead (P. guadelupense)
are tree species. Excluding these species, marsh rabbits generally fed on the
most abundant species available.
The 4 most abundant species in the fecal pellets, S. virginicus.
spartinae. B. frutescens. and L. racemosa. were used in the univariate analysis.
Univariate analysis indicated a significant between-season difference in the
relative density of S,. spartinae in the fecal pellets, and a significant season-by
sex interactions in presence of virginicus and of spartinae (Table 5.3).
There were no significant differences found among sites or between sexes for
any of the plant species, despite the fact that the statistical power of the
univariate analyses were high (0.48 0.98).
Resource Use and Availability
Eight of the original 17 sites on Boca Chica/Geiger Key were used to
determine habitat selection (Table 5.4). Nine sites could not be included in the
analysis because either there were no pellets present during 1 of the surveys or
1 of the habitat types was missing. To simplify the analysis data, the 2 years of
data were combined and habitat selection was determined by season.
The majority of the overall chi-square goodness-of-fit statistics were
significant (P < 0.05) for all of the sites during the 3 seasons. Goodness-of-fit


104
took the longest time to decay, whereas those in mud decayed in the shortest
time.
Fifty-three rabbits (19 females and 34 males) each produced 10 intact
pellets and were included in the regression of body mass and pellet area. The
relationship was strong and significant for both males (F = 95.8, P < 0.0001, r2
= 0.75) and females (F = 198.6, P < 0.0001, r2 = 0.92). The regression lines
were very similar, and both regression equations produced a pellet area of 0.4
cm for a body mass of 1000 g (Figure 5.3). This pellet area was used to
distinguish between adult and juvenile pellets on the pellet grids. It was
determined in Chapter 3 (Growth and Morphology) that most rabbits were
sexually mature by 1000 g and considered to be adults.
Metapopulation Structure
Marsh rabbits were found at sites on Boca Chica, Saddlebunch,
Sugarloaf, and Big Pine Keys (Figure 5.4). Between Sugarloaf and Big Pine
Key, a gap in the distribution of marsh rabbit populations occured, despite the
presence of transition-zone habitat.
Of the 59 patches of transition zone habitat used in the analysis (Figure
5.5), 20 had pellets present during all of the surveys (occupied patches), 22 had
pellets present during at least 1 survey (variable patches), and 17 never had any
pellets present (empty). When these patches were plotted by log area and


APPENDICES
A DESCRIPTIVE INFORMATION ABOUT THE
HABITAT PATCHES AND MAPS 203
B VARIABLES USED IN THE DISCRIMINANT FUNCTION
ANALYSIS 213
C MARSH RABBIT DENSITY ESTIMATES 216
D MARSH RABBIT PATCH OCCUPANCY 219
REFERENCES 222
BIOGRAPHICAL SKETCH 244
vi


APPENDIX B
VARIABLES USED IN THE DISCRIMINANT FUNCTION ANALYSIS


Year
Number of Occupied Populations
^ ^ hO hO CO CO
ocnocnocnocn
Year
Total Estimated Abundance
ro cd co o
o o o o o
o o o o o o
1200


Estimated Number of Adults
175
Figure 6.1-The number of adult Lower Keys marsh rabbits (Svlvilagus palustris
hefneri) estimated during the 8 pellet-counting sessions (March 1991 July 1993), by
Key group (see Appendix A).


91
Figure 4.9The core areas of 4 sympatric Lower Keys marsh rabbits
(Sylvilagus palustris hefneri). Rabbits A55M and A59M are adult males,
A57F and A58F are adult females.


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
METAPOPULATIONS OF MARSH RABBITS: A POPULATION
VIABILITY ANALYSIS OF THE LOWER KEYS MARSH RABBIT
SYLVILAGUS PALUSTRIS HEFNERH
By
Elizabeth A. Forys
May, 1995
Chairman: Stephen R. Humphrey
Major Department: Wildlife Ecology and Conservation (Forest Resources and
Conservation)
The Lower Keys marsh rabbit (Svlvilagus palustris hefneri) is a state
and federally endangered subspecies that historically ranged from Big Pine
Key to the southernmost of the Florida Keys, Key West. Lower Keys marsh
rabbits inhabit the marsh transition zone, an area that is currently highly
fragmented due to development. This dissertation incorporates data collected
during a 2.5-year study of the population biology, habitat requirements, and
spatial structure of the populations of the Lower Keys marsh rabbit.
Within and between patch dynamics were studied by live-trapping
marsh rabbits and fitting them with radio-collars at 6 habitat patches. Genetic
Vll


87
Figure 4.5--A comparison of the wet and dry season home ranges using the
95% harmonic mean (outer boundary) and the core area (inner boundary) for
an adult male at site 10.


235
Long, R. W. and O. Lakela. 1971. A flora of tropical Florida. University of Miami
Press, Coral Gables, Florida, 962 pp.
Lord, R. D., Jr. 1963. The cottontail rabbit in Illinois. Illinois Department of
Conservation Technical Bulletin Number 3. Southern Illinois University
Press, Carbondale, 94 pp.
Lord, R. D., Jr. 1960. Litter size and latitude in North American mammals.
American Midland. Naturalist 64:488-499.
Lovejoy, T. E., Bierregaard, R. O. Jr., Rylands, A. B., Malcolm, J. R., Quinela, C.
E., Harper, L. H., Brown, K. S. Jr., Powell, A. H., Powell, B. V. N., Schubart
H. O. R. and M. B. Hays. 1986. Edge and other effects of isolation on
Amazon forest fragments. Pages. 257-285 in Conservation biology: the
science of scarcity and diversity (M. E. Soul, ed.), Sinauer Associates, Inc.,
Sunderland, Massachusetts.
MacArthur, R. H. and E. O. Wilson. 1967. The theory of island biogeography.
Princeton University Press, Princeton, New Jersey.
Mader, H. J. 1984. Animal habitat isolation by roads and agricultural fields.
Biological Conservation 29:81-96.
Maguire, L. A. 1991. Risk analysis for conservation biologists. Conservation
Biology 5:123-125.
Marcot, B. G. and R. Holthausen. 1987. Analyzing population viability of the
spotted owl in the Pacific Northwest. Transactions of the North American
Wildlife and Natural Resources Conference 52:333-347.
Marsden, H. M. and N. R. Holler, 1964. Social behavior in confined populations of
the cottontail and swamp rabbit. Wildlife Monograph 13:1-39.
McAlpine, P. J., Shows, T. B., Miller, R. L. and A. J. Pakstis. 1985. The 1985
catalog of mapped genes and report of the nomenclature committee. Human
Gene Mapping 8. Cytogenetics and Cell Genetics 40:8-66.
McCreedy, C. D. and H. P. Weeks, Jr. 1993. Growth of cottontail rabbits
(Sylvilagus floridanus) in response to ancillary sodium. Journal of
Mammalogy 74:217-223.


APPENDIX A
DESCRIPTIVE INFORMATION ABOUT THE HABITAT PATCHES AND MAPS
Each transition zone site in the Lower Keys of Florida is listed by the Key it occurs
on, the area of the habitat patch, the ownership, and whether the patch was included in the
density estimates. Maps with each site located by number follow this appendix.
Key
Site #
Area
Ownership*
Density
Boca Chica
1
1.05
USN
Yes
Boca Chica
2
1.22
USN
Yes
Boca Chica
3
1.62
USN
Yes
Boca Chica
4
2.92
USN
Yes
Boca Chica
5
2.43
USN
Yes
Boca Chica
6
2.75
USN
Yes
Boca Chica
7
3.89
USN
Yes
Boca Chica
8
3.90
USN
Yes
Boca Chica
9
4.86
USN
Yes
Boca Chica
10
2.27
USN
Yes
Boca Chica
11
5.18
USN
Yes
Boca Chica
12
2.33
USN
Yes
Boca Chica
13
3.40
USN
Yes
Boca Chica
14
5.83
USN
Yes
Boca Chica
15
1.15
USN
Yes
Boca Chica
16
0.69
USN
Yes
Boca Chica
17
1.04
USN
Yes
Boca Chica
18
0.55
USN
No
Boca Chica
19
1.50
USN
No
Boca Chica
20
1.00
USN
No
Boca Chica
21
3.00
USN
No
Boca Chica
22
0.30
USN
No
Boca Chica
23
0.50
USN
No
Boca Chica
24
0.50
USN
No
Boca Chica
25
1.56
?
Yes
Boca Chica
26
0.45
USN
No
Boca Chica
27
0.30
?
No
203


Core Area (ha)
83
Figure 4.1The cumulative core area measurements of 8 radio-collared male
Lower Keys marsh rabbits fSylvilagus palustris hefneriV


14
remaining two-thirds of the rainfall occurs during the wet season which begins
in May and ends in October (Figure 2.2). Relative humidity remains high
throughout the year, averaging nearly 75% (NOAA 1993).
Hurricane season occurs mainly during the wet season, from June until
November. The chance that a hurricane (winds >74 mph) will hit the Lower
Keys is slightly less than 10% each year. Historically, most hurricanes have
occurred during the months of September and October. Tropical storms (i.e.,
winds 39-74) occur more frequently, but do not produce the large storm surges
and destruction that hurricanes are capable of producing.
Flora and Fauna
The flora of the Lower Keys is derived from 4 sources: the Caribbean,
the eastern U.S. coastal plain, endemic taxa evolved in place, and exotic taxa
from cosmopolitan sources (Long 1974). Four plant associations predominate:
mangrove, transitional salt-marsh, pineland, and hardwood hammock.
Mangrove community is dominated by 3 saline-tolerant, unrelated
species: red mangrove (Rhizophora mangle), black mangrove (Avicennia
germinans), and white mangrove (Laguncularia racemosa). These trees occur
in areas that are either continually submerged or tidally inundated.
Transitional salt-marsh (also called the transition zone) is a grassy,
nearly treeless marsh area that generally occurs from 1 to 3 m above sea level.


Table 5.7--A comparison of habitat measurements between occupied and empty marsh rabbit habitat in the Lower Keys of Florida.
Habitat variable (Unit)
Occupied habitat (n=42)
Empty habitat (n=17)
Test statistic*
P
X(SE~)
Untransformed
X(SE)
Untransformed
GCover(%)
7.7(0.2)
59.3
8.0(0.3)
64.0
1.12 (T)
0.27
CCover(%)
3.7(0.2)
13.7
3.8(0.5)
14.4
-0.44 (U)
0.66
Borrichia(%)
2.3(0.2)
5.3
1 5(0.5)
2.3
-2.36 (U)
0.02
Clump(%)
4.4(0.3)
19.4
4.5(0.6)
20.3
0.15 (T)
0.88
MaxHgt(cm)
22.2(1.7)
22.2
28.7(4.2)
28.7
1.10(U)
0.27
H(#)
1.4(0.1)
1.4
1.3(0.1)
1.3
-1.35 (T)
0.19
Area(ha)
1.6(0.3)
4.8
0.6(0.1)
1.8
-1.15 (T)
0.25
DPopulation(m)
5.8(0.2)
330.3
8.1(0.1)
3294.5
5.78 (U)
0.001
DResidence(m)
5.7(0.4)
298.9
2.9(0.8)
18.2
-3.10 (T)
0.005
DWater(m)
3.0(0.4)
20.1
2.9(0.6)
18.2
-0.20 (U)
0.84
* For all normally distributed variables (or transformed variables) a two-sample t-test (T) was used, a Mann-Whitney U-test using normal approximation (U)
was used for all non-normally distributed variables.


174
Table 6.6.The number of years until the Lower Keys marsh rabbit (Sylvilagus
palustris hefneri) goes extinct as predicted by the PVA simulation model under
4 different scenarios. The mortality rate for each stage was taken from the
original life table.
Scenario Number of years until extinction
-1 SE Average +1SE
Basic
21
28
33
4a. Mild disease
15
20
25
4b. Severe disease
2
3
4
5a. Class 3 hurricane
22
25
28
5b. Class 4 hurricane
21
23
25
5c. Class 5 hurricane
20
22
24
6a. Reduce migration (25%)
21
30
34
6b. Reduce migration (50%)
19
30
34
7a. Destroy largest patch
20
27
32
7b. Destroy 5 largest patches
21
27
33
7c. Destroy 10 smallest patches
21
28
33


212
Figure A.8--A map of Noname Key. The habitat patches used in the study are numbered.


231
Hawthorne, D. W. 1980. Wildlife damage and control techniques. Pages 411-440
in Wildlife management techniques, 4th edition (S. D. Schemnitz, ed.), The
Wildlife Society, Washington, D.C.
Hayne, D. W. 1949. Calculation of size of home range. Journal of Mammalogy
39:190-206.
Hilbom, R., Redfield, J. A. and C. J. Krebs. 1976. On the reliability of
enumberation for mark and recapture census of voles. Canadian Journal of
Zoology 54:1019-1024.
Hill, E. P. 1969. Observations of imported fire ant predation on nesting cottontails.
Proceedings of the Annual Conference of the Southeastern Association of
Game and Fish Commission 23:171-181.
Hill, E. P. 1972. The cottontail rabbit in Alabama. Bulletin 440. Agricultural
Experimentation Station, Auburn University, Auburn, Alabama, 103 pp.
Hoffmeister, J. E. 1974. Land from the sea, the geologic story of South Florida.
University of Miami Press, Coral Gable, Florida.
Holler, N. R. and M. F. Sorenson. 1969. Changes in behavior of a male swamp
rabbit. Journal of Mammalogy 50:832-833.
Holler, N. R., and C. H. Conaway. 1979. Reproduction of the marsh rabbit
(Sylvilagus palustris) in south Florida. Journal of Mammalogy 60:768-777.
Holling, C. S. 1978. Adaptive environmental assessment and management, Wiley,
New York, 377 pp.
Holt, R. D. 1993. The influence of regional processes on local communities. Pages
77-88 in Ecological communities: historical and geologic perspectives (R. E.
Ricklefs and D. Schluter, eds.), University of Chicago Press, Chicago.
Howe, S. E. 1988. Lower Keys marsh rabbit status survey. Report to U.S. Fish
and Wildlife Service, 3100 University Blvd. South, Suite 120, Jacksonville.
8 pp.
Humphrey, S. R. 1992. Rare and endangered biota of Florida: Volume 1.
Mammals. University Press of Florida, Gainesville, Florida.


141
formal listing, habitat destruction has abated somewhat, although development
of the rabbits habitat is still possible.
Methods
The RAMAS/metapop (Ak9akaya 1994) simulation model was used for
the PVA because of its flexibility and ability to incorporate a large number of
populations. Processes occurring at all 3 spatial levels were incoiporated in the
model. At the level of the population, a life-history table in the form of a
Leslie matrix (Leslie 1945) was completed, the impact of demographic and
genetic stochasticity assessed, the form of the population growth curve was
estimated, and the initial abundances for each population were estimated. At
the between-population level, the number and age of the individuals moving
between populations was determined, and the spatial structure of the
metapopulation and the impact this has on inter-population movements was
estimated. At the metapopulation level, the degree of correlation between
population density was estimated.
Population-level Parameters
A stage rather than age Leslie matrix was used based on the results of
Chapter 3, which suggested that S. p. hefneri differs in sum val and
reproduction depending on the developmental stage. Five stages were used:


7
Debate currently exists over how many examples of true
metapopulations exist in nature (Harrison 1994). This may be due in part to
the lack of studies that embody the spatial and temporal scales necessary for
studying metapopulations.
Metapopulation Persistence
For a metapopulation to persist, the rate of patches being colonized and
established must exceed the rate of patches going extinct (Levins 1970, Hanski
1989). A metapopulation is said to be at equilibrium when these rates are
equal. Departures from equilibrium towards increasing extinction rate may
ultimately result in a regional extinction. There may be a threshold number
and configuration of occupied patches, below which the species is not likely to
rebound (Hanski 1991), similar to the concept of a minimum viable population
of individuals (MVP; Shaffer 1981).
Persistence of individual patches is determined by dynamics that occur
both within the patch and by movements between patches. Population
dynamics, internal to patch dynamics, such as natality, mortality, sex ratio, and
age structure, interacting with patch size, habitat quality, and predator density
determine the size of the local population and its variability.
Between-patch dynamics are shaped by a species' home range, spacing
and movement patterns, especially dispersal. Movements between patches are


142
nestlings (0-3 months), juveniles (4-7), subadults (8-10), first-year adults, and
second- (and older-) year adults. Survivorship was estimated using a
combination of live-trapping and radio-telemetry, as was reproductive effect
per female (see Chapter 3). Males and females were combined when
determining all rates.
Demographic stochasticity (e.g., sex ration, demographic structure) was
studied by monitoring marsh rabbit populations at 5 patches of habitat.
Genetic variation was determined by taking blood from the ears of marsh
rabbits and performing starch-gel enzyme electrophoresis (as described in
Chapter 3).
Determining whether population growth of S. p. hefneri is dependent or
independent of density provided a challenge. Detecting density dependence in
natural populations is difficult because of the variation caused by stochastic
demographic and environmental processes (Burgman et al. 1993). Few studies
have actually documented that density-dependence occurs, despite a profusion
of studies. Several methods of detecting density dependence using sequential
censuses have been developed, but none of these methods reliably test density
dependence (Hassell 1986). An experimental manipulation of population
density over time is the best way to determine density dependence. However,
when working with endangered species, manipulations may be difficult
because they may affect the population and because the necessary time may not


Proportion
132
Dispersal (m)
Figure 5.7 Average distance between a patch and the nearest occupied patch (patch
isolation) for the 59 patches, in comparison to the dispersal distances of the 6 female
Lower Keys marsh rabbits (Svlvilaeus palustris hefneri) and the 11 male.


105
isolation (Figure 5.6), area did not have a large effect on whether a patch had
rabbits present (variable or occupied). There did appear to be a maximum
isolation (inter-patch distance) beyond which no rabbits occurred on patches.
This maximum isolation occurred around 2000 m (2 km) and is similar to the
maximum dispersal distance for males (Figure 5.7).
When the metapopulation structure of the S. p. hefneri was compared to
the predicted structures (Figure 5.1), it appeared that it did not fit 1 pattern
perfectly. The structure was most similar to the classical metapopulation
structure. An examination of patch extinction and recolonization on Big Pine
(Figure 5.8) demonstrates that patches of all sizes went extinct during the study
and were often recolonized. Of the 22 sites that had had pellets present for at
least 1 but not all of the surveys, 11 were observed to go extinct during the
study. Four of those local extinctions were recolonized.
The Lower Keys marsh rabbit metapopulation was least similar to the
mainland-island structure. In the mainland-island pattern, only large patches
are permanently occupied; yet in the observed pattern, patches of all size
appeared to be permanently occupied and some of the largest sites were
variably occupied.
In the source-sink model, low-quality patches that are near source
patches are often variably occupied, while low-quality patches that are further
away are empty. This appeared to be partially true in the marsh rabbit


6
Between these 2 extremes are species that spend at least a portion of
their life in 1 habitat patch but are capable of moving to other habitat patches
during their lifetime. For populations of these species, local extinctions may
be counteracted by colonizations from nearby patches, and the overall
persistence time of the species may be collectively longer than the persistence
time of any one patch (Levins 1969). Populations of species that exhibit this
dynamic are said to exist together as a metapopulation (Levins 1970).
Metapopulations
Recently the term "true or classic metapopulation" has been used to
distinguish between this type of population configuration and others that
outwardly appear to be metapopulations such as mainland-island and source-
sink configurations (Harrison 1991). In the mainland-island configuration,
most movement occurs from a larger patch to smaller patches. These
metapopulations do not function in the "true" reciprocal colonization pattern
unless the mainland population goes extinct (Thomas and Jones 1993) and is
recolonized by the smaller "islands. In a source-sink metapopulation, the
higher quality source habitat patch produces a surplus of descendants (rate of
population growth or A. > 0), whereas the low quality patches may be sinks that
produce a deficit (A < 0) (Pulliam 1988, Pulliam and Danielson 1991).


153
(c). The 10 smallest patches are destroyed. The majority of these
patches are on Boca Chica and Saddlebunch Keys. This scenario is very likely,
the patches are all owned by the Navy or privately owned.
Results
Population Parameter Estimations
The Leslie matrix was completed using estimates of survivorship and
reproduction given in Chapter 3. Survivorship of nestlings and the standard
error was 0.28 (0.17), for juveniles it was higher at 0.75 (0.10), for subadults it
was 0.67 (0.14), for first-year adults it was 0.52 (0.09) and for second-year and
older adults it was 0.10 (0.12). Reproductive rates were assumed to be the
same for first- and second- year adults. The reproductive rate was obtained by
multiplying the average female fecundity rate by the survivorship of the stage
class and then dividing by 2 to account for males. The rate for first-year adults
was 2.19 (0.07) and 1.70 (0.10) for second-year adults.
The marsh rabbit populations did not exhibit evidence of inbreeding
depression or low genetic variation. Sylvilagus palustris hefneri had a level of
genetic variation similar to large populations of cottontail (Scribner and Warren
1986); genetic effects were not factored into the model. Demographic
stochasticity did appear in the marsh rabbit populations. Several populations
were so small that at times they occasionally only supported members of the


CHAPTER 5
METAPOPULATION DYNAMICS: PATCH OCCUPANCY AND
HABITAT QUALITY
Introduction
In Levins (1969) metapopulation model, all habitat patches were
identical in size and quality. More recent models (Taylor 1991, Thomas et al.
1992, Hanski 1994, Hanski et al. 1994), have taken an incidence-function
approach (Diamond 1975) that predicts the probability of a species occurrence
based on the area of the habitat and the distance of the habitat patch to other
patches. In these models, all patches are assumed to be inhabitable past a
threshold size or minimum isolation (Hanski 1994). For species whose natural
history is relatively unknown, an important aspect of a metapopulation
occupancy model is determining if unoccupied habitats are vacant due to lack
of suitable habitat or from past extinctions unrelated to the habitat patch
quality.
Metapopulation Structure
Before studying habitat use, metapopulation structure should be studied
to determine how the subpopulations interact with each other. In a classical
92


32
1969). Raccoons are more likely than cats to eat the breast, crop, and entrails
of their prey. They may also carry portions of the prey to water (Anderson
1969).
Results
A total of 54 Lower Keys marsh rabbits was caught and examined
during this study (Table 3.1). Forty-three were caught on the 6 main grid sites
(41 on Boca Chica, 2 on Saddlebunch), 5 rabbits were caught while trying to
recapture collared dispersed rabbits, and 6 were caught on other sites in the
Lower Keys. Rabbits were only examined once per trapping period. Data
from the 54 rabbits was recorded 130 times during 8 trapping sessions (June
1991 May 1993).
Forty-three (28 male, 15 female) of the rabbits caught were fitted with
radio-collars, and 7 of these were juveniles. More juveniles were caught
before the break-away collar technology was complete. Forty-one of the radio-
collared rabbits were on Boca Chica Key and two were on Saddlebunch Key.
Growth and Morphology
Mass of the 54 marsh rabbits ranged from 100 to 1400 grams (including
pregnant females). It was judged that the 100-gram rabbits had just left the
nest and were approximately 1 month old. Several young rabbits were causght,


232
Humphrey, S. R. 1994. Endangered mammals of Florida: is Mammalogy credible?
Journal of Mammalogy 75:551-552.
Humphrey, S. R. and D. B. Barbour. 1981. Status and habitat of three subspecies
of Peromysus polionotus in Florida. Journal of Mammalogy 62:840-844.
Ingles, L. G. 1941. Natural history observations on the Audubon cottontail. Journal
of Mammalogy 22:227-250.
Jellison, W. L., Owen, C. R., Bell, J. F., and G. M. Kohls. 1961. Tularemia and
animal populations: ecology and epizootiology. Wildlife Diseases 17:1-22.
Jennrich, R. I., and F. B. Turner. 1969. Measurements of noncircular home range.
Journal of Theoretical Biology 22:227-237.
Johnson, D. H. 1980. The comparison of usage and availability measurements for
evaluating resource preference. Ecology 61:65-71.
Johnson, M. K. and H. A. Pearson. 1981. Esophageal, fecal and exclosure
estimates of cattle diets on a longleaf pine-bluestem range. Journal of Range
Management 34:232-234.
Johnson, M. K., Wofford, H. and H. A. Pearson. 1983. Microhistological
techniques for food habits analyses. Research paper SO-199, U.S.
Department of Agriculture, Southern Forest Experimental Station, New
Orleans, Louisiana, 40 pp.
Jones, E., and B. J. Coman. 1981. Ecology of the feral cat, Felis catus (L.), in
south-eastern Australia. Australian Wildlife Research. 8:537-547.
Jurewicz, R. L., Cary, J. R and O. J. Ronstad. 1981. Spatial relationships of
breeding female cottontial rabbits in southwestern Wisconsin. Pages 295-
309 in Proceedings of the world lagomorph conference (E. Myers, and C. D.
Machines, eds.), Guelph University Press, Guelph, Ontario.
Karr, J. R. 1982. Avian based extinction of Barro Colorado Island, Panama: a
reassessment. American Naturalist, 119:220-239.
Kilpatrick, C. W. 1981. Genetic structure of insular populations. Pages 28-59 in
Mammalian population genetics (M. H. Smith and J. Joule, eds.), University
of Georgia Press, Athens, Georgia.


26
the variability of these parameters, and the extrinsic factors that influence this
variability. This information about the population biology was used in the final
model predicting the future of the Lower Keys marsh rabbit presented in
Chapter 6.
Methods
Two methods were employed to study the population biology of the
Lower Keys marsh rabbit: live-trapping and radio-telemetry. Marsh rabbits
were studied at a subsample of 6 habitat patches. All of the patches chosen
were on Navy-owned land; a diversity of patch size and shape was sampled.
Five habitat patches were selected on Boca Chica and 1 on Saddlebunch Key.
Additional sites throughout the Lower Keys were used for the portion of the
study that examined the genetic composition of the marsh rabbits.
Trapping Grids
Individual marsh rabbits were examined by trapping the 6 main sites.
Trapping occurred twice during the wet season (June November), and twice
during the dry season (December May). For the 5 sites on Boca Chica Key
(Figure 3.1), trapping occurred from June 1991 to May 1993 (8 trapping
sessions). On Saddlebunch, trapping was conducted from June 1992 to May
1993 (4 trapping sessions). Trapping grids were placed on each site, using


36
Natality
Eleven adult female marsh rabbits were radio-collared, followed for >1
month, and used in this portion of the analysis. Length of time followed
ranged from 2 to 22 months (X = 9.09 months, SE = 1.94 ) for a total of >100
months. Thirty-one nesting events were recorded; all females were observed to
nest and produce a litter at least once. The number of young observed per nest
ranged from 1 to 3 with an average of 1.77 kittens per nest (SE = 0.09).
All nests were made in clump grasses (thick grasses and sedges), with
22 of the nests predominately in Spartina spartinae and the other 9 in
Fimbrystvlis castanae. In general, the nests consisted of a main chamber with
several smaller chambers and exit/entry routes. None of the nests were
obviously lined with fur as reported in the northern subspecies (Tomkins
1935). Only 2 of the females used the same nesting area more than once, and
none of the female rabbits used the nest of another rabbit during this study.
There was little apparent seasonal pattern in the reproduction of the
marsh rabbit (Figure 3.4). Combining data from the 2 years of the study, the
proportion of rabbits that produced litters each month ranged from 0-56%. The
highest proportion of females with litters was seen in March and September;
the lowest proportion was seen in April and December. The average number of
litters produced during the wet and diy seasons did not differ significantly (G^
= 0.15, df = 1, P > 0.05). Although reproduction in most cottontail species is


115
to be closest to the classic metapopulation in structure. Implications of these
conclusions will be further examined in Chapter 7.


67
of the radio-collared rabbits were on Boca Chica Key and 2 were on
Saddlebunch Key.
Home Range
Sufficient data were available for seven juveniles, 13 adult males, and
10 adult females for the home-range analysis (Table 4.3). Adult males were
followed an average of >8 months (98 locations) and adult females were
followed an average of >10 months (124 locations). Comparisons using the
Wilcoxon test for differences between adult males and females did not show
significant differences between the average distance moved between locations
(Z = -1.71, P = 0.09), the 95% harmonic-mean home range (Z = -1.33, P =
0.18), or the core area (Z = -1.21, P =0.23). Resident adult males and females
were similar with respect to their home-range shape and size (Figure 4.4).
Both sexes appeared to establish a permanent home range shortly after maturity
(9-10 months), and both resided in that area for the duration of their lives.
Combining both sexes, the average 95% harmonic mean was 3.96 ha (SE =
0.65) and the average core area was 1.21 ha (SE = 0.87). Home-range size
estimates for juvenile rabbits were not significantly different from adults, but 3
of the juveniles did not develop core areas. The lack of core area formation
may be an attribute of juvenile home ranges or an artifact of the low number of
locations used in the analysis.


oo
Figure 4.3To determine if a dispersing Lower Keys marsh rabbit (Sylvilagus palustris hefneri) used certain habitat as a
corridor, the habitat type the rabbit moved through was compared to the habitat available. Using the centerpoint of the
rabbits natal home range and the distance dispersed as a radius, a circle of available habitat was drawn. The arrow
indicates the actual patch the rabbit took.


110
Discussion
Metapopulation Structure
In the comparison of occupied sites (consistently and variably) and
empty patches, the metapopulation structure of S. p. hefheri conforms to the
classic definition of a metapopulation (Levins 1969, Hanski and Gilpin 1991).
Nearly all patches that are below a maximum inter-patch distance are occupied
and patches of all areas are occupied. Lack of a minimum area effect may be
due to the methods for patch identification; habitat patches were included in
the study only when they exceeded 0.5 ha.
Few good examples of classic metapopulations in nature exist in the
literature (Harrison 1991, 1994). This may be a function of the duration of
most population studies and scarcity of larger-scale research; otherwise, most
systems studied are not true metapopulations. Species that have been found to
exist in a classic metapopulation may have some unifying life-histories and
behaviors (Murphy et al. 1990), although debate exists as to what these
characteristics include (see Harrison 1994). Small-bodied, short-lived species
with high reproductive rates (r-selected species) appear to be the most likely
candidates for classic metapopulations (Murphy et al. 1990), because such
species generally are good colonists and have exceptional dispersal abilities
(Harrison 1994). The Lower Keys marsh rabbit fits this description well, as do
many species of small mammals, invertebrates, and annual plants.


167
Minimum Viable Population, Area, and Number of Patches
Using the current Leslie matrix values, it does not seem that there is any
minimum population size (Soul 1980) that could reasonably be called viable
for even short-term persistence. With the 25% increase in survivorship, the
effective population size (Ne) varied around 167 (or 334 adults if a 1:1 sex ratio
is assumed). This is above the hypothesized minimum level of 50 to avoid
inbreeding problems, but below the level of 500 individuals, which is the level
necessary to maintain genetic diversity (see Lande and Barrowclough 1987).
Minimum viable area seems less appropriate in this situation than
minimum number of patches. The number of occupied patches was seen to be
more important than the size of the patches, although the connectivity of the
patches might also be important. Judging from the results of scenario #7
(habitat loss), this minimum number of patches for metapopulation persistence
is between 20 and 30 patches spread over at least 3 keys.
These results suggest that although S. p. hefneri is probably currently
moving toward extinction, an increase in survivorship might cause the
subspecies to persist in a classic equilibrium metapopulation. For the
Lower Keys marsh rabbit, the metapopulation paradigm can form a useful
framework from which to build recovery actions.


61
the traps were open, 2 nights with the traps closed and another 5 nights with
traps open. Traps were checked twice daily, once in the morning and once in
the evening and were covered in burlap for shade. All rabbits caught were
sexed, weighed, and tagged (Monel no. 3, National Band and Tag, Newport,
KY).
Radio-telemetry
Each rabbit weighing >1,000 grams was fitted with a radio-collar and a
transmitter with an estimated 10-month operational life (Telonics, Inc., Mesa,
AZ). Smaller rabbits (300-1,000 g) were fitted with a similar radio-collar, but
with a velcro break-away device added to allow the animals to lose the
equipment as their bodies grew larger. Collars were replaced as the animals
aged.
Collared rabbits were located on separate days 3 times a week, once in
the early morning (7-9 a.m.), once at mid-day (11 a.m. 1 p.m.), and once in
the evening (4-6 p.m.). Signals were followed until the animal could be seen
or the exact location was found. All locations were made >24 hours apart to
ensure independence of observations. Locations were plotted on 1:2400 aerial
maps with 10-m2 grid overlays. All road and water crossings were recorded
and the distance of the water crossing was measured.


54
(/)
0
ro
n
o
0
_Q
E
3
Vehicle Cat Cat/Rac Snake Rac Poach
Cause of Mortality
Figure 3.6Cause of death for 24 Lower Keys marsh rabbits (Sylvilagus
palustris hefneri). (Vehicle car, truck, or plane; Cat = domestic cat; Cat/Rac
= unable to distinguish if predator was a cat or a raccoon; Snake = eastern
diamond back rattlesnake; Rac = raccoon; Poach = death by gun)


60
within the patch and individuals only interact with others living at the same
patch. There may be occasional movements between patches and/or some of
the dispersing individuals will successfully move to new patches. The areas
surrounding the patches act as a barrier to most movements and is not included
in the home range of the individuals. These patches contain sub or local
populations that exist within a greater metapopulation (Levins 1970). These
patches are not necessarily genetically distinct but are demographically
isolated. Matings between individuals at different patches does not occur.
3. Lower Keys marsh rabbits move regularly between patches and may
use several patches at once. The home ranges encompass several patches at
one period of time and individuals interact with others from other patches
regularly. All of the patches are part of one large population, which inhabits a
highly fragmented environment.
Methods
Marsh rabbits were trapped at the 5 main sites on Boca Chica. Trapping
occurred twice during the wet season (May October), and twice during the
dry season (November April) from May 1991 to May 1993 (8 trapping
sessions). Trapping grids were placed on each site, using unbaited collapsible
National live traps (80 x 30 x 30 cm), placed in a 6 x 6 array, spaced
approximately 25 m apart. Each trapping session consisted of 5 nights where


71
the males that dispersed left patches were the male present could have been
their father, but the other 5 left patches were there either were no males
present, or a male too young to be their parent was present. The 1 subadult
male that did not disperse occupied a patch where there were no other adult
males.
Corridor Use
Subadult rabbits traveled through a variety of habitats between their
natal and permanent home ranges. Three rabbits crossed a dirt road, 7 crossed
a 2-lane road and 3 rabbits were observed crossing taxiways and runways.
None of the rabbits crossed the 4-lane highway (US 1), but none of the
dispersal radii encompassed or were adjacent to the highway. Two rabbits
swam across ditches, 1 across a canal, and 1 crossed a (12 m) body of water.
In general, most of the subadult rabbits traveled through areas with
dense ground cover. These marsh rabbits were recorded traveling through
mangroves, upland hardwood hammocks, and in the vegetation between the
shoulder of the road and the water. The narrowest strip of plant cover used
(corridor) by a dispersing marsh rabbit was 3-5 m wide. The Johnson test that
compared the amount of each habitat traveled through compared to its use was
significant (JF = 6.62, df = 3,14, P = 0.005). A Waller-Duncan comparison
found that rabbits used areas of mangrove, hardwood hammock, and transition


76
resident fitness hypothesis as being the sole explanation. Only the inbreeding
avoidance hypothesis is consistent with all of the data collected. It is possible
that dispersal may be motivated by several factors (Dobson and Jones 1985)
and that there may be variance between individuals. This interpretation is
based only on current dispersal patterns. Current dispersal behavior may be a
response to population structure in evolutionary time. The marsh rabbit
population structure has probably undergone dramatic changes since the
colonization of man in the Keys.
Conclusions
These results imply that the S. p. hefneri should be managed as a
metapopulation. Each local population is socially isolated from the other
populations. Interchange generally occurs by movement of subadult males and
this movement is facilitated by the occurrence of habitat corridors. The impact
these conclusions has on the persistence of S._g. hefneri will be addressed in
chapter 6.


Average Temperature ( F)
21
1 2 3 4 5 6 7 8 9 10 11 12
Month
10
9
8
7
6
5
4
3
2
1
0
Figure 2.2-The average temperature (line, left axis) and average precipitation
(bars, right axis) during the past 25(1965-1990) years measured at Key West
airport, Key West, Florida.
Average Precipitation (in.)


CHAPTER 7
CONCLUSIONS AND MANAGEMENT RECOMMENDATIONS
Population Viability Analysis (PVA) and Metapopulation Dynamics
Recently, the political and legal battle over the spotted owl (Strix
occidentalis occidentalis) and other listed species have spurred debate about
the uses of PVA (Boyce 1992, Harrison 1994). A PVA model can have vastly
different outcomes depending on the parameters and algorithms used, and
therefore might not be legally defensible. Because of these legal problems,
Harrison (1994) suggests that we should seek sensible alternatives to PVA
such as comparing the abundances of species of concern in forests (or other
habitats) of varying degrees of fragmentation. However, it is difficult to see
how correlative studies will exceed the power of a good PVA and how this will
solve the legal problems associated with endangered species.
At its worst, conducting a PVA can demonstrate where more data is
needed, and the results can be interpreted knowing the assumptions of the
model. This will provide more information than simple estimates of minimum
viable population size (MVP) and extinction probabilities associated with MVP
(Boyce 1992).
193


56
Home-range size varies greatly among individuals, populations and
species of the genus Svlvilagus (see Chapman et al. 1982); Only 1 study has
attempted to estimate home-range size for the marsh rabbit (S. palustris). Blair
(1936) estimated a linear home range based on a maximum home range
width of 183 m from trapping data.
It is generally assumed that most cottontails are not territorial (Chapman
et al. 1982). This assumption is based largely on studies of the eastern
cottontail (S. floridanus; Haugen 1942; Chapman and Trethewey 1972, Dixon
et al. 1981) and the brush rabbit (S. bachmani; Shields 1960, Chapman 1971).
Yet, Trent and Rongstad (1974) reported that female eastern cottontails did not
have overlapping home ranges during the breeding season. Jurewicz et al.
(1981) found small amounts of overlap in the nocturnal home ranges of
breeding female eastern cottontails, and suggested that a spacing mechanism
was in operation. Male swamp rabbits (S. aquaticus) have exhibited linear
dominance hierarchies involving non-overlapping, defended home ranges in
several studies (Marsden and Holler 1964, Sorenson et al. 1968, Holler and
Sorenson 1969). Additionally, Kjolhaug and Woolf (1988) found no overlap
between individuals of the same sex in both male and female swamp rabbits.
Information on the spacing behavior of the other 10 species of cottontails
(genus Svlvilagus) is lacking.


Table D. 1--Continued.
Site
code
Adult3
Juv.3
Adult4
Juv.4
Adult4
Juv.j5
Adult6
Juv.6
Adult7
Juv.7
Adult8
Juv.8
31
2
1
0
1
0
1
0
1
0
1
0
1
1
32
2
1
0
1
0
1
0
1
0
1
0
1
0
33
2
1
1
1
1
1
1
1
0
1
1
1
1
34
0
1
0
0
0
0
0
0
0
0
0
0
0
35
1
0
0
0
0
0
0
0
0
0
0
1
0
36
0
0
0
0
0
0
0
0
0
0
0
0
0
37
0
0
0
0
0
0
0
0
0
0
0
0
0
38
0
0
0
0
0
0
0
0
0
0
0
0
0
39
0
0
0
0
0
0
0
0
0
0
0
0
0
40
0
0
0
0
0
0
0
0
0
0
0
0
0
41
0
0
0
0
0
0
0
0
0
0
0
0
0
42
0
0
0
0
0
0
0
0
0
0
0
0
0
43
0
0
0
0
0
0
0
0
0
0
0
0
0
44
0
0
0
0
0
0
0
0
0
0
0
0
0
45
0
0
0
0
0
0
0
0
0
0
0
0
0
46
0
0
0
0
0
0
0
0
0
0
0
0
0
47
0
0
0
0
0
0
0
0
0
0
0
0
0
48
0
0
0
0
0
0
0
0
0
0
0
0
0
49
0
0
0
0
0
0
0
0
0
0
0
0
0
50
0
0
0
0
0
0
0
0
0
0
0
0
0
51
0
0
0
0
0
0
0
0
0
0
0
0
0
52
1
0
0
0
0
0
0
0
0
0
0
1
0
53
2
1
1
1
1
1
0
1
0
1
1
1
0
54
1
1
0
1
1
1
0
0
0
0
0
0
0
55
1
1
0
1
0
0
0
0
0
0
0
0
0
56
1
1
0
1
1
1
0
0
0
0
0
0
0
57
2
1
1
1
1
1
0
1
1
1
0
1
1
58
1
1
0
0
0
0
0
0
0
1
0
1
1
59
0
0
0
0
0
0
0
0
0
0
0
0
0


29
equipment as their bodies grew larger. Collars were replaced as the animals
aged. Previous studies on cottontail rabbits (S. floridanus) found that mortality
rates were not statistically different for radio-telemetered rabbits and rabbits
marked using other methods (Trent and Rongstad 1974, Rose 1977).
Collared rabbits were located on separate days three times a week, once
in the early morning (7-9 a.m.), once at mid-day (11 a.m. 1 p.m.), and once in
the evening (4-6 p.m.). Signals were followed until the animal could be seen
or the exact location was found. All locations were made >24 hours apart to
ensure independence of observations. Locations were plotted on 1:2400 aerial
maps with 10-m2 grid overlays. Because the 5 sites on Boca Chica were
studied for 2 years and the Saddlebunch site was only studied during the final
year, natality and mortality portions of the study only used data from the 5 sites
on Boca Chica Key (Figure 3.1).
Natality
Natality was studied by examining females during each trap session for
pregnancy (by palpating the uterus) and lactation. Nesting was determined by
following collared female rabbits until they centered their activities around one
small area. Nest confirmation was made by observing the young through a
tunnel in the grass or by finding deposits of juvenile pellets near the nest area.
Where possible, the number of young observed in or fleeing the main chamber


244
BIOGRAPHICAL SKETCH
Elizabeth A. Forys was bom September 12, 1966, in Berkeley, California.
She attended the University of Virginia in Charlottesville, where she received a
B.A. in environmental sciences and biology, in 1988. In 1990, Elizabeth received
her M.S. in environmental sciences with a concentration in ecology. Her thesis title
was The effect of immigration on the demographic and genetic composition of
Oryzomvs palustris on the Virginia Barrier Islands. In August of 1990, she entered
the graduate program in the Department of Wildlife and Range Sciences (currently
the Department of Wildlife Ecology and Conservation) and will receive her Ph.D. in
May of 1995. After graduation, Beth is returning to the Lower Keys to work on the
recovery efforts of several endangered species.


APPENDIX D
MARSH RABBIT PATCH OCCUPANCY


171
Table 6.3 Fitting the frequency distribution of the number of pellets per
sampling unit to the Poisson model for sample units using 0.5-m and 1-m radii.
The degree of correspondence is measured using a chi-square test. A
significant result indicates that the distribution is not random.
0.5-m radius 1.0-m radius
Site
Chi-square d.f.
P < 0.05
Chi-square
d.f.
P<0.
1
12.46
6
no
86.48
6
yes
2
3.88
2
no
7.48
2
yes
3
0.07
1
no
0.11
1
no
4
16.66
9
no
36.22
10
yes
5
0
0
0
0
6
2.70
3
no
7.18
2
yes
7
3.52
9
no
3.88
6
no
8
17.73
6
yes
101.31
6
yes
9
24.70
9
yes
26.79
9
yes
10
31.76
7
yes
37.68
7
yes
11
34.71
8
yes
61.58
8
yes
12
7.47
4
no
12.35
9
yes
13
60.64
9
yes
77.73
9
yes
14
0.84
4
no
8.96
3
yes


237
Opdam, P. 1990. Metapopulation theory and habitat fragmentation: a review of
holartic breeding bird studies. Landscape Ecology 5:93-106.
Oxley, D. J., Fenton, M. B. and G. R. Carmody. 1974. The effects of roads on
small mammals. Journal of Applied Ecology, 11:51-59.
Packer, C. 1979. Inter-troup transfer and inbreeding avoidance in Papio anubis.
Animal Behaviour 27:1-36.
Payne, N. F. 1975. Range extension of the marsh rabbit in Virginia. Chesapeake
Science 16:77-78.
Pietz, P. J. and J. R. Tester. 1983. Habitat selection by snowshoe hares in North
Central Minnesota. Journal of Wildlife Management 47:686-696.
Pimm, S. L. 1991. The balance of nature? Ecological issues in the conservation of
species and communities. University of Chicago Press, Chicago, Illinois.
434 pp.
Pollock, K. H., Nichols, J. D., Brownie, C. and J. E. Hines. 1990. Statistical
inference for capture-recapture experiments. Wildlife Monographs 107:1-97.
Proulx, G. 1988. Control of urban wildlife predation by cats through public
education. Environmental Education 15:358-359.
Pulliam, H. R. 1988. Sources, sinks, and population regulation. American
Naturalist 132:652-661.
Pulliam, H. R. and B. J. Danielson. 1991. Sources, sinks and habitat selection: A
landscape perspective on population dynamics. American Naturalist
137:S50-S66.
Pusey, 1987. Sex-biased dispersal and inbreeding avoidance in birds and mammals.
Trends in Ecology and Evolution. 2:295-299.
Quinn, J. F. and A. Hastings. 1987. Extinction in subdivided habitats.
Conservation Biology 1:198-208.
Ralls, K. and J. Ballou. 1983. Extinction: lessons from zoos. Pages 164-184 in
Genetics and Conservation: a reference for managing wild animal and plant
populations (C. M. Schonewald-Cox, S. M. Chambers, B. MacBryde, and L.
Thomas, eds.), Benjamin/Cummings, Menlo Park, California.


150
and a reduction in mortalities attributable to raccoons. Raccoon population
reduction might occur when supplemental food sources are eliminated (i.e.,
open dumpsters, outdoor pet food).
Scenario #2 Decrease Road-kills.
All vehicular deaths are prevented. This would mean a roughly 25%
increase in the survivorship of subadults and 10% increase in the number of
adults. Implementing this management strategy would require strict speed limit
reductions, or an over(under) pass for the rabbits to use.
Scenario #3- Reintroduce Rabbits.
Rabbits are reintroduced to all vacant habitat patches. This would entail
using captive breeding to produce rabbits to be released to the 17 empty habitat
patches (see Chapter 5). The number and stage of the rabbits to be introduced
would be determined by calculating carrying capacity for each patch and
assuming a stable age distribution.
Scenario #4 Disease.
(a). A relatively mild disease (e.g. tularemia) that kills some of the
infected individuals spreads through the keys. The disease kills animals in the
stages that have social contact (i.e., the nestlings and adults). The disease


158
Increasing the survival rate of the juveniles and subadults and increasing
reproduction did not produce large changes in the fate of the population.
In scenario #1, the increase in survivorship in all stages by 25 %
resulted in a stabilization of the predicted population size and metapopulation
occupancy (Figure 6.4). The population varied around 1,200 total individuals
occupying approximately 30 patches. When survivorship was increased by
50%, the population exponentially grew to nearly 8,000 individuals, and all
patches were eventually occupied.
In scenario #2, elimination of vehicular mortalities only prolonged the
persistence time by approximately 10 years (Figure 6.5). Similarly, in scenario
#3, when all the currently empty patches were recolonized by a carrying
capacity level of marsh rabbits in scenario #1, the population only persisted
slightly longer than without the recolonizations (Figure 6.6). The average time
to extinction was 25 years.
Because scenarios 4-7 represented further threats to the persistence of
the marsh rabbit, these simulations were run both with the original life table
and using the life history parameters in scenario #l(a) (survivorship was
increased by 25% in all stages). When scenarios 4-7 were run using the actual
life table survivorship estimates, 1 of the scenarios slightly increased the
number of years before extinction and the other 3 scenarios hastened the
amount of time before extinction (Table 6.6). Reducing migration 25% and


181
Figure 6.5Results of Scenario #2 (all vehicular deaths are avoided) of the Lower
Keys marsh rabbit (Sylvilagus palustris hefhen) PVA model, where survivorship was
increased in the subadult stage by 25% and 10% in the adult 1 stage.


30
of the nest was recorded. Although this method is approximate, it provided an
estimate of the number of individuals surviving at the time of nest discovery.
Number of litters for each female was calculated for both the wet and dry
seasons and compared using a likelihood-ratio Chi-square (G^). The nest area
was mapped, and the dominant vegetation was identified.
Mortality
Percent mortality was calculated for 5 age classes: nestling, juvenile,
subadult, first-year adult, and second-year adult. Nestling mortality (0-3
months) was calculated by comparing the number of young observed in nests to
the number of juveniles caught on the trapping grid. Juvenile mortality (4-7
months) was estimated using telemetry for individuals >300 g. Subadult (8-10
months), first-year and second-year adult mortality was determined from the
telemetry data alone. First-year adults were rabbits that had been followed as
an adult for 1 year; most had been collared as juvenile or subadults. Second-
year adults were rabbits that had been followed as adults for 2 years. First- and
second-year adult classes may have contained individuals older than 2 and 3
years old.
All mortalities were located within 2 days of death. When possible,
field necropsies were performed as outlined by Wobeser and Spraker (1980),


152
(b). Same scenario as (a), only the hurricane is class 4 and therefore the
mortality is higher. Class 4 hurricanes occur less frequently than class 3. The
probability of this strength of a hurricane is <5%.
(c). Same scenario as (a) and (b), but the hurricane is class 5 and
therefore will cause the greatest amount of mortality. This type of hurricane
has a <1% chance of occurring each year.
Scenario #6 Corridor Destruction.
(a). Corridor habitat is destroyed and this decreases migration by 25%.
All patches are affected.
(b). Corridor habitat is destroyed and decreases migration by 50%. All
patches are affected.
Scenario #7 Habitat Destruction.
(a). The largest patch (site 33, 31.5 ha) is destroyed. This patch is
privately owned and could potentially be developed.
(b). The 5 largest patches are destroyed. This would include site #33
and 4 large areas on Big Pine. This is a less likely scenario because the 4 large
areas on Big Pine curr ently are protected.


65
dispersal. A hierarchical hypothesis design was used (Table 4.1) to attempt to
narrow the potential cause of dispersal.
First, a binomial probability test was used to determine if the sex ratio
of the dispersers was significantly different from unity. If this sex-ratio
showed a significant bias, then trapping and radio-telemetry data were
examined to determine if other adult males inhabited the dispersers natal
patch. Trapping records of the adult males were further examined to determine
if these males could potentially be the father of the dispersers. Potential
fatherhood was assumed if the adult male was present at the dispersers natal
patch around the time when the dispersers were believed to have been
conceived.
Corridor Use
To determine if dispersing individuals were randomly leaving a patch or
if they were influenced by potential corridors, paths of all individuals making
one-way long distance movements were plotted. The proportion of each
movers trip (measured in meters) that covered different habitat types and the
number of roads, runways, and bodies of water they crossed was measured.
Table 4.2 lists the habitat types and features compared.
Using this distance as a radius with the arithmetic centerpoint of the
natal home range as the center, a circle was drawn for each moving individual