<%BANNER%>

Effects of off-road vehicles on small mammals in Big Cypress National Preserve

Permanent Link: http://ufdc.ufl.edu/UFE0024505/00001

Material Information

Title: Effects of off-road vehicles on small mammals in Big Cypress National Preserve
Physical Description: 1 online resource (102 p.)
Language: english
Creator: Jeffery, Brian
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2009

Subjects

Subjects / Keywords: bicy, florida, occupancy, orv, recapture, rodent
Interdisciplinary Ecology -- Dissertations, Academic -- UF
Genre: Interdisciplinary Ecology thesis, M.S.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Effects of off-road vehicles (ORV) on the environment have been debated as many public lands have restricted or closed areas to off-road vehicles. The use of ORVs in Big Cypress National Preserve has prompted managers to evaluate their use and impacts on natural resources. I studied small mammal and vegetation response to off-road vehicles. The vegetation structure of ORV areas was different from that in non-ORV areas. I found that percent cover and average graminoid height in ORV areas were significantly less than those in non-ORV areas. Abundance of small mammals was different for each species. Hispid cotton rats had higher abundance in non-ORV areas than in ORV areas; marsh rice rats showed no difference in abundance between areas; and cotton mice showed higher abundance in ORV areas than in non-ORV areas. Occupancy of the three species of small mammals was estimated in relation to ORV use. Off-road vehicle use was a significant factor in site occupancy for hispid cotton rats, while percent cover and habitat type were better indicators for cotton mice. Slight differences in survival between ORV and non-ORV areas were not significant. Small mammals showed some relationship to ORV use and with additional research could be good indicators of ORV impacts.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Brian Jeffery.
Thesis: Thesis (M.S.)--University of Florida, 2009.
Local: Adviser: Mazzotti, Frank J.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2009
System ID: UFE0024505:00001

Permanent Link: http://ufdc.ufl.edu/UFE0024505/00001

Material Information

Title: Effects of off-road vehicles on small mammals in Big Cypress National Preserve
Physical Description: 1 online resource (102 p.)
Language: english
Creator: Jeffery, Brian
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2009

Subjects

Subjects / Keywords: bicy, florida, occupancy, orv, recapture, rodent
Interdisciplinary Ecology -- Dissertations, Academic -- UF
Genre: Interdisciplinary Ecology thesis, M.S.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Effects of off-road vehicles (ORV) on the environment have been debated as many public lands have restricted or closed areas to off-road vehicles. The use of ORVs in Big Cypress National Preserve has prompted managers to evaluate their use and impacts on natural resources. I studied small mammal and vegetation response to off-road vehicles. The vegetation structure of ORV areas was different from that in non-ORV areas. I found that percent cover and average graminoid height in ORV areas were significantly less than those in non-ORV areas. Abundance of small mammals was different for each species. Hispid cotton rats had higher abundance in non-ORV areas than in ORV areas; marsh rice rats showed no difference in abundance between areas; and cotton mice showed higher abundance in ORV areas than in non-ORV areas. Occupancy of the three species of small mammals was estimated in relation to ORV use. Off-road vehicle use was a significant factor in site occupancy for hispid cotton rats, while percent cover and habitat type were better indicators for cotton mice. Slight differences in survival between ORV and non-ORV areas were not significant. Small mammals showed some relationship to ORV use and with additional research could be good indicators of ORV impacts.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Brian Jeffery.
Thesis: Thesis (M.S.)--University of Florida, 2009.
Local: Adviser: Mazzotti, Frank J.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2009
System ID: UFE0024505:00001


This item has the following downloads:


Full Text

PAGE 1

1 EFFECTS OF OFF-ROAD VEHICLES ON SMALL MAMMALS IN BIG CYPRESS NATIONAL PRESERVE By BRIAN MATTHEW JEFFERY A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2009

PAGE 2

2 2009 Brian Matthew Jeffery

PAGE 3

3 To my parents, David and Mary Jeffery, who ha ve encouraged and supported me to do what I enjoy

PAGE 4

4 ACKNOWLEDGMENTS I am very grateful to my committee members for taking the time and patience to work with me: Frank Mazzotti for the opportunity and support that I needed to be successful throughout my graduate career, Ken Rice for the encouragemen t and guidance when needed, and Doria Gordon for her inspiration and insightful suggestions. Franklin Percival was especially helpful fo r direction and support wh ile in Gainesville. Hardin Waddle was instrumental for stimulating conversations, helping me in the field and for mentoring me. I would also like to thank those that have helped me in the field: Meghan Riley, Deborah Kramp and Aletris Neils, various volunteers from Student for Conservation Association and VIP personnel from Big Cypress National Pr eserve. All methods for handling and marking small mammals have been approved by the Universi ty of Florida Instituti onal Animal Care and Use Committee (IACUC). Big Cypress National Preserve staff provide d an office, housing, logistics and granted research permit. Jim Snyder of the USGS, for sh aring of his knowledge and for his support in Big Cypress National Preserve. Finally, I thank my parents. With their co ntinued encouragement, support, and most importantly, love, I was able to obtain my dream.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS...............................................................................................................4LIST OF TABLES................................................................................................................. ..........7LIST OF FIGURES.......................................................................................................................10ABSTRACT...................................................................................................................................11 CHAP TER 1 INTRODUCTION..................................................................................................................12Off-Road Vehicles..................................................................................................................12Types of ORVs................................................................................................................13Swamp buggies........................................................................................................ 13Street-Legal vehicles................................................................................................14All-Terrain vehicles.................................................................................................. 14Airboats.................................................................................................................... 14Uses of ORVs..................................................................................................................14Ecology of Small Mammals in Southern Florida................................................................... 15Sigmodon hispidus ...........................................................................................................15Oryzomys palustris ..........................................................................................................16Peromyscus gossypinus ...................................................................................................17Blarina carolinensis ........................................................................................................17Potential Impacts to Wildlife.................................................................................................. 18Habitat Modification....................................................................................................... 19Species Composition and Structure.................................................................................20Alteration of Behavior.....................................................................................................20Objectives...............................................................................................................................212 RESPONSE OF VEGETATION TO OFF-ROAD VEHICLES ............................................ 26Introduction................................................................................................................... ..........26Study Area..............................................................................................................................29Methods..................................................................................................................................30Results.....................................................................................................................................32Discussion...............................................................................................................................323 ECOLOGY OF SMALL MAMMALS IN BIG CYPRESS NATIONAL PRESERVE IN AREAS AFFECTED BY ORV USE ..................................................................................... 41Introduction................................................................................................................... ..........41Methods..................................................................................................................................42Study Area.......................................................................................................................42

PAGE 6

6 Site Selection...................................................................................................................44Small Mammal Surveys.................................................................................................. 44Statistical Analysis.......................................................................................................... 45Results.....................................................................................................................................48Species Distribution and Abundance............................................................................... 48Weight and Size Comparisons......................................................................................... 49Body Condition...............................................................................................................51Mark-Recapture...............................................................................................................52Discussion...............................................................................................................................544 CONCLUSIONS.................................................................................................................... 90Introduction................................................................................................................... ..........90ORV Impacts on Wildlife.......................................................................................................90Habitat Modification....................................................................................................... 90Species Composition and Structure.................................................................................91Small Mammals as an Indicator Species......................................................................... 92Conclusions.............................................................................................................................93LIST OF REFERENCES...............................................................................................................94BIOGRAPHICAL SKETCH.......................................................................................................102

PAGE 7

7 LIST OF TABLES Table page 2-1. Overall variation in hei ght am ong three vegetation classifications (forb, graminoid, and shrub) and two impact levels (off-road vehicle (ORV) and non-off-road vehicle (Non-ORV) at Big Cypre ss National Preserve.................................................................. 352-2. Variation in height among three vegetation classifications (forb, graminoid, and shrub) and two impact levels (off-road vehicle (ORV) and non-off-road vehicle (Non-ORV) in pine habitat at Big Cy press National Preserve............................................................... 362-3. Variation in height among three vegetation classifications (forb, graminoid, and shrub) and two impact levels (off-road vehicle (ORV) and non-off-road vehicle (Non-ORV) in prairie habitat at Big Cypress National Preserve........................................................... 372-4. Variation in vegetative percent ground cove r between habitats (p ine and prairie) and impacts (off-road vehicle (ORV) and nonoff-road vehicle (Non-ORV) (Percent cover data was arcsin transformed for analysis)................................................................ 382-5. Variation in vegetative vertical cover am ong pine habitats and impacts (off-road vehicle (ORV) and non-off-road vehicle (Non-ORV)...................................................................392-6. Variation in vegetative vertical cover among prairie habitats and impacts (off-road vehicle (ORV) and non-off-ro ad vehicle (Non-ORV)....................................................... 403-1. Combinations of the 3 site covariates and one sampling covari ates that were used in the occupancy analysis for each species. Each set of site covariates was modeled along with each set of sampling covariates for a total of 16 models for each species................ 633-2. List of 25 models analyzed in program MARK for captures of cotton rat, marsh rice rat, and cotton mouse in Big Cypress National Preserve. Explanation defines each model in terms of the effects of time (t) and ORV group (g), on apparent survival ( ) and capture probability (p)........................................................................................................643-4. Number of captures by hispid cotton rat, marsh rice rat, cotton mouse and short-tailed shrew on impact (off-road vehicle (ORV ) and non-off-road vehicle (Non-ORV)) treatment areas. Individuals captured more than once within the same trapping event were not counted................................................................................................................663-5. Total captures and individual captures of hispid cotton rat, marsh rice rat, cotton mouse and short-tailed shrew by off-road ve hicle (ORV) and non-off-road vehicle (NonORV) areas.........................................................................................................................673-6. Total captures of hispid cotton rat, marsh rice rat, cotton mouse, and short-tailed shrew by pine and prairie habitats................................................................................................ 68

PAGE 8

8 3-7. Mean SD captures of hispid cotton rat, m arsh rice rat, cotton mouse, and short-tailed shrew by impact (off-road vehicle (ORV) and non-off-road vehicle (Non-ORV)), habitat (pine and prairie), and sex (male and female)........................................................ 693-8. Mean weight SD of adult, non-pregnant hispid co tton rat, marsh rice rat, and cotton mouse by impact (off-road vehicle (ORV ) and non-off-road vehicle (Non-ORV)), habitat (pine and prairie), and sex (male and female)........................................................ 703-9. Mean length SD of hispid cotton rat, marsh ri ce rat, and cotton mouse by impact (offroad vehicle (ORV) and non-off-road ve hicle (Non-ORV)), habitat (pine and prairie), and sex (male and female).................................................................................... 713-10. Linear regressions of ln (body weight) over ln(right hind foot))of hispid cotton rat, marsh rice rat, and cotton mouse....................................................................................... 723-11. Mean body condition indices ( SD) of hisp id cotton rat, marsh rice rat, and cotton mouse by impact (off-road vehicle (ORV ) and non-off-road vehicle (Non-ORV)), habitat (pine and prairie), and sex (male and female)........................................................ 733-12. Number of detections by cotton rat, mars h rice rat, and cotton mouse and proportion of sites at which a detection o ccurred (nave occupancy) during small mammal surveys..... 743-13. Program PRESENCE model selection resu lts for the hispid cotton rat, including Akaikes Information Criterion (AIC) and the delta AIC and AIC weights for all models with any weight.....................................................................................................753-14. Program PRESENCE beta estimates, standa rd errors (S.E.), and lower and upper 95% confidence intervals for the hispid cotton ra t. The best model and second best model are shown...........................................................................................................................763-15. Program PRESENCE model selection re sults for the marsh rice rat, including Akaikes Information Criterion (AIC) and the delta AIC and AIC weights for all models with any weight.....................................................................................................773-16. Program PRESENCE beta estimates, standa rd errors (S.E.), and lower and upper 95% confidence intervals for the marsh rice rat. The best model and second best model are shown.................................................................................................................................783-17. Program PRESENCE model selection result s for the cotton mouse, including Akaikes Information Criterion (AIC) and the delta AIC and AIC weights for all models with any weight..........................................................................................................................793-18. Program PRESENCE beta estimates, standa rd errors (S.E.), and lower and upper 95% confidence intervals for the best model of the cotton mouse............................................. 803-19. The number of individuals marked, recap tured and the return rate (proportion of marked individuals recaptured at least once) for the hispid cotton rat, marsh rice rat,

PAGE 9

9 and cotton mouse in off-road vehicle (O RV) and non-off-road vehicle (Non-ORV) areas.......................................................................................................................... .........813-20. Model selection table for Cormack-Jo lly-Seber closed population mark-recapture model of the cotton rat, including Quasilikelihood Akaikes Information Criterion for small sample sizes (QAICc), model weights based on QAICc, the number of parameter in each model, and the model deviance. Model structure includes the effects of time (t), and off-road vehicl e (ORV) group (g) on apparent survival ( ) and capture probability (p)................................................................................................. 823-21. Estimates, standard error (S.E.), and 95% co nfidence interval (C.I.) of the beta values for the off-road vehicle (ORV) effect on apparent survival ( ) and capture probability (p) on the cotton rat, ma rsh rice rat, and cotton mouse................................... 833-22. Model selection table for Cormack-Jo lly-Seber closed population mark-recapture model of the marsh rice rat, including Quasi-likelihood Akaikes Information Criterion for small sample sizes (QAICc ), model weights based on QAICc, the number of parameter in each model, and the model deviance. Model structure includes the effects of time (t), and off -road vehicle (ORV) group (g) on apparent survival ( ) and capture probability (p)............................................................................ 843-23. Model selection table for Cormack-Jo lly-Seber closed population mark-recapture model of the cotton mouse, including Quasi-likelihood Akaikes Information Criterion for small sample sizes (QAICc ), model weights based on QAICc, the number of parameter in each model, and the model deviance. Model structure includes the effects of time (t), and off -road vehicle (ORV) group (g) on apparent survival ( ) and capture probability (p)............................................................................ 85

PAGE 10

10 LIST OF FIGURES Figure page 1-1. Image of a swamp buggy commonly used in Big Cypress National Preserve, Ochopee, Florida. ....................................................................................................................... ........221-2. Image of a typical street -legal 4 used in Big Cypress National Preserve, Ochopee, Florida........................................................................................................................ ........231-3. Image of an all-terrain vehicle (ATV ) commonly used in Big Cypress National Preserve, Ochopee, Florida................................................................................................241-4. Image of an airboat commonly used in Big Cypress National Preserve, Ochopee, Florida........................................................................................................................ ........253-1. Total number of captures for hispid cott on rat, marsh rice rat and cotton mouse from October 2004 to September 2005...................................................................................... 863-2. Apparent survival ( ) and 95% confidence interval of cotton rat by off-road vehicle (ORV) group and total rainfall for the 10 mont hly survival intervals. Estimates for cotton rats are averaged across models as no model had a majority of QAICc weight (Burnham and Anderson 1998).......................................................................................... 873-3. Apparent survival ( ) and 95% confidence interval of marsh rice rat by off-road vehicle (ORV) group for the 10 monthly sampling interv als. Estimates for marsh rice rats are averaged across models as no model had a majority of QAICc weight (Burnham and Anderson 1998)..................................................................................................................883-4. Survival ( ) and 95% confidence interval of co tton mouse for the 10 monthly sampling intervals...................................................................................................................... ........89

PAGE 11

11 Abstract of Thesis Presen ted to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science THE EFFECTS OF OFF-ROAD VEHICLES ON THE SMALL MAMMALS OF BIG CYPRESS NATIONAL PRESERVE By Brian Matthew Jeffery May 2009 Chair: Frank J. Mazzotti Cochair: Kenneth G. Rice Major: Interdisciplinary Ecology Effects of off-road vehicles (ORV) on the e nvironment have been debated as many public lands have restricted or closed areas to off-road vehicles. Th e use of ORVs in Big Cypress National Preserve has prompted managers to eval uate their use and impact s on natural resources. For this thesis, I studied small mammal and vegetation response to off-road vehicles. The vegetation structure of ORV areas was different from that in non-ORV areas. I found that percent cover and average graminoid height in ORV areas were significantly le ss than those in non-ORV areas. Abundance of small mammals was different for each species. Hispid cotton rats had higher abundance in non-ORV areas than in ORV areas; marsh rice rats showed no difference in abundance between areas; and co tton mice showed higher abundance in ORV areas than in nonORV areas. Occupancy of the three species of sm all mammals was estimated in relation to ORV use. Off-road vehicle use was a significant factor in site occupancy for hispid cotton rats, while percent cover and habitat type were better indi cators for cotton mice. Slight differences in survival between ORV and non-ORV areas were not significant. Small mammals showed some relationship to ORV use and with additional resear ch could be good indicators of ORV impacts.

PAGE 12

12 CHAPTER 1 INTRODUCTION Accelerating population growth in southern Florida has incr eased recreational pressure on Big Cypress National Preserve (BCNP) (National Park Service 2000). Land managers are confronted with a difficult si tuation of protectin g wildlife and habitat while demands for recreational access to public lands have increased (Cordell and Bergstrom 1991; English et al. 1993; Knight and Gutzwiller 1995). It has been difficult for off-road vehicle (ORV) enthusiasts to find areas in which it is legal to ride. Larg e expanses of public and private lands are no longer available for ORV use in southern Florida. In north Florida, many ar eas are open to ORV use including Ocala National Forest, Osceola National Forest, Apalachicol a National Forest, and Croom Motorcycle Area in Withlacoochee State Forest. Big Cypress National Preserve is the only large tract of public land remaining in so uthern Florida in which ORV enthusiasts can legally ride. The large area of pr airie and pine habitats within BCNP make it an ideal place to hunt, but this also exposes these habitats to ORV use. There is little known about impacts of ORVs on wildlife, especially small mammals, in BCNP. For this reason I discuss the ecology of small mammals in southern Florida, historical us es of ORVs in BCNP, what types of ORVs are used, and potential consequences of ORV use on wildlife. Finally, I desc ribe my research in BCNP and outline the rest of the thesis. Off-Road Vehicles The use of motorized v ehicles to explore the Big Cypress Swamp region began in the 1920s in association with logging, farming, oil exploration, and development of the Tamiami Trail (Tebeau 1997). Recreationa l use of ORVs was not common until the 1940s. By the 1950s, logging activities produced ORV trails which remained even afte r logging (Duever et al. 1986).

PAGE 13

13 An increase in hunting in creased use of these trails and e xpanded the trail system to over 1,100 km by 1973 (Duever et al. 1986). On October 11, 1974, Big Cypress National Pres erve was created (16 USC 698) in order to assure the preservation, conservation, and protection of the natural, scenic, hydrologic, floral and faunal, and recreational values of the Big Cypress Watershed in the State of Florida and to provide for the enhancement and public enjoym ent thereof. The recreational values of Big Cypress Watershed include the use of ORVs Widespread ORV use on many federal lands concerned environmentalists in th e 1970s and prompted research. The majority of the research was focused on sandy coasts (Steiner and Leatherman 1981, Wolcott and Wolcott 1984, Anders and Leatherman 1987) and arid landscapes of th e southwest (Vollmer et al. 1976, Eckert et al. 1979, Webb 1983). Executive Orders were enac ted in 1972 (Order # 11644) and 1977 (Order # 11989) for control of ORV use. These executive orders allow closures of ORV trails that cause harm to an ecosystem. Types of ORVs Several types of ORVs are used in B CNP in cluding swamp buggies, st reet-legal 4s, allterrain vehicles, and airboats. Trail bikes, track vehicles, and vehicl es with chains on the tires are prohibited from use within the Preserve (Nationa l Park Service 2000). All vehicles must past a detailed safety and compliance inspection prefor med by the Preserve before a permit is issued (National Park Service 2000). Swamp buggies There are many varieties of swa mp buggies, all of which are constructe d from a four-wheel drive automotive frame (Figure 1-1). Completed buggies usually have an elevated platform constructed over an engine and frame. Tires vary from aircraft to street -legal brands and must meet a minimum width of nine inches (National Park Serv ice 2000). Swamp buggies can access

PAGE 14

14 every part of the Preserve except for areas design ated for airboats or environmentally sensitive habitats. Swamp buggies account for approxima tely 30 percent of ORV permits in BCNP (National Park Service 2000). Street-Legal vehicles Street-legal vehicles are fou r-wheel drive veh icles that ar e commercially manufactured and sold. Street-legal vehicles must also have a minimum tire width of nine inches (Figure 1-2; National Park Service 2000). Street-legal vehicles are limited to the Bear Island Unit of the Preserve and account for approximately 18 perc ent of ORV permits in BCNP (National Park Service 2000). All-Terrain vehicles All-terrain v ehicles (ATVs) are small, commercially manufactured vehicles that are designed for off-highway use (Figur e 1-3). The ATV is 50 inches or less in width and has a total dry weight of less than 900 lbs. ATVs account for approximately 39 perc ent of ORV permits in BCNP (National Park Service 2000). Airboats Airboats a re commercially manufactured or ha nd built (Figure 1-4). They consist of a fiberglass or aluminum hull powered by an aircraft or automotive engine the drives an aircraft propeller. Airboats are restricted to Zone 4 which is a deep wa ter area (National Park Service 2000). Airboats account for approximately 13 per cent of ORV permits in BCNP (National Park Service 2000). Uses of ORVs Hunting and hunting-related activities, such as preseason scouting, are predom inant uses of ORVs within the Preserve (National Park Service 2000). Non-hunti ng activities include camping, pleasure driving, fishing, sightseeing, wildlife viewing and photography, socializing,

PAGE 15

15 picnicking, and swimming (National Park Servic e 2000). The Preserve utilizes ORVs for law enforcement, research, and resource management. Ecology of Small Mammals in Southern Florida Sm all mammal communities often respond rapidly to changes in plant composition (Kaufman et al. 1983) and habitat structure (Abramsky 1978; Kaufman et al. 1983). Because small mammals occupy vital positio ns in food webs (Kaufman et al. 1998), they are useful indicators of biological change (Dale and Beyeler 2001). Small mammals can also respond quickly to disturbance (Clark et al. 1989). Sma ll mammals are ubiquitous and prolific; they can be used by managers as a tool for assessing conditions across landscap es (Dale and Beyeler 2001). I selected Sigmodon hispidus (hispid cotton rat), Oryzomys palustris (marsh rice rat) Peromyscus gossypinus (cotton mouse) and Blarina carolinensis (short-tailed shrew) as study animals because they are commonly found in pine and prairie habitats that are found throughout BCNP. Sigmodon hispidus The hispid cotton rat is a m edium-sized rodent with a tail shorter th an its body. Ears are blackish, medium sized, rounded and buried in th e neck fur. The upper pelage has a coarse, grizzled appearance and is a combination of br own, black, and light tan banded hairs. The belly is grayish or light brown. The tail is sparsely ha ired and weakly bicolore d, black above and light brown below. Hispid cotton rats are one of the most common mammals in Florida and are abundant statewide (Brown 1997). This species is found throughout the southern United States as well as Central and South America. Hispid cotton rats occur in a wide variety of open and semiopen habitats. They are most abundant in old fields composed of de nse grasses. In south Florida, hispid cotton rats occupy both prairie and uplan d habitats (Bigler and Jenkins 1975; Smith and Vrieze 1979; Mazzotti et al. 1981). During the dry s eason, hispid cotton rats are present in dry

PAGE 16

16 prairie and upland habitats (Bigler and Jenkins 1975; Smith and Vrieze 1979). But during the wet season, hispid cotton rats were found more frequently in upland habitats because of increased water levels (Bigler and Jenkins 1975; Smith and Vrieze 1979). Hispid cotton rats make well-defined runw ays through grasses which are good indicators of presence. A small bulbous nest is constructed of dry grass or fibers stripped off the stems of larger plants. It is usually placed underground in a shallow tunnel, but oc casionally the nest is built on the surface in dense vegetati on. Hispid cotton rats eat leaves, stems, roots, and seeds of many grasses, sedges, legumes, and other herbac eous plants. They also consume some insects and bird eggs, and will even feed on carcasses. Oryzomys palustris The m arsh rice rat is a small-sized rodent with a slender, poorly hair ed tail that is about equal to the size of its body. The upper pelage is light brown sli ghtly scattered with black. The color is darkest down the middle of the back w ith more brown on the sides. The underparts are grayish white and the feet are white. The tail is bicolored, brownish above and grayish white below. The body fur is coarse and fairly long. The ears are medium sized and round, and extend beyond the fur. Marsh rice rats are found in the southeastern United States. They are most abundant in freshwater marshe s, saltwater marshes, and wet grassy meadows. They are sometimes found along wet ditches and the open e dges of lakes and streams and are very good swimmers. In south Florida, marsh rice rats are abundant in freshwater prairies and sloughs (Smith and Vrieze 1979; Mazzotti et al. 1981). In the wet se ason, marsh rice rats can commonly be found in prairie habitats (Smith and Vrieze 1979; Mazzotti et al. 1981). During the dry season, marsh rice rats abandon the prairie habitats for more mesi c sites (Smith and Vrieze 1979). Marsh rice rats are active mostly at night and to some extent during the day. They eat seeds of grasses and

PAGE 17

17 sedges, tender green plants, berr ies, fruits, fungi, snails, insect s, crustaceans, and bird eggs (Brown 1997). Marsh rice rats ma ke extensive runway systems where the vegetation is thick enough (Brown 1997). Their nests are grapefruit-s ized spherical masses suspended in thick vegetation over water. Feeding platforms of matted vegetation are sometimes constructed in marshy environments. Peromyscus gossypinus The cotton mouse is a fairly la rge deer m ouse. The upper pelage can vary from chestnut brown to dark brown or grayish brown. The middle of the back is darker than the sides. The underparts are white and there is no orange wash down the sides where the white meets the dark upper pelage. The tail is equal to or slightly l onger than the body. It is sparsely haired and distinctly bicolored with th e upper side brown and the unders ide white. The ears are large, rounded and gray. Cotton mice occur throughout Fl orida and are most abundant in mature hardwood forests (Brown 1997). They the most abundant mouse found in Florida (Brown 1997). In south Florida, cotton mice are most abundant in upland habitats (Smith and Vrieze 1979; Mazzotti et al 1981). In the dry season, cotton mice were primarily found in upland habitats, but some were caught in prairie habitats (Smith and Vrieze 1979). During the wet season, cotton mice are found in upland habitats (Bigler and Jenkins 1975; Smith and Vrieze 1979; Mazzotti et al. 1981). Cotton mice are active at night and often nest in logs, hollow stumps and holes in the ground. They also climb trees and build a spherica l nest of leaves and plant fibers in hollow cavities. Food includes acorns, nuts, and seeds of trees, as well as snails, slugs, spiders, and insects. They have also been known to eat several types of mushrooms and other fungi. Blarina carolinensis The southern short-tailed shrew is an insecti vorous rodent that has degenerate eyes and sm all ears hidden by fur. Short-tailed shrews rang e is in the southeastern United States and is

PAGE 18

18 found throughout Florida, except for the Keys. Shor t-tailed shrews are most common in wooded areas, but are sometimes found in old fields, brushy areas and forest openings Their fur is velvet like and grey, and their tails are short and ba rely reach beyond the extended hind foot. Their metabolic rate is high and they spend most of their time foraging for food. They have a heart rate of 160 beats per minute while respiration is around 150 breaths per minute (Brown 1997). The diet of short-tailed shrews consists of mice, small snakes, snails, slugs, insects, spiders, and earthworms. This species is considered one of the most aggressive shrews and will attack and kill prey much larger than itself (Brown 1997). It has a submaxillary salivary gland that produces a neurotoxin and a hemotoxin that is introduced through a bite. The poison causes the preys breathing and heart rate to slow, and the prey is th en either eaten or stored. Short-tailed shrew nests are co mprised of dried grasses and le aves and are usually built in stumps, logs, or underground. The nests are six to ei ght inches in diameter and may have several openings. Breeding primarily occurs from early sp ring to late fall with two or three litters producing two to eight young. The ge station period is 21 days and the young are weaned at four weeks and start to breed at three months. Potential Impacts to Wildlife Hum an intrusions into wildlife habitats can cau se various types and le vels of changes in wildlife and habitats (H ammitt and Cole 1998). Wildlife can re spond to ORVs in different ways depending upon type of activity, behavior of re creationists, frequency and magnitude of ORV use, timing, and location (Knight and Cole 1995). In teractions may be dire ct (Cole and Knight 1990) or indirect (Skagen et al. 1991; Pfister et al. 1992). Direct interactions involve various levels of disturbance, harassment, and harvesting of species, while indire ct interactions modify habitat and cause stress (e.g. noise) (Hammitt and Cole 1998).

PAGE 19

19 Habitat Modification Vegetation can be classif ied into three types of strata; ground cover, shrubs and saplings, and mature trees (Hammitt and Cole 1998). Ground cover is directly affected where ORV use breaks and crushes plants. Most plants e xhibit reduced abundance, height, vigor, and reproductive capacity in impacted sites (Hamm itt and Cole 1998). Cole (1985) found a linear relationship between certain vegetation type covers and amount of use. Cover loss is increased rapidly with initial incr eases in use. Low lying shrubs and saplings are part of the ground cover and are susceptible to tramp ling. McEwen and Tocher (1976) found 76 saplings per acre compared to 338 per acre in adjacent unused section. Possibly the most serious effect is its long term impact on recruitment. Tree reproduction is negligible as a result of trampling (Hammitt and Cole 1998). Major impacts on mature trees resu lt from vehicular impacts. Roots are exposed from erosion due to vehicular traffic (personal observation). Trees with exposed roots are more at risk of wind throw in major wind events, such as hurricanes. Habitat change can affect behavior, distribu tion, survivorship and re production of wildlife (Hammit and Cole 1998). Not only do ORVs cause physical damage to animals through the collapse of burrows and tunnels, they also reduce animals means to escape extreme temperatures. Snowmobiles compact snow on wh ich mice, shrews, and voles rely for its insulating properties (Schmid 1971; Stace-Smith 1978) Removal of shrubs and hazardous trees from campgrounds eliminates sources of food an d shelter for birds and small mammals (Webb 1968). ORV use at the beach mixes dry upper sand with wet lower sand e liminating a natural moisture barrier and increasing desiccation of the sand (Brodhead and Godfrey 1979). This lack of moisture dried out gills of the Ghost Crab (O cypote quadrata) causing mortality (Steiner and Leatherman 1981).

PAGE 20

20 Species Composition and Structure Population declines of the desert tortoise (Gopherus agasizii ) (Bury 1980) and Couchs spadefoot ( Scaphiopus couchi ) (Berry 1980) have been linke d to ORVs. Likewise, beluga whales (Delphinapterus leucas) abu ndance declined in the St. Lawrence Estuary due to increased boating activity over several years (Caron and Sergeant 1988). In the Netherlands, negative relationships between intensity of recr eation and population density were recorded for eight of 13 avian species; in areas where r ecreational use was common, 11 of the 12 most common species exhibited lower numbe rs than those in areas with fewer visitors (Van der Zande and Vos 1984). Similarly, in the west Nether lands where primary songs were affected by disturbances, birds appeared reluctant to establ ish territories (Reijnen et al. 1994). Adult bald eagles were more sensitive to disturbance than younger birds and preferre d areas of lower human activity (Stalmaster and Newman 1978). Townsends solitaires ( Myadestes townsedni) exhibited reduced numbers as far as away as 100 m from trails (Mi ller et al., 1998). Alteration of Behavior Disturbance can cause behavioral changes am ong wild anim als that range from slight modifications to extirpation (Hammitt and Co le 1998). Wildlife may react to disturbance by elevated metabolism, lowered body weight, reduced fetus survival, and withdrawal from suitable habitat (Geist 1971; Moen 1976). White-tailed deer ( Odocoileus virginianus) moved farther away from trails as snowmobile activity increased (Do rrance et al. 1975). Displacement of deer from suitable habitat during stressful winter c onditions may lead to changes in their energy budget, which could be detrimental (Do rrance et al. 1975). Florida panthers ( Puma concolor corys) were found farther from ORV trails during the hunting season in BCNP, which may be linked to prey behavior or di sturbance (Janis and Clark 2002). ORV disturbance at beaches reduced reproduction and recruitmen t in ghost crabs (Steiner and Leatherman 1981). Bald eagles

PAGE 21

21 ( Haliaeetu leucocephalus ) avoid using areas of high activity and were not as successful in feeding attempts where existing human activities were present (Stalmaster and Newman 1978). Feeding was the activity most sensitive to human disturbance. Birds th at were disturbed by ORVs did not return to feeding sites until several hours after the disturbance (Stalmaster and Newman 1978). Cetaceans such as bottlenose dolphins ( Tursiops truncatus ), killer whales ( Orcinus orca ) and harbor porpoises ( Phocoena phocoena) increased swimming speed (Kruse 1991), changed surfacing patterns (J anik and Thompson 1996), adjusted dive length (Evans et al. 1992) and altered foraging habitat selection (A llen and Read 2000) in response to boats. Objectives Objectives of this study we re to determ ine if ORV use influenced small mammal population and community structure and to inform managers in BCNP about the impacts of ORVs on small mammals. Chapter 2 describes in fluences that ORVs had on the vegetative communities that make up small mammal habita t in BCNP. Chapter 3 describes population responses of small mammals to ORVs, and appl ies the method of site occupancy estimation (MacKenzie et al. 2002) to descri be distribution of the three sm all mammal species in BCNP. Chapter 4 discusses the overall usefulness of sm all mammals as indicators of ORV impacts in southern Florida, and provides recommendations for monitoring of small mammals in BCNP.

PAGE 22

22 Figure 1-1. Image of a swamp buggy commonly used in Big Cypress National Preserve, Ochopee, Florida.

PAGE 23

23 Figure 1-2. Image of a typical str eet-legal 4 used in Big Cypress National Preserve, Ochopee, Florida.

PAGE 24

24 Figure 1-3. Image of an all-terrain vehicle (ATV) commonly used in Big Cypress National Preserve, Ochopee, Florida.

PAGE 25

25 Figure 1-4. Image of an airboat commonly used in Big Cypress National Preserve, Ochopee, Florida.

PAGE 26

26 CHAPTER 2 RESPONSE OF VEGETATION TO OFF-ROAD VEHICLE S Introduction Big Cypress National P reserve (BCNP) was established in 1974 to ensure preservation, conservation, and protection of natural scenic, floral and faunal, and recreational values of the Big Cypress Watershed in southern Florida. BCNP was the first national preserve integrated into the National Park Service and its establishment he lped to protect the hydrology of the southern Everglades ecosystem. BCNP is comprised of a mixture of pinelands, hardwood hammocks, prairies, mangrove forests, and cypress strands and domes (Duever et al. 1986). Many visitors take advantage of various r ecreational activities including hi king, nature viewing, canoeing, camping, hunting, and operating off-road vehicles. Off-road vehicles (ORVs) are used in many r ecreational activities, a nd as a result, there can be impacts to the surrounding environment. A number of studies have attempted to quantify damage caused by ORVs. ORVs have been shown to affect vegetation by reducing plant cover and height, decreasing species diversity, a nd altering community composition (Bates 1935, Chappell et al. 1971, Trew 1973, Liddle and Gr eig-Smith 1975, Boorman and Fuller 1977). Research on the effects of ORV use on vegetation has been conducte d in several habitat types. Impacts on dunes include reduction of total number of species, species diversity and total vegetation cover (Brodhead and Godfrey 1977, Ho sier and Eaton 1980). Grasslands have exhibited slow recovery and growth and reduced vegetative cover from impacts (Foresman et al. 1976, Wilshire et al. 1978, Webb 19 83). Bulk density and macropore space decreased in desert scrub habitats from impacts (Davidson and Fox 1974, Vollmer et al. 1976, Bury et al. 1977, Lathrop 1978). Snow compaction and decrease d subniviean airspace from snowmobiles damaged samplings in the tundra (Neuma nn and Merriam 1972, Wooding and Sparrow 1978,

PAGE 27

27 Emers et al. 1995). Off-road vehicl es striped vegetation and reduced soil moisture in chaparrals (Wilshire et al. 1978). Habitat fragmentation occurs when habitat loss or anthropogenic influences divide an area of relatively contiguous habitat in to smaller, fractured tracts. Fragmentation changes the amount of habitats, isol ates patches from other patches of the same habitats, and alters adjoining habitats by cha nging their spatial characteristics (Garrison 2005). However, few studies have examined the respon se of vegetation to ORV impacts in Floridas ecosystems. Wheeled ORVs, which are familiar in Florida swamps and marshes, are a common mode of transportation because they can traverse remote wetland areas Rapid population growth and resulting increase in recreation pres sure in southern Florida have led to concern about damage to natural resources. Land managers are confronted with a difficult and controversial situation of protecting wildlife and habitat while juggling demands for increasing recreational access to public lands (Cordell and Bergstrom 1991, Eng lish et at.1993, Knight and Gutzwiller 1995) including BCNP. ORVs were used in the Big Cypress Swamp re gion before the creation of BCNP; therefore that tradition was incorporated into the enabling legislation. The enabling legislation allows use of ORVs, but stipulates that ORV use will be controlled in a manner that will not harm resources of BCNP. Concerns over potential impacts on wetla nd ecosystems have been escalating because of increased use of ORVs. Possible negative eff ects include disturbance of wildlife, vegetation, soils, habitat fragmentation and changes in hydrology. ORV use can lead to habitat fragmentation by tire ruts, isolati ng patches of vegetation. In addi tion, there have been conflicts among recreationists, hunters, and environmen talists over perceptions of BCNPs natural resource use policy. These conflicts prompted BCNP to develop an ORV use management plan

PAGE 28

28 to assess potential outcomes of various management alternatives. One major concern is effects of ORV travel on vegetation. Pine a nd prairie habitats are relativel y easy to traverse with ORVs compared to cypress and hammock habitats. Cypress domes and strands have high density of trees and cypress knees that could cause damage to ORV drive train, while hammocks are typically composed of hardwood species that grow to large sizes that cannot be pushed over by ORVs permitted by BCNP. Prairie and pine habitats are ideal locations for finding game species, and as a result, hunters using ORVs utilize these habitats heavily. The only study to examine ORV use in BCNP ha bitats was conducted by Duever et al. in 1981 and 1986. Duever et al. (1981) detailed experiments conducted with four major ORV types (3-wheeled all-terrain cycles (ATCs), airboats, tracked vehicl es, and swamp buggies) used in BCNP and examined their impacts on four majo r vegetation associations (marshes with sand, marl and peat substrates; cypress; and pine habita ts). Experimental lanes were created in each of the habitats and impact levels were determined by the number of passes by a vehicle. One pass by a vehicle was categorized as a light impact, me dium impact sites were categorized by multiple passes that impacted vegetation but not soil, and heavy impact sites were categorized by multiple passes that severely impacted vegetation and cr eated ruts in the soil. Dwarf cypress was the habitat type found to be most sensitive to wheeled ORV impacts, while pine was most resistant. All impacted habitats were significantly diffe rent from control plot s (Duever et al. 1981). Vegetation recovery rates for the first year va ried from less than 25 percent to 50 percent, depending on the habitat and the type of vehicle used (Duever et al. 1981). After six years, there was still visible damage in most habitats (Dueve r et al. 1986). My study di ffers from Duevers in that I quantify the response of vegetation in response to ORV use by examining percent ground cover and vertical cover.

PAGE 29

29 Since the original studies by Duever a nd others, ORV use in BCNP has persisted, increased, and become very controversial. The National Park Service was sued by the Florida Biodiversity Project, U.S. Department of Interior, U.S. Fish and Wildlife Service, U.S. Environmental Protection Agency, U.S. Department of the Army and the U.S. Army Corps of Engineers in 1995 for failure to implement an ORV management plan. The purpose of the current study is to investigate impacts to vege tation structure in response to ORV use in BCNP. Based on previous studies (Duever et al. 1981 and 1986), we hypothesize that ORV use reduces percent ground cover, height, and vertical cover of vegetation. Study Area BCNP is located in south west Florida. The Bi g Cypress regional climate is affected by both tropical and temperate influences (Duever et al. 1986). The region is characterized by hot, humid summers and mild, dry wint ers. Southwest Florida has been classified as a tropical savanna due to its temperat ure and rainfall (Hela 1952). The landscape of BCNP contains 295,000 ha of a heterogeneous mixture of pine forests, prairies, cypress dome/strands, marshes, and is olated hardwood hammock s (Duever et al. 1986). Pine forests are open areas th at are composed primarily of South Florida slash pine ( Pinus elliottii var. densa ) and are usually bordered by wet prairies A fire frequency of three to seven years is required to prevent succession from pi ne forest to hardwood hammock (Hofstetter 1984). Pine forests are elevated from a few centimet ers to a meter or more above the surrounding lowlands, which results in a gr adient in hydroperiod. The variety densa developed longer taproots and smaller needle size than the northern variety of elliottii in response to different water conditions (McMinn and McNab 1971). Variety densa is also highly fire tolerant and young seedlings sprout needles at the root collar below the burn ed growing tip (Ketcham and Bethune 1963). Cabbage palms ( Sabal palmetto ) are often abundant in pine forests. They tolerate

PAGE 30

30 flooding by producing adventitious roots (Brown 1973) and have an embedded bud and a fireresistant trunk (Myers 1977). Saw palmettos ( Serenoa repens ) are often associated with pine forests and are well adapted to fire but do not withstand inundation as well as cabbage palms. Hardwood shrubs and trees can be found scattered throughout pine forests where fire has not been present. Ground cover is usually do minated by grasses such as bluestem ( Andropogon floridanus) and panic grass (Panicum hemitomon ). Sedges, rushes, and composites may also be present. Prairie communities are dominated by he rbaceous vegetation and grow along a hydroperiod gradient. They have mixed grasses, sedges, and other herbac eous plants with few trees. Prairies may be seasonally inundated or infrequently inundated depending upon elevation. Prairie habitats are maintained by fire and burn about every one to three years to prevent the invasion of woody vegetation (Wade et al 1980). Common sp ecies include maidencane ( Panicum hemitomon ), blackhead rush ( Schoenus nigricans), muhly ( Muhlendergia batatifolia ), and scattered sawgrass ( Cladium jamaicensis ). Maximum water levels rarely exceed 20 cm and are inundated approximately 50-150 da ys per year (Duever et al. 1975). Methods The study was conducted between October 2004 and September 2005. Study sites were located in B CNP near the Concho Billie ORV access area and Upper Wagonwheel Road (Collier County Road 837). The Concho Billie ORV access site has been open to ORVs since the inception of BCNP, and we designated it as the ORV area in our study. Upper Wagonwheel Road has been closed to ORVs for more than 20 years and was designated as the non-ORV area. A total of eight sample sites were randomly chosen from aerial maps that had at least 100 m of pine and an adjacent 100 m of prairie habita ts. In October 2004, four sites were established: two in ORV and two in non-ORV areas. In Januar y 2005, four more sites were established: two

PAGE 31

31 in ORV and two in non-ORV areas. At each sample site, a 200 m transect was established with 100 m of the transect in pine ha bitat and 100 m in adjacent prairie habitat. Sample points were located every 10 m along the transect for a total of 21 points per transect. Vegetation groups were categor ized as graminoids (sedge s and grasses), forbs (nongraminoids) or shrubs. Non-vegetation groups were categorized as bare gr ound or litter. At each sample point, a meter stick was used to measure average plant height to the nearest millimeter for each vegetation group. Percent ground cover wa s measured using a 1 m 1m quadrat made of PVC. Percent ground cover was used to estim ate the amount of ground covered by vegetation within the quadrat. Bare ground was recorded when no vegetative group was touched. Twine was laced to create 100 intersections 10 cm apart creating a one m2 quadrat. A wire was lowered at each intersection and the vegetation group first t ouched was recorded. Percent cover plots were measured 1 m from the sample location using random compass readings created from a random point generator ranging from 1 to 360. Percen t ground cover was calculate d by summing the total number of intersections at which each plant group occurred and dividing by 100. A cover board (Nudds 1977) was used to assess vertical cover. A board 2.5 m in height and 30.48 cm wide was marked alternately with white and black bands at 0.5 m intervals. The board was divided into five categories (0-20%, 2140%, 41-60%, 61-80%, and 81-100%). The vertical cover was assessed at each location by viewing th e board from 15 m away in a randomly chosen direction. All vegetation measurements were r ecorded during the dry season (January through March). Analysis of variance was used to determine if there were any differences in percent ground cover, vertical cover and plant height between ORV and non-ORV sites. All percent values were arcsine-transformed prior to analysis because th e data were not normally distributed (Zar 1984).

PAGE 32

32 Results Variation in vegetation height differed depending upon habitat type and O RV use. Overall graminoid and shrub height was significantly higher in non-ORV areas than in ORV areas (Table 2-1). Height values for forbs were not signifi cantly different between ORV and non-ORV areas. In pine habitats, graminoid heig ht was significantly higher in non-ORV areas than in ORV sites (Table 2-2). Forb and shrub heights were not significantly different from one another in nonORV and ORV pine sites. In prairie habitats, forb height was signifi cantly higher in non-ORV areas than in ORV areas (Table 2-3). Shrub height was significantly higher in ORV areas than in non-ORV areas (Table 2-3). There was no significan t difference in graminoid height in prairie habitats. Percent ground cover was significantly lower in ORV than in non-ORV areas (Table 2-4). ORV areas showed an 11.19 percent decrease in mean percent cover from non-ORV areas. Percent ground covers in pine habitats in ORV areas were significantly lower than in non-ORV areas. ORV areas showed a 13.01 percent decrease in mean percent ground cover from non-ORV areas (Table 2-4). Percent ground covers of pr airie habitats in ORV areas were significantly lower than non-ORV areas. An 8.51 percent d ecrease in mean percent ground cover in ORV from non-ORV prairie habitats was significant (Table 2-4). Vertical cover of pine vegetation in the 0.0m to 0.5m and 1.0m to 1.5m height classes were significantly higher in non-ORV than ORV sites (Table 2-5). There we re no significant diffe rences found in any other vertical cover comparisons (Table 2-5 and Table 2-6). Discussion The objectives of this study were to i nvestigate impacts ORVs have on vegetation communities. I hypothes ized that forb, graminoi d, and shrub heights would be lower in ORV sites than in non-ORV sites due to the increased traffic from vehicles. I found that different

PAGE 33

33 vegetation types respond differently to ORV imp acts. Overall graminoid height was higher in non-ORV sites than in ORV sites. In prairie habi tats, forb height was higher in non-ORV sites than ORV sites. On the other hand, in pine habi tats, graminoid heights and prairie shrub heights were higher in ORV sites than in non-ORV sites. Duever et al. (1981) found that initial ORV impact was significantly decrease d plant height from the control group; after one year without ORV use, they found that pine habitat had rec overed completely and marl marsh had recovered except for heavily impacted sites. A possible explanation for my finding that the graminoid height was higher in ORV pine si tes is that constant trampling from ORVs may cause the pine forest floor to revert to an early successional stag e. Shrub heights in ORV s ites in prairie habitats might be higher due to ORV users. Off-road ve hicle operators traveli ng through marl prairies would want to travel through a path of least re sistance and avoid anything that would slow down their momentum. Off-road vehicle operators would avoid hitting shrubs and would trample surrounding vegetation, which may benefit shrubs. My results also supported the hypothesis that percent ground cover in ORV sites would be less than in non-ORV sites. This finding applied to both pine and prairie site s. This is consistent with Duever et al.s (1981) finding that per cent ground cover in pine and marl prairie was significantly lower in ORV-impacted sites than in control sites. Duever et al. (1981) also found that increased use of ORVs decreased the percen t ground cover in marl prairies, but did not find any clear trends in the pine habitats. Reduced ground cover could impact wildlife. Many species rely on cover from vegetation to hide from predators. I hypothesized that vertical cover would be lower in ORV sites than non-ORV sites. Vertical structure is important to wildlife as it provides both hiding cover and thermal cover. The 0.0-0.5m and 1.0-1.5m height classes showed a difference between ORV and non-ORV sites.

PAGE 34

34 Long-term monitoring plots could be establishe d in current ORV use areas, recovering ORV use areas, and non-ORV use areas. Monitoring plots would help resource managers understand the effects of ORVs on vegetative communities and recovery of vegetation from ORV impacts. During this study, I did observe that ORV sites had a loss of continuous habitat from ORV ruts. These ruts may lead to habitat fragmentation. Habitat fragmentation is a set of mechanisms (i.e. ORV use) leading to the discontinuity in spatial distribution of resources and conditions present in an area at a given scal e that affects occupancy, reproduc tion, and survival of particular species (Franklin et al. 2002). Habitat fragmentation from ORVs may reduce size of natural patches and increase distance be tween patches. Fragmentation e ffects can be magnified by edge effects with remaining habitats (Lindenmayer et al. 1999). Duever et al. (1981) documented soil displacement and rut depth from ORVs in th eir experiment and documented its effects on vegetation. Ruts created by ORVs can alter depth and period of inundation of areas, resulting in the loss of vegetation and increased ponding in ru ts and artificial depressions (Duever et al. 1981). But effects on wildlife are unknown. Larger animals, such as the Florida panther ( Puma concolor coryi ) and the American alligator ( Alligator mississippiensis ), use ORV trails to travel and may benefit from ORV trails because it he lps them transverse dense habitat (personal observation). However, effects of ORV ruts on smaller animals may be more detrimental and may influence behavior patterns. Managers shou ld take into consideration the impacts ORVs have on vegetation. The decrease ground cover w ould leave many smaller sp ecies susceptible to larger predators. The results of this study raise additional questions for future research: To what extend does ORV use fragment pine and prairie habitats? Does the species composition change from non-ORV to ORV areas?

PAGE 35

35 Table 2-1. Overall variation in height among th ree vegetation classificat ions (forb, graminoid, and shrub) and two impact levels (off-ro ad vehicle (ORV) and non-off-road vehicle (Non-ORV) at Big Cypre ss National Preserve. Vegetation Type Impact n Mean Height F-Value p-value Forb ORV 87 199.00 0.00112 0.9 Non-ORV 88 198.43 Graminoid ORV 87 528.69 4.08 <0.05 Non-ORV 88 578.00 Shrub ORV 45 280.35 2.48 0.1 Non-ORV 36 275.14

PAGE 36

36 Table 2-2. Variation in height among three ve getation classifications (forb, graminoid, and shrub) and two impact levels (off-road vehicle (ORV) and non-off-road vehicle (NonORV) in pine habitat at Bi g Cypress National Preserve. Vegetation Type Impact n Mean Height F-Value p-value Forb ORV 38 3.50 0.0227 0.8 Non-ORV 32 3.64 Graminoid ORV 43 501.16 6.25 <0.05 Non-ORV 44 594.64 Shrub ORV 34 614.02 1.04 0.3 Non-ORV 33 509.80

PAGE 37

37 Table 2-3. Variation in height among three ve getation classifications (forb, graminoid, and shrub) and two impact levels (off-road vehicle (ORV) and non-off-road vehicle (NonORV) in prairie habitat at Big Cypress National Preserve. Vegetation Type Impact n Mean Height F-Value p-value Forb ORV 30 1.84 6.86 <0.05 Non-ORV 40 3.14 Graminoid ORV 44 556.23 0.011 0.9 Non-ORV 44 559.32 Shrub ORV 11 146.68 4.49 <0.05 Non-ORV 3 40.48

PAGE 38

38 Table 2-4. Variation in vegetativ e percent ground cover between ha bitats (pine and prairie) and impacts (off-road vehicle (ORV) and non-off -road vehicle (Non-ORV) (Percent cover data was arcsin transformed for analysis). Habitat Impact n Mean Percent Cover F-Value p-value Overall ORV 88 75.13 13.64 <0.001 Non-ORV 88 84.16 Pine ORV 44 72.34 10.26 <0.01 Non-ORV 44 83.16 Prairie ORV 44 77.91 4.41 <0.05 Non-ORV 44 85.16

PAGE 39

39 Table 2-5. Variation in vegetativ e vertical cover among pine ha bitats and impacts (off-road vehicle (ORV) and non-off -road vehicle (Non-ORV). Layer Impact n Mean Cover F-Value p-value 0.0-0.5 ORV 44 2.25 2.88 0.1 NonORV 47 4.00 0.5-1.0 ORV 18 1.25 0.628 0.5 NonORV 17 2.00 1.0-1.5 ORV 14 3.50 3.857 0.1 NonORV 20 5.00 1.5-2.0 ORV 4 1.00 0.0588 0.8 NonORV 5 1.25 >2.0 ORV 5 1.25 1.174 0.3 NonORV 2 0.50

PAGE 40

40 Table 2-6. Variation in vegetative vertical cove r among prairie habitats and impacts (off-road vehicle (ORV) and non-off -road vehicle (Non-ORV). Layer Impact n Mean Cover F-Value p-value 0.0-0.5 ORV 31 7.75 0.338 0.6 NonORV 35 8.75 0.5-1.0 ORV 13 3.25 0.338 0.6 NonORV 9 2.25 1.0-1.5 ORV 0 NonORV 0 1.5-2.0 ORV 0 NonORV 0 >2.0 ORV 0 NonORV 0

PAGE 41

41 CHAPTER 3 ECOLOGY OF SMALL MAMMALS IN BIG CY PRES S NATIONAL PRESERVE IN AREAS AFFECTED BY ORV USE Introduction Big Cypress National P reserve (BCNP) was established to include recreational ORV use (16 USC 698). Off-road vehicles (ORVs) ar e commonly used throughout the Preserve for recreation, especially hunting. As a result of fragmentation, once contiguous habitat has increased into small patches delimited by tire ruts. Duever et al. (1981) de monstrated that ORVs altered vegetation composition and hydrology. Fragme ntation of natural habitats and vegetation changes has potentially negative effects on natural populations of small mammals such as change of behavior or change in body c ondition. There is evidence that OR V use altered the behavior of the endangered Florida panther in BCNP (Janis and Clark 2002). Understanding these changes has important implications for land-use policies th at promote long-term survival and persistence of natural populations. Small mammals may be an ideal taxonomic gr oup for addressing quest ions of impacts of ORV use to wildlife populations in BCNP. Deta iled information regarding biology and natural history of small mammals at the organismal population, and community levels is readily available (Chase et al. 2000; Mares and Ernest 1995). Small mammals are easily marked to monitor their movements, behavior, and study life history parameters such as survival and occupancy. Live-trapping studies have revealed dispersal behavi or and an understanding of why a particular species selects a particular habitat (Gaines and McClenaghan 1980). Small mammals live in relatively small areas and have shor t lives. They typically disperse upon reaching adulthood and exhibit behavioral responses to environmental and anthropogenic changes. I selected pine and prairie habitats because these habitats are used by ORV users the most (Duever

PAGE 42

42 et al. 1981). Pine and prairie habitats are also easily traver sed than cypress and hammock habitats because of low density of trees and less obstacles. The objective of this study was to determine: 1) if ORV or habitat use influence small mammal abundance, 2) if ORV or habitat use changes body condition of small mammals, 3) does ORV use change survival of small mammals, 4) does occupancy change in ORV use areas. I hypothesized that captures of Sigmodon hispidus (hispid cotton rat), Oryzomys palustris (marsh rice rat) and Peromyscus gossypinus (cotton mouse) in non-ORV s ites are greater than in ORV sites because changes in vegetation would not make suitable habitat. Abundances and body condition of hispid cotton rats and cotton mice should be higher in pine habitats, while that of marsh rice rats are higher in prairie habitats be cause those are their preferred habitats. I also predict that body condition of a ll the small mammal species will be greater in non-ORV sites than in ORV sites because the ORV sites may not have the necessary vegetation for cover and food. Males should have a highe r body condition than females regardless of ORV presence because females will have young to take care of. Occupancy of all specie s will be greater in non-ORV sites than ORV sites, as non-ORV s ites are more continuous habitat. Finally, I hypothesize that survival and captu re probabilities of hispid co tton rats, marsh rice rats, and cotton mice are lower in ORV than in non-ORV s ites due to sparse vegetation in ORV sites based on my findings in Chapter 2. Methods Study Area Big Cypress National P reserve is located in southwest Florida. The Big Cypress regional climate is affected by both tropical and temper ate influences (Duever et al. 1986). The region is characterized by hot, humid summers (30-35C) and mild, dry winters (19-24C). Southwest

PAGE 43

43 Florida has been classified as a tropical savanna due to its temperat ure and rainfall (148.8 cm/year) (Hela 1952). The landscape of BCNP contains 295,000 ha of a heterogeneous mixture of pine forests, prairies, cypress dome/strands, marshes, and is olated hardwood hammock s (Duever et al. 1986). Pine and prairie habitats are the most abunda nt communities in BCNP and are most used by ORV enthusiasts (National Park Service 2000). Pi ne forests are open areas that are composed primarily of South Florida slash pine ( Pinus elliottii var. densa ) and are usually bordered by wet prairies. A fire frequency of three to seven y ears is required to preven t succession from pine forest to hardwood hammock (Hofstetter 1984). Pine forests are elevated from a few centimeters to a meter or more above the surrounding lowlan ds, which results in a gradient in hydroperiod. The variety densa developed longer taproots and smaller ne edle size than the northern variety of P. elliottii in response to different water cond itions (McMinn and McNab 1971). Variety densa is also highly fire tolerant and young seedlings sprout needles at the root collar below the burned growing tip (Ketcham and Be thune 1963). Cabbage palms ( Sabal palmetto ) are often abundant in pine forests. They tolerate flooding by producing adventitious roots (Brown 1973) and have an embedded bud and a fire-resistant trunk (Myers 1977) Prairie habitats are maintained by fire and burn about every one to three years to preven t the invasion of woody ve getation (Wade et al 1980). Saw palmettos ( Serenoa repens ) are often associated with pine forests and are well adapted to fire but do not withstand inundation as well as cabbage palms. Hardwood shrubs and trees can be found scattered thr oughout pine forests where fire has not been present. Ground cover is usually dominated by grasses such as bluestem (Andropogon floridanus) and panic grass ( Panicum hemitomon). Sedges, rushes, and composites may also be present.

PAGE 44

44 Prairie communities are dominated by he rbaceous vegetation and grow along a hydroperiod gradient. They have mixed grasses, sedges, and other herbac eous plants with few trees. Prairies may be seasonally inundated or infrequently inundated depending upon elevation. Common species include maidencane ( Panicum hemitomon ), blackhead rush ( Schoenus nigricans ), muhly ( Muhlendergia batatifolia ), and scattered sawgrass ( Cladium jamaicensis ). Maximum water levels rarely exceed 20 cm and are inundated approximately 50-150 days per year (Duever et al. 1975). Site Selection A total of eight sam ple sites were randomly c hosen from aerial maps that had at least 100 m of pine adjacent to 100 m of pine habitats. ORV sites were established in areas where ORV use is permitted and non-ORV sites were established in areas where ORVs are excluded for at least 20 years. On October 2004, four sites were established: two in ORV and two in non-ORV areas. On January 2005 four more sites were esta blished: two in ORV and two in non-ORV. At each sample site, a 200 m transect was establis hed with 100 m in pine habitat and 100 m in prairie habitat. Small Mammal Surveys We captured four species of s mall mammal in BCNP: hispid cotton rat, marsh rice rat, cotton mouse, and southern short-tailed shrew ( Blarina carolinensis ). Captures of southern shorttailed shrews were not included in the analysis because of their low numbers. Rodents were live trapped from October 2004 to September 2005 using Sherman live traps (3 3.5 9 inch) baited with sunflower seeds. A total of 44 traps were set on each transect with two traps set every 10 m (n=176 traps). When temperatures were for ecasted below 20 degrees Celsius, cotton was provided to prevent hypothermia. The absence/presence of standing water was recorded for each trap location at a wooden stake. During the we t season, traps were placed on 30cm 30 cm

PAGE 45

45 Styrofoam floats. Sampling of the original f our sites occurred every other week. When four new sites were established in January 2005, at wh ich point sampling frequency decreased to once per month. During each sampling session, sites were trapped for three or four consecutive nights depending upon weather. Traps were closed afte r the morning inspection to ensure that no animals were caught and exposed to the after noon heat. Traps were reopened before sunset. When the traps were not in use, they were removed from the study area and were thoroughly cleaned with bleach to remove uneaten food and pr event transmission of disease and parasites. When animals were captured, they were carefully removed from the trap and placed into a bag to avoid harm and reduce stress. The cap tured animals were identified by species, age (adult/juvenile), sex (male/female), and reproductive status. Males were considered reproductive if they had descended testes and females were considered reproductive if they were pregnant, had a perforated vagina, or had lactating ma mmary glands. Measurements of head-body length, head length, and right hind foot length were measur ed with a ruler to the nearest millimeter. Each individual was weighed using a hang ing scale to the nearest gram. Captured animals were individually marked on both ears using National Band and Tag Company Small Mammal Ear Tag #1005-1 (h ttp://www.nationalband.com/1005b.htm). All animals were released as soon as possible af ter capture at the orig inal capture site. Statistical Analysis Com parisons of capture numb ers in ORV/non-ORV sites and pine/prairie habitat were made using a Chi2 test. Mean weight, mean length, a nd body condition in ORV/non-ORV sites and pine/prairie habitat were made using Wilcoxon Rank Sign test Body condition was assessed by regressing body we ight against a right hind foot length. Data were transformed using a natural log to m eet assumptions of the regression analysis. Body condition was assessed using Fultons K:

PAGE 46

46 K = 104W/L3, where W = weight (g), L3 = skeletal measurement length (mm) and is used to assume growth is isometric. Isometric growth is when growth occu rs at the same rate for all parts of an organism so that its shape is consistent throughout development. The key assumptions are that condition is independent of length, that the measure used for le ngth is an accurate meas ure of structural size, that length measurements are not subject to error, and that the relationship between body size and mass is linear (Green 2001). Th e condition index was compared using Wilcoxon Rank Sign test for the observational groups from each species (O RV vs. Non-ORV, male vs. female, pine vs. prairie). Only non-pregnant adults were used in the body condition analysis. Pregnant females may show better body condition as the fetus may increa se the weight of the female. There is little information on when juvenile rodents stop suckling and the addition of female milk to the diet of juveniles may bias the body condi tion analysis of juvenile individuals. Individuals were considered adults when their hind foot measurem ents were greater than 28 mm for hispid cotton rats (Cameron and Spencer 1981), 28 mm for marsh rice rats (Wolfe 1982), and 16 mm for cotton mice (Wolfe and Linzey 1977). I selected = 0.1 for all statistical testing to help better understand the biological cha nges that could be happening. Detection probabilities for all small mammal species were assumed a priori to be less than one. By estimating detection probabilities, it was po ssible to estimate the true site occupancy rate of each species, while taking into account effects of environmenta l variables on behavior of the animals. Detection rates were not assumed to be constant. Howeve r, if a species was present, a detection probability greater than zero was assumed. Also, sites were assumed to be closed to changes in occupancy between subsequent samples.

PAGE 47

47 The single season model in program PRESENCE was used to look at the impacts of ORVs on small mammal site occupancy rates and dete ction probabilities (MacK enzie et al. 2002; MacKenzie et al. 2006). The singl e season model assumes that sites are closed to changes in occupancy and that the detection of a species at a site is indepe ndent of detections from other sites. This method also assumes that species are not falsely detected, but species may or may not be detected when present. This method was cons idered appropriate for the small mammal species due to low dispersal and difficulty of detection. Sample sites were defined by using the trap locations from the eight 200 m transects, resulting in 176 sample sites. Presence/absence was recorded for each sample site when the traps were checked each day. Site-specific covariables that could directly affect the estimate of occupancy ( ) were ORV use (ORV/Non-ORV site), habitat type (p ine/ prairie), and pe rcent ground cover (see Chapter 2 for methods). A sampling occasion covariate that could affect detection probability ( p ) was presence of standing water. For each sp ecies, we considered 16 models that were combinations of the covariables thought to be bi ologically important (Table 3-1). The best model was chosen as the one with the lowest value for Akaikes information criterion (AIC) or the model with the best fit for the fewest parame ters (Burnham and Anderson 1998). The effect of ORV use on species occupancy was established using model selection to determine the AIC weight of all models including the ORV use cova riate and by examining the beta estimates for ORV use in the models. Mark-recapture data were analyzed by usi ng the Cormack-Jolly-Seber closed population recapture model performed in program MARK (W hite and Burnham 1999) to estimate survival for the three species. Individuals of all species were divided into two groups for analysis: ORV and non-ORV areas. A series of 26 models representing different hypotheses about the effects of

PAGE 48

48 time and group on apparent survival ( ) and capture probability ( p ) were fit for all species (Table 3-2). Goodness of fit was assessed by estimating (variance inflation factor) using the parametric bootstrap method in program MARK (White and Burnham 1999) on the most general model, g*tpg*t. Model selection was conducted using th e information-theoretic approach of Burnham and Anderson (1998) with the Quas i-likelihood Akaikes Information Criterion adjusted for over dispersion of data and small sample sizes (QAICc). Results Species Distribution and Abundance From October 2004 to September 2005, I samp led on 14,784 trap nights at eight sites. Trapping yielded 412 individuals from four diff erent species (Table 3-3). I had 1,293 recaptures of these individuals for a total of 1,710 captu res. Most captures we re cotton rats (57.5%) followed by marsh rice rats (29.1%), cotton mice (12.8%) and short-tailed shrews (0.6%) (Table 3-4). Hispid cotton rat captures peaked in April during the beginning of the wet season (Figure 3-1). Marsh rice rat and cotton mouse captu res peaked in March af ter heavy rain events (Figure 3-1). Captures decrease in all small ma mmal species after a major rain event in May (Figure 3-3). The hispid cotton rat was captured the most frequently, with 118 and 83 individuals captured, respectively, in ORV and non-ORV sites (T able 3-5). The hispid cotton rat also had the highest captures in pine habitats with 312 captu res (Table 3-6). The marsh rice rat was the second most abundant species, with 75 indivi duals captured in ORV-impacted areas and 81 individuals captured in non-ORV areas (Table 3-5). The marsh rice rat was also the most frequent species in the prairie habitat with 190 captures and the second most abundant in pine

PAGE 49

49 habitats with 72 captures (Table 3-6). The cott on mouse was the third most abundant species, with 46 individuals captured in ORV sites and fi ve individuals captured in non-ORV sites (Table 3-5). Cotton mouse captures in pine habitats were 57 and in prairie habitats were 35 (Table 3-6). Short-tailed shrew captures in pine habitats were 4 and none were found in prairie habitats (Table 3-6). I exclude short-ta iled shrews from the analysis due to their low numbers. Mean hispid cotton rat captures in ORV sites (n=268) were significantly lower than in nonORV sites (n=152, 2=32.038, p-value <0.001) (Table 3-7). Ho wever, mean marsh rice rat captures showed no significant differences be tween ORV (n=101) and non-ORV sites (n=107 2=0.173, p-value = 0.7). In contrast, mean cotton mi ce captures were significantly higher in ORV sites (n=82) than in non-ORV sites (n=10, 2=56.348, p -value <0.001) (Table 3-7). Mean hispid cotton rat captures in pine habitats (n=321) were signif icantly higher than in prairie habitats (n=99, 2=117.3, p-value <0.001) (Table 3-7). On the other hand, mean marsh rice rat captures in prairie habitats (n=184) were significantly higher than in pine habitats (n=24, 2=123.1, p-value <0.001) (Table 3-7). Conversely, m ean cotton mice captures in pine habitats (n=57) were significantly higher th an in prairie habitats (n=35, 2=5.621, p-value = 0.02) (Table 3-7). Mean hispid cotton rat male captures (n =232) were not significantly higher than female captures (n=233, 2=6.565, p-value = 0.3) (Table 3-7). Mean marsh rice rat male captures (n=150) were not significantly gr eater than female captures (n=94, 2=8.565, p-value = 0.3) (Table 3-7). Mean cotton mouse male captures (n=59) were not signifi cantly different from female captures (n=39, 2=10.915, p -value = 0.09) (Table 3-7). Weight and Size Comparisons The m ean weight of adult, non-pregnant hispid cotton rats was 113.96 7.59 g (mean SD). Male hispid cotton rats had significantly higher mean weights (125.55 8.57 g, p< 0.001) than females (97.32 10.76 g) (Table 3-8). The mean weights of hispid cotton rats for the

PAGE 50

50 ORV/non-ORV and pine/prairie comparisons were not significantly different (Table 3-8). The mean adult head-body length of adult hispid cotton rats was 125.08 3.98 mm. The mean adult head-body length of male hispid cotton rats (129.78 4.59 mm, p< 0.001) was significantly higher than that of female hispid cotton rats (117.84 7.64 mm) (Table 3-9). The mean adult head-body lengths of hispid cotton rats for the ORV/non-ORV and pine/prairie comparisons were not significantly different (Table 3-9). The mean weight of adult, non-pregnant marsh rice rats was 56.77 5.88 g. Male marsh rice rats mean weight (62.61 7.43 g, p<0.01) was significantly highe r than that of females (48.61 7.59 g) (Table 3-8). The mean weights of marsh rice rats for the ORV/non-ORV and pine/prairie comparisons were not significantly different (Table 3-8). The mean adult head-body length of adult marsh rice rats was 103.89 5.98 mm. The mean adult head-body lengths for the males (107.06 8.35 cm, p=0.08) were significantly larger than females (101.28 6.78 cm). Marsh rice rat lengths in ORV/non-ORV, and pine /prairie comparisons were not significantly different (Table 3-9). The mean weight of adult, non-pregnant cotton mice was 27.76 3.09 g. There were no significant differences in average weights for male/female, ORV/non-ORV, and pine/prairie comparisons (Table 3-8). The mean headbody length of adult cotton mice was 83.68 4.25 mm. There were significant differences in average head-body length for the male/female and pine prairie comparisons (Table 3-9) Male lengths (84.65 4.99 cm, p= 0.1) were significantly larger than females (80.26 7.64 cm). Cotton mice in pine habitats (84.60 5.39, p=0.1) were longer than prairie habitats (81.21 5.73).

PAGE 51

51 Body Condition The regression of ln (weight) on ln (hind foot length) for hispid cotton rats, m arsh rice rats, and cotton mice yielded significan t regressions (Table 3-10). Fo r all three species, the mean condition index was independent of ORV presence and habitat type (Table 3-11). Hispid cotton rat mean condition index score for females was si gnificantly lower than that for males; female and male condition indices were equivalent for both marsh rice rats and cotton mice (Table 311). A total of 176 sites were vi sited for a total of 48 sampling visits from October 2004 to September 2005. Of the 176 sites, 88 were located in prairie habitat and 88 were located in pine habitat. Also, of the 176 sites, 88 were located in ORV areas and 88 were located in non-ORV areas. Hispid cotton rats marsh rice rats, and cotton mice we re detected between 78-132 times, and nave occupancy rates (propor tion of sites at which a detection occurred) varied from 44 to 75 percent (Table 3-12). The best model is recognized by using th e lowest AIC value (Burnham and Anderson 1998). The two best models for the cotton rat incl uded ORV use, habitat and percent cover as a site covariate (Table 3-13). ORV use had a nega tive beta estimate for occupancy in the second best model but the 95% confidence interval overl aps 0 and is not significant (Table 3-14). The best model had a habitat beta of 1.4233 (S.E. = 0.4233) and the second best model had a habitat beta estimate of 1.4298 (S.E. = 0.4283) indicating a positive associ ation with pine habitats (Table 3-14). The best model had percent c over beta estimate of 0.8812 (S.E. = 0.2804) and the second best model had a percent cover beta estimate of 0.9906 (S.E. = 0.3521) indicating a positive association with higher horizontal vegeta tive percent cover (Table 3-14). Detection probability beta estimate for the cotton rat was -2.335 (S.E. = 0.1661), which suggests a strong negative detection with pres ence of standing water.

PAGE 52

52 The two best models for the rice rat included OR V use, habitat, and percent cover as site covariates (Table 3-15). The best model had an ORV use beta estimate of 0.7212 (S.E = 0.3732) but the 95% confidence interval ove rlaps 0 and is not significant; th is is because we did not have a large enough sample size to detect a trend (T able 3-16). The best model had a habitat beta estimate of -2.6835 (S.E. = 0.4127) and the second best model had a habitat beta of -1.9206 (S.E. = 0.3847), indicating a positive detecti on with prairie habitat (Table 3-16). The best model had a percent cover beta estimate of 1.684 (S.E. = 0.4028) and the second best model had a beta of 1.9526 (S.E. = 0.3947), indicating a positive associat ion with higher horizontal vegetative percent cover. The best model for the cotton mouse included ORV use, habitat, and horizontal vegetative percent cover as site covariates (Table 3-17). The ORV use had a beta estimate of 2.7672 (S.E. = 0.5600), indicating a strong association with ORV us e (Table 3-18). Habitat had a beta estimate of 1.0869 (S.E. = 0.5060), which indicates a positive association with pine habitat (Table 3-18). Percent cover had a beta estimate of -2.0566 (S .E. = 0.4980), indicating a negative association with horizontal vegetative pe rcent cover (Table 3-18). Mark-Recapture During the sam pling period, a total of 447 co tton rats, 218 marsh rice rats, and 97 cotton mice were captured. These captures included reca ptures of 205 indivi dual cotton rats, 157 individual marsh rice rats, and 51 individual cotton mice. The retu rn rates (proportion of marked individuals recaptured at least once) for the co tton rat and the cotton mouse were higher in ORV areas than in non-ORV areas (Table 3-19). The retu rn rates for the marsh rice rat were lower in ORV areas than in non-ORV areas (Table 3-19). The parametric bootstrap of the most genera l Cormack-Jolly-Seber model for the cotton rat produced an estimate of = 1.3206. Three models had delta QAICc values less than 2 (Table 3-

PAGE 53

53 20). Two of the three best models included ORV group effect on survival. Models that included ORV group effect on survival had 71.58% of th e QAICc weight among the set of candidate models, and models that included ORV group e ffect on capture probability had 61.76% of the model weight. The estimated beta for the best model that ha d ORV group effect on survival in cotton rats was 0.7567 (S.E. = 0.1817; Table 3-21). There was a m ean decrease in survival of 7.98% from the non-ORV areas to the ORV areas (Figure 3-2) The estimate beta for the best model that included ORV effect on capture probability in cotton rats was 1.5163 (S.E. = 0.2951). Cotton rat recapture probability in non-ORV areas was 8.35 % higher than that in ORV areas. It is interesting to note that survival numbers decr eased dramatically afte r August. The drop of survival coincides with Hurricane Katrina (Cate gory 1) which went through southern Florida August 23, 2005. For the marsh rice rat, the estimate of from the parametric bootstrap was 1.699. Four models had delta QAICc values less than 2 (Tab le 3-22). Two of the top four models included ORV effect on survival and on capture probabili ty. Models that included the ORV effect on survival for the marsh rice rat received 43.59% of the QAICc weight, and models that included the ORV effect on capture probability acc ounted for 45.85% of the total weight. The estimated beta for the best model that included ORV effect on survival in the marsh rice rat was -0.6200 (S.E. = 0.4045), which indicate s a slightly negative effect of ORV use (Table 3-21). But the 95% confidence interval includes 0, which means we did not have a large enough sample size to detect a trend. There was a mean decrease in survival of 5.47% from the non-ORV areas to the ORV areas (Figure 3-3). The estimate of beta for the best model that included ORV effect on capture probability of the marsh rice rat was -1.2581 (S.E. = 0.7744),

PAGE 54

54 which indicates a negative effect of ORV use (Tab le 3-21). But the 95% confidence interval for this parameter also includes 0, which means we di d not have a large enough sample size to detect a trend. Marsh rice rat recapture probabilities were 8.38% higher in non-ORV areas than ORV areas. For the cotton mouse, the estimate of from the parametric bootstrap was 1.7045. Only the model (.) p (.) had a delta QAICc value less than 2, and therefore had strong support (Table 323). Two of the remaining top th ree models had ORV effect on su rvival and capture probability. Models that included the ORV effect on survival for the cotton mouse received 27.89% of the QAICc weight, and models that included ORV effect on capture probability accounted for 25.27% of the total weight. The estimated beta for the best model that had ORV effect on survival was -0.2866 (S.E. = 0.8184), which indicates a negative effect of ORV use (Table 3-21; Figure 3-4). But the 95% confidence interval includes 0, which means we did not have a large enough sample size to detect a trend. The estimate of beta for the be st model that included ORV effect on capture probability of the cotton mouse was -0.1419 (S.E. = 1.5411), which indicates a negative effect of ORV use (Table 3-21). But the 95% confidence interval for this parameter also included 0, which means we did not have a large enough sa mple size large enough to detect a trend. Discussion This study addresses the lack of baseline da ta concerning the eff ects of ORVs on s mall mammals in BCNP. I captured four species of sm all mammals while trapping in pine and prairie habitats. Hispid cotton rats were the most abundant specie s and appear to be habitat generalists, as they were found at all the trapping sites. Marsh rice rats were the dominant species found in prairie habitats, while they were only captured in pine habitats that were flooded. The increase of water levels may reduce the amount of ground cover and suitable nesting refugia in the prairie

PAGE 55

55 habitats, driving the marsh rice rats into the pine habitats. Cotton mice were mostly found in pine habitats, but were also captured in prairie habitats during the dry season. Short-tailed shrews were captured only in pine habitats. Capture comparisons of species based on OR V use showed mixed results. I expected captures and densities in ORV sites to be lo wer than in non-ORV sites. I found that small mammal species respond differently to ORV impact s. I did find that mean hispid cotton rat captures and densities were higher in non-OR V sites than in ORV sites, supporting my hypothesis. Mean cotton mice captures and dens ities displayed the opposite effect, with ORV sites having higher captures than non-ORV site s. Mean marsh rice rat captures, on the other hand, showed no difference between ORV and non-ORV sites. As discussed in Chapter 2, habitat fragmentation may affect species behaviors, dispersal abilities, life history, trophic level, social relationships, and overall responses to change in habitat size, connectivity, and type of habitat matrix (Hammitt and Cole 1998). Hispid cotton rats are habitat generalists, but they di splay preferences for grassland or early successional habitats high in vegetative cover (Lidicker et al. 1992, Cothran et al. 1991). The reduced patch size and increased distance between habitat patches caused by ORVs seem to have an effect on hispid cotton rat captures. Diffendorfer et al. (1995) found a 40% decrease in abundance in hispid cotton rats in fragmented habitats compared to continuous habitats There is also an increase of transient hispid cotton rats observed in fragment ed habitats (Diffendorfer et al. 1995). This can lead to animals staying for a short period of time and leaving quickly, making for an unstable population. Significant genetic diffe rentiation has been documented for populations of whitefooted mice (Peromyscus leucopus ) 500 m apart (Mossman and Waser 2001), and Allegheny woodrats ( Neotoma magister ) living within 2 km (Castleberry et al. 2002). Genetic diversity

PAGE 56

56 may be lost is a population that experience s a long-term bottleneck of low population size (Monty et al. 2003). A loss of genetic heterozygos ity in a population may l ead to the expression of deleterious genes that can result in lower survival rates and hastens population declines (Frankham 1995; Lacy 1993; O Brien et al. 1985). Constant use of ORVs in prairie habitats re duces vegetative cover (Cha pter 2) preferred by hispid cotton rats and re duces their cover to travel between habitat patches. Hispid cotton rats also exhibit sexual segregation in specific habita t requirements during all li fe stages (Lidicker at al. 1992). Female hispid cotton rats are territori al and may be limited by breeding sites in ORVimpacted sites. Habitat fragmentation could lead to a smaller proportion of sexually active females reducing the overall population. Cotton mice, on the other hand, have higher captur es in ORV sites, which suggest that they may be more tolerant of fragmented habitats With cotton mice smaller body size and home range, movements within habitat patches that we re created by ORV use may not be affected. Peromyscus maniculatus also showed a positive response to a habitat fragmentation experiment conducted by Diffendorfer et al. (1995). He found that abundances were three times higher in fragmented sites. Increased habitat and reduced competition from the larger rodents abandoning the smaller fragments is theorized for the in crease in abundances (Di ffendorfer et al. 1995). Cotton mice females can endure an increase in de nsity when space is limited, this display of tolerance may explain why we find more co tton mice in ORV impacted areas (Wolff 1994). Hispid cotton rats have lower captures in frag mented ORV sites presumably because the smaller patches do not have enough habitat to sustain i ndividuals. Cotton mice may therefore reach their highest densities on smaller, fragmented patches of habitat. I found that hispid cotton rats and cotton mice mean captures in pine habitats were significantly higher than in prairie habitats,

PAGE 57

57 supporting my hypothesis. Hispid co tton rats are well adapted to survive in many different types of habitat and have been observed swimming. Hispid cotton rats on tree islands in the Everglades showed little movement in hardwood hammocks and high dispersion in pr airie habitats during interface of wet and dry seasons (Smith and Vrieze 1975). This suggests that hispid cotton rats prefer the higher elevation habitats of pine and hammock habitats. Cotton mice often nest in logs and holes in the ground, and they ar e one of only two Florida mice species to frequently nest in trees. Prairie habitats do not offer the same foragi ng and nesting opportunities as pine habitats for the cotton mouse. As expected, marsh rice rats were predominantly found in prairie sites. Smith and Vrieze (1975) observed the same pattern in Ev erglades with the mars h rice rats moving to hardwood hammocks in the wet season. This specie s is an adept swimme r often spending most of its foraging time in the water. It can readily dive and is capable of swimming underwater to forage or escape predation. Hispid cotton rat males were found to be signifi cantly longer than females, contrary to my hypothesis. A possible explanati on might be that hispid cott on rats are sexually dimorphic (Cameron and Spencer 1981). Males may grow larg er to compete for females and territory. There were no differences in length between ORV and non-ORV sites contrary to my hypothesis. This might be because the suitability of habitat in ORV a nd non-ORV sites may not inhibit growth. Hispid cotton rats, marsh rice rats a nd cotton mice showed no differences in body condition between ORV and non-ORV sites. These fi ndings contradict my original expectations that body condition of small mammal species w ould be better in non-ORV sites. A possible explanation might be the ideal free distribution theory. Th is theory states that if several habitat types (e.g., disturbed vs. undisturbed) are available but differ in their basic suitability, then each

PAGE 58

58 individual is free to settle where their expected fi tness is highest. If the population is distributed in accordance with this theory, densities will be hi ghest in the most suitable habitat and lowest in the least suitable, resulting in a decreased comp etition for scarce resources. The distribution and densities of the population might have created an equilibrium that would not affect body condition. My data supports this theory for the hispid cotton rat as I saw higher capture numbers in non-ORV sites, but body condition was not affected. Another possibility is that the disturbance at ORV sites keeps the habitat in early successional stage, producing more food. Undisturbe d sites may offer a stable food source that is higher in quality but lower in quantity. On the other hand, disturbed sites may offer food in the form of early successional plants that are lowe r in quality, but higher in quantity. This dynamic relationship may account for why I did not see any differences in body condition between ORV and non-ORV sites. My data showed that co tton mice had higher counts in ORV than non-ORV sites, but body condition was not affected, this may be due to the smaller body size and home range of the cotton mice. I did find a difference in body condition for the hispid cotton rat betw een habitats. Cotton rats in pine habitats had signi ficantly higher body condition than pr airie habitats. Contrary to my hypothesis, I found that there were no differences in body condition index scores between pine and prairie habitats for rice rats and cotton mice. I predicted that cotton mice would have a better condition index score in pine habita ts than in prairie and that mars h rice rats would have a better condition index score in prai rie habitats than in pine habitats. Rodents are very ad aptable to their surrounding environment. They are able to adjust to different food sour ces in a variety of habitats. Even though we observed the rodents in their secondary habitats, they were able to maintain their body condition.

PAGE 59

59 Male hispid cotton rats body c ondition index scores were significantly higher than female. Males may be more selective in the food they eat. Higher quality food could be in short supply and males could be willing to travel great distances to obtain it. Smith and Vrieze (1975) observed female cotton rats moved farther than males. Females may have to travel great distances because of their young. In order to get the required am ount of nutrition to feed the young, longer distances would have to be traveled to get the proper food quality Traveling long distances for higher quality food could produce significantly different body condition from gathering lower quality food at a short distance. Marsh rice rats and cotton mice had no signi ficant difference in body condition index scores between males and females. This finding does not support my hypothesis. Marsh rice rats and cotton mice have smaller home ranges than hi spid cotton rats and may be able to find enough nutrition within their limited territory, thus not showing a difference in body condition. The results from the site occupancy analysis i ndicate that ORV use is not a strong predictor of occupancy for the cotton rat and marsh rice rat, but it may be a strong predictor for the cotton mouse. Two of the three speci es included the ORV use covari ate in the best model for occupancy, and the sum of the model AIC weight was highest for those models that included ORV use. This indicates that occupancy for so me species may depend more on ORV use than on habitat or horizontal vegetative percent cover. I predicted that these small mammals would be negatively influenced by ORV use due to ground level disturbance of vegeta tion. Two of the three species of rodents had beta values for ORV use that were positive, indicating positive associations with ORV use. One species, however, the cotton rat, was negativ ely associated with ORV use. A lthough this is counter to the original prediction, altered hydrology and morpho logy might explain the difference in response

PAGE 60

60 to ORV use. The marsh rice rat is a medium-sized rodent that is associated with water (Brown 1997). ORVs can alter the hydopattern of areas, resulting in the creation of artificial depressions and the increase of ponding in ruts (Duever et al 1981). The marsh rice rats may take advantage of the increased temporal and spatial extent of standing water for c onsumptive purposes. The cotton mouse is a small-sized rodent that may thrive in areas where ORVs have created a fragmented habitat because of a potential competitive release from cotton rats. For all species, horizontal vegetative percent cover was an important predictor in site occupancy. This may be due to the fact that sm all mammals are regular prey for many species. The cotton rat and marsh rice rat had positive beta values indicating positive associations with horizontal vegetative percent cove r. Without proper cover, nesti ng and foraging areas would be exposed to predation and weather. The cotton ra t and the marsh rice rat may take advantage of the increased vegetative cover to evade predator s, to have sufficient cover to rear young, to forage, or a combination of these factors (e.g., rear young and forage). For all species, habitat was an important predic tor in site occupancy. The cotton rat prefers dry habitat that is composed of dense grasse s, and is common in brush and palmettos where some grass is present, although we did catch any in the marsh fringes during high water (Cameron and Spencer 1981). The marsh rice rat pref ers habitats that are wet and seldom occur in dry areas (Wolfe 1982). I observed the marsh rice rat in both the prairie and pine habitats, but they were in the pine habitat only when ther e was high water or the pine habitat was flooded. The cotton mouse prefers wooded ha bitats as it is very arboreal and nests in logs and hollow stumps (Wolfe and Linzey 1977). Hurricane Katrina had an impact on survival of hispid cotton rats in both ORV and nonORV sites. There was a drop of captures post-hurricane but it is unclear if the decline is from the

PAGE 61

61 hurricane or normal population fl uctuations. After Hurricane Andr ew in 1992, Everglades whitetailed deer populations in five wildlife mana gement areas had a 67% decline in population and had a 13 fold decrease in fawn production (Labisky et al. 1999). A major natural event such as a hurricane could put more pressure on an already stressed small mammal population in BCNP. Swilling et al. (1998) found that the effect s of Hurricane Opal (1995) on beach mice ( Peromyscus polionotus ammobates ) were delayed until the next summer. Beach mice populations decreased 30% from pre-hurricane levels and utili zed more transitional habitat (Swilling et al. 1998). The population of the sma ll mammals in BCNP could also show a lag effect from Hurricane Katrina, where effects may not be seen for many months. Significant redcockaded woodpecker ( Picoides borealis ) nesting cavities were destroyed in BCNP after Hurricane Andrew (Loope et al. 1994). Although Hurrican e Katrina was a category 1 hurricane and did not cause much damage in BICY, a major hurricane could devastate small mammal populations in ORV areas. There is clear evidence of impacts from ORV use, but the long-term implications are unclear. The number species of small mammal s did not change between ORV and non-ORV areas, but there were changes of species composition. Similar results were observed by Waddle (2006) for anurans. Three of the four anurans st udied had a negative relationship with ORV use, but the southern toad ( Bufo terrestris ) had a positive relationship. The southern toad prefers longer hydroperiod as their tadpoles require a longer developmental period. The ruts and modified hydrology caused by ORV use created a be neficial habitat for the southern toad (Waddle 2006). This is a concern for specie s that rely upon small mammals as prey. The Florida panther has been shown to move away from ORV use ar eas during hunting season due to prey avoiding

PAGE 62

62 hunters (Janis and Clark 2002). A nother example is Bobcats ( Lynx rufus). Bobcats could be affected by the shift of species composition caused by ORV use. Preying on the larger hispid cotton rat would give more energy per effort th an preying on cotton mice. In ORV areas where hispid cotton rats are not abunda nt, bobcats would have to prey on the smaller cotton mouse. The bobcat would have to capture multiple cotton mice to get the same nutritional requirements for a hispid cotton rat. It is not possible to determine the mechanisms that ORV use has on influencing the small mammal community due to the observational nature of this study. However, this study should help promote more research on the use of ORVs. Is there evidence that ORV use influences small mammal species? Does the species co mposition shift from non-ORV to ORV affect predator species? Long-term m onitoring plots could be establis hed in current ORV use areas, recovering ORV use areas, and non-O RV use areas using the technique s of this study to observe any changes in small mammal population structure. This study provides ba seline data on small mammal species in BCNP that managers can use to make management decisions or to develop future research hypothesis.

PAGE 63

63 Table 3-1. Combinations of the 3 site covariates and one sampling covariates that were used in the occupancy analysis for each species. Each set of site covariates was modeled along with each set of sampling covariates for a total of 16 models for each species. Site Covariates Sampling Covariates Constant Constant ORV Water Habitat Percent Cover ORV, Habitat ORV, Percent Cover Habitat, Percent Cover ORV, Habitat, Percent Cover

PAGE 64

64 Table 3-2. List of 25 models analyzed in program MARK for captures of cotton rat, marsh rice rat, and cotton mouse in Big Cypress National Preserve. Explanation defines each model in terms of the effects of time (t) and ORV group (g), on apparent survival ( ) and capture probability (p). Model Explanation (.) p(.) Survival and capture pr obability constant throughout study (.) p(g) Survival is constant; capture varies by ORV (.) p(t) Survival is constant; capture varies by time (.) p(g*t) Survival is constant; capture is an interaction of ORV and time (.) p(g+t) Survival is constant; capture is an additive effect of ORV and time (g) p(.) Survival varies by ORV; capture is constant (g) p(g) Survival varies by ORV; capture varies by ORV (g) p(t) Survival varies by ORV; capture varies by time (g) p(g*t) Survival varies by ORV; captu re is an interaction of ORV and time (g) p(g+t) Survival varies by ORV; capture is an additive effect of ORV and time (t) p(.) Survival varies by time; capture is constant (t) p(g) Survival varies by time; capture varies by ORV (t) p(t) Survival varies by time; capture varies by time (t) p(g*t) Survival varies by time; capture is an interaction of ORV and time (t) p(g+t) Survival varies by time; capture is an additive effect of ORV and time (g*t) p(.) Survival is an interactio n of ORV and time; capture is constant (g*t) p(g) Survival is an interaction of ORV and time; capture varies by ORV (g*t) p(t) Survival is an interaction of ORV and time; capture varies by time (g*t) p(g*t) Survival and capture are both an interactive effect on ORV and time (g*t) p(g+t) Survival is and interaction of ORV and time; capture is an additive effect of ORV and time (g+t) p(.) Survival is an additive effect of ORV and time; capture is constant (g+t) p(g) Survival is an additive effect of ORV and time; capture varies by ORV (g+t) p(t) Survival is an additive effect of ORV and time; capture varies by time (g+t) p(g*t) Survival is an additive effect of ORV and time; capture is an interaction of ORV and time (g+t) p(g+t) Survival and capture are both the additive effect of ORV and time

PAGE 65

65 Table 3-3. Number of captures and recaptures for hispid cotton ra ts, marsh rice rats, cotton mice, and short-tailed shrews from 14,784 trap nights. Number of Times Captured Hispid cotton ratMarsh rice rat Cotton mouse Short-tailed shrew 1 118 128 29 4 2 39 20 7 0 3 20 5 10 0 4 12 3 1 0 5 7 4 3 0 6 6 0 1 0 7 2 0 0 0 8 1 0 0 0 9 4 0 0 0 10 0 0 0 0 11 0 0 0 0 12 0 0 0 0 Number of Individuals 201 156 51 4 Total Individuals 412 Total Recaptures 304 Total Captures 716

PAGE 66

66 Table 3-4. Number of captures by hispid cotton rat, marsh rice rat, cotton mouse and short-tailed shrew on impact (off-road vehicle (ORV ) and non-off-road vehicle (Non-ORV)) treatment areas. Individuals captured more than once within the same trapping event were not counted. Site Impact Hispid cotton rat Marsh rice rat Cotton mouse Short-tailed shrew Total Proportion of Total 1 ORV 41 38 23 4 106 0.148 2 ORV 73 41 33 0 147 0.205 5 ORV 11 5 16 0 32 0.045 6 ORV 19 17 10 0 46 0.064 3 Non-ORV 98 32 3 0 133 0.186 4 Non-ORV 71 39 0 0 110 0.154 7 Non-ORV 52 17 4 0 73 0.102 8 Non-ORV 47 19 3 0 69 0.096 Total 412 208 92 4 716 Proportion of Total 0.575 0.291 0.128 0.006

PAGE 67

67 Table 3-5. Total captures and individual captures of hispid cotton rat, marsh rice rat, cotton mouse and short-tailed shrew by off-road vehicle (ORV) and non-off-road vehicle (Non-ORV) areas. Total Captures Species ORV Non-ORV Hispid cotton rat 166 281 Marsh rice rat 128 134 Cotton mouse 86 11 Short-tailed shrew 4 0 Individual Captures Species ORV Non-ORV Hispid cotton rat 83 118 Marsh rice rat 81 75 Cotton mouse 46 5 Short-tailed shrew 4 0

PAGE 68

68 Table 3-6. Total captures of hispid cotton rat, marsh rice rat, cotton mouse, and short-tailed shrew by pine and prairie habitats. Species Pine Prairie Hispid cotton rat 312 135 Marsh rice rat 72 190 Cotton mouse 59 34 Short-tailed shrew 4 0

PAGE 69

69 Table 3-7. Mean SD captures of hispid cotton rat, marsh rice rat, cotton mouse, and short-tailed shrew by impact (off-road vehicle (ORV) and non-off-road vehicle (Non-ORV)), habitat (pine and prairie), and sex (male and female). ORV Species ORV Non-ORV p-value df Hispid cotton rat 38.5 15.76 67.0 11.55 <0.001 1 Marsh rice rat 25.5 16.62 26.8 10.53 0.7 1 Cotton mouse 20.8 9.74 2.5 1.75 <0.001 1 Habitat Species Pine Prairie p-value df Hispid cotton rat 40.5 24.68 12.3 7.81 <0.001 1 Marsh rice rat 4.6 5.21 21.5 10.73 <0.001 1 Cotton mouse 7.1 6.94 4.5 5.15 <0.05 1 Sex Species Male Female p-value df Hispid cotton rat 52.88 26.71 62.88 7.81 0.1 1 Marsh rice rat 34.38 16.14 24.13 20.64 <0.05 1 Cotton mouse 16.75 16.79 10.75 11.31 0.06 1

PAGE 70

70 Table 3-8. Mean weight SD of adult, non-pregnant hispid cotton rat, marsh rice rat, and cotton mouse by impact (off-road vehicle (ORV ) and non-off-road vehicle (Non-ORV)), habitat (pine and prairie), and sex (male and female). Hispid Cotton Rat Mean Weight SD n p-value ORV 111.84 9.49 134 0.1 Non-ORV 116.07 5.69 228 Pine 113.10 14.69 271 0.12 Prairie 107.73 10.15 91 Female 97.32 10.76 172 <0.001 Male 125.55 8.57 190 Marsh Rice Rat Mean Weight SD n p-value ORV 57.14 8.46 101 0.3 Non-ORV 54.12 5.08 106 Pine 53.97 18.11 26 0.95 Prairie 57.58 5.77 181 Female 48.61 7.59 84 <0.001 Male 62.61 7.43 122 Cotton Mouse Mean Weight SD n p-value ORV 29.50 3.26 75 0.2 Non-ORV 25.92 2.77 10 Pine 27.72 2.62 52 0.5 Prairie 27.71 7.10 33 Female 27.58 3.89 35 0.3 Male 28.33 3.10 50

PAGE 71

71 Table 3-9. Mean length SD of hispid cotton rat, marsh ri ce rat, and cotton mouse by impact (off-road vehicle (ORV) and non-off-road vehicle (Non-ORV)), habitat (pine and prairie), and sex (male and female). Hispid Cotton Rat Mean Length SD n p-value ORV 123.37 4.90 134 0.18 Non-ORV 126.79 2.30 228 Pine 125.72 5.91 271 0.4 Prairie 122.36 6.55 91 Female 117.84 7.64 172 <0.01 Male 129.78 4.59 190 Marsh Rice Rat Mean Length SD n p-value ORV 104.43 6.21 101 0.4 Non-ORV 103.36 6.65 106 Pine 102.49 13.91 26 0.3 Prairie 104.36 5.26 181 Female 101.28 6.78 84 0.08 Male 107.06 8.35 122 Cotton Mouse Mean Length SD n p-value ORV 81.57 3.50 75 0.3 Non-ORV 85.11 5.44 10 Pine 84.60 5.39 52 0.1 Prairie 81.21 5.73 33 Female 80.26 7.64 35 0.1 Male 84.65 4.99 50

PAGE 72

72 Table 3-10. Linear regressions of ln(body weight) over ln(right hi nd foot))of hispid cotton rat, marsh rice rat, and cotton mouse. Skeletal measurement Slope r2 p-value Cotton Rat y=3.28x-6.29 0.438 <0.0001 Marsh Rice Rat y=1.79x-1.99 0.266 <0.0001 Cotton Mouse y=1.06x+0.08890.0640<0.05

PAGE 73

73 Table 3-11. Mean body condition indices ( SD) of hispid cotton rat, marsh rice rat, and cotton mouse by impact (off-road vehicle (ORV ) and non-off-road vehicle (Non-ORV)), habitat (pine and prairie), and sex (male and female). ORV Impact Species ORV Non-ORV df p-value Cotton rat 46.34 5.82 51.59 2.81 3 0.2 Marsh rice rat 24.11 3.12 21.68 2.21 3 0.3 Cotton mouse 12.01 1.30 11.26 1.21 2 0.6 Habitat Species Pine Prairie df p-value Cotton rat 49.61 5.51 44.79 10.80 6 0.3 Marsh rice rat 26.88 6.93 22.42 2.74 4 0.2 Cotton mouse 11.69 1.31 11.94 1.77 5 0.8 Sex Species Female Male df p-value Cotton rat 44.62 9.98 50.83 8.13 6 0.2 Marsh rice rat 21.43 4.29 23.81 2.96 5 0.2 Cotton mouse 12.37 1.50 11.52 1.36 5 0.2

PAGE 74

74 Table 3-12. Number of detections by cotton rat, marsh rice rat, and cotton mouse and proportion of sites at which a detection occurred (nave occupancy) during small mammal surveys. Species # Detections Nave Occupancy Cotton Rat 132 75.57% Marsh Rice Rat 105 59.66% Cotton Mouse 78 44.57%

PAGE 75

75 Table 3-13. Program PRESENCE m odel selection results for the hispid cotton rat, including Akaikes Information Criterion (AIC) and the delta AIC and AIC weights for all models with any weight. Model AIC delta AIC AIC weight (Percent Cover, Habitat),p(Water) 1809.7 0 0.6874 (ORV, Percent Cover, Habitat),p(Water) 1811.431.73 0.2894 (ORV, Habitat),p(Water) 1817.978.27 0.011 (Habitat),p(Water) 1818.448.74 0.0087 (Percent Cover),p(Water) 1821.0711.37 0.0023 (ORV, Percent Cover),p(Water) 1823.0113.31 0.0009 (.),p(Water) 1825.2715.57 0.0003

PAGE 76

76 Table 3-14. Program PRESENCE beta estimates, standard errors (S.E.), and lower and upper 95% confidence intervals for the hispid co tton rat. The best model and second best model are shown. Best Model Covariate Beta Estimate S.E. Lower 95% C.I. Upper 95% C.I. Habitat 1.42330.4283 0.58382.2628 Percent Cover 0.88120.2804 0.33161.4308 Second Best Model Covariate Beta Estimate S.E. Lower 95% C.I. Upper 95% C.I. ORV -0.19360.3681-0.91510.5279 Habitat 1.42980.4283 0.59032.2693 Percent Cover 0.99060.3521 0.30051.6807

PAGE 77

77 Table 3-15. Program PRESENCE m odel selection results for th e marsh rice rat, including Akaikes Information Criterion (AIC) and the delta AIC and AIC weights for all models with any weight. Model AIC delta AIC AIC weight (ORV, Habitat, Percent Cover),p(.) 1589.640 0.731 (Habitat, Percent Cover),p(.) 1591.642 0.2689

PAGE 78

78 Table 3-16. Program PRESENCE beta estimates, standard errors (S.E.), and lower and upper 95% confidence intervals for the marsh rice rat. The best model and second best model are shown. Best Model Parameter Beta EstimateS.E.Lower 95% C.I. Upper 95% C.I. ORV 0.72120.3732-0.01027 1.4528 Habitat -2.68350.4127-3.4924 -1.8746 Percent Cover 1.6840.42080.8592 2.5088 Second Best Model Parameter Beta EstimateS.E.Lower 95% C.I. Upper 95% C.I. Habitat -1.92060.3847-2.6746 -1.1667 Percent Cover 1.95260.3947 1.1790 2.7262

PAGE 79

79 Table 3-17. Program PRESENCE model selection results for the cotton mouse, including Akaikes Information Criterion (AIC) and the delta AIC and AIC weights for all models with any weight. Model AIC delta AIC AIC weight (ORV, Habitat, Percent Cover),p(.) 1000.72 0 0.8422 (ORV, Percent Cover),p(.) 1004.073.35 0.1578

PAGE 80

80 Table 3-18. Program PRESENCE beta estimates, standard errors (S.E.), and lower and upper 95% confidence intervals for the best model of the cotton mouse. Covariate Beta EstimateS.E.Lower 95% C.I.Upper 95% C.I. ORV 2.76720.56 1.66883.8656 Habitat 1.08690.5060.095732.0781 Percent Cover -2.05660.498-3.0327-1.0805

PAGE 81

81 Table 3-19. The number of individuals marked, r ecaptured and the return rate (proportion of marked individuals recaptured at least once) for the hispid cotton rat, marsh rice rat, and cotton mouse in off-road vehicle (O RV) and non-off-road vehicle (Non-ORV) areas. Species Treatment # Marked # Recaptured Return Rate Cotton Rat ORV 88 4652.27% Non-ORV 121 5847.93% Marsh Rice Rat ORV 83 1416.87% Non-ORV 77 1823.38% Cotton Mouse ORV 46 2043.48% Non-ORV 5 240.00%

PAGE 82

82 Table 3-20. Model selection table for Cormack-J olly-Seber closed population mark-recapture model of the cotton rat, including Quasilikelihood Akaikes Information Criterion for small sample sizes (QAICc), model weights based on QAICc, the number of parameter in each model, and the model deviance. Model structure includes the effects of time (t), and off-road vehicl e (ORV) group (g) on apparent survival ( ) and capture probability (p). Model QAICc Delta QAICc QAICc Weights # Par QDeviance (g*t) p(.) 531.842 00.2997719 148.5903 (.) p(g*t) 532.036 0.19390.2720716 155.33299 (g) p(g*t) 532.76 0.9183 0.189417 153.88531 (g*t) p(g) 533.885 2.04320.1079220 148.42883 (t+g)p(.) 535.026 3.1844 0.06113 164.77562 (g*t) p(t) 537.107 5.26530.0215526 138.18746 (t+g)p(g) 537.115 5.27330.0214614 164.72445 (t) p(g*t) 538.597 6.75480.0102324 144.21006 (t+g)p(t) 539.099 7.25680.0079620 153.64243 (t+g)p(t+g) 541.314 9.47240.0026321 153.64221 (.)p(t+g) 541.931 10.08960.0019311 175.92973 (g)p(t+g) 542.078 10.23620.0017912 173.95707 (g*t) p(g*t) 542.716 10.8742 0.001330 134.59058 (t) p(.) 544.847 13.00530.0004512 176.72616 (t) p(g) 545.827 13.98490.0002813 175.57612 (g) p(t) 547.506 15.66430.0001212 179.38521 (t)p(t+g) 548.908 17.06650.0000620 163.4521 (.) p(t) 549.152 17.30970.0000510 185.25872 (t) p(t) 552.055 20.21360.0000120 166.59917 (g) p(.) 557.407 25.5652 03 207.9932 (g) p(g) 559.339 27.4976 04 207.88701 (.) p(.) 560.475 28.6332 02 213.0901 (.) p(g) 561.672 29.8303 03 212.25836 (g+t) p(g*t) 205.391 30.4477 026 49.792671 (t) p(g*t) 205.652 30.7091 026 50.054092

PAGE 83

83 Table 3-21. Estimates, standard error (S.E.), and 95% confidence interval (C.I.) of the beta values for the off-road vehicle (O RV) effect on appa rent survival ( ) and capture probability (p) on the cotton rat, marsh rice rat, and cotton mouse. Species Parameter Beta S.E. Lower 95% C.I. Upper 95% C.I. Cotton Rat 0.75670.18170.39760.9449 p 1.51630.2951 0.94492.0876 Marsh Rice Rat -0.62000.4045-1.41290.1728 p -1.25810.7744-2.77590.2597 Cotton Mouse -0.28660.8184-1.89071.3174 p -0.14191.5411-3.16242.8786

PAGE 84

84 Table 3-22. Model selection table for Cormack-J olly-Seber closed population mark-recapture model of the marsh rice rat, including Quasi-likelihood Akaikes Information Criterion for small sample sizes (QAICc ), model weights based on QAICc, the number of parameter in each model, and the model deviance. Model structure includes the effects of time (t), and off -road vehicle (ORV) group (g) on apparent survival ( ) and capture probability (p). Model QAICc Delta QAICc QAI Cc Weights # Par QDeviance (.) p(g) 174.943 00.312373 72.98424 (g) p(.) 175.105 0.16170.288113 73.145939 (.) p(.) 175.442 0.49850.243462 75.541814 (g) p(g) 176.525 1.58180.141644 72.486676 (g) p(t) 183.506 8.56320.0043211 64.31812 (.) p(t) 183.577 8.63350.0041710 66.618638 (g) p(g+t) 185.196 10.25270.0018512 63.754602 (.) p(g+t) 185.775 10.8320.0013911 66.586976 (t) p(g) 186.78 11.83650.0008413 63.062065 (t) p(.) 187.358 12.41510.0006312 65.91702 (g+t) p(.) 187.887 12.94360.0004813 64.169162 (g+t) p(t) 188.934 13.99050.0002917 55.871341 (g+t) p(g) 189.003 14.060.0002814 62.98574 (g+t) p(g+t) 191.33 16.3870.0000918 55.869869 (t) p(t) 191.93 16.98630.0000617 58.867126 (g*t) p(.) 195.766 20.82240.0000121 52.956533 (g*t) p(g) 197.426 22.4829 022 52.114477 (t) p(g+t) 197.588 22.6448 020 57.25474 (.) p(g*t) 198.527 23.5835 019 60.642879 (g*t) p(t) 201.597 26.6542 026 45.999209 (g) p(g*t) 202.133 27.1901 021 59.324288 (g*t) p(g+t) 204.232 29.2888 027 45.990648 (g*t) p(g*t) 204.823 29.8796 028 43.908864 (g+t) p(g*t) 205.391 30.4477 026 49.792671 (t) p(g*t) 205.652 30.7091 026 50.054092

PAGE 85

85 Table 3-23. Model selection table for Cormack-J olly-Seber closed population mark-recapture model of the cotton mouse, including Quasi-likelihood Akaikes Information Criterion for small sample sizes (QAICc ), model weights based on QAICc, the number of parameter in each model, and the model deviance. Model structure includes the effects of time (t), and off -road vehicle (ORV) group (g) on apparent survival ( ) and capture probability (p). Model QAICc Delta QAICc QAICc Weights # Par QDeviance (.) p(.) 107.885 00.515022 49.434354 (g) p(.) 109.89 2.0049 0.1893 49.310253 (.) p(g) 110.006 2.12030.178413 49.425677 (g) p(g) 112.065 4.17940.063724 49.309945 (g*t) p(.) 115.256 7.37050.0129210 38.402476 (.) p(t) 115.496 7.61050.011468 43.553207 (t) p(.) 115.583 7.69820.010978 43.640895 (g) p(t) 117.684 9.79840.003849 43.313673 (g*t) p(g) 117.714 9.82910.0037811 38.319981 (g+t) p(.) 117.742 9.85710.003739 43.372329 (.) p(g+t) 117.878 9.99260.003489 43.507867 (g) p(g+t) 120.167 12.28170.0011110 43.313637 (t) p(g) 120.489 12.60330.0009410 43.635216 (.) p(g*t) 121.141 13.25540.0006811 41.746326 (g+t) p(g) 122.763 14.8777 0.000311 43.368629 (g) p(g*t) 123.05 15.16450.0002612 41.054579 (g*t) p(t) 123.453 15.56820.0002114 36.068665 (t) p(t) 125.703 17.81770.0000713 41.045051 (g*t) p(g*t) 127.984 20.09920.0000216 34.943859 (g+t) p(t) 128.161 20.2760.0000214 40.77646 (t) p(g+t) 128.22 20.33490.0000214 40.83531 (g*t) p(g+t) 128.996 21.11090.0000116 35.955543 (g+t) p(g+t) 130.83 22.94470.0000115 40.651702 (t) p(g*t) 133.163 25.2777 016 40.122317 (g+t) p(g*t) 134.873 26.9879 017 38.898625

PAGE 86

86 Figure 3-1. Total number of captures for hispid co tton rat, marsh rice rat and cotton mouse from October 2004 to September 2005. Hispid Cotton Rat Cotton Mouse Marsh Rice

PAGE 87

87 Figure 3-2. Apparent survival ( ) and 95% confidence interval of cotton rat by off-road vehicle (ORV) group and total rainfall for the 10 mont hly survival intervals. Estimates for cotton rats are averaged across models as no model had a majority of QAICc weight (Burnham and Anderson 1998).

PAGE 88

88 Figure 3-3. Apparent survival ( ) and 95% confidence interval of marsh rice rat by off-road vehicle (ORV) group for the 10 monthly samp ling intervals. Estimates for marsh rice rats are averaged across models as no model had a majority of QAICc weight (Burnham and Anderson 1998).

PAGE 89

89 Figure 3-4. Survival ( ) and 95% confidence interval of cotton mouse for the 10 monthly sampling intervals.

PAGE 90

90 CHAPTER 4 CONCLUSIONS Introduction In the beginning of this thesis I outlined the potential harm ORVs may have on wildlife and explained why small mammals would be useful indicators of ORV impacts. In this chapter, I will address the major problems that ORVs cause for wildlife in BCNP and explain whether the results of this research support the use of small mammals as indicators of ORV impacts. ORV Impacts on Wildlife Habitat Modification I observed in Chapter 2 that the direct im pacts of ORVs have influenced characteristics of vegetative communities that make up small mamma l habitat in BCNP. I found that vegetation height in ORV areas were significantly less than non-ORV areas. Duev er et al. (1981) found similar results where there was a decreased trend in height with increased ORV activity. Duever et al. (1981) also showed little r ecovery in height after one year from the heavily impacted sites. The continual use of heavily used trails may dela y recovery time from im pacts or not recover at all. Recovery may also change species compos ition in vegetation. Duever et al. (1981) reported that Panicum species and Muhlenbergia species declined and Utricularia species increased from the increased water levels from ruts. The shif t from aquatic grasses and sedges to submerged aquatic species could have a dramatic e ffect on how wildlife u tilizes the area. I also observed changes in percent cover, where ORV sites had less than non-ORV sites. Duever et al. (1981) also saw changes in percent cover as well in his experiment. While Duever et al. (1981) saw a decrease in marsh habitats, he did not see any differen ce in pine habitats. I observed a decrease in both marsh and pine habi tats. The difference may come from the duration of impacts at experimental sites. Duever et al. (1981) experimented at sites where no impacts

PAGE 91

91 were observed and conducted experiments on intensity of impacts. My sites were established in continuous ORV use areas. This shows that not only does intensity of impacts matter, but duration of impacts are impor tant to monitor as well. Ruts cutting through pine and prairie hab itats have a potential to create habitat fragmentation. Ruts could create small patches of habitat that could affect wildlife. Ruts may decrease patch size and increase distances between habitats. Habitat fragmentation has adverse effects on wildlife populations th at may influence local abundance (Fahrig and Paloheimo 1988), viability (Fahrig and Merriam 1985; Lande 1987; Roff 1974), and community organization (Holt 1985, 1993). On the other hand, the total species dive rsity in a landscape may increase when new patches of habitat are created from habitat fragmentation. Although we were not studying the effects of fragmentation of hab itat by ORVs, examining the potential of fragmentation on small mammals should be considered. Studies could be designed for small mammal to examine effects of patch size, distances between patches, and movement. Species Composition and Structure There was a significant change in sm all mammal population structure in terms of abundances in disturbed areas. Di sturbance preferences of the th ree small mammal species show three types of distribution pa tterns: ORV preference (cotton mouse), ORV avoidance (hispid cotton rat), and no preference (marsh rice rat). Waddle (2006) showed similar results with anurans where all species were present in OR V and non-ORV sites, three of four species preferred non-ORV areas and the remaining species preferred ORV areas. This suggests that examining species richness may not be the best method for monitoring impact of ORVs. Monitoring species assemblages could be a more efficient way. Species assemblages can serve as indicators for monitoring composition, structur e, and function in natural areas (Kremen 1992). Monitoring species assemblages al lows the most direct assessme nt of viable populations and

PAGE 92

92 biodiversity. Monitoring at the population or community level can provide direct or indirect assessments of ecological change (Gilbert 1980). When monitoring populations are conducted within the context of know environmental change (i.e. ORV use), it can provide a base for management decisions (Kremen 1992). Species a ssemblages should provide a finer estimate of biotic response than single species. Assemblage s that include species co vering a wide range of movement and distribution could be expected to display a greater ra nge of sensitivities to habitat modification or fragmentation over time (T erborgh 1974). Monitoring programs could use separate indicator species assemblages, such as small mammals and amphibians, to monitor different environmental impacts (Kremen 1992). Small Mammals as an Indicator Species An ideal indicator species m ust be abundant and/or cost effective to sample. Small mammals in southern Florida meet this requiremen t. They have niches and microhabitats that are small and easily sampled. Sampling small mammal s using live trapping techniques, as described in Chapter 3, was an efficient way to sample several species at once. Another requirement for an indicator species is that it provides early warning of natural responses to environmental impacts (N oss 1990; Munn 1993; Woodley 1996). Chapter 3 demonstrated that two of the three species of small mammals were sensitive to ORV use. Indicator species should indicate the cause of change rather than the existence of change (Herricks and Schaeffer 1985). The responses are us eful for managers to predict changes of the whole system by monitoring community levels. Sm all mammals are suitable to detect changes over the BCNP ecosystem. They are found throughout the terrestrial habitats in BCNP and are the base of the food chain for many species. I be lieve this study has shown how small mammals responded to anthropogenic stre sses in the short term.

PAGE 93

93 Conclusions I conclude that sm all mammals could serve as indicators of ORV effects. There is detailed information about their biology and natural hist ory on the levels of population and community. Their roles and niches are well known in their habitats and ecosyste ms. Small mammals are easily marked for identification and we can follow their survivorship, reproduction, and population fluctuations from commu nity to ecosystem levels. Th e spatial scale on which small mammals live is relatively small. They have shor t life spans and display behavioral responses to seasonal and environmental changes. In Chapter 3, I showed that species react differently to ORV impacts. While my study was for one year, there may be climatic effect on the small mammals. We need to determine what the real pattern is with multiple year study on small mammals. With the proper network of sites, lo cal populations can be studied intensively and modeled appropriately to allow managers to track changes in popul ation in relation to management decisions. At the landscape level, s ite occupancy modeling can be used on all small mammal species to monitor the ch anges in colonization and extin ction of sites throughout the Big Cypress National Preserve. My comparisons of ORV and non-ORV areas have raised a numbe r of interesting questions. With the larger rodent species ha ving lower captures in the ORV areas, how are predatory species responding to this change? How much and how fast are ORVs changing the vegetative community? To what extent is habi tat fragmentation affecting local wildlife populations? I feel that more can be gained by the continued monitori ng of the small mammal community, given that current data only spans one year. There may be population fluctuations due to hydrologic or seasonal changes over many y ears that this project did not observe. These and other research questions can be addressed using small mammals as indicator species.

PAGE 94

94 LIST OF REFERENCES Abramsky, Z. 1978. Small mammal community eco logy: Changes in species diversity in response to manipulated pr oductivity. Oecologia 34: 113. Allen, M.C. and A.J. Read. 2000. Habitat selection for foraging bottlenose dolphins in relation to boat density near Clearwater, Florid a. Marine Mammal Science 16: 815-824. Anders, F.J. and S.P. Leatherman. 1987. Disturbance of beach sediment by off-road vehicles. Environmental Geology and Wa ter Sciences 93: 183-189. Andren, H. 1994. Effects of habitat fragmenta tion on birds and mammal s in landscapes with different proportions of suitable hab itat: A review. Oikos 71 (3): 355-366. Bates, G.H. 1935. The vegetation of footpaths, sidewalks, cart tracks a nd gateways. Journal of Ecology 23: 470-487. Berry, K.H. 1980. The effects of four-wh eel vehicles on biol ogical resources. in R.N.L. Andrews and P. Nowak eds., Off-road vehicle use: A management challenge. U.S. Department of Agriculture, Office of Environm ental Quality, Washington, D.C. Brodhead, J.M.B. and P.J. Godfrey. 1979. Effects of off-road vehicles on coastal dune vegetation in the Province Lands, Cape Cod National Seashore, Massachusetts. Report 32, National Park Service Cooperative Research Unit, Un iversity of Massachsetts, MA, USA. Brown, K.E. 1973. Biological life history and geographical distribution of the cabbage palm, Sabal palmetto Ph.D. Dissertation, Department of Botany, North Carolina State University, Raleigh, North Carolina. Brown, L.N. 1997. Mammals of Florida. Wi ndward Publishing, Inc., Miami, FL. Boorman, L.A. and R.M. Fuller. 1977. Studies on the impact of paths on dune vegetation at Winterton, Norfolk, England. Biological Conservation 12: 203-215. Burnham, K.P., and D.R. Anderson. 1998. Mode l selection and multi-model inference: a practical information-theoretic approac h. Springer-Verlag, New York, NY, USA. Bury, R.B. 1980. What we know and dont know about off-road vehicle impacts on wildlife. Pages 110-123. in R.N.L. Andrews and P. Nowak eds., Off-road vehicle use: A management challenge. U.S. Department of Agriculture, Office of Environmental Quality, Washington, D.C. _____, R.A. Luckenbach, and S.D. Bussack. 1977. Effect s of off-road vehicles on vertebrates in the California Desert. U.S. Fish and Wildlife Service, Wildlife Research Report, 8: 1-23. Cameron, G.N. and S.R. Spencer. 1981. Sigmodo n hispidus. Mammalian Species 158: 1-9.

PAGE 95

95 Caron, L.M.J. and D.E. Sergeant. 1988. Yearly variation in the frequency of passage of beluga whales ( Delphinapterus leucas ) at the mouth of the Saguenay River, Quebec, over the past decade. Naturalist e Canadien. 115: 111-116. Castleberry, S.B., T.L. King, P.B. Wood, and W. M. Ford. 2002. Microsatellite DNA markers for the study of Allegheny woodrat ( Neotoma magister ) populations and cross-specific amplification in the genus Neotoma. Molecular Ecology 9: 824-826. Chappell, H., J. Ainsworth, R. Cameron, and M. Redfern. 1971. The effects of trampling on a chalk grassland ecosystem. Journal of Applied Ecology. 8: 869-882. Chase, M.K., W.B. Kristan III, A.J. Lynam, M.V. Price, and J.T. Rotenberry. 2000. Single species as indicators of species richness and composition in californi a coastal sage scrub birds and small mammals. Conservation Biology 14 (2): 474-487. Clark, B. K., Kaufman, D. W., Finck, E. J., & Kaufman, G. A.1989. Small mammals in tallgrass prairie: Patterns associated with gr azing and burning. Prairi e Naturalist 21: 177. Cothran, G.E., M.H. Smith, J.O. Wolff, and J. B. Gentry. 1991. Mammals of the Savannah River Site. Savannah River Site National Environmental Research Park Program. Cole, D.N. 1985. Recreational tram pling effects on six habitat types in Western Montana. USDA Forest Service Research Paper INT350. Cole, D.N. and R.L. Knight. 1990. Impacts of re creation on biodiversity in wilderness. Pages 3340 in Proceedings from the Symposium on Wilder ness Areas: Their Impacts. Utah State University, Logan, Utah USA. Cordell, H.K. and J.C. Bergstrom. 1991. A methodology for assess ing national outdoor recreation demand and supply trends. Leisure Sciences 13: 1-20. Dale, V. H., & Beyeler, S. C. 2001. Challenge s in the development and use of ecological indicators. Ecological Indicators 1: 3. Davidson, E. and M. Fox. 1973. Effects of off-ro ad motorcycle activity on Mojave Desert vegetation and soil. Madrono 22: 381-390. Diffendorfer, J.E., N.A. Slade, M.S. Gaines R.D. Holt. 1995. Population dynamics of small mammals in fragmented and continu ous old-field habitat. Pages175-199 in Lidicker, W.Z. Jr., ed. Landscape approaches in mamm alian ecology and conservation. University of Minnesota Press, Minneapolis, MN. Dorrance, M.J. P.J. Savage, and D.E. Huff. 1975. Effects of snowmobiles on White-tailed Deer. Journal of Wildlife Mana gement 39(3): 563-569. Duever, M.J., J.E. Carlson, and L.A. Riopelle. 19 81. Off-road vehicles an d their Impacts in the Big Cypress National Preserve. National Audubon Society, Ecosystem Research Unit.

PAGE 96

96 _____, J.E. Carlson, and L.A. Riopelle. 1975. Ecosys tem analyses at Corkscrew Swamp. Pages 627-725 in H.T. Odum, K.C. Ewel, J.W. Ordway and M.K. Johnston, eds., Cypress wetlands for water management, recycling and conservation. Second annual Report to National Science Foundation and Rockeller Found ation. Center for Wetlands, University of Florida, Gainesville, Florida. _____, L.A. Riopelle and J.M. McCollum. 1986. Long term recovery of experimental off-road vehicle impacts and abandoned old trails in the Big Cypress National Preserve. South Florida Research Center Report SFRC-86/09. Ecker, R.E. Jr., M.K. Wood, W.H. Blackburn an d F.F. Petersen. 1979. Impacts of off-road vehicles of infiltration and sediment produc tion of two desert soils. Journal of Range Management 32: 394-397. Emers, M., J.C. Jorgenson, and M.K. Raynolds 1995. Recreational respon se of arctic tundra plant communities to winter vehicle distur bance. Canadian Journal of Botany 73: 905917. English, D.B.K., C.J. Betz, J.M. Young, J.C. Bergstrom, and H.K. Cordell. 1993. Regional demand and supply projections for outdoor recreation. USDA Forest Service Genereal Technical Report RM-230. Fort Collins, Colorado: Rocky Mountain Forest and Range Experiment Station. Evans, P.G.H., P.J. Canwell and E.J. Lewis. 1992. An experimental study of the effects of pleasure craft noise upon bottl e-nosed dolphins in Cardigan Bay, West Wales. Pages 4346 in P.G.H. Evans, eds. European research on cetaceans. European Cetacean Society, Cambridge. Fahrig, L. and G. Merriam. 1985. Habitat patc h connectivity and populat ion survival. Ecology 66(6): 1762-1768. _____ and J. Paloheimo. 1988. Effect of spatial arrangement of habitat patches on local population size. Ecology 69(2): 468-475. Frankham, R. 1995. Conservation genetics. Annual Review of Genetics 29: 305-327. Franklin, A.B., B.R. Noon, and T.L. George. 2002. What is habitat fragme ntation? Studies in Avian Biology 25: 20-29. Foresman, C.L., D.K. Ryerson, R.N. Walejko, W.H. Paulson and J.W. Pendleton. 1976. Effect of snowmobile traffic on bluegrass ( Poa pratensis ). Journal of Environmental Quality 5(2): 129-130. Gaines, M.S. and L.R. McClenaghan, Jr. 1980. Dispersal in small mammals. Annual Review of Ecology and Systematics 11(1): 163-196. Garrison, B.A. 2005. Fragmentation of terrestrial habitat: An ove rview for wildlife biologist. Transactions of the West ern Section of the Wildlife Society 41: 42-58.

PAGE 97

97 Geist, V. 1971. Bighorn sheep ecology. Wildlife Society News 136: 20-25. Gilbert, L.E. 1980. Food web organization and the conservation of ne otropical diversity. In M.E. Soule and B.A. Wilcox, eds. Conservation biology: an evolu tionary-ecological perspective. Sinauer Associates, Sund erland, Massachusetts, USA. Pp. 11-34. Hammitt W.E. and D.N. Cole. 1998. Wildland recreation: ecology and management. 2nd ed. John Wiley and Sons, Inc., New York, USA. Hela, I. 1952. Remarks on the climate of southern Florida. Bulletin of Marine Science Gulf Caribbean 2: 438-447. Herricks, E.E. and D.J. Schaeffer. 1985. Ca n we optimize biomonitoring? Environmental Management 9(6): 487-492. Hofstetter, R.H. 1984. The effect of fie on the pineland and sawgrass communities of southern Florida. In: Gleason, P.J. ed. Environments of South Fl orida Present and Past II. Coral Gables FL: Miami Geologi cal Society, pp. 201-212. Holt, R.D. 1985. Population dynamics in two-patch environments: some anomalous consequences of an optimal habitat dist ribution. Theoretical P opulation Biology 28: 181208. _____. 1993. Ecology at the mesoscale: the infl uence of regional processes on local communities. In : Ricklefs, R. and D. Schlutter eds. Species diversity in ecological communities. Chicago IL: Chicago University Press, pp. 77-88. Hosier, P.E. and T.E. Eaton. 1980. The impact of vehicles on dune and grassland vegetation on the south-eastern North Carolina barrier beac h. Journal of Applied Ecology 17: 173-182. Janik, V.M., and P.M.Thompson. 1996. Changes in su rfacing patterns of bo ttlenose dolphins in response to boat traffic. Mari ne Mammal Science 12: 597-602. Janis, M.W. and J.D. Clark. 2002. Responses of Florida panthers to re creational deer and hog hunting. Journal of Wildlife Management 66(3): 839-848. Kaufman, D.W., G.A. Kaufman,, P.A. Fay, J.L. Zimmerman, J. L., and E.W. Evans. 1998. Animal populations and communities. In A. K. Knapp, J. M. Briggs, D. C. Hartnett, & S.L. Collins (Eds.), Grassland dynamics (pp. 113). New York: Oxford University Press. _____, G.A. Kaufman,, andE.J. Finck, 1983. Effects of fire on rodents in tallgrass prairie of the flint hills region of eastern Kansas. Prairi e Naturalist 15: 49. Ketcham, D.E. and J.E. Bethune. 1963. Fire resist ance of the south Florida slash pine. Journal of Forestry 61: 529-530.

PAGE 98

98 Knight, R.L. and D.N. Cole. 1995. Wildlife response to recreationists. Pages 51-70 in R.L. Knight and K.J. Gutzwiller eds. Wildlif e and recreationists: Coexistence through management and research. Island Press, Washington, D.C., USA. ______ and K.J. Gutzwiller. 1995. Wildlife and re creationist: Coexistence through management and research. Island Pres s, Washington, D.C. Kremen, C. 1992. Assessing the indi cator properties of species assemblages for natural areas monitoring. Ecological App lications 2(2): 203-217. Kruse, S. 1991. The interactions between killer wh ales and boats in Johnstone Strait, B.C. Pages 149-159 in K.S. Norris and K. Pryor, eds. Dolphi n societies: Discoveries and puzzles. University of California Press, Berkeley, CA. Labisky, R.F., K.E. Miller, and C.S. Hartless. 1999. Effect of Hurrican Andrew on survival and movements of white-tailed deer in the Ev erglades. Journal of Wildlife Management 63(3): 872-879. Lacy, R.C. 1993. Impacts of inbreeding in natu ral and captive populati ons of vertebrates: implications for conservation. Perspec tives in Biology and Medicine 36: 480-496. Lande, R. 1987. Demographic models of the northern spotted owl ( Strix occidentalis caurina ). Oecologia 75(4): 601-607. Lathrop, E.W. 1983. The effect of ve hicle use on desert vegetation. in Environmental effects of off-road vehicles, R.H. Webb and H.G. W ilshire eds. Pages 153-166. Springer-Verlag, New York. Liddle, M.J. and P. Greig-Smith. 1975. A survey of tracks and paths in a sand dune ecosystem. II. Vegetation. Journal of Applied Ecology 12: 909-930. Lidicker, Z.W. Jr., J.O. Wolff, L.N. Lidicker, and M. H. Sm ith. 1992. Utilization of a habitat mosaic by cotton rats during a population decline. Landscape Ec ology 6(4): 259-268. Lindenmayer, D.B., R.B. Cunningham, M.L. Pope, and C.F. Donnelly, 1999. The response of arboreal marsupials to landscape context: a large-scale fragmentat ion study. Ecological Applicaitions 9: 594-611. Loope, L, M. Duever, A. Herndon, J. Snyder, an d D. Jansen. 1994. Hurricane impact on uplands and freshwater swamp forest BioScience 44(4): 238-246. Lopez, R.R., N.J. Silvy, and P.A. Frank. 2003. Hu rricane impacts on key deer in the Florida Keys. The Journal of Wildlife Management 67(2): 280-288. MacKenzie, D.I., J.D. Nichols, G.B. Lachman, S. Droege, J.A. Royle, and C.A. Langtimm. 2002. Estimating site occupancy rate s when detection probabilities are less than one. Ecology 83: 2248-2256.

PAGE 99

99 _____, J.D. Nichols, J.A. Royle, K.H. Pollock, L.L. Bailey, and J.E. Hines. 2006. Occupancy estimation and modeling: inferring pattern s and dynamics of species occurrence. Academic Press, Burlington, MA, USA. Mares, M.A. and K.A. Ernest. 1995. Population and community ecology of small mammals in a gallery forest of central Brazil. Journal of Mammalogy 76(3): 750-768. McEwen, D. and S.R. Tocher. 1976. Zone manageme nt: Key to controlling recreational impact in developed campsites. Journal of Forestry 74: 90-93. McMinn, J.W. and W.H. McNab. 1971. Early grow th and development of slash pine under drought and flooding. U.S. Forest Serv ice Research Paper No. SE-89. Miller, S.G, R.L. Knight, and C.K. Miller. 1998. Influence of recreational trails on breeding bird communities. Ecological Applications 8(1): 162-169. Moen, A.N. 1976. Energy conservation by White-taile d Deer in the winter. Ecology 57(1): 192198. Monty, A.M. 1997. The eastern woodrat (Neotoma floridana ) in southern Illinois: population assessment and genetic variation. Ph.D. di ssertation, Southern Illinois University, Carbondale, IL. Mossman, C.A. and P.M. Waser. 2001. Effects of habitat fragmentation on population genetic structure in the white-footed mouse ( Peromyscus leucopus ). Canadian Journal of Zoology 79: 285-295. Munn, R.E. Monitoring for ecosystem integrity. in Woodley, S, K. James, and G. Francis eds. Ecological integrity and the management of ecosystems. Pages 105-116. St.Lucie Press, Florida. Myers, R.L. 1977. A preliminary study of the Sabal palmetto forest on Little Corkscrew Island, Florida. Interim Report. National Audubon So ciety Ecosystem Rese arch Unit, Naples, Florida. National Park Service. 2000. Fina l recreational off-road vehicle management plan supplemental environmental impact statement. USNPS, Big Cypress National Preserve. Ochopee, Florida. 603 pp. Neumann, P.W. and H.G. Merriam. 1972. Ecological effects of snowmobiles. Canadian Field Naturalist 86(3): 207-212. Noss, R.F. 1990. Indicators for monitoring biodiversity: A hierarchical approach. Conservation Biology 4(4): 355-364. Nudds, T.D. 1977. Quantifying the vegetative struct ure of wildlife cover. The Wildlife Society Bulletin 5(3): 113-117.

PAGE 100

100 OBrien, S.J., M.E. Roelke, L. Marker, A. Newm an, C.A. Winkler, D. Metzler, L. Colly, J.F. Evermann, M. Bush, and D.E. Wildt. 1985. Genetic basis for species vulnerability in the cheetah. Science 227: 1428-1434. Pfister, C., B.A. Harrington, and M. Lavine 1992. The impact of human disturbance on shorebird migration staging area. Biological Conservation 60: 115-126. Reijnen, R., R. Foppen and G. Veenbass. 1994. Th e effects of car traffic on breeding bird populations in woodland. I. Evidence of reduced habitat quality for Willow Warbler ( Pylloscopus trochilius ) breeding close to a highway. Journal of Applied Ecology 31:8594. Roff, D.A. 1974. Spatial heterogeneity and the pe rsistence of populations. Oecologia 15(3): 245258. Schmid, W.D. 1971. Modification of the subnive an microclimate by snowmobiles, Pages 251257. in A.O. Haugen ed. Proceedings of the snow and ice in relation to wildlife and recreation symposium. Iowa Cooperative Wildlife Research Unit, Iowa, USA. Skagen, S.K., R.L. Knight, and G.H Orians. 199 1. Human disturbance of an avian scavenging guild. Ecological Applications 1: 215-225. Smith, A.T. and J.M. Vrieze. 1979. Population struct ure of Everglades rodents: responses to a patchy environment. Journal of Mammalogy 60(4): 778-794. Stace-Smith, R. 1975. The misuse of snowmobiles against wildlife in Canada. Nature Canada 4(4): 3-10. Stalmaster, M.V. and J.R. Newman. 1978. Behavior responses of mintering bald eagles to human acitivity. Journal of Wildlife Mangament 42: 506-513. Steiner, A.J. and S.P. Leatherman. 1981. Recreat ional impacts on the dist ribution of ghost crabs Ocypode quadrata Fab. Biological Conservation 20: 111. Swilling, W.R. Jr., M.C. Wooten, N.R. Holler, and W.J. Lynn. 1998. Population dynamics of Alabama beach mice ( Peromyscus polionotus ammobates ) following Hurricane Opal. American Midland Naturalist 140(2): 287-298. Tebeau, C.W. 1966. Floridas last frontier: Th e history of Collier County, rev ed. Hallmark Press, Inc. Miami,FL. Terborgh, J. 1974. Preservation of natural diversit y: the problem of extinction prone species. BioScience 24: 715-722. Trew, M.J. 1973. The effects and management of trampling on coastal sand dunes. Journal of Environmental Planning and Pollution Control 1: 38-49.

PAGE 101

101 Vollmer, A.T., B.G. Maza, P.A. Medica, F.B. Turner, and S.A. Bamburg. 1976. The impact of off-road vehicles on a desert ecosyste m. Environmental Management 1: 115-129. Wade, D., J. Ewel, and R. Hofstetter. 1980. Fire in South Florida ecosystems. U.S. Forest Service General Technical Report No. SE-17. Southeast Forest Experimental Station, Asheville, North Carolina. Webb, R.H. 1983. Compaction of desert soils by off-road vehicles in Environmental effects of off-road vehicles, R.H. Webb and H.G. W ilshire eds. Pages 51-166. Springer-Verlag, New York. Webb, W.L. 1968. Public use of fore st wildlife: Quantity and quality considerations. Journal of Forestry 66: 106-110. White, G.C., and K.P. Burnham. 1999. Program Ma rk: survival estimation from populations of marked animals. Bird Study 46 supplement: 120-138. Wilshire, H.G., J.K. Nakata, S. Shipley, and K. Prestegaard. 1978. Impacts of vehicles on natural terrain at seven sites in the San Francisc o Bay area. Environmental Geology 2(5): 295319. Wolcott, T.G. and D.L. Wolcott. 1984. Impact of off-road vehicles on macroinvertebrates of a mid-Atlantic beach. Biological Conservation. 29: 217-240. Wolfe, J.L. 1982. Oryzomys palustri s. Mammalian Species 176: 1-5. _____ and A.V. Linzey. 1977. Peromyscus gossypinus. Mammalian Species 76: 1-5. Wolff, J.O. 1994. More on juvenile di spersal in mammals. Oikos 71:349-352. Wooding, F.J. and S.D. Sparrow. 1978. An assessment of damage caused by off-road vehicle traffic on subarctic tundra in the Denali Hi ghway area of Alaska. U.S. Forest Service Recreational Impact on Wildlands Conference. 89-93. Woodley, S. 1996. A scheme for ecological monito ring in national parks and protected areas. Environments 23(3): 50-74. Woodward, T.N., R.J. Gutierrez, and W.H. Ru therford. 1974. Bighorn ram production, survival, and mortality in southcentral Colorado. J ournal of Wildlife Management 38(4): 771-774. Zande, A.N. van der, and P. Vos. 1984. Impact of semi experimental increase in recreation intensity on the dens ities of birds in groves and hedges on a lake shore in The Netherlands. Biological C onservation 30: 237-259. Zar, J.H. 1984. Biostatistical Analysis, 2nd ed. Prentice-Hall, Inc. Englewood Cliffs, NJ.

PAGE 102

102 BIOGRAPHICAL SKETCH Brian Jeffery was born in Ft. Lauderdale, Florida in 1977. Brian rece ived his high school diploma from South Plantation High School in1996. He then attended Florida International University and received his Bach elor of Science in environmen tal studies in 2001. While he was an undergrad, Brian worked as a student technician in the Aquatic Ecology Lab at FIU, working for Joel Trexler. After graduation, Brian joined FART (Florida Alligator Research Team) and then worked as a technician helping a graduate student doing alligator research in A.R.M. Loxahatchee National Wildlife Refuge for six months. In the beginning of 2002, Brian worked in Big Cypress National Preserve doing reptile and amphibian inventory. In 2004, Brian enrolled at the University of Florida to pursue a Master of Science in School of Natural Resources and Environment.