<%BANNER%>

Effects of Habitat Type and Structure on Detection Probabilities of American Alligators (alligator Mississippiensis) dur...

Permanent Link: http://ufdc.ufl.edu/UFE0041794/00001

Material Information

Title: Effects of Habitat Type and Structure on Detection Probabilities of American Alligators (alligator Mississippiensis) during Night-Light Counts
Physical Description: 1 online resource (62 p.)
Language: english
Creator: Carter, Cameron
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: alligator, detectability, estimates, habitat, population, sightability, survey
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: Obtaining reliable population estimates is one of the most challenging issues in wildlife ecology today. Population surveys are an attempt to obtain an index of population size but rarely achieve the total number of all individuals in the population. These estimations are affected by many variables that lead to bias and variability in counts, i.e. the probability of detecting an individual animal is < 1 and this probability varies. One of the most important steps in increasing precision and accuracy in estimations is to design surveys so detection probabilities can be estimated. The objective of this thesis was to determine the effect of habitat type, vegetation height, visual obstruction, water depth, distance from transect, survey boat seat height, and survey speed on detection rates of American alligators (Alligator mississippiensis) during night light surveys. I quantified effects on detection due to habitat types by establishing transects in different alligator habitats with reflective markers to simulate alligators eye reflections and then conducted mock surveys with experienced observers to determine the proportion of markers detected. Results from these surveys were then modeled using PROGRAM MARK to determine which variables had the most influence on alligator detectability. Detectability was found to be 0.53 (SE = 0.0611), 0.63 (SE = 0.0498), and 0.51 (SE = 0.0483) in the habitats of sawgrass marsh, slough, and wet prairie respectively. The variable that had the greatest effect on detection rates was visual obstruction, followed by distance from the transect, vegetation height, water depth, and seat height. Detectability functions for the various habitats were obtained and varied slightly for each habitat by the variables included in the optimum model. These functions can be used to make more reliable population estimates during alligator population monitoring in Florida. These methods could also be used as a model to determine the effect of habitat on detectability for crocodilian species worldwide. Increased precision of estimates will allow alligator researchers to increase response time to population changes and will provide a basis for making better management decisions.
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 Cameron Carter.
Thesis: Thesis (M.S.)--University of Florida, 2010.
Local: Adviser: Percival, Henry F.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0041794:00001

Permanent Link: http://ufdc.ufl.edu/UFE0041794/00001

Material Information

Title: Effects of Habitat Type and Structure on Detection Probabilities of American Alligators (alligator Mississippiensis) during Night-Light Counts
Physical Description: 1 online resource (62 p.)
Language: english
Creator: Carter, Cameron
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: alligator, detectability, estimates, habitat, population, sightability, survey
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: Obtaining reliable population estimates is one of the most challenging issues in wildlife ecology today. Population surveys are an attempt to obtain an index of population size but rarely achieve the total number of all individuals in the population. These estimations are affected by many variables that lead to bias and variability in counts, i.e. the probability of detecting an individual animal is < 1 and this probability varies. One of the most important steps in increasing precision and accuracy in estimations is to design surveys so detection probabilities can be estimated. The objective of this thesis was to determine the effect of habitat type, vegetation height, visual obstruction, water depth, distance from transect, survey boat seat height, and survey speed on detection rates of American alligators (Alligator mississippiensis) during night light surveys. I quantified effects on detection due to habitat types by establishing transects in different alligator habitats with reflective markers to simulate alligators eye reflections and then conducted mock surveys with experienced observers to determine the proportion of markers detected. Results from these surveys were then modeled using PROGRAM MARK to determine which variables had the most influence on alligator detectability. Detectability was found to be 0.53 (SE = 0.0611), 0.63 (SE = 0.0498), and 0.51 (SE = 0.0483) in the habitats of sawgrass marsh, slough, and wet prairie respectively. The variable that had the greatest effect on detection rates was visual obstruction, followed by distance from the transect, vegetation height, water depth, and seat height. Detectability functions for the various habitats were obtained and varied slightly for each habitat by the variables included in the optimum model. These functions can be used to make more reliable population estimates during alligator population monitoring in Florida. These methods could also be used as a model to determine the effect of habitat on detectability for crocodilian species worldwide. Increased precision of estimates will allow alligator researchers to increase response time to population changes and will provide a basis for making better management decisions.
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 Cameron Carter.
Thesis: Thesis (M.S.)--University of Florida, 2010.
Local: Adviser: Percival, Henry F.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0041794:00001


This item has the following downloads:


Full Text

PAGE 1

1 EFFECT S OF HABITAT TYPE AND STRUCTURE ON DETECTION PROBABILITIES OF AMERICAN ALLIGATOR S (Alligator mississippiensis) DURING NIGHT LIGHT COUNTS By CAMERON BLAIR CARTER A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVE RSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2010

PAGE 2

2 2010 Cameron Blair Carter

PAGE 3

3 To my family, friends, and colleagues

PAGE 4

4 ACKNOWLEDGMENTS I thank Allan Woodward, Ken Rice, Franklin Percival, and Frank Mazzotti for the opportunity and their guidance throughout the project. I thank researchers and technicians who helped with the project; they include, Brian Jeffery, Chris Bugbee, Joe Kern, J ustin Davis, Mark Parry, Mike Cherkiss, Amanda Rice Waddle, Hardin Waddle, Arnold Brunell, Lindsey Hord, Dwayne Carbonneau, Patrick Delaney, and Rio Throm. I thank Mark Miller, Erin Leone, and Richard Kiltie for statistical guidance and support.

PAGE 5

5 TABLE OF CONTENTS p age ACKNOWLEDGMENTS .................................................................................................................... 4 LIST OF TABLES ................................................................................................................................ 7 LIST OF FIGURES .............................................................................................................................. 8 ABSTRACT .......................................................................................................................................... 9 CHAPTER 1 INTRODUCTION ....................................................................................................................... 11 Population Estimates ................................................................................................................... 11 Detectability ................................................................................................................................. 12 Alligator Night Light Surveys .................................................................................................... 13 FWC Alligator Surveys ....................................................................................................... 14 Everglades Alligator Surveys .............................................................................................. 15 2 EFFECTS OF HABITAT TYPE AND STRUCTURE ON DETECTION PROBABILITIES OF AMERICAN ALLIGATORS .............................................................. 17 Factors Affecting Alligator Night light Surveys ....................................................................... 17 Wariness ............................................................................................................................... 17 Emergence Dynamics .......................................................................................................... 18 Habitat Characteristics ......................................................................................................... 18 Objectives .................................................................................................................................... 19 Habitat Related Det ection Probabilities .................................................................................... 19 Anticipated Results and Benefits ............................................................................................... 19 Materials/Methods ....................................................................................................................... 20 Study Area ............................................................................................................................ 20 Sawgrass Marsh ................................................................................................................... 20 Slough ................................................................................................................................... 21 Wet P rairie ........................................................................................................................... 21 Air Cannon ........................................................................................................................... 22 Reflective Markers ............................................................................................................... 22 Mock Surveys ...................................................................................................................... 23 Habitat Variables ................................................................................................................. 24 Modeling .............................................................................................................................. 25 Model Selection ................................................................................................................... 26 Results .......................................................................................................................................... 26 Discussion .................................................................................................................................... 28 Visual Obstruction ............................................................................................................... 30 Distance from Transect ....................................................................................................... 30 Vegetation Height ................................................................................................................ 30

PAGE 6

6 Water Depth ......................................................................................................................... 31 Seat Height ........................................................................................................................... 32 Management Implications ................................................................................................... 32 3 CONCLUSIONS ......................................................................................................................... 36 APPENDI X: APPLICATION OF ALLIGATOR DETECTABILITY MODELS ....................... 53 LIST OF REFERENCES ................................................................................................................... 57 BIOGRAPHICAL SKETCH ............................................................................................................. 62

PAGE 7

7 LIST OF TABLES Table page 2 1 Model set used to model American alligator detectability as a function of distance (D), visual obstruction (VO), ve getation height (VE), water depth (W), seat height (Seat), and survey speed (Speed). ......................................................................................... 39 2 2 in sawgrass, slo ugh and wet prairie habitats. ....................................................................... 40 2 3 Beta parameters for optimum alligator detectability models for sawgrass (SG), slough (SL), and wet prairie (WP) habitat. Beta is represented by LOGIT link function parameters. ............................................................................................................... 41

PAGE 8

8 LIST OF FIGURES Figure page 2 1 Locations of sawgrass, slough, and wet prairie habitats used for alligator detection su rveys in Florida. .................................................................................................................. 42 2 2 The air cannon, which was constructed to distribute reflective markers in selected habitats in Florida wetlands. The pressure gauge and firing angles were used to obta in the desired distance from the transect. ...................................................................... 43 2 3 Reflective markers used to simulate alligator eye reflections during night light counts. ..................................................................................................................................... 44 2 4 Relationship of visual obstruction (VO, dm) on alligator detection probabilities in sawgrass, slough, and wet prairie habitats during night light counts ................................. 45 2 5 Relationship o f distance from transect (D, m) on alligator detection probabilities in sawgrass, slough, and wet prairie habitats during night light counts ................................. 46 2 6 Relationship of vegetation height (VE, cm) o n alligator detection probabilities in sawgrass habitats during night light counts .......................................................................... 47 2 7 Relationship of water depth (W, cm) on alligator detection probabilities in slough, and wet prairie ha bitats during night light counts. .............................................................. 48 2 8 Relationship of seat height (Seat, cm) on alligator detection probabilities in sawgrass, slough, and wet prairie habitats during night light counts .................................................. 49 2 9 Factors affecting alligator detection probabilities during night light counts in a sawgrass habitat ...................................................................................................................... 50 2 10 Factors affecti ng alligator detection probabilities during night light counts in a slough habitat .......................................................................................................................... 51 2 11 Factors affecting alligator detection detection probabilities during night -light counts in a wet prai rie habitat ............................................................................................................ 52

PAGE 9

9 Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science EFFECT S OF HABITAT TYPE AND S TRUCTURE ON DETECTION PROBABILITIES OF AMERICAN ALLIGATORS (Alligator mississippiensis) DURING NIGHT LIGHT COUNTS By Cameron B lair Carter May 2010 Chair: H. Franklin Percival Major: Interdisciplinary Ecology Obtaining reliable population estimates is one of the most challenging issues in wildlife ecology today. Population surveys are an attempt to obtain an index of population size but rarely achieve the total number of all individuals in the population. These estimations are affected by many variabl es that lead to bias and variability in counts i.e. the probability of detecting an individual animal is < 1 and this probability varies One of the most important steps in increasing precision and accuracy in estimations is to design surveys so detection probabilities can be estimated The objective of this thesis wa s to determine the effect of habitat type, vegetation height, visual obstruction water depth distance from transect, survey boat seat height, and survey speed on detection rates of American alligators ( Alligator mississippiensis ) during night light surveys. I quantified effects on detection due to habitat types by establishing transects in different alligator habitats with reflective markers to simulate alligators eye reflections and then c onducted mock surveys with experienced observers to determine the proportion of markers detected. Results from these surveys were then modeled using PROGRAM MARK to determine which variables had the most influence on alligator detectability. Detectability was found to be 0.5 3 (SE = 0.0611) 0.6 3 (SE = 0.0498) and 0.5 1 (SE = 0.0483) in the habitats of sawgrass

PAGE 10

10 marsh, slough, and wet prairie respectively. The variable that had the greatest effect on detection rates was visual obstruction, followed by distance from the transect vegetation height, water depth, and seat height. Detectability functions for the various habitats were obtained and varied slightly for each habitat by the variables included in the optimum model These functions can be used to make more reliable population estimates during alligator population monitoring in Florida These methods could also be used as a model to determine the effect of habitat on detectability for crocodilian species worldwide. Increased precision of estimates will allow alligator researcher s to increase response time to population changes and will provide a basis for making better management decisions.

PAGE 11

11 CHAPTER 1 INTRODUCTION Population Estimates Obtaining population estimates is one of the most challenging obst acles in wildlife ecology. Population surveys are an attempt to obtain an index of population abundance but they rarely represent actual population size Scientists conducting population surveys are interested in the total number of individuals, structur e (sex ratios, age structure) and size of populations, and population distribution throughout landscapes or study area s (Samuel and Pollock 1981, Lancia et al. 2005). There are many factors to consider when designing surveys to obtain population estimate s, such as species characteristics size of the survey area, vegetation, terrain, and resources needed (Krebs 2002). Population estimates can be developed in many ways. The first is a census of the population, or a complete count of the individuals in t he population ( Williams et al. 2002 Lancia et al. 2005). Second, there are population estimates, which are estimates of total abundance based on a sample survey of individuals, nests, or other indicators, and adjusted for detection probabilities (William s et al. 2002, Lancia et al. 2005). Lastly, there are population indices which assume a known proportion of the population observed during surveys ( Williams et al. 2002, Lancia et al. 2005 ). When attempting population counts it is very difficult to obtai n complete counts due to uncertain detection rates, large survey areas and resource availability, so a sample of the population is normally used. Population monitoring is essential for successful and effective manage ment of wildlife populations. Populati on monitoring programs serve two management roles (Nichols et al. 1995). To provide periodic assessments of the relative population size and demographic characteristics (sex ratios, age distribution ) at possible decision points in the management process and to

PAGE 12

12 provide knowledge of system responses to possible management alternatives. Some components of a sound monitoring program include; established objectives and goals, established protocols, and determining the scale of the program. Failure to address any of the crucial factors upon initiating a monitoring program can lead to problems with analysis and interpretation of the data. For example, t wo large -scale, longterm monitoring studies in the United States, the Mourning Dove Survey (Dolton 1996) and the Breeding Bird Survey (Peterjohn et al. 1996) have major fundamental problems (Pollock et al. 2002, Rosenstock et al. 2002). The primary source s of error are nonrandom placement of sampling points or sampling variance which can be controlled by rando mly or systematically selecting sample units and establishing uniform survey protocols (Pollock et al. 2002) and visibility bias, which is related to detection probabilities of the species, habitat characteristics and environmental variables ( Caughley 1974, Cook and Martin 1974, Samuel and Pollock 1981, Steinhorst and Samuel 1989) Detectability Thompson (1992) defined detectability as the probability tha t an object in a selected unit is observed whether seen, caught, heard, or detected by some other means. There are three types of detectability that are associated with wildlife surveys and population estimates : 1 C omplete detectability All individuals w ithin the population are completely detected over time, space, and other dimensions of the survey unit (Williams et al. 2002 ). If complete detectability is achieved over an entire population, the survey count is identical to the actual population size, an d error free comparisons can be made within the population. Complete detectability is rarely achieved in wildlife population surveys (MacKenzie et al 2004). 2 Less than complete but constant detectability T his is a biased estimate of the population (Williams et al. 200 2 ). However, t he bias is constant over time and space and, if the detection probability can be estimated, the count can be adjusted to provide an unbiased estimate of the population (Thompson and Seber 1994, Williams et al. 200 2 ). 3 Variable d etectability T he survey count is a biased estimate of the population and the bias is not constant over time and space (Williams et al 200 2 ). In some cases, detection probabilities can be influenced by treatments, management decisions, or environmental

PAGE 13

13 va riables and, if not estimated can lead to results that might not reflect the status of the actual population. Detection probabilities should be estimated for each survey, t o account for variable detectability S urveys of wildlife populations typically h ave detectability problems (Thompson and Seber 1994, Rosenstock et al. 2002). Imperfect detectability, if not adjusted will lead to underestimat es of the population (Thompson and Seber 1994, Ramsey and Harrison 2004). Population estimates are affected by many variables that lead to variability and bias in counts. Factors that influence detectability can be classified as environmental, species related, or human induced factors (Rosenstock et al. 2002) Environmental factors that affect detection rates are associated with habitat, topography, and weather conditions for a given survey area (Nupp and Swihart 1996, Rosenstock et al. 2002). Species related influences on detectability can be related to changes in activity levels, shifts in movement patterns, an d other natural behaviors. Potential h uman induced factors include wariness, survey techniques, and observer abilities (Rosenstock et al. 2002) One of the most important steps in increasing accuracy and precision in population estimations is to design su rveys to account for detection probabilities and the most critical variables that cause bias and variability (Thompson and Seber 1994). A lligator Night -Light Surveys Night spotlight counts (night -light counts) are commonly used as an index of relative size of crocodilian populations (Magnusson 1982, Wood et al. 1985, Bayliss 1987, Hutton and W oolhouse 1989, King et al. 1990) Indices assume that a constant or known proportion of the population is counted during each survey over time If counts change, the population is expected to change proportionally. However, this assumes that detection is constant or known across space and over time. Departure of detection probabilities from some mean can be categorized as either variability (fluctuation around the mean due to uncontrolled variation of environmental variables observer effectiveness, or survey methods) or changing bias (the systematic change in

PAGE 14

14 detection probabilities in one direction due to changes in habitat, increased wariness, shifts in behavior, or observer bias ). When conducting analysis of change in populations, variability affec ts uncertainty about estimates whereas changing bias can affect inferences about population changes. Past studies have shown that only a small proportion of alligators are observed during a night light survey (Murphy 1977, Brandt 1989, Woodward et al. 1996) As much as 91% of the total alligator population may go undetected during night light counts (Woodward et al. 1996). The proportion of alligators detected during night light counts is dependent on habitat, wariness, natural behavior shifts, and obse rver efficacy (Graham and Bell 1969, Murphy 1977, Brandt 1989, Woodward and Linda 1993). Detection probabilities of alligators may also vary with changes in environmental variables such as, water temperature, wave action, and vegetation cover (Murphy 1977, Woodward et al. 1978). Another source of variability is alligator availability, the movement of individuals in and out of the survey area. Some of the factors affecting availability are water level, hydrilla ( Hydrilla verticicllata ) coverage, and seasonal movements (Woodward et al. 1978, Woodward and Moore 1990). Higher water levels increase the accessible habitat for alligator s but these areas may be inaccessible to surveyors, which leads to reduced counts. FWC Alligator Surveys In 19 87, the Florida Fish and Wildlife Conservation Commission (FWC) established an alligator harvest program to manage alligator population s at preharvest levels; increase economic value of wild alligators; generate revenue for management, research, and conservation ; and provide alligator hunting opportunities to the public (Hines and Abercrombie 1987, Woodward et al. 1987, Wiley and Jennings 1990). Population monitoring is conducted to ensure harvests on alligator management units (AMU) are sustainable and populations are no t in decline

PAGE 15

15 (Woodward et al. 1987) Surveys are conducted on 120 AMUs every year to monitor populations. Current population models have established harvest quota levels of adult ( 183 cm.) alligators depending on population estimates and trends The har vest quota s are set at 15%, 12%, 6%, 3%, and 0%, depending on the change in population estimates from year 0 of hunt initiation Alligator night -light surveys are a critical component to the success of this harvest program and ensure populations are harves ted at sustainable levels. Survey routes are standardized and typically follow the perimeter of a lake along the open water -shoreline/marsh interface (Woodward and Marion 1978), or middle/centerline of a river section (depending on river width). Airboats and outboard -motored boats are used to conduct surveys, but an effort is made to use the same craft type for any given area to maintain consistency in survey methodologies (Woodward and Marion 1978). Spring surveys are conducted from May to midJune when adult alligator activity is the greatest (Woodward and Marion 1978) whereas summer surveys are conducted from July to mid-August. Surveys are conducted at a speed of approximately 20 25 km/hr unless the driver must slow to properly record dense concent rations of alligators, maintain safe operating speed, or for navigational constraints. Spotlights (200,000 c.p.) are used to locate alligator eye reflections by working the light in a 180 degree arc in front of the vessel. The size of an alligator is estim ated to the nearest 1 ft. (30 cm) if possible. In cases where the exact size cannot be estimated alligators are recorded in broader size categories 0 2 ft (0 60 cm), 2 4 ft (61121 cm), 4 6 ft (122182 cm), ft ( sized. Data from these surveys are used to analyze trends on an area byarea basis. Everglades Alligator Surveys A network of survey routes has been established througho ut the Everglades e cosystem to monitor alligator population trends during Everglades restoration efforts Survey routes include

PAGE 16

16 both marsh and canal habitats and were established based on accessibility, hydrological characteristics, and orientation around marsh and canal habitats Surveys are conducted along established survey routes, and alligators observed out to 50 -m from the route are recorded. At every alligator observed, a GPS location is recorded along with, estimated size, water depth muck depth, and habitat/ vegetation type. Other variables recorded for surveys are air temperature (C), water temperature (C), wind (mph), cloud cover (%), and visible moon. Size estimates are in 0.25 -m size classes, and alligators unable to be placed into a size cl ass are recorder in broader size categories hatchlings (<0.25 m), small (<1.25 m), medium (1.25 <1.75 m), large ( m) and unknown. This monitoring program was designed to assess the success of E verglades restoration. The American alligator has been chosen as an indicator species for the restoration of the Greater Everglades Ecosystem due to their sens itivity to hydrologic changes, ecological importance, and sensitivity to habitat and system productivity (Rice et al. 2005). Restoration efforts will be assessed through performance measures of alligator populations including alligator abundance and dist ribution, nest production, body condition, and alligator hole occupancy (Rice et al. 2005). Long-term monitoring of alligator populations is part of the Monitoring and Assessment Plan (MAP) of the Restoration Coordination and Verification (RECOVER) team formed to provide assessment of the success of the Comprehensive Everglades Restoration Plan (CERP).

PAGE 17

17 CHAPTER 2 EFFECT S OF HABITAT TYPE AND STRUCTURE ON DETECTI ON PROBABILITIES OF AMERICAN ALLIGATOR S Factors Affecting Alligator Night -light Surveys Prio r to 1978, night light surveys in Florida followed the guidelines recommended by Charbreck (1966). He recommended surveys be conducted on low moonlight nights with minimal wind during the months of April and May. The se guidelines were later updated by the Alligator Recovery Team (Chabreck 1976), which recommended beginning surveys at least one hour after sunset during May October with air temperature s abo ve 21 C. Woodward and Marion (1978) examined the effect of different environmental variables and survey procedures on night -light counts and found alligator counts were positively correlated with water temperatures but leveled off at higher temperatures (Woodward and Marion 1978). They also found wave height was negatively correlated with counts and var iability was related to increased wave heights (Woodward and Marion 1978). Contradictory results were found between Woodward and Marion (1978) and Chabreck (1966) in relation to moon light effects on counts. Precipitation, air temperature, and percent cl oud cover showed minimal effects on counts (Woodward and Marion 1978). In more recent years, wariness, emergence behavior, and habitat characteristics have been identified as factors that should be examined as potentially affecting counts. Wariness Changes in detection rates of crocodilians due to natural avoidance behavior from previous contact with boats and or surveyors, can affect survey results (Webb and Messel 1979). Increase d wariness was observed in the Black caiman ( Melanosuchus niger ) and the Yac are caiman ( Caiman yacare ) following human interactions (Pacheco 1996). Preliminary investigations by Spratt (1997), however, found little effect of harvest on alligator wariness in

PAGE 18

18 central Florida lakes. Wariness can influence night -light counts by increa sing the distance at which an alligator submerges when a boat is approaching This will reduce the probability that the alligator is counted and increase bias of counts Emergence Dynamics It is thought that alligators typically hunt for prey below th e water surface at night and surface to move around, breathe, and conduct courtship activities. The proportion of time they spend at the surface, and are therefore visible during night light surveys are thought to depend on feeding behavior, courtship act ivities, water temperature, and surface disturbance such as waves. Natural emergence behavior can be altered by human disturbance (Pacheco 1996). Ambient air temperature has been shown to influence submergence behavior in alligators (Murphy 1977, Brandt 1 989). Bugbee (2008) found that adult alligators in the Everglades spend about 2/3 of the day submerged. V ariables found to affect emergence behavior were hour of night, season, moon phase, water depth, water temperature, air temperature, rain, and wind ( Bugbee 2008). Understanding alligator emergence behavior will help to determine what percentage of the population is actually available for detection during night -light surveys. Habitat Characteristics Habitat type, composition, and structure can influence detection probabilities of alligators during night light surveys. Detectability of crocodilian species has been found to vary with habitat conditions, such as presence and density of vegetation, as well as complexity, shape, and size of the survey area ( Wood et al. 1985, Thorbjarnarson 1988). Crocodilian detection probabilities have been suggested to decrease in areas with vegetation cover (Bayliss 1987, Woodward et al. 1996, Da Silvera et al. 1997, Cherkiss et al. 2006). Although vegetation characteris tics are thought to affect detection probabilities for alligator surveys, few studies have quantified the effects.

PAGE 19

19 Objectives My objectives we re to determine the effect of habitat type and structure (as defined by vegetation height and density, and water depth) and survey procedure (as defined by survey speed observer height and distance from transect ) on detection rates of alligators The three habitats examined were sawgrass, slough, and wet prairie Habitat -Related Detection Probabilities I hypothesize d that habitat type influences detection rates of the American alligator s during night light surveys with slough having the highest detectability followed by wet prairie then sawgrass. I hypothesize d that with increasing vegetation height, distance from transect, visual obstruction, and survey speed will result in a decrease in detection probabilities. I also hypothesize d that increased water depths and seat height will result in a n increase in detection probabilities. The alternative hypothesis was that alligator detection is constant in all habitat types. Anticipated Results and Benefits Information gained from this study will increase the ability of alligator researchers to detect changes in FWC and Everglades alligator populations Determining detect ion rates for habitat variables critical to alligator population surveys is essential for increasing the reliability of population estimates. Accounting for sources of variation in detection rates will improve precision, which will allow population trends to be detected sooner than current population analysis methods thus lead to enhanced response time for research to observe population changes. As with all harvested species it is vital to have reliable population estimates in order to harvest in a susta inable manner. This project could be a model for assessing effects of habitat characteristics on night light surveys of crocodilian species worldwide.

PAGE 20

20 Materials /Methods Study Area I used three habitats to assess effects of vegetation characteristics on all igator detectability : sawgrass marsh, open slough, and wet prairie. The sawgrass and slough habitats were located within Water Conservation Area 2A in the Everglades Ecosystem, whereas the wet prairie transect w as located on Lake Kissimmee ( Fig. 1 1 ). Th e sawgrass marsh and open slough habitats were selected for habitat detectability models because they are the most abundant habitats within the Everglades Ecosystem and comprise the majority of Everglades alligator survey routes. The wet prairie habitat w as selected on Lake Kissimmee because there were larger contiguou s tracts of wet prairie habitat which is similar in structure to Everglades wet prairie (Kushlan 1990) Sawgrass Marsh Sawgrass marsh is the most abundant marsh habitat in the Everglades accounting for about 65 70% of the remaining wetlands habitat in that ecosystem (Loveless 1959, Kushlan 1990). Sawgrass marshes are considered a n emergent marsh dominated by sawgrass ( Cladium jamaicense ), which accounts for (93%) of the plant biomass (Jor dan et al. 1997). Vegetation s pecies diversity is considered very low but other species occurring in sawgrass marsh habitats include maidencane ( Panicum hemitomon), arrowhead (Sagittaria spp.), spikerush ( Eleocharis spp.) and various floating leave speci es [e.g. spatterdock ( Nuphar luteum ) and fragrant waterlily (Nymphaea ordata)] in open areas. Sawgrass marshes usually occur on slightly higher elevat ions or areas that have shallower water depth than open slough or wet prairie habitats. Vertical structur e of sawgrass marshes can reach heights as great as 3 m The mean vegetation height for a sawgrass marsh is around 140 cm (Jordan et al. 1997). Sawgrass marshes are usually categorized into either very dense stand s of tall plants or sparse stands of short plants ( Loveless

PAGE 21

21 1959, Kushlan 1990, Wood and Tanner 1990, Gunderson 1994). Habitat structure is greatly influenced by hydroperiod and related water depths. Longer hydroperiods with greater water depths tend to produce taller and denser stands of sawgras s (Kushlan 1990) Conversely shorter hydroperiods and shallower water depths tend to limit growth and favor less dense and shorter stands of sawgrass ( Kushlan 1990, Newman et al. 1996) Alligator detection rates in sawgrass habitats may be influenced by vertical structure and density of the stands and fluctuation s in hydroperiods Slough Slough habitats in the E verglades occur in areas that have the greatest water depths. Many sloughs in the Everglades occur in the main water passages such as Shark Ri ver slough to the west and Taylor slough to the east. Because sloughs are in the deepest water of the Everglades, they remain inundated for the longest periods and have a higher diversity of plants species than sawgrass Slough habitats are dominated (88 %) by floating macrophytes and submerged vegetation, and have a mean vegetation height of less than 30 cm (Loveless 1959, Gunderson 1994, Jordan et al. 1997). Some of the major species of slough habitats are fragrant water lily, spatterdock, and floating hearts ( Nymphoides aquaticaum ). Alligator d etection rates in slough habitats are expected to be greater than in other habitats as a result of less vertical structure and a greater proportion of open water. Wet Prairie Wet prairie habitats occur in the in termediate water depths between the deeper slough habitat and the shallower sawgrass habitats within the Everglades Wet prairies also occur on the fringe of lakes and in flatwoods of central Florida (Kushlan 1990). Wet prairies include a collection of gra minoid (grasslike) plants of low -stature ( Kushlan 1990, Gunderson 1994). Wet prairies have a mean vegetation height of about 70 cm (Jordan et al. 1997). Wet prairies have a

PAGE 22

22 shorter hydroperiod, but greater plant diversity then the slough or sawgrass habi tats (Kushlan 1990). Some of the major species that occur in wet prairies are beakrush ( Rhynchospora tracyi ), maidencane, spikerush and maidencane ( Loveless 1959, Kushlan 1990, Wood and Tanner 1990, Gunderson 1994). The vegetative structure of wet prairi e undergoes significant changes as water depths vary, b eakrush and spikerush both exhibited taller canopy heights with greater water depths (Busch et al. 2004). Wet prairies within the Everglades and Lake Kissimmee are simi lar in structure but have diff er ences in the dominate grass species (Kushlan 1990) Lake Kissimmee is dominated by maidencane and beakrush, whereas, the Everglades is dominated by beakrush and spikerush (Kushlan 1990). Changes in structure and density of vegetation in wet prairie habita t with varying in water depths may lead to lower alligator detection rates. Air Cannon An air cannon was built to distribute reflective markers throughout study areas without leaving an airboat trail to attract the attention of surveyors. The cannon was constructed of schedule 40 PVC pipe and a sprinkler valve was used for the firing mechanism. The cannon was equipped with a pressure gauge and firing angles ranging from 0 to 85 degrees ( Fig. 1 2 ). Air pressure was added using a portable air compressor that charged the cannon to the desired pressure. Prior to employing the cannon to distribute markers for the study, I fired it at different air pressures and angles to calibrate the horizontal distances of the reflective markers in respect to pressure and firing angle. Reflective Markers Reflective markers were constructed of a 3.2 cm diameter wooden dowel, 0.9 cm diameter steel rod, monofilament fishing line, heavy duty washers, and white reflective tape. The dowel was cut to 18 cm with a hole drilled in the bottom, and a 7.5 cm section of steel rod was attached to the bottom of the dowel with epoxy adhesive, and with monofilament line

PAGE 23

23 attached to the washer. The steel rod acted as a keel and kept the reflective marker floating upright while a washer anch ored it in place. The marker was painted black and the reflective tape was attached to the top of the dowel (Fig. 1 3) White reflective tape was used to help observers distinguish markers from actual alligators. Two observers used spotlights at 10 m int ervals out to 50 m and t here was no difference observed in detecting the white reflective tape versus an actual alligator eye shine at the various distances Mock Surveys Survey t ransects were established through areas most characteristic of the habitat type. Random points (60) were generated to distribute reflective markers along transects Two sets of random numbers were used to distribute reflective markers: 1) To determine distance (0 200 m.) between markers along the transect, and 2) to determine di stance (0 50 m.) from transect and whether the marker should be deployed on the right or left side of the transect Transect lengths ranged from 4.18 km in the wet prairie habitat to 6.38 km in the sawgrass marsh habitat. Transect lengths varied in lengt h depending on the random distances between marker locations Markers were launched perpendicular to transects by air cannon, and GPS waypoints were taken as well as direction (left or right) and distance from the transect for each reflective marker. Refl ective markers less than 5 m from the transect were hand -thrown from the front of the airboat. Observers were considered skilled if they h a d conducted alligator night -light surveys in the past year or according to established protocols The protocols con sisted of a training period of participating on surveys as a recorded and exhibiting proficiency at size estimations. Once reflective markers were randomly distributed throughout each habitat surveys were conducted by observers according to methods established by Woodward and Marion (1978), except observers didnt slow down to estimate sizes. Survey replicates were conducted 5 8 times on

PAGE 24

24 each habitat. Only five surveys from the most experience observers were used for modeling effects of habitat type and survey procedures on alligator detectability due to unequal replicates between habitats A t each marker detected, observers recorded a GPS waypoint perpendicular to the reflective marker on the transect, indicated a direction (left or right), and estima ted the distance from the transect to the reflective marker. When surveys were completed, all markers were collected and markers not recovered or not detectable were removed from the survey sample. E xamples of reflectors removed from the study included t hose cases w h ere the marker was laying on its side or did not re -surface after being launched Both instances above would have reduced the probability of observing markers and would have bias ed survey results. Retained survey samples for the sawgrass, slo ugh, and wet prairie habitats were 49, 49, and 54. Habitat Variables After all surveys were completed v ariables such as vegetation height (VE) visual obstruction (VO) water depth (W) and distance (D) from transect were recorded at each marker. Vegetat ion height (cm) at the marker was measured from the surface of the water to the top of the vegetation. Visual obstruction, an indication of vertical vegetation density, was measured by 1 dm classes using the Robel Pole method (Robel 1970). Water depth (cm ) was measured from the top of the substrate layer to the surface of the water. Airboat seat height (cm) and survey speed (km \ hr) were also recorded for each observer /boat combination Waypoints were loaded into Garmin Map Source (Garmin International In corporation, Olathe, Kansas ) and were examined by estimated distance and direction (right or left) to see what proportion of markers were observed by surveyors. Known locations were compared with observed markers to develop a mark recapture data set for mo deling detection probabilities.

PAGE 25

25 Model ing Count data were analyzed using Program MARK (White and Burnham 1999) to determine the most suitable model for determining habitat -specific detection probabilities of alligators during night -light counts. The mode l used for detectability models was the Huggins Closed Capture Model (Huggins 1989). This model allowed estimation of a closed population size ( N ) from initial capture probabilities (p ) and recapture probabilities ( c ) (Huggins 1989, Huggins 1991). One condition of the model was animals are captured or recaptured at least once during the study and it allowed for individual covariates to be used to model p and c The initial capture probability ( p ) for use in modeling detection probabilities was set to one This represented the initial time markers were deployed and all markers are known to be available for detection within the study area As discussed earlier, markers unavailable for detection were removed from the data set Recapture probabilities ( c ) represent ed detection probabilities of reflective markers. Models were created based on biological factors and other significant variables that influence night light counts. I selected distance from transect, visual obstruction, vegetation height, water depth, survey speed, and airboat seat height as covariates to be included in model. Other considerations for model development were whether to group habitats together or keep them separate and examine possible interactions among variables. Separating ha bitats would allow me to determine if the covariates affected habitats equally or independently If the effect of covariates was equal among the habitats, there would be no difference that could not be explained by the individual covariates. Interaction ef fects were examined because some covariates could be dependent on others For example, an increase in water depth may lead to a decrease in vegetation height. A one -way non -parametric analysis was conducted to determine differences in habitat detectability rates. T hirty -one models were examined based on biological

PAGE 26

26 variables or combination s of variables thought to affect alligator detection rates in the three habitats (Table 1 1). Model Selection The Akaike Information Criterion (AICc) was used to determi ne which variables or mixture of variables resulted in the best fit for modeling habitat effects on alligator detectability during night light counts (Burnham and Anderson 2002, Pollock et al. 2002). Anderson (2008) suggested using AICc over AIC, because AICc is adjusted for small sample size AICc determines the most parsimonious of models by taking into account the number of parameters in the model (Anderson 2008) AICc penalizes additional parameters and therefore identifies the best fitting model with a minimal numbers of parameters. AICc weight represents the probability that the model is the best fit given the tested data and models (Anderson 2008) LOGIT link functions were used to estimate optimum models beta parameters. Beta parameters were d etermined to be significant if the upper and lower 95% confidence intervals did not encompass zero. Detectability functions were derived from beta parameters resulting from the best fitting model, and were used to graphically represent the effects of the habitat and survey variables on alligator detectability in the three habitats. Results Covariates that had the most influence on habitat detectability were visual obstruction (VO), vegetation height (VE), distance from transect (D), water depth (W) and air boat seat height (Seat) (Table 1 2 ). Survey speed in the range we tested had minimal effect on alligator detectability There was no evidence of interaction effects between model covariates. Differences were found in the detection probabilities between the three habitat types and I found it more appropriate to model them separately as opposed to combining habitat types for a uni versal detectability function. The best model ( Model 3; AICc Weight = 0. 54), include d

PAGE 27

27 distance from transect, visual obstruction, vegetation height, water depth, and airboat seat height (Table 1 2). Models 4 and 5 showed some evidence for support with AICc weights of 0.13 and 0.10. I considered m odel averaging but due to the clear separation in AICc weights between the top compe ting models (0.54 to 0.13) and the fact beta parameters for models ranked 2 4 were not significant, I decided against model averaging. The optimum model estimated detection probabilities for available alligators (alligators that are on the surface availab le for detection) in sawgrass marsh, slough, and wet prairie habitats as 0.5 3 0.6 3 and 0.5 1 The only significant difference in detection rates was found between the wet prairie and slough habitat s ( P = 0.0274) All b eta parameters for habitat variables for the optimum model were significant (Table 1 3). Alligator Detectability Functions for each habitat type : P Sawgrass = 1/(1+e^ ( ( 1.0518 ) (0.0400 D) (0.559 9 VO) (0.013 6 VE)+(0.0131 Seat) ) P Slough = 1/(1+e^ ( ( 3.0797 ) (0.0637 D) ( 0.559 9 VO)+(0.0577 W)+ (0.0131 Seat) ) P Wet Prairie = 1/(1+e^ ( ( 0.2896 )+(0.0372 D) (0.598 9 VO ) (0.0144 W)+(0.0131 Seat) ) D etection probabilities from the above functions can be obtained by simply inserting various covariate values (D = m, VO = dm, VE = cm, W = cm, and Seat = cm) for the habitat of interest. 0.5599) had the greatest negative effect on alligator detectability, and affected all habitats equally ( Fig. 1 4 ). Although visual obstruction exhibited the same effect on all habitats, detecta bility curves behaved differently due to other variables in the habitat specific models 0.0400) and .0372) (Fig. 1 5 ). Vegetation height was important only for the sawgrass habitat and water depth was important only for slough and wet prairie habitats. Vegetation height had a negative effect on 0.0136) (Fig. 1 6) and w ater depth had a positive effect in slough

PAGE 28

28 in 0.0144) ( Fig. 1 7 ). Airboat Fig. 1 8 ). Cumulative graphs for the sawgrass ( Fig. 1 9 ), slough ( Fig. 1 10), and wet prairie (Fig. 1 1 1 ) show the relationship of all models variables with the other variables set to minimum values. Discussion I found differences in alligator detection rates among the three habita t types (sawgrass, slough, wet prairie) tested Detection probabilities varied from p = 0.5 1 in the wet prairie habitat to p = 0.6 3 in the slough habitat. The sawgrass habitat had a slightly higher detection probability than the wet prairie with a p = 0. 5 3 The only difference ( P = 0.0274) was between wet prairie and slough habitat. This could possibly be due to a location effect and suggest more replicates in different geographic regions should be conducted within habitat types Although the other habit at combinations did not show significant difference s in alligator detection rates, the final model variables were discernibly different among habitats and therefore they were modeled independently. The slough habitat was expected to have the highest det ection rate because they generally ha d greater water depths, l ower vegetation height, and the greatest amount of open water (Loveless 1959, Kushlan 1990, Jordan et al. 1997 ). F actors affecting detection probabilities in the slough habitat in order of grea test positive to negative effect were water depth (+), seat height (+), distance from transect ( -), and visual obstruction ( ). Generally, s awgrass marshes consist ed of dense stands of sawgrass with a substantial amount of vertical structure (Loveless 1959, Kushlan 1990), which was also true for the study area Sawgrass marshes ha d the lowest water depths, highest vegetation height, and the greatest plant biomass of habitats tested ( Kushlan 1990, Jordan et al. 1997). F actors affecting detection probabil ities on sawgrass habitat from greatest positive to negative effect were ; seat height (+),

PAGE 29

29 vegetation height ( ), distance from transect ( -), and visual obstruction ( ). Sawgrass habitat had only one positive relationship, which was seat height. Wet pra irie habitats include intermediate characteristics between the slough and sawgrass habitat in relation to water depth, vegetation height and plant biomass (Jordan et al. 1997). Factors affecting detection probabilities on the wet prairie habitat in order from greatest positive to negative effect were distance from transect (+), seat height (+), water depth ( ), and visual obstruction ( -). Based on habitat characteristics vegetation height and visual obstruction, I expect ed the sawgrass marsh to have the lowest alligator detectability, but this was not the case. The lowest detectability was observed in the wet prairie habitat, which could be caused by other confounding factors that will be discussed later. When examining composite graphs ( Fig. 1 9 1 1 1 ), the relationships between positive and negative variable effects of models were more evident. For example, in sawgrass habitat, when all covariates were set to minimum values, the detectability function yielded a detection probability of p = 0.93. The probability is very high at the minimum values because most of the sawgrass covariates had negative effects except for seat height, which had a small positive effect. Conversely when all covariates of slough habitat w e re set to minimum values the det ection probability was only p = 0.42. This probability starts out fairly low because water depth and seat height in slough habitat had positive effects on detectability. For the detectability functions to be useful to alligator researcher s and behave cor rectly, all critical variables in habitat models must be measured and incorporated into the model Although visual obstruction and seat height had the same effect on all three habitats, distance from transect, vegetation height, and water depth differed g reatly in their influence on alligator detectability.

PAGE 30

30 Visual Obstruction Visual Obstruction had the greatest influence on detection rates and affected all habitats equally. This was expected because, as vertical vegetation density increases, detectabili ty decreases regardless of other habitat characteristics If vegetation is obstructing an observer s view and an o bserver cannot see the surface of the water he is unlikely to see an alligator eye reflection Detection probability decline d quickly as vis ual obstruction increased ( Fig. 1 4) Distance from Transect I expected distance from transect to be negatively correlated with detectability for all habitats, but I found mixed results Distance had a negative effect on alligator detectability in sawgr ass and slough habitats but a positive effect on detectability in the wet prairie ( Fig. 1 5) I n the slough habitat distance had a greater negative effect on detectability than in the sawgrass habitat. However, the positive effect of distance on detecta bility in the wet prairie habitat goes against the basic principle of distance sampling. One of the major assumptions of distance sampling is that all animals on the line should be completely detected ( p = 1) (Buckland et al. 1993). However, I believe th at vegetation type can explain this counterintuitive finding. In wet prairies, t he primary vegetation consisted of maidencane and various other grass species, which was relatively dense. T he light beam reflecting off wet grass may sometime obscure alliga tor eye shines in denser grass I frequently observed alligator eye refelections to be more difficult to detect at close range in the dew covered vegetation on wet prairies Frequently, alligators are observed from far distances in wet prairies but disappe ar when trying to approach the alligator. Vegetation Height Vegetation height affected detectability only in sawgrass habitat ( Fig. 1 6) This is probably because the sawgrass habitat had a greater vertical vegetation structure when compared to slough and wet prairie habitats Sawgrass stands can have mean canopy heights of 1.4 m

PAGE 31

31 (Jordan et al. 1997) and maximum heights of 3 m (Kushlan 1990) whereas vegetation in other habitats is generally shorter An interaction between vegetation height and water depth was expected but did not emerge as a significant effect in model sets. Water Depth Water depth affected alligator detectability in slough and wet prairie habitat s but did not have a significant e ffect in sawgrass habitat (Fig. 1 7) Detectability w as positively associated with water depth within the slough habitat and negatively associated with water depth in the wet prairie habitat. The positive effect of water depth on detectability in slough habitat was expected. With higher water levels, slough habitat would have more open areas with less and shorter emergent vegetation, which should increase the chances of detecting an eye reflection On the other hand, the negative effect of water depth on detectability in the wet prairie habitat is a little harder to understand, and was unexpected. One possible alternative could be that vegetation structure in wet prairies changes dramatically with differing water depths, thus leading to decreased detectability. Busch et al. (2004), found two wet prairie species, spikerush and beakrush, had significant differences in vegetation height between flooded and drained conditions. Spikerush height was greater in flooded treatments (83 cm) compared to (59 and 61 cm) in drained treatments (Busch et al. 2004) Bea krush also had greater height in flooded treatment (67 cm) compared to the drained treatment (59 cm) (Busch et al. 2004) I collected v egetation data along survey transects but data derived from th i s was insufficient to assess changes in vegetation structu re with varying water depths. Another alternative hypothesis is that greater water depths could have confounded the problem of spot light glare on the taller grass species.

PAGE 32

32 Seat Height Airboat seat height was positively correlated with detection probabilities on all habitat types (Fig. 1 8) This seems intuitive, because the higher the observer is, the less obstructions encountered along the vision line and the greater the chances of detecting an eye reflection Higher seat height give s observers a be tter line of sight angle to detect alligators behind taller vegetation. This would suggest that i n order to increase detection rates on alligator surveys, alligator researcher s should use an airboat with the highest observer seat as safely possible. At a minimum, standardization of seat height among survey crafts would reduce variability due to that variable. Management Implications Habitat detectability functions can be applied to future alligator nightlight surveys in the Everglades to obtain an improv ed estimate of the alligator population. For the three habitats the various covariates could be measured and used in the detectability functions to obtain a detection probability of available (on the surface) alligators for a given survey. Distance from transect, c ould be set to a constant 25 m, since Everglades alligator surveys only count alligators detected < 50 m from either side of the survey transect. One assumption associated with setting the distance from transect to a constant 25 m would be tha t alligators are equally distributed throughout the 50 m wide swath on both sides of the survey route. By setting the distance from transect to 25 m the model will be estimating the mean detection probability. Random points should be sample d out to 50 m along the survey route to obtain mean vegetation height, water depth, and visual obstruction measurements. Due to variation that can occur within habitat variables over time such as, dry season vs. wet season, vegetation composition shifts, and other proc esses that could alter the habitat characteristics, it is helpful to obtain habitat measurements for each survey. If variability is small, mean water depth for a given survey route may be able to

PAGE 33

33 be correlated with an established water gauge within the Ev erglades Depth Estimation Network (EDEN) in close proximity. One problem associated with using existing water gauges is that in extreme dry seasons, the marsh water levels may lose connectivity with the gauge and would be considered an unreliable measurem ent for the survey route. It is strongly suggested that since vegetation height and visual obstruction measurements should be taken for each survey, water depth measurements should also be recorded. I recommend obtain ing habitat measurements for each sur vey for the first several years After several years of collecting habitat data variability of the habitat measurements could be analyzed, and if variability is small frequency of collection of habitat data could be reduced. If variability is great wit hin the habitat characteristics, I recommend calculating detection probabilities for each random sampling point, and then taking the average of the individual detection probabilities to obtain the detection probability for the survey route. On the other h and, if variability is small among the random sampling points, the mean of habitat characteristics could be used to obtain the detection probability for the survey route. Although, habitat detectability models are useful for determining detection probabi lities of alligators on the surface, they do not take into account the proportion of alligators that are submerged or unavailable for detection. Bugbee (2008) conducted a study looking into the effects of various environmental variables and their effect on emergence behavior of adult alligators. The product from Bugbee (2008) was an adult alligator emergence model, which incorporated various environmental variables to estimate the proportion of alligators emerged, or the proportion of alligators available for detection. V ariables included in the adult alligator emergence model were: hour of night, season, moon phase, water depth, water temperature, air temperature, rain, a nd wind, which have all been documented to affect crocodilian night light

PAGE 34

34 counts ( Murp hy 1977, Woodward and Marion 1978, Montague 1983, Bayliss 1987, Mazzotti 1989, Pacheco 1996). To further improve alligator population estimates, the combination of habitat detectability models and the adult alligator emergence model (Bugbee 2008), could be used to adjust night light counts to gain more reliable alligator population estimates in the Everglades (Appendix A ) Although, habitat detectability models and the adult alligator emergence model account for the majority of variability in alligator det ection rates, there are still some gaps in the information. Applying detectability functions to alligator surveys on alligator management units (AMU) administered by the FWC would be a n expensive and time consuming task. AMUs do not have rigid survey tr ansects but rather have survey routes that cover the majority of occupied habitat within an AMU. C hanges in water levels and vegetation types and densities would require annual assessment s of habitat characteristics The FWC alligator surveys also inclu de many more habitats than were examined in the study and FWC alligator surveys do not record locations for each alligator observed A nother reason it would be difficult would be the amount of survey areas ( approximately 120) that are surveyed yearly. In order to apply habitat detectability functions to FWC alligator survey s more research would need to be conducted and replicated for more habitats and areas across the state. If detectability functions are obtained for the multiple habitats statewide, a co mbination of the use of Geographic Information Systems (GIS) and changes to night light survey recording procedures (obtaining locations for alligators encountered) could result in feasible application of the habitat detectability models statewide. Previ ous estimate have indicated detectability of alligators 122 cm in North Central Florida lakes during night -light counts to average 0.09 on Lake Woodruff and 0.19 on Orange Lake (Woodward et al. 1996). Other estimate s from Par Pond, South Carolina indicat ed

PAGE 35

35 detectability range of 0.3 0 0.35 for all sizes (Murphy 1977, Brandt 1989) and detectability of 0.09 for 183 cm alligators in impounded wetlands in South Carolina (Rhodes and Wilkinson 1994). With the combination of the Bugbee (2008) emergence model and the habitat detectability functions detectability ranged from 0.15 to 0.21 for adult alligators (Appendix A ) but this estimate does not account for wariness This gives some support for the current universal correction factor of 0.14 being used by the FWC when adjusting night light counts

PAGE 36

36 CHAPTER 3 CONCLUSIONS Improving reliability and reducing uncertainty in population estimates can enhance management capabilities for wildlife populations. In recent years, researchers have improved the capability a nd implemented techniques to account for detectability during wildlife population surveys. Incorporating detectability into population estimates will enable researcher s to obtain a better understanding of trends or changes occurring within populations. There are many methods that can be used to account for varying detectability rates such as double sampling, removal methods, capture recapture, and distance sampling (Pollock et al. 2002). However, many of these approaches are very expensive when compared t o using traditional indices (Pollock et al. 2002). Simulated surveys are another method to assess detectability under varying conditions and treatments. Some studies on other species have constructed replicas or plaster models of animals then conducted s urveys to obtain detection rates (Gardner et al. 1999, Cherkiss et al. 2006, Pearse et al. 2007). The use of simulated surveys has proven to be a valid technique to account for and quantify the effects of different habitat variables on alligator detection probabilities during night light counts. Although the habitat detectability models in this study accounted for the majority of habitat variables, additional variables might improve and enhance modeling abilities for different habitats. Vegetation composi tion, horizontal structure, and species of vegetation could be included in future model sets to determine effects and enhance estimations of detectability. One issue that was not addressed in this study is the spatial relationship between reflective mar kers. In areas with higher densities of markers, observers may tend to fixate on an observed marker and reduce or eliminate scanning for other markers, which could lead to missed

PAGE 37

37 markers that should otherwise be detected. This also occurs in actual alligator surveys, particularly in areas where alligator density is high. The problem can also exist in areas with difficult terrain to navigate during surveys, where observers are concentrating on avoiding obstacles rather than scanning for eye reflections Th is problem could be minimal for some observers but could be higher for other observers, depending on survey experience, survey techniques, and driving ability. At the very least, this adds variability to detection rates with associated uncertainty. This p roblem supports the need to establish strict protocols for how surveys are conducted and ensure all observers are conducting surveys by the same methodology. There should be standard methods on how to handle situations that occur during surveys, such as e ncountering high alligator densities, new stands of vegetation cover that reduce visibility, and other cases. Addressing these problems can result in a reduction in observer associated variability in survey counts. Another problem not addressed in the s tudy was replication of habitat types statewide and under varying hydrologic conditions. Surveys were replicated by observers but habitats were not replicated. Without replication, inferences should not be made for habitats statewide but rather made by ar ea. Future work could be conducted on habitats in different areas to test detection models and determine if these findings are consistent for all areas within these habitat types. It would be instructive to examine other common alligator habitats should throughout the state, such as shrub swamp, spatterdock, and other types of emergent marshes. During the study period, hydrologic conditions were relatively stable, and the models may not account for influence that could appear during varying hydrologic condi tions. Future work should examine and analyze model predicted detection probabilities under varying hydrologic conditions, to

PAGE 38

38 determine the effect of hydrology and assess whether the current model variables account for fluctuations. Determining the eff ect of human induced wariness on alligator night light counts needs to be explored further. Preliminary data examining effects of hunting on alligator wariness, determined there was no significant difference between harvested and unharvested areas; howeve r results were confounded by water levels and temperatures (Spratt 1997). In more recent years, more data have been added and another analysis conducted. The recent analysis of harvested and unharvested areas has shown a significant difference for alliga tors >1.8 m, but only at or below median water levels. The average depressive effect of wariness on nightlight counts was estimate to be 42% ( 60% to 16% CI95%) at median water level, and 76% ( 88% to 50% CI95%) at 0.6 m below the median ( R. A. Kiltie Florida Fish and Wildlife Conservation Commission, unpublished report ). Future research is needed to examine the effect of changing wariness on detection rates and a practical way of applying wariness detectability coefficient to alligator surveys. Alligator night light surveys will continue to be the most viable and efficient method for monitoring alligator populations, and a better understanding of factors that influence detectability during night light counts will lead to better estimates of the p opulation. Reducing variation in population estimates will allow researchers to detect changes in alligator populations over a shorter period of time and provide more reliable information for setting harvest quotas. Improved population estimates also wil l allow researchers to better determine the effects of Everglades restoration activities on alligator populations.

PAGE 39

39 Table 2 1. Model set used to model American alligator detectability as a function of distance (D), visual obstruction (VO), vegetation he ight (VE), water depth (W), seat height (Seat), and survey speed (Speed). Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model 8 D VO VE W Speed Seat Habitat D VO VE W Seat D VO VE (SG) W (SL,WP) Seat Habitat D VO VE W Seat D VO VE W Seat Habi tat D VO VE W Seat Habitat D VO VE W(SL,WP) Seat Habitat D VO VE(SL,WP) W (SG) Seat Habitat Model 9 Model 10 Model 11 Model 12 Model 13 Model 14 Model 15 Model 16 D VO VE W Seat D VO W Seat Habitat D VO VE Seat Habitat D VO VE W Seat Habitat D VE W Hab itat D VO VE W(SG,WP) Seat Habitat D VO VE Speed Seat Habitat D VO VE Seat Model 17 Model 18 Model 19 Model 20 Model 21 Model 22 Model 23 Model 24 D VO VE Speed Habitat D VO VE W Seat Habitat D VO VE *W D VO VE Seat Habitat D VO VE W Habitat D VO VE Habitat D VO VE Individual VO VE W Habitat Model 25 Model 26 Model 27 Model 28 Model 29 Model 30 Model 31 D VO Habitat D VO VE W Speed Seat Individual D VO VE W Individual D VO VO*W Individual D VO W VO*W Individual D VE VO*VE Habitat D VE W Habitat Note: SG, SL, and WP next to a variable indicates that it only appli es to those specific habitats. Variables in BOLD indicates that the variable applies to the all habitats

PAGE 40

40 Table 2 2 to describe detectabili ty of alligators in sawgrass, slough and wet prairie habitat s Model AICc Model Likelihood # Par Deviance 3 622.61 0 .00 0.54 1.00 11 600.32 4 625.39 2.78 0.13 0.25 12 601.04 5 626.02 3.41 0.10 0.18 14 597.55 6 627.17 4.56 0.06 0.10 16 594 .56 7 627.37 4.76 0.05 0.09 11 605.08 9 627.76 5.15 0.04 0.08 14 599.29 2 628.26 5.65 0.03 0.06 14 599.79 1 628.51 5.9 0 0.03 0.05 17 593.82 12 629.70 7.09 0.02 0.03 16 597.09 10 632.36 9.75 0.00 0.01 11 610.06 14 634.29 11.68 0.00 0.00 13 607.89 11 635.54 12.93 0.00 0.00 11 613.25 8 645.25 22.64 0.00 0.00 11 622.96 20 654.33 31.72 0.00 0.00 7 640.21 15 654.91 32.3 0 0.00 0.00 8 638.75 18 656.37 33.76 0.00 0.00 8 640.21 16 660.66 38.05 0.00 0.00 5 650.59 22 661.44 38.83 0.00 0.00 6 649.35 17 66 1.73 39.12 0.00 0.00 7 647.61 21 663.47 40.86 0.00 0.00 7 649.34 24 666.55 43.94 0.00 0.00 6 654.46 25 666.76 44.15 0.00 0.00 5 656.69 19 668.23 45.62 0.00 0.00 4 660.19 23 668.36 45.75 0.00 0.00 15 637.82 27 670.43 47.82 0.00 0.00 16 637.82 26 671. 04 48.43 0.00 0.00 18 634.28 28 674.81 52.2 0 0.00 0.00 15 644.28 29 676.77 54.15 0.00 0.00 16 644.16 30 797.18 174.58 0.00 0.00 6 785.10 13 844.76 222.15 0.00 0.00 6 832.66 Note: the probability that the models is the most parsimonious out of the set.

PAGE 41

41 Table 2 3 Beta parameters for optimum alligator detectability models for sawgrass (SG), slough (SL), and wet prairie (WP) habitat Beta is represented by LOGIT link function parameters. 95% Confidence Interval Parameter Beta Standard Error Lower Upper Sawgrass (SG) 1.0518 0.9365 0.7837 2.8873 Slough (SL) 3.0796 1.5991 6.2139 0.0545 Wet Prairie (WP) 0. 2896 0.8182 1.8932 1.3141 SG Distance (D) 0.0400 0.0152 0.0698 0.0103 SL Distance (D) 0.0637 0.0182 0.0993 0.0281 WP Distance (D ) 0.0372 0.0141 0.0096 0.0648 AH Visual Ob. (VO) 0.5599 0.0470 0.6519 0.4679 SG Vegetation (VE) 0.0136 0.0042 0.0218 0.0054 SL Water Depth (W) 0.0577 0.0200 0.0185 0.0970 WP Water Depth (W) 0.0144 0.0069 0.0280 0.0009 AH Seat Height (Seat) 0.0131 0.0043 0.0047 0.0216 Note: AH represents the variable applies to all three habitat types equally

PAGE 42

42 Figure 2 1. Locations of sawgrass, sl ough, and wet prairie habitats used for alligator detection surveys in Florida.

PAGE 43

43 Figure 2 2 The air cannon which was constructed to distribute reflective markers in selected habitats in Florida wetlands The pressure gauge and firing angles were used to obtain the desired distance from the transect.

PAGE 44

44 Figure 2 3. Reflective markers used to simulate alligator eye reflections during night -light counts.

PAGE 45

45 Figure 2 4 Relationship of v isual o bstruction (VO, dm) on a lligator detection probabil ities in s awgrass, s lough, and w et p rairie habitats during n ight light c ounts. All other variables set at minimum values; d istance (D = 0 m), v egetation h eight (VE = 0 cm), w ater d epth (W = 20 cm) and s eat h eight (Seat = 120 cm)

PAGE 46

46 Figure 2 5 Relationsh ip of d istance from t ransect (D, m) on a lligator d etection p robabilities in s awgrass, s lough, and w et p rairie h abitats during n ight light c ounts. All other variables set at minimum values; v isual o bstruction (VO = 0 dm), v egetation h eight (VE = 0 cm), w at er d epth (W = 20 cm) and s eat h eight (Seat = 120 cm)

PAGE 47

47 Figure 2 6 Relationship of v egetation h eight (VE, cm) on a lligator d etection p robabilities in s awgrass h abitats during n ight light counts. All other variables set at minimum values; v isual o bstruc tion (VO = 0 dm), d istance (D = 0 m) and s eat h eight (Seat = 120 cm)

PAGE 48

48 Figure 2 7. Relationship of water depth (W, cm) on alligator detection probabilities in slough, and wet prairie habitats during night light counts. All other variables set at minimu m values; v isual o bstruction (VO = 0 dm), d istance (D = 0 m) and s eat h eight (Seat = 120 cm)

PAGE 49

49 Figure 2 8 Relationship of s eat h eight (Seat, cm) on a lligator d etection p robabilities in s awgrass, s lough, and w et p rairie h abitats during n ight light c o unts. All other variables set at minimum values; v isual o bstruction (V O = 0 dm), d istance (D = 0 m), v egetation h eight (VE = 0 cm), and w ater d epth (W = 20 cm)

PAGE 50

50 Figure 2 9 Factors a ffecting a lligator d etection p robabilities during n ight light c ount s in a s awgrass h abitat. Mean a lligator d etection p robability for a s awgrass h abitat was estimated to be 0.5224. Detectability variables are from minimum to maximum values with others set at minimum values ; visual obstruction (0 10 dm), distance (0 50 m), vegetation height (0 300 cm), and seat height (120 220 cm).

PAGE 51

51 Figure 2 10. Factors a ffecting a lligator d etection p robabilities during n ight -light c ounts in a s lough h abitat. Mean a lligator d etection p robability for a s lough h abitat was es timated to be 0.6261. Detectability variables are from minimum to maximum values with others set at minimum values ; v isual obstruction (0 10 dm), distance (0 50 m), water depth (20 120 cm), and seat height (120 220 cm).

PAGE 52

52 Figure 2 1 1 Factors affecting alligator detection detection probabilities during night -light counts in a wet prairie habitat Mean alligator detection probability for a wet prairie habitat was estimated to be 0.5088. Detectability variables are from minimum to maximum val ues with others set at minimum values ; visual obstruction (0 10 dm), distance (0 50 m), water depth (20 120 cm), and seat height (120 220 cm).

PAGE 53

53 APPENDIX APPLICATION OF ALLIG ATOR DETECTABILITY M ODELS Alligator detectability The probability of obse rving an alligator in a defined population during a night light survey can be expressed as the product of alligator availability and the probability of detecting an available alligator. P t otal = P available P detected Pother For this application Pava ilable is represented by whether an alligator is emerged (on the surface) or submerged (below the surface), and Pdetected is represented by the probability an available alligator will be detected in a given habitat (sawgrass, slough, or wet prairie) Poth er represent s factors that may influence detection rate s such as environmental variables or observer ability, but have not been quantified, and are not included in the model. P t otal = P available P detected Pother = P e mergence P h abitat Pother Alligator e mergence m odel Bugbee (2008) developed an emergence model to determine the availability of alligators during night light surveys. The equation for the alligator emergence model is represented below : P e mergence = 1 / (1+e^ (( 0.1241) (0.007 hr1) (0.187 hr2) (0.067 hr3) + (0.105 hr5) + (0.010 hr6) + (0.095 hr7) (0.037 hr8) + (0.025 hr9) (0.744 AU) + (0.419 SP) (0.190 MQ) + (0.060 MH) (0.0085 WD) (0.012 WT) + (0.016 AT) (0.003 R) (0.026 WI)) w he re hour after sunset equals hr1 to hr9, season equals autumn (AU) and spring (SP), moon phase is quarter moon (MQ) and half moon (MH), water depth (WD) is surface of water to top of substrate water temperature (WT) at the surface air temperature (AT), rain fall (R) during

PAGE 54

54 survey and wind speed (WI) during survey. Hr4, summer and Moon (full) were considered reference values were assigned values of 0, and were not included in the equation. To apply the alligator emergence model to a given survey, one sh ould input the values in the above equation. Inputs of 1 or 0 for hour after sunset, season, and moon phase indicate the variable applies (1) or does not apply (0) to a survey. For example if a survey was conducted in the first and second hour after sunset hr1 and hr2 would get a 1 while hr3 hr9 would get a 0. Habitat d etectability m odels Habitat detectability models were developed for a sawgrass, slough, and wet prairie habitat. The equation s for the habitat detectability model s are represented below : P s awgrass = 1 / (1+e^ ((1.0518) (0.0400 D) (0.5599 VO) (0.0136 VE) + (0.0131 Seat)) P s lough = 1 / (1+e^ (( 3.0797) (0.0637 D) (0.5599 VO) + (0.0577 W) + (0.0131 Seat)) P w et prairie = 1 / (1+e^ (( 0.2896) + (0.0372 D) (0.5989 VO) (0.0144 W) + (0.0131 Seat)) where distance from transect (D, visual obstruction (VO) or vegetation vertical density vegetati on height (VE) water depth (W) is surface to top of the substrate and airboat seat height (Seat). A practical examp le Take for example a n alligator night light survey that was conducted in spring under a quarter moon and took three hours to complete. Hypothetical environmental measurements associated with the survey were; 70 cm water depth, 29 C water temperature, 30 C air temperature, 0 cm/hr rain, and a wind speed of 2 km/hr. According to the alligator emergence model the estimated probability of an alligator emerged would be 0. 34. The survey conditions for the sawgrass, slough, and wet prairie habitat can be assu med the same for these examples

PAGE 55

55 P e mergence = 1 / (1+e^ (( 0.1241) (0.007 1 ) (0.187 1 ) (0.067 1 ) + (0.105 0 ) + (0.010 0 ) + (0.095 0 ) (0.037 0 ) + (0.025 0 ) (0.744 0 ) + (0.419 1 ) (0.190 1 ) + (0.060 0 ) (0.0085 50) (0 .012 29) + (0.016 30) (0.003 0 ) (0.026 2 )) = 0. 34 Sawgrass habitat Hypothetical m ean habitat variables associated with the sawgrass habitat were; 22 m distance from transect, 2 dm visual obstruction, 75 cm vegetation height, and an airboat seat height of 160 cm. Therefore the probability of detecting an available alligator in the sawgrass habitat was P s awgrass = 1 / (1 + e^ ((1.0518) (0.0400 22) (0.5599 2 ) (0.0136 75) + (0.0131 160 )) = 0.51 The combined estimated detection probab ility for the alligator emergence model and th e sawgrass habitat detectability model were P t otal = P available P detected = P e mergence P sawgrass = 0. 34 0.5 1 = 0.1 7 Slough habitat Hypothetical mean habitat variables associated with the slough habi tat were; 22 m distance from transect, 2 dm visual obstruction, 70 cm water depth, and an airboat seat height of 160 cm. Therefore the probability of detecting an available alligator in the slough habitat was P s lough = 1 / (1 + e^ (( 3.0797) (0.0637 22) (0.5599 2 ) + (0.0577 7 0 ) + (0.0131 160 )) = 0.6 3 The combined estimated detection probability for the alligator emergence model and the slough habitat detectability model were P t otal = P available P detected = P e mergence P slough = 0. 34 0.6 3 = 0. 21

PAGE 56

56 Wet prairie habitat The mean habitat variables associated with the wet prairie habitat were; 22 m distance from transect, 2 dm visual obstruction, 70 cm water depth, and an airboat seat height of 160 cm. Therefore the probability of detecting an available alligator in the wet prairie habitat was P w et prairie = 1 / (1 + e^ (( 0.2896) + (0.0372 22 ) (0.5989 ) (0.0144 70) + (0.0131 160 )) = 0.6 2 The combined estimated detection probability for the alligator emergence model and the wet prairie habitat detectability model were P t otal = P available P detected = P e mergence P wet prairie = 0. 34 0.6 2 = 0. 2 1

PAGE 57

57 LIST OF REFERENCES Anderson, D. R. 2008. Model b ased i nference in the l ife s ciences, A p rimer on e vidence. Springer Science and Busi ness Media. New York, New York, USA Bayliss, P. 1987. Survey methods and monitoring within crocodile management programmes. Pages 157175 in G. J. W. Webb, S. C. Manolis and P. J. Whitehead, editors. Wildlife Management: crocodiles and alligators. Surrey Beatty and Sons Pty Ltd, Chipping Norton, Australia. Brandt, L A. 1989. The status and ecology of the American alligator ( Alligator mississippiensis ) in Par Pond, Savannah River Site. Thesis, Florida International University, Miami, Florida, USA. Buc kland, S. T., D. R. Anderson, K. P. Burnham, and J. L. Laake. 1993. Distance sampling. Chapman and Hall. London Bugbee C. D. 2008. Emergence dynamics of American alligators (Alligator mississippiensis) in Arthur R. Marshall Loxahatchee National Wildlife Refuge: life history and application to statewide alligator surveys Thesi s University of Florida, Gainesville, Florida, USA. Burnham, K. P., and D. R. Anderson. 2002. Model selection and multimodel inference: a practical information -theoretic approach. 2nd Edition. Springer -Verlag, New York, New York, USA. Busch, J., I. A. Mendelssohn, B. Lorenzen, H. Brix, and S. Miao. 2004. Growth responses of everglades wet prairie species Eleocharis cellulose and Rhynchospora tracyi to water level and phosphate a vailability. Aquatic Botany 78: 37 54. Caughley, G. 19 74. Bias in aerial surveys. Journal of Wildlife Management 38:921933. Chabreck, R. H. 1966. Methods of determining the size and composition of alligator populations in Louisiana. Proceedings of the 20th Annual Conference of Southeastern Association of Game and Fish Commissioners 20: 105112. Chabreck, R. H. 1976. Cooperative surveys of population trends in the American alligator, 19711975. Proceedings of the 3rd Working Meeting, Crocodile Specialis t Group, IUCN Supplementary Paper Cherkiss M. S., F. J. Mazzotti, and K. G. Rice. 2006. Effects of shoreline vegetation on visibility of American crocodiles (Crocodylus acutus ) during spotlight surveys Herpetological Review 37:37 40. Cook R. D. and F. B. M artin 1974. A model for quadrat sampling with "visibility bias." J ournal of America Statistical Association 69:345349.

PAGE 58

58 Da Silveira, R., W. E. Magnusson, and Z. Campos. 1997. Monitoring the d istribution, a bundance and b reeding a reas of Caiman croc odilus crocodilus and Melanosuchus niger in the Anavilhanas Archipelago, Central Amazonia, Brazil. Journal of Herpetology 31:514 520 Dolton, D. D. 1996. Mourning dove breeding population status. U.S. Fish and Wildlife Service, Washington, DC. Gardner, S. E. H. W Baard, and N. J. Le Roux. 1999. Estimating the detection probability of the geometric tortoise. South African Journal of Wildlife Research 29: 6271. Graham, A., and R. Bell. 1969. Factors influencing the countability of animals. East Afr ican Ag ricultural and Forestry Journal 34:3843. Gunderson L. H. 1994. Vegetation of the Everglades: d eterminants of community compostion. Pages 323340 in S. Davis and J. Ogden, editors. Everglades: the ecosystem and its restoration. St. Lucie Press, Delray Be ach, Florida, USA. Hines, T. C. and C. L. Abercrombie. 1987. The management of alligators in Florida, USA. Pages 4347 in G. J. W. Webb, S. C. Manolis and P. J. Whitehead, editors. Wildlife Management: crocodiles and alligators. Surrey Beatty and Sons Pty Ltd, Chipping Norton, Australia. Hutton, J. M. and M. E. J. Woolhouse. 1989. Mark recapture to assess factors affecting the proportion of a Nile crocodile population seen during spotlight counts at Ngezi, Zimbabwe, and the use of spotlight counts to moni tor crocodile abundance. Journal of Applied Ecology 26: 381395. Huggins R. M. 1989. On the statistical analysis of capture experiments. Biometrika 76:133140. Huggins R. M. 1991. Some practical aspects of a conditional likelihood approach to capture e xperiments. Biometrics 47:725732. Jordan, F., H. L. Jelks, and W. M. Kitchens. 1997. Habitat structure and plant community composition in a northern everglades wetland landscape. Wetlands 17:275283. King, F. W., M. Espinal, and C. Cerrato. 1990. Distri bution and status of the crocodilians of Honduras in Crocodiles. Proceedings of the10th working meeting, Crocodile Specialist Group. IUCN The World Conserv ation Union. Gland, Switzerland 313354. Krebs, C. J. 2002. Ecological Methodology. 2nd Edition Ad disonWesley. Menlo Park, California, USA. Kushlan, J. A. 1990. Freshwater marshes. Pages 324363 in R. L. Meyers and J. J. Ewel, editors. Ecosystems of Florida. University of Central Florida Press, Orlando, FL, USA.

PAGE 59

59 Lancia, R. A., W. L. Kendall, K. H. Pollock, and J. D. Nichols.2005. Estimating the number of animals in wildlife populations P ages 106146 in C.E. Braun editor Techniques for Wildlife investigations and management. The Wildlife Society, Bethesda, Maryland, USA. Loveless, C. M. 1959. A s tudy of the vegetation in the Florida everglades. Ecology 40:29. Mackenzie, D. I., L. L. Bailey, and J. D. Nichols. 2004. Investigating species co -occurrence patterns when species are detected imperfectly. Journal of Animal Ecology 73:546 555. Magnusson W. E. 1982. Techniques of surveying for crocodilians in Crocodiles. Proceedings of the 5th working meeting, Crocodile Specialist Group, Species Survival Commission. IUCN The World Conservation Union. Gland, Switz e rland 389403. Mazzotti, F. J. 1989. S tructure and function. Pages 4257 in C.A. Ross and S. Garnett, editors. Crocodiles and Alligators. Weldon Owen Pty. Ltd., Australia. Murphy, T. M. 1977. Distribution, movement, and population dynamics of the American alligator in a thermally altered rese rvoir. Thesis. University of Georgia, Athens, GA, USA. Newman, S., J. Grace, and J. Koebel. 1996. Effects of nutrients and hydroperiod on Typha, Cladium and Eleocharis : Implications for Everglades restoration. Ecological Applications 6:774783 Nichols J D, F. A. Johnson, and B. K. Williams. 1995. Managing North American waterfowl in the face of uncertainty. Annual Review of Ecology and Systematics 26:177199. Nupp, T. E., and R. K. Swihart. 1996. Effect of f orest patch area on population attributes of white -footed mice (Peromyscus leucopus) in fragmented land scapes. Canadian Journal of Zoology 74:467472. Pacheco, L. F. 1996. Effects of environmental variables on black caiman counts in Bolivia. Wildlife Society Bulletin 24: 44 49. Pearse, A. T., P. D. Gerard, S. J. Dinsmore, R. M. Kaminski, and K. J. Reinecke. 2008. Estimation and c orrection of v isibility b ias in a erial s urveys of w intering d ucks. Journal of Wildlife Management 72:808813. Peterjohn, B. G., J. R. Saurer, and W. A. Link. 1996. The 1 994 and 1995 summary of the n orth American b reeding b ird s urvey. Bird Populations 4:4866. Pollock, K. H., J. D. Nichols, T. R. Simons, G. L. Farnsworth, L. L. Bailey, and J. R. Saurer. 2002. Large scale wildlife monitoring studies: statistical methods fr om design and analysis. Environmetrics. 13:105119.

PAGE 60

60 Ramsey, F. L., and K. Harrison. 2004. A closer look at detectability. Environmental and Ecological Statistics 11:73 84. Rhodes, W. E. and P. M. Wilkinson. 1994. Alligator night light surveys of impoundm ent habitats in coastal South Carolina a preliminary validation. Pages 66 73 in Proceedings of the 12th Working Meeting. Crocodile Specialist Group, Vol. 2. IUCN, Gland Switzerland. Rice, K. G., F. J. Mazzotti, L. A. Brandt. 2005. Status of the American a lligator in southern Florida and its role in measuring restoration success. Pages 145153 i n W. E. Meshaka, and K. J. Babbitt editors Amphibians and Reptiles: status and conservation in Florida Krieger Publishing Co., Malibar, Florida, USA. Robel, R. J ., J. N. Briggs, A. D. Dayton, and L. C. Hulbert 1970. Relationship between visual obstruction measurements and weight of grassland vegetation. Journal of Range Management 23:295297 Rosenstock, S. S., D. R. Anderson, K. M. Giesen, T. Leukering, and M. F. Carter 2002. Landbird counting techniques: current practices and an alternative. Auk 119: 4653. Samuel, M. D. and K. H. Pollock. 1981. Correction of visibility bias in aerial surveys where animals occur in groups Journal of Wildlife Management 45:993997 Spratt, R. G. 1997. Harvest induced wariness in American alligators in Florida. Thesis, University of Florida, Gainesville, FL, USA Steinhorst R. K. and M. D. Samuel. 1989. Sightability adjustment methods for aeri al surveys of wildlife populations. Biometrics Journal 45:415426. Thompson, S. K. 1992. Sampling. Wiley. New York, New York, USA. Thompson, S. K. and G. A. F. Seber. 1994. Detectability in conventional and adaptive sampling. Biometrics 50: 712724. Thorbjarnarson, J.B., 1988. The status and ecology of the American crocodile in Haiti. Bulletin Florida State Museum, Biological Sciences 33:1 86. Webb, G. J. W., and H. Messel. 1979. Wariness in Crocodylus porosus Australian Wildlife Research 6: 227 234. White, G. C., and K. P. Burnham. 1999. Program MARK: Survival estimation from populations of marked animals. Bird Study 46 Supplement, 120138. Wiley, E. N., and M. Jennings. 1990. An overview of alligator management in Florida. Pages 274285 in Proceedin gs of the 10th Working Meeting. Crocodile Specialist Group, Vol. 2. IUCN, Gland Switzerland.

PAGE 61

61 Williams, B. K., J. D. Nichols, and M. J. Conroy. 2002. Analysis and management of animal populations. Academic Press, San Diego, California, USA. Wood, J. M., A R. Woodward, S. C. Humphrey, and T. C. Hines. 1985. Night counts as an index of American alligator population trends. Wildlife Society Bulletin 13:262273. Wood, J. M., and G. W. Tanner. 1990. Graminoid community composition and structure within four everglades management areas. Wetlands 10: 127149. Woodward, A. R., and W. R. Marion. 1978. An evaluation of factors affecting night -light counts of alligators. Proceedings of the Annual Conference of the Southeastern Association of Fish and Wildlife Agen cies 32:291302. Woodward, A. R., D. N. David, and T. C. Hines. 1987. American alligator management in Florida. Pages 98113 in R. R. Odom, K. A. Riddleberger, and S. C. Ozier, editors. Proceedings of the Third Southeast Nongame and Endangered Wildlife Symposium. Athens, GA, USA. Woodward, A. R., and C. T. Moore. 1990. Statewide alligator surveys. Bureau of Wildlife Research, Florida Game and Fresh Water Fish Commission, Tallahassee, Florida. USA. Woodward, A. R., and S. B. Linda. 1993. Alligator populat ion estimation. Final Report, Fl orida Game and Freshwater Fish Comm ission Tallahassee, Florida USA Woodward, A. R., K. G. Rice, and S. B. Linda. 1996. Estimating sighting proportions of American alligators during night light and aerial helicopter surve ys. Proceedings of the Annual Conference of the Southeastern Association of Fish and Wildlife Agencies 50:509 519.

PAGE 62

62 BIOGRAPHICAL SKETCH Cameron Blair Carter was born on February 5, 1980 in Orlando, Florida. He grew up in Lake Mary, Florida, graduat ed from Lake Mary High School in 1998. He spent much of his free time while growing up, exploring the lakes and rivers throughout Florida where he developed an interest in wildlife. He earned a B.S. in n atural r esource c onservation with a minor in w ildl ife e cology and forestry from the University of Florida in 2003. Upon graduation in 2003, Cameron began working with the Florida Fish and Wildlife Conservation Commission (FWC) in the Reptiles and Amphibian Research Subsection of the Fish and Wildlife Res earch Institute. He is currently a Biological Scientist with FWC and focuses on alligator research and management.