Embed Size (px)
Citation preview
ECOLOGY AND HABITAT USE OF BATS IN SOUTHERN FLORIDA, WITH SPECIAL EMPHASIS ON THE FLORIDA BONNETED BAT (EUMOPS FLORIDANUS)
By
AMANDA M. BAILEY
A THESIS PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF SCIENCE
UNIVERSITY OF FLORIDA
2016
© 2016 Amanda M. Bailey
To my grandfathers. Thanks for always supporting me even when you did not understand what I was doing, and for pushing me to always do my best.
4
ACKNOWLEDGMENTS
First, I would like to thank my advisors, Dr. Robert McCleery and Dr. Holly Ober, for
their patience and guidance throughout this project, even when I up and left for a
summer. I thank them for not giving up on me and for always pushing me to do my best.
I thank all of my lab mates for always making time at the office enjoyable, and for
always being willing to assist in any way possible.
I thank Florida Fish and Wildlife Conservation Commission for providing the grant
supporting my research. I thank their staff for making the demographic study possible,
and for assisting in various ways with the acoustic surveys. I especially would like to
thank Jeff Gore, Kathleen Smith, Jennifer Myer, Terry Doonan, Jeanette Parker and
Rachel Young for all of their assistance throughout this project.
I thank Tommy Bell for putting up with all of my time 1000 miles away and for
forgiving me for working throughout most of our visits. I thank my wonderful friend Amy,
who supported me and was always the best friend that I could ask for, even though she
was literally as far away as possible from me while remaining in the US. You are going
to do great things, my friend! I would like to thank my former NYSDEC supervisors, Alan
Hicks and Carl Herzog, who first introduced me to bat research and mentored me as I
began my career. Finally, I thank my parents and brother for all of their support during
this project. There are not many other people that I know that are willing to drop
everything that they are doing to take a phone call from me and coach me through
anything that is wrong at the time. I would not be the person that I am today without all
of the love and support that I have received throughout my life.
5
TABLE OF CONTENTS page
ACKNOWLEDGMENTS .................................................................................................. 4
LIST OF TABLES ............................................................................................................ 7
LIST OF FIGURES .......................................................................................................... 8
LIST OF ABBREVIATIONS ........................................................................................... 10
ABSTRACT ................................................................................................................... 11
CHAPTER
1 INTRODUCTION .................................................................................................... 13
Background ............................................................................................................. 13
Objectives ............................................................................................................... 16
2 DEMOGRAPHIC RATES OF THE FEDERALLY ENDANGERED FLORIDA BONNETED BAT ON A WILDLIFE MANAGEMENT AREA ................................... 18
Synopsis ................................................................................................................. 18 Background ............................................................................................................. 19
Methods .................................................................................................................. 21 Study Area ........................................................................................................ 21
Data Collection ................................................................................................. 21 Data Analysis ................................................................................................... 23 Simulations ....................................................................................................... 24
Results .................................................................................................................... 25 Mark-Recapture Study...................................................................................... 25
Simulations ....................................................................................................... 26 Discussion .............................................................................................................. 27
3 DISTRIBUTION OF THE FLORIDA BONNETED BAT (EUMOPS FLORIDANUS) IN SOUTHERN FLORIDA ............................................................. 41
Synopsis ................................................................................................................. 41 Background ............................................................................................................. 42 Methods .................................................................................................................. 44
Study Area ........................................................................................................ 44 Site Selection ................................................................................................... 47 Acoustic Surveys .............................................................................................. 48 Covariate Sampling .......................................................................................... 49
Detection covariates .................................................................................. 49 Occupancy Covariates ............................................................................... 49
6
Data Analysis ................................................................................................... 51
Results .................................................................................................................... 54 Discussion .............................................................................................................. 55
4 ECOLOGY OF BATS IN HUMAN-DOMINATED HABITATS IN SOUTHERN FLORIDA ................................................................................................................ 78
Synopsis ................................................................................................................. 78 Background ............................................................................................................. 78 Methods .................................................................................................................. 84
Study Species .................................................................................................. 84 Study Area ........................................................................................................ 85 Site Selection ................................................................................................... 87
Acoustic Surveys .............................................................................................. 88 Detection Covariates ........................................................................................ 89 Occupancy Covariates ..................................................................................... 90
Broad-scale analysis .................................................................................. 90 Fine-scale analysis .................................................................................... 91
Data Analysis ................................................................................................... 92 Broad-scale model ..................................................................................... 95 Fine-scale model ........................................................................................ 96
Results .................................................................................................................... 97 Broad-scale model ........................................................................................... 97
Fine-scale model .............................................................................................. 98 Discussion .............................................................................................................. 99
5 CONCLUSION ...................................................................................................... 120
APPENDIX
A RESULTS OF SIMULATIONS IN CHAPTER 2 .................................................... 123
B SPECIES MAPS IN RELATION TO DEVELOPMENT AND CROP-BASED AGRICULTURE .................................................................................................... 125
LIST OF REFERENCES ............................................................................................. 133
BIOGRAPHICAL SKETCH .......................................................................................... 149
7
LIST OF TABLES
Table page 2-1 Breakdown of each capture event of bonneted bats ........................................... 34
2-2 Capture probabilities (p) of adult and juvenile bonneted bats ............................. 35
2-3 Model comparisons for capture probability (p), apparent survival (φ) and population growth rate (λ) for bonneted bats ...................................................... 36
2-4 Model comparisons for recruitment (f) of bonneted bats .................................... 39
2-5 Relationship between adult survival (φ) and recruitment (𝑓) on derived population growth (λ) of bonneted bats .............................................................. 40
3-1 Florida Natural Areas Inventory (FNAI) state-level land cover types grouped into broad land covers ........................................................................................ 60
3-2 Description of covariates used in occupancy models for the Florida bonneted bat ...................................................................................................................... 62
3-3 Beta estimates of final model predicting bonneted bat occupancy, including variables affecting detection (p) and occupancy (psi). ........................................ 64
4-1 Florida Natural Areas Inventory (FNAI) state-level land cover types grouped into broad land covers ...................................................................................... 106
4-2 Covariates used in the fine-scale Bayesian hierarchical model for each bat species ............................................................................................................. 107
4-3 Beta estimates and credible intervals for detection covariates included in the final broad-scale Bayesian hierarchical model for each bat species................. 108
4-4 Beta estimates and credible intervals for occupancy covariates included in the final broad-scale Bayesian hierarchical model for each bat species ........... 109
4-5 Beta estimates and credible intervals for covariates included in the final fine-scale Bayesian hierarchical model for each bat species recorded in human-dominated landscapes ...................................................................................... 115
8
LIST OF FIGURES
Figure page 1-1 Locations where bonneted bats were previously recorded during acoustic
surveys from 2006 – 2012. Adapted from Marks and Marks 2012. .................... 17
2-1 Location of bat houses on Babcock-Webb Wildlife Management Area, Charlotte Co, FL. ................................................................................................ 33
2-2 Apparent survival rates of bonneted bats (Eumops floridanus) .......................... 37
2-3 Average annual growth rate (λ) estimates for male and female bonneted bats .. 38
3-1 Tentative range map of the Florida bonneted bat. Adapted from Marks and Marks (2012). ..................................................................................................... 59
3-2 Koppen-Geiger Climate Classification map ........................................................ 61
3-3 Land cover map of our study area in southern Florida ....................................... 63
3-4 Effect of Julian date on detection probability of bonneted bats ........................... 65
3-5 Effect of minimum temperature during each survey night on detection probability of bonneted bats ............................................................................... 66
3-6 The relationship between P*, the probability to detect bonneted bats at a site acoustically at least once during n surveys ........................................................ 67
3-7 Effect of the percent of grid cell classified as developed on the occupancy probability of bonneted ba .................................................................................. 68
3-8 Map of percent of each grid cell in the study area covered by development ...... 69
3-9 Effect of average annual precipitation on the occupancy probability of bonneted bats ..................................................................................................... 70
3-10 Effect of average annual spring precipitation on the occupancy probability of bonneted bats ..................................................................................................... 71
3-11 Map of average annual precipitation from 1981 – 2010 of each grid cell within our study area ..................................................................................................... 72
3-12 Effect of the average spring minimum temperature on the occupancy probability of bonneted bats ............................................................................... 73
3-13 Map of average spring minimum temperature from 1981 – 2010 of each grid cell in the study area ........................................................................................... 74
9
3-14 Effect of the percent of grid cell classified as crop-dominated agriculture on the occupancy probability of bonneted bats ....................................................... 75
3-15 Map of percent of each grid cell in the study area covered by crop-dominated agriculture ........................................................................................................... 76
3-16 Posterior distribution of the number of sites occupied by the bonneted bat based on acoustic surveys ................................................................................. 77
4-1 Map of percent of each grid cell in the study area covered by crop-dominated agriculture ......................................................................................................... 104
4-2 Map of percent of each grid cell in the study area covered by development .... 105
4-3 Bat species richness in grid cells of each major land cover type. ..................... 110
4-4 Effect of developed land cover on occupancy probabilities of species of bats . 111
4-5 Effect of crop-based agriculture land cover on occupancy probabilities of species of bats .................................................................................................. 112
4-6 Effect of amount of forested wetlands within sampled grid cells on occupancy probabilities of species of bats.......................................................................... 113
4-7 Effect of amount of forested uplands within sampled grid cells on occupancy probabilities of species of bats.......................................................................... 114
4-8 Effect of distance to undeveloped habitats in human-dominated landscapes on occupancy probabilities of bats .................................................................... 116
4-9 Response of bat species to distance to undeveloped habitat in human-dominated landscapes ...................................................................................... 117
4-10 Effect of canopy cover in human-dominated landscapes on occupancy probabilities of bat species ............................................................................... 118
4-11 Response of bat species to canopy cover in human-dominated landscapes ... 119
10
LIST OF ABBREVIATIONS
AICc Akaike’s Information Criterion accounting for small sample size
BWWMA Babcock-Webb Wildlife Management Area, Charlotte County, Florida.
ESA Endangered Species Act
FWC Florida Fish and Wildlife Conservation Commission
USFWS United States Fish and Wildlife Service
USDA United States Department of Agriculture
CORA Rafinesque’s big-eared bat (Corynorhinus rafinesquii)
EUFL Florida bonneted bat (Eumops floridanus)
LASE Seminole bat (Lasiurus seminolus)
LABO Eastern red bat (Lasiurus borealis)
LAIN Northern yellow bat (Lasiurus intermedius)
EPFU Big brown bat (Eptesicus fuscus)
MYAU Southeastern myotis (Myotis austroriparius)
NYHU Evening bat (Nycticeius humeralis)
PESU Tri-colored bat (Perimyotis subflavus)
TABR Brazilian free-tailed bat (Tadarida brasiliensis)
11
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
ECOLOGY AND HABITAT USE OF BATS IN SOUTHERN FLORIDA, WITH SPECIAL
EMPHASIS ON THE FLORIDA BONNETED BAT (EUMOPS FLORIDANUS)
By
Amanda M. Bailey
May 2016
Chair: Holly K. Ober Cochair: Robert A. McCleery Major: Wildlife Ecology and Conservation
The tropical climate and landscape of southern Florida allow this region to
support a large number of endemic species. The Florida bonneted bat (Eumops
floridanus) is a federally endangered species that is endemic to this area. Very little is
known about this species. We investigated demographic parameters of a population of
bonneted bats using a wildlife management area by conducting a mark-recapture study.
We found that bonneted bats had low survival rates compared to other species of bats.
In addition, juveniles had relatively low survival rates, likely the result of dispersal. We
also conducted an acoustic survey across southern Florida to investigate the distribution
of bonneted bats. The distribution of the species was limited by the southern Florida
climate, with the bats preferring warm areas with high levels of precipitation.
Additionally, bonneted bats avoided areas with high amounts of human development
and appeared to be positively impacted by the amount of crop-based agriculture in an
area.
While the southern Florida landscape supports a number of unique species, it is
also highly threatened by a growing human population and increasing urbanization. We
12
conducted an acoustic survey throughout all of southern Florida to investigate bat use of
human-dominated landscapes. All bat species were negatively influenced by the
percentage of the landscape covered by development, and most species were
negatively influenced by the percentage of the landscape covered by crop-dominated
agriculture. Within human-dominated landscapes, most bat species were positively
impacted by the amount of tree canopy cover and the proximity to natural areas.
13
CHAPTER 1 INTRODUCTION
Background
Extinction rates of species over the past decade are 3 to 4 times background
levels (Barnosky et al. 2011), largely as a result of anthropogenic disturbances
(Valiente-Banuet et al. 2015). Unfortunately, the number of threatened species
outnumbers the available resources for conservation (Ehrlich 1994, Myers 1996, Myers
et al. 2000). In order to cope with this, a number of conservation priority strategies have
been developed to determine where conservation efforts should be focused. Most
strategies incorporate metrics of irreplaceability and vulnerability (Brooks et al. 2006).
Irreplaceability measures the number of endemic species and uniqueness of major
vegetative communities (Brooks et al. 2006). Vulnerability considers the amount of
habitat loss and other threats facing an area, as well as the number of threatened
species (Brooks et al. 2006). ‘Hotspots’ where there are a number of endemic and
threatened species facing high levels of anthropogenic threats (Myers et al. 2000) are
often considered the top priority for conservation actions.
The southern portion of the Florida peninsula has been recognized as part of the
Caribbean biodiversity hotspot, containing over 5% of the endemic species identified
globally (Myers et al. 2000). Southern Florida’s mixture of subtropical humid and tropical
savanna conditions creates a climate similar to the tropics (Kottek et al. 2006). As a
result of this, the area is not only a hotspot for endemic species, but also supports
populations of tropical species that are found nowhere else in the country (Snyder et al.
1990, Webb 1999).
14
Historically, southern Florida was dominated by oligotrophic wetlands
interspersed with upland environments (Gunderson 2001). The area featured a number
of unique habitats, including the vast sawgrass prairies of the Everglades, the sandy
ridges of the Lake Wales Ridge and the pine rocklands of the Atlantic Ridge (Webb
1999). Much of southern Florida was considered largely inhospitable until the 20th
century, when the human population began to grow rapidly (Gunderson 2001). As the
population grew from around 32,000 in 1900 to over 6 million in 2010 (US Census
Bureau 2010), the region experienced drastic changes. Over 65% of the Kissimmee –
Everglades wetlands were drained for development and agriculture, the pine rocklands
were reduced to approximately 2% of their historic range, and much of southern
Florida’s dry prairies and scrub were converted to agricultural areas (Noss et al. 1995,
Noss and Peters 1995). Today, southern Florida is considered one of the most
endangered landscapes in the United States (Noss and Peters 1995), and is home to
around 70 threatened and endangered species (Brown et al. 2006).
One of the threatened species found in southern Florida is the Florida bonneted
bat (Eumops floridanus), which was federally listed as endangered in 2013 (US Fish
and Wildlife Service [USFWS] 2013). Endemic to southern Florida, this species is
believed to have one of the most limited geographic ranges of any bat species in North
America (Belwood 1992, Timm and Genoways 2004). The Florida bonneted bat
(hereafter; bonneted bat) was considered common throughout southeastern Florida in
the early part of the 20th century (Belwood 1981, Belwood 1992, Timm and Genoways
2004); however, records of the species from after 1965 are sparse (Timm and
Genoways 2004). The bonneted bat was first identified in southwestern Florida in 1979,
15
when a colony was found inhabiting a cavity in a longleaf pine (Pinus palustris) near
Punta Gorda (Belwood 1981). Recent acoustic surveys throughout southern Florida
have found the species in extreme southeastern Florida, at a number of locations in
southwestern Florida, and at 2 locations north of Lake Okeechobee (Figure 1-1).
Arguably, one of the largest threats to the Florida bonneted bat is a lack of
information on their ecology and life history (USFWS 2013). There have been no
estimates of demographic rates for this species. Additionally, the range of the species is
uncertain, with recent acoustic surveys for the bonneted bat being limited by sampling
pine-dominated areas and using varying methodology. As the bonneted bat has only
been identified from a small number of localities, environmental associations for this
species are virtually unknown. This lack of information inhibits the development of
effective conservation and recovery plans, and obscures potential causes of decline.
There is an urgent need to better understand these factors, as the southern Florida
landscape is expected to continue to change with increasing development pressure and
a growing human population (Zwick and Carr 2006).
The bonneted bat is not the only species of bat that is likely to be affected by the
dramatic human influence across southern Florida. This area is home to 10 species of
bats, 9 of which are considered species of greatest conservation need (SGCN) in
Florida (Florida Fish and Wildlife Conservation Commission [FWC] 2005). Southern
Florida is predicted to become mostly urbanized with some areas of intensive
agriculture in the coming decades (Zwick and Carr 2006), which is likely to have drastic
effects on native species. Because insectivorous bats provide important ecosystem
services by limiting arthropod populations (Jones et al. 2009, Kunz et al. 2011) and are
16
sensitive to a range of stressors that may be affecting other taxa, they have been
recognized as important bioindicators (Jones et al. 2009). By monitoring the responses
of bats to increasing levels of human influence, it is possible to get an idea of how other
taxa may also respond.
Objectives
In this study, we investigate the ecology of bats in southern Florida, with a
particular focus on the federally endangered Florida bonneted bat. We examine
demographic rates and environmental associations of the bonneted bat, and examine
how varying levels of human influence impact the bonneted bat and other bat species
throughout southern Florida. The specific objectives are (1) to determine the apparent
survival and recruitment rates of a population of bonneted bats on Babcock-Webb
Wildlife Management Area, (2) to investigate the range and environmental associations
of the Florida bonneted bat, and (3) to explore the effects of human-dominated
landscapes on the bat community of southern Florida.
17
Figure 1-1. Locations where bonneted bats were previously recorded during acoustic surveys from 2006 – 2012. Adapted from Marks and Marks 2012.
18
CHAPTER 2 DEMOGRAPHIC RATES OF THE FEDERALLY ENDANGERED FLORIDA
BONNETED BAT ON A WILDLIFE MANAGEMENT AREA
Synopsis
Knowledge of demographic rates is crucial for endangered species management.
An understanding of survival and recruitment rates of imperiled species can help to
elucidate causes of decline and assist with the development of recovery and
management plans. The Florida bonneted bat (Eumops floridanus) was federally listed
as endangered in 2013, but no estimates of demographic rates of this species have yet
been made. We conducted a mark-recapture study on a population of bonneted bats
using a number of bat houses on Babcock-Webb Wildlife Management Area, Charlotte
County, FL. We captured 175 individual bonneted bats during 6 capture events that
recurred every 4 months between April 2014 and December 2015. Passive integrated
transponders (PIT tags) were implanted in all captured individuals. We used Pradel’s
models to estimate apparent survival, recruitment, and population growth rates of this
population. Survival estimates were lower than expected based on estimates of other
bat species. Juveniles [φ = 0.094 (95% confidence interval: 0.038 – 0.193)] had lower
annual apparent survival than adults [φ = 0.462 (0.355 – 0.566)]. Recruitment was
constant between sexes and over time [f = 0.484 (0.231 - 0.745)]. Population growth
rate models of the bonneted bats using the bat houses on Babcock-Webb Wildlife
Management Area showed a stable to potentially declining population trend [λ = 0.893
(0.652 – 1.225)]. Any potential declines in the population may be more pronounced in
adult females [λ = 0.810 (0.584 – 1.125)]; however, 95% CI of λ include 1, indicating a
need for additional research to derive more precise estimates. This work represents the
first estimates of demographic parameters of this endangered species.
19
Background
Knowledge of demographic rates is crucial for endangered species management.
An understanding of survival and recruitment rates of imperiled species can help to
elucidate causes of decline and assist with the development of recovery and
management plans (Morris and Doak 2002, Bakker et al. 2009). Knowledge of which
demographic rates have the largest effect on population growth allows managers to set
more realistic and effective targets for recovery of threatened species (Johnson et al.
2010). Unfortunately, estimates of demographic parameters are often difficult to obtain
for rare and elusive species, providing challenges to conservationists and managers
(Tear et al. 1995, Fieberg and Ellner 2001).
Over 30% of bats (Order Chiroptera) are considered either threatened or data
deficient by the IUCN (IUCN 2015); however, relatively few studies have investigated
survival and recruitment rates of bats (O’Shea et al. 2004). Compared to other
mammals of their size, bats have the characteristics of a ‘slow’ species along the fast-
slow continuum (O’Shea et al. 2004). They are relatively long-lived (Tuttle and
Stevenson 1982, Bielby et al. 2007) and have low reproductive outputs (Barclay and
Harder 2003). Populations of slow species are expected to be sensitive to fluctuations in
adult survival (O’Shea et al. 2004, Pryde et al. 2005, Schorcht, Bontadina and Schaub
2009, Frick et al. 2010, Hayman et al. 2012). Apparent survival of adult bats in North
America generally ranges from 0.7 – 0.8, and there is some evidence that female
survival is higher than males in bats (O’Shea et al. 2004, Frick et al. 2010).
Higher extinction risks are associated with species that are long-lived, have a low
reproductive output and a small range (Purvis et al. 2000). The Florida bonneted bat
(Eumops floridanus) is one species that falls into this category. The bonneted bat was
20
federally listed as endangered in 2013 and is considered critically endangered by IUCN
(U.S. Fish and Wildlife Service [USFWS] 2013, IUCN 2015). Endemic to south Florida,
the Florida bonneted bat (hereafter bonneted bat) is believed to have one of the most
limited geographic ranges of any bat species in North America (Belwood 1992); their
range is estimated at approximately 12,000 km2 or less (Florida Fish and Wildlife
Conservation Commission [FWC] 2011, USFWS 2011). Although the bonneted bat is
believed to have a small, declining population (Marks and Marks 2012, USFWS 2013,
IUCN 2015), no actual estimates of demographic parameters exist. This lack of
information makes it impossible to understand how threats and management efforts are
affecting the population, and can potentially lead to the development of ineffective
recovery and management plans (Romesburg 1981, Campbell et al. 2002).
To increase our understanding of bonneted bat demographic rates and the
factors influencing them, we conducted a mark-recapture study of bonneted bats
roosting in bat houses on Babcock-Webb Wildlife Management Area, Charlotte County,
FL. Our objectives for this study were to 1) estimate rates of apparent survival and
recruitment of the population and population growth and 2) determine if demographic
rates were influenced by age and sex of bonneted bats or varied over time, and 3)
compare the relative influence of survival and recruitment on the population growth rate
of bonneted bats. We hypothesized that apparent survival would be higher in females,
and apparent survival of juveniles (< 1 year) would be lower than adults, as has been
documented in other species of bats (O’Shea et al. 2004). We also hypothesized that
adult survival would have a greater effect on population growth rate than recruitment,
which is typical of ‘slow’ species.
21
Methods
Study Area
We conducted all research on Fred C. Babcock-Cecil M. Webb Wildlife
Management Area (BWWMA), a 30,456 ha parcel of land in Charlotte and Lee
Counties, Florida. The climate is subtropical, with an annual average summer
temperature of 27.6º C in July, and average winter temperature of 17.8º C in January
(FWC 2003). BWWMA averages 125.3 cm of rainfall annually, with a wet season from
July - September and dry season extending through the winter (FWC 2003).
The major natural vegetation communities on BWWMA are dry prairie and hydric
pine flatwoods (FWC 2003). The pinelands of BWWMA are dominated by an open
canopy of slash pine. They are poorly drained, and on BWWMA remain flooded
throughout the wet season (FWC 2003). Around 40% of BWWMA is freshwater
marshes, sloughs and seasonal ponds (FWC 2003). BWWMA maintains a short fire
return interval, with most areas burned every 2 – 3 years (J. Birchfield, FWC,
pers.comm.). The wildlife management area is managed for bobwhite quail, and
supports a population of red-cockaded woodpeckers (FWC 2003).
Bonneted bats were first detected in BWWMA in 2006. Wildlife management
area staff erected 8 paired 1- or 3-chamber bat houses (Bat Conservation International,
Austin, TX) in 2007 – 2008. An additional 5 bat houses were constructed in 2012. We
conducted our research on all 7 of the paired bat houses used by bonneted bats during
our study period (Figure 2-1).
Data Collection
We conducted a mark-recapture study on the colony of bonneted bats using the
bat houses on Babcock-Webb Wildlife Management Area (BWWMA). The study
22
consisted of six 2 to 3 day capture events separated by 4 months: 22 – 25 April 2014,
27 – 30 August 2014, 15 – 17 December 2014, 20 – 24 April 2015, 24 – 26 August
2015, and 14 – 16 December 2015. Each bat house was trapped once during each
capture event; we used stacked mist nets to capture bonneted bats as they emerged
from each occupied house. We set up triple-high or double-high mist nets (Avinet, Inc.,
Dryden, NY) in a triangle or square shape around each house; houses were completely
encircled to minimize chances of bats emerging without being captured. We opened
mist nets at sunset, and kept the nets open for a maximum of 3 hours. The nets were
monitored continuously; when a bat was captured it was carefully removed from the net
and placed in a cotton bag.
We determined sex, age (adult or juvenile), reproductive status, body mass and
forearm length of each bat. Juveniles were distinguished by the partial or complete
fusing of phalangeal cartilage and/or the status of the genitals and mammae (Davis and
Hitchcock 1965). We marked each bat with a 12.1 mm, 134.2 kHz FDXB Passive
integrated transponder (PIT) tag (Biomark Inc., Boise, ID). During the first 2 capture
events, we sterilized all tags using 70% ethanol before insertion into the lower lumbar
region. In later capture sessions, we used PIT tags that were pre-sterilized and pre-
loaded into individual needles. Prior to insertion of each PIT tag, the desired injection
site was swabbed with chlorohexidine solution. Whenever possible, PIT tags were
implanted subcutaneously in the lower right lumbar region (as recommended by F.
Ridgeley, DVM). After injection, the entry site was sealed with a small amount of Skin
Bond (Montreal Ostomy, St. Louis, MO). After insertion, the bats were scanned with a
PIT tag reader to ensure that the PIT tag was properly implanted. During the initial
23
capture event in April 2014, every captured bonneted bat had a PIT tag implanted.
During subsequent recapture events, all bats were scanned with a PIT tag reader when
they first reached the handling station. All unmarked individuals had a PIT tag inserted
using the above methods. Once processing was complete, bats were released at the
capture location. All capture and handling of animals was conducted in accordance with
U.S. Fish and Wildlife Service permit #TE 23583B-0, Florida Fish and Wildlife
Conservation Commission permit # SUO-49616 and was approved by the University of
Florida IACUC (#201308070).
Data Analysis
We used Pradel’s temporal symmetry framework to develop estimates of
detection probability (p), apparent survival (φ), recruitment (f) and population growth
rate (λ) (Pradel 1996, Hines and Nichols 2002). We performed all analyses using
Program MARK version 3.0.3 (White and Burnham 1999). We used a sequential
approach to analyze all data, first developing models using the φ and λ
parameterization of the Pradel model (McCleery et al. 2013). We used Akaike’s
information criterion corrected for small sample size (AICc) to compare and select
models. Models with the lowest AICc value were considered the most parsimonious,
while models with an AICc within 2 of the most parsimonious model were considered
competing models. We reported all results as rates between capture events (4 month
vital rates), unless otherwise specified. We also calculated annual φ to compare rates to
other bat populations.
We fixed apparent survival and population growth rate to constant [φ(.)λ(.)] to
examine the effects of sex, age and time between capture events (CE) on p. We then
fixed p based on the most parsimonious model identified above, and made λ constant.
24
To determine the factors influencing φ, we allowed φ to vary as a function of CE, sex
and age and selected the most parsimonious model. Finally, we fixed p and φ based on
the estimates from the most parsimonious models previously, and allowed λ to vary by
CE and sex.
When estimating recruitment (f), we used an alternative parametrization of
Pradel’s model that estimated φ, p, and f (Pradel 1996). We fixed φ and p to the models
that were identified as most parsimonious above, and f was allowed to vary by CE and
sex. Because the Pradel model framework does not allow for the transition of juveniles
to adults within the model (White and Bufrnham 1999), age effects were not
investigated for λ or f.
Simulations
To test the robustness of the model and its response to violations of
assumptions, we ran a number of simulations. These simulations assess the effects that
tag loss and changes in survival would have on our results. We used GENCAPH1
(written by J.E. Hines, http://www.mbr-pwrc.usgs.gov/software.html) to generate capture
histories and first analyzed the effect of tag loss by setting φ constant at 0.70, p
constant at 0.60 and λ = 1.25. We generated capture histories with tag retention rates of
0.90 and 0.50 between each capture event. We then imported the simulated capture
histories into MARK, and ran Pradel’s model to test how the results compared to the
original variables.
We also analyzed the effect that varying survival rates would have on λ. We used
GENCAPH1 to generate capture histories at high (0.90), medium (0.70), low (0.50), and
very low (0.30) levels of φ. We held p constant at 0.60, tag retention constant at 1.0,
and λ at 1.25. We generated capture histories for 6 different capture events. We then
25
imported the simulated capture histories into MARK, and ran Pradel’s models to test
how λ varied with φ.
Results
Mark-Recapture Study
We captured a total of 175 individual bonneted bats during all of the capture
sessions; 139 bats were recaptured at least once (Table 2-1). The most parsimonious
model for p included temporal and age effects, with juveniles having a lower p than
adults (Table 2-2). The lowest overall p occurred in December 2014 [p = 0.585 (95%
confidence interval: 0.437 – 0.720) for adults, p = 0.340 (0.133 – 0.634) for juveniles,
Table 2-2]. Capture probabilities for the first and last capture events were confounded
(White and Burnham 1999). No juveniles were captured in April 2014, confounding the
estimate for juvenile detection probability in the following capture event (August 2014).
The most parsimonious model for apparent survival (φ) included additive
temporal and age effects (Table 2-3). Overall, φ of adults between capture events [φ =
0.773 (0.708 – 0.827)] was higher than φ of juveniles [φ = 0.454 (0.336 – 0.578)]
(Figure 2-2). The annual φ for adults was 0.462 (0.355 – 0.566), while the annual φ for
juveniles was 0.094 (0.038 – 0.193). Adults had the highest φ between April and August
2015 [φ = 0.863 (0.691 – 0.947)]; interestingly, they had the lowest φ between April and
August 2014 [φ = 0.689 (0.525 – 0.816)]. Juvenile φ approached 1.000 between August
and December 2014, but was low (around 0.155) between the other capture events
(Figure 2-2). Lowest overall φ was between December and April 2015 (φ = 0.696
(0.527 – 0.825), Figure 2-2). There was also some support for the model including age
and sex effects (Table 2-3), providing evidence that sex also accounts for some
variation in φ. This model estimated that adult females have the highest apparent
26
survival (φ = 0.791 (0.733 – 0.839)), followed by adult males (φ = 0.682 (0.554 –
0.787)), juvenile females (φ = 0.528 (0.363 – 0.686)) and juvenile males with the lowest
[(φ = 0.377 (0.226 – 0.556)).
The model for λ that received the most support included an effect of sex, but
there was no evidence of temporal variation (Table 2-3). According to this model,
growth between capture events was 0.932 (0.836 – 1.040) for females and 1.036 (0.903
– 1.189) for males, which corresponded to an annual λ of 0.810 (0.584 – 1.125) for
females and 1.112 (0.737 – 1.685) for males (Figure 2-3). Overall, the population of
bonneted bats using the bat houses on BWWMA showed a stable to declining trend
[annual λ of 0.893 (0.652 – 1.225)].
The constant model for f received the most support (Table 2-4). The annual
recruitment estimates according to this model were 0.484 (0.231 – 0.745) new animals
per individual in the population. The proportional contribution of φ on λ was larger than
the proportional contribution of f on population growth rate (Table 2-5).
Simulations
High tag loss could lead to a severe underestimation of apparent survival
probabilities (Appendix A). When we simulated that half of the tags were lost between
capture intervals, φ was estimated to be 0.315 (0.211 – 0.442), compared to the actual
φ of 0.70. When 90% of the tags were retained between capture events, the estimated
φ was very close to the actual value [estimated φ = 0.731 (0.630 – 0.812)]. In contrast, λ
was relatively robust to variations in tag retention, with a predicted λ of 1.218 (1.124 –
1.321) when tag retention was 90% and a predicted λ of 1.219 (1.119 – 1.329) when tag
retention was 50% (compared to the actual λ of 1.250).
27
The estimated φ probabilities were very close to the actual φ probabilities
(Appendix A). Low rates of φ led to an underestimation of λ, with an estimated λ of
1.077 (0.965 – 1.200) when φ = 0.300. When φ = 0.70, which was similar to rates
observed in this study, λ was slightly underestimated [λ = 1.188 (1.095 – 1.288)].
Discussion
Our study found apparent survival rates for bonneted bats using the bat houses
on BWWMA that were lower than rates reported for other species of bats in the western
hemisphere (see review in O’Shea et al. 2004). There are a couple of potential
explanations for the low apparent survival rates we observed. The majority of
demographic studies have been conducted on hibernating bats; hibernation is known to
have a significant effect on survival (Turbill et al. 2011). Turbill et al. (2011) found that
annual apparent survival of hibernating species is 15% higher than non-hibernating
species of similar body sizes. Bonneted bats do not hibernate, thus we would expect
lower survival rates than have been estimated for hibernating species of bats. A
demographic study on the related non-hibernating Molossid Molossus molossus also
found relatively low apparent survival rates. Monthly survival was ≈ 0.95 (Gager et al.
2016), with an extrapolated annual survival ≈ 0.540, similar to the 0.462 apparent
survival rate estimated in our study. Apparent survival for bonneted bats was near its
lowest levels around the winter months, providing some support that this period may
significantly contribute to the low apparent survival rates observed.
Another potential explanation for the low annual survival rates and the depressed
apparent survival rates observed in winter has to do with potential edge-of-range
effects. The Florida bonneted bat was considered a subspecies of Eumops glaucinus
until 2004, when it received species status based on morphology (Timm and Genoways
28
2004). Recent genetic evidence suggests that species status may not have been
warranted (Bartlett et al. 2013). E. glaucinus is a Neotropical bat species found
throughout Central and South America to the north of Argentina (Best et al. 1997; IUCN
2015). The depressed winter survival that we observed could be related to cold-
intolerance in a tropical species. Anecdotal evidence suggests that bonneted bats are
sensitive to cold. In 2010, a cold spell resulted in the permanent disappearance and
presumed mortality of half of the bonneted bats using a bat house in North Fort Myers
(USFWS 2013). There is some support for a link between fitness of vertebrates and the
location within a distribution; a meta-analysis by Sexton et al. (2009) found 64 of 112
papers showed reduced fitness of populations at the edge of their ranges.
With the apparent survival rates calculated in this study, it is impossible to know if
animals died or emigrated from the study area. High site fidelity has been documented
for many species of bats (Lewis 1995), and research on Molossus molossus found high
site fidelity for this related species, particularly for adult females (Gager et al. 2016).
However, behavior and social structure of the bonneted bat has not been formally
researched, so it is unknown whether they exhibit high degrees of roost fidelity. There is
some evidence that points to the potential that bats may be dispersing, at least
temporarily, from bat houses. 15 bats (~9% of individuals captured and tagged) were
not captured during at least 1 capture period but were subsequently recaptured. Seven
of these (4% of individuals tagged) were absent for at least 2 consecutive periods
before being recaptured, and 1 was absent for 3 consecutive periods before being
recaptured. The Pradel model allows for random temporary emigration, which is
accounted for by a decreased detection probability (Pradel 1996, Pradel et al. 1997);
29
however, the short duration of this study likely led to an underestimation of survival
rates.
It is also possible that tag loss contributed to the low apparent survival rates. One
assumption of the Pradel model is that no tags are lost for the duration of the study
(Pradel 1996, Pradel et al. 1997). We chose to mark all bats with PIT tags because of
high retention rates observed in other studies (Ellison et al. 2007). However, 11 PIT
tags were recovered underneath the occupied bat houses between April 2014 and April
2016. It is unknown if bats that lost PIT tags were recaptured after the tag was lost and
marked as new individuals. If this occurred, survival estimates would be biased low. We
were taking wing biopsies of every untagged individual captured; once the genetic
analysis is completed, we will have a better understanding of how many individuals lost
their tags, and can revise demographic estimates if necessary.
Apparent survival of juvenile bonneted bats using the bat houses on BWWMA
was significantly lower than apparent survival of adults, which was expected. Decreased
survival rates of juveniles less than 1 year of age has been observed for many different
species of bats (reviewed by O’Shea 2004). It is also possible that the juveniles are
dispersing after they become volant. Throughout the study, juveniles showed much
lower site fidelity than adults, with no juveniles being recaptured in their natal roosts
after eight months (Ober et al., in press). Juvenile Molossus molossus (which have a
similar social structure to bonneted bats) were found to disperse from natal roosts by at
least 7 months of age (Gager et al. 2016). Bonneted bats have a harem social structure
(Ober et al., in press). The dispersal of juvenile males would be expected to reduce
mate competition with the dominant adult male (Moore and Ali 1984). Juvenile female
30
dispersal is rare in mammals, but has been documented in several species of tropical
bats (McCracken 2010, Nagy et al. 2013, Gager et al. 2016). The fact that juvenile
survival declined rapidly over 1 time period provides support to this potential
explanation.
This study confirmed anecdotal evidence suggesting that the bonneted bat had
an extended breeding season, with recruitment of juveniles remaining constant
throughout the year. Corroborating this was the presence of juveniles during every
capture event with the exception of April 2014. In addition, BWWMA staff conducted
emergence counts from May – December 2014 and documented non-volant young in
houses during every month (Ober et al., in press). Reproductive activity has been
documented throughout most of the year in the closely related E. glaucinus (Best et al.
1997). However, our models do not completely line up with our observations. During
April 2014 and April 2015, nearly all adult female bonneted bats captured were
pregnant; however, the expected influx of juveniles from this pregnancy peak was never
observed. This is likely a factor of our capture intervals being too far apart, leading to
missed trends in recruitment. Gager et al. (2016) documented a similar trend in
reproduction in M. molossus in Panama, with a pregnancy peak occurring in April. They
monitored a subset of juveniles that were tagged in July and September of that year; the
juveniles all disappeared from the roost between July of that year and February of the
next year (Gager et al. 2016). With the low apparent survival rates documented for
juveniles during this study, it is likely that a number of juveniles born soon after the April
capture session would either have died or dispersed from their natal roosts before our
31
August capture event. More frequent captures will be necessary to document these
finer-scale trends in recruitment.
Although the bonneted bat was federally listed in 2013, no previous studies
attempted to determine population trends. Our study suggests that the population of
bonneted bats using the bat houses on BWWMA has a stable to slightly declining trend.
Most concerning is the apparent loss of females from the population. Although the
confidence intervals of female population growth crossed 1.0 (stable populations), our
results suggest that the population of female bonneted bats using bat houses on
BWWMA may be declining. Population growth was driven by adult survival, which was
expected based on previous demographic studies on bats (O’Shea et al. 2004).
Although adult females had the highest survival rates, the population is also heavily
female-biased (Ober et al., in press). The low apparent survival rates observed in this
study may not maintain the population of females. A loss of females also leads to a loss
of recruitment, which has the potential to be detrimental to this endangered population.
It is important to take the assumptions of Pradel’s model into consideration when
interpreting our results. Hines and Nichols (2002) found that trap response of individuals
led to a substantial bias in the estimation of λ. We did not witness evidence of trap
response during our study; although bats occasionally did not emerge from the bat
house, the number of individuals observed in the houses after nets were closed
appeared to be random throughout the study. Even if a trap response occurred, or
individuals had heterogeneous capture probabilities, Hines and Nichols (2002)
recommend using only models that do not look at time effects on λ. Our selected
models did not include an effect of time on λ, making our estimates relatively robust to
32
biases from heterogeneous capture probabilities or a trap response (Hines and Nichols
2002). Losses on capture were also found to lead to a biased estimation of λ (Hines and
Nichols 2002). Again, we had no indication that there was any trap- or tag-related
mortality of captured animals. The major assumption that was likely violated during our
study was that marks were not lost. Our simulations showed that λ was relatively robust
to high levels of tag loss, indicating that this assumption violation may not have resulted
in substantial bias to our λ estimate.
This study represents the first estimates of demographic parameters of the
federally endangered Florida bonneted bat. We documented surprisingly low apparent
survival rates compared to those estimated for other species of bats. In a population
that is sensitive to adult survival, this is especially concerning. Most alarming was the
potentially declining female population of bonneted bats, which could limit the ability of
this population to recover. While this study provides a good baseline of demographic
information, it also highlights the need for further research. Additional research focused
on the dispersal of individuals will be critical to elucidate whether the apparent survival
rates are low due to mortality or because bats are using roosts that we are not yet
aware of. If the latter is the case, the conservation of these roosts will be crucial for the
survival and recovery of this population.
33
Figure 2-1. Location of bat houses on Babcock-Webb Wildlife Management Area, Charlotte Co, FL.
34
Table 2-1. Breakdown of each capture event of Florida bonneted bats (Eumops floridanus) captured from occupied bat houses on Babcock-Webb Wildlife Management Area, Charlotte Co, FL during a demographic study from April 2014 – December 2015.
Capture Event
Number Captured
Number of Recaptures
Newly Marked
Total Previously Marked
Apr-14 50 0 50 0
Aug-14 61 25 36 50
Dec-14 42 31 11 86
Apr-15 61 40 21 97
Aug-15 63 32 31 118
Dec-15 79 53 26 149
35
Table 2-2. Capture probabilities (p) of adult and juvenile Florida bonneted bats (Eumops floridanus) captured from occupied bat houses on Babcock-Webb Wildlife Management Area, Charlotte Co, FL during a demographic study from April 2014 – December 2015. Estimates from the first and last capture events were confounded and not included.
Age
Capture Event
Estimate
95% Confidence Interval
Lower Upper
Adult
Aug-14 0.739 0.562 0.862
Dec-14 0.585 0.437 0.72
Apr-15 0.848 0.695 0.932
Aug-15 0.621 0.451 0.766
Juvenile
Aug-14 0.415 0.159 0.727
Dec-14 0.340 0.133 0.634
Apr-15 0.414 0.186 0.685
Aug-15 0.755 0.321 0.952
36
Table 2-3. Model comparisons of number of parameters (K), AICc values, change in AICc from the selected model (Delta AICc) and model weight (weight) for capture probability (p), apparent survival (φ) and population growth rate (λ) for Florida bonneted bats (Eumops floridanus) occupying bat houses on Babcock-Webb Wildlife Management Area, Charlotte Co, FL from April 2014 – December 2015.
Models K AICc Delta AICc Weight
Capture probability (p)
φ(.) P(CE + age) λ(.) 12 1047.971 0.000 0.675
φ(.) P(CE + age + sex) λ(.) 23 1049.432 1.461 0.325
φ(.) P(age + sex) λ(.) 6 1111.401 63.430 0.000
φ(.) P(age) λ(.) 4 1113.134 65.163 0.000
φ(.) P(CE + sex) λ(.) 13 1113.873 65.902 0.000
φ(.) P(sex) λ(.) 4 1118.511 70.540 0.000
φ(.) P(CE) λ(.) 8 1124.372 76.401 0.000
φ(.) P(.) λ(.) 3 1125.229 77.258 0.000
Survival rate (φ)
φ(CE + age) P(CE + age) λ(.) 22 1027.896 0.000 0.735
φ(age + sex) P(CE + age) λ(.) 17 1030.665 2.769 0.184
φ(age) P(CE + age) λ(.) 15 1033.589 5.693 0.043
φ(CE + age + sex) P(CE + age) λ(.) 31 1033.850 5.954 0.037
φ (sex) P(CE + age) λ(.) 15 1046.430 18.534 0.001
φ(.) P(CE + age) λ(.) 14 1047.971 20.075 0.000
φ(CE + sex) P(CE + age) λ(.) 23 1054.696 26.800 0.000
φ(CE) P(CE + age) λ(.) 18 1055.360 27.464 0.000
Population growth rate (λ)
φ(CE + age) P(CE + age) λ(sex) 19 1027.041 0.000 0.589
φ(CE + age) P(CE + age) λ(.) 18 1027.896 0.855 0.378
φ(CE + age) P(CE + age) λ(CE) 21 1032.496 5.4551 0.038
φ(CE + age) P(CE + age) λ(CE + sex) 26 1037.348 10.3070 0.003
37
Figure 2-2. Apparent survival rates of Florida bonneted bats (Eumops floridanus) between four-month capture intervals in Babcock-Webb Wildlife Management Area, Charlotte Co, FL. The intervals represent April – August 2014, August – December 2014, December 2014 – April 2015, April – August 2015 and August – December 2015.
38
Figure 2-3. Average annual growth rate (λ) estimates for male and female Florida bonneted bats (Eumops floridanus) using bat houses on Babcock-Webb Wildlife Management Area, Charlotte Co, FL. Estimates are the results of a demographic study that took place from April 2014 – December 2015.
39
Table 2-4. Model comparisons of number of parameters (K), AICc values, change in AICc from the selected model (Delta AICc) and model weight (weight)for recruitment (f) of Florida bonneted bats (Eumops floridanus) using bat houses on Babcock-Webb Wildlife Management Area, Charlotte Co, FL from April 2014 – December 2015.
Models for Recruitment (f) K AICc Delta AICc Weight
φ(CE + age) P(CE + age) f(.) 18 1044.530 0.000 0.519
φ(CE + age) P(CE + age) f(sex) 19 1045.498 0.968 0.320
φ(CE + age) P(CE + age) f(CE) 20 1046.921 2.391 0.157
φ(CE + age) P(CE + age) f(CE + sex) 25 1054.548 10.018 0.000
40
Table 2-5. Relationship between adult survival (φ̂) and recruitment (𝑓) on derived
population growth (λ̂) of Florida bonneted bats (Eumops floridanus) using bat houses on Babcock-Webb Wildlife Management Area, Charlotte Co, FL from
April 2014 – December 2015. φ̂
λ̂ represents influence of adult survival on
population growth rate, while �̂�
λ̂ represents the influence of recruitment on
population growth rate.
Capture Event
λ̂
φ̂
𝑓
φ̂
λ̂
𝑓
λ̂
Apr - Aug 2014 0.963 0.682 0.126 0.708 0.131
Aug - Dec 2014 0.963 0.799 0.126 0.830 0.131
Dec - Apr 2015 0.963 0.786 0.126 0.816 0.131
Apr - Aug 2015 0.963 0.848 0.126 0.881 0.131
Aug - Dec 2015 0.963 0.772 0.126 0.802 0.131
41
CHAPTER 3 DISTRIBUTION OF THE FLORIDA BONNETED BAT (EUMOPS FLORIDANUS) IN
SOUTHERN FLORIDA
Synopsis
The modeling of species distributions is critical in conservation and management
of wildlife. Habitat models are particularly useful for predicting habitat relationships of
rare or elusive species that may be difficult to sample. The federally endangered Florida
bonneted bat (Eumops floridanus) is endemic to southern Florida. Little is known about
the range and habitat requirements of this species, but it is believed to have one of the
most limited geographic distributions of any bat in the United States. We conducted a
large-scale acoustic survey of 330 sites spread across approximately 38,000 km2 of
southern Florida over a 2 year period. We used a hierarchical Bayesian approach
accounting for imperfect detection to model the distribution and environmental
associations of the bonneted bat. Bonneted bat occupancy was negatively correlated
with the amount of development within 5 km of the sampling point; occupancy was
positively correlated with the amount of crop-based agriculture within 5 km of the
sampling point. Bonneted bat occupancy probabilities were higher in areas with higher
minimum spring temperatures and lower spring precipitation levels; however, average
annual precipitation was positively correlated with bonneted bat occupancy rates.
Bonneted bat detection was positively influenced by Julian date, and also minimum
temperature of the survey night. This study is the first to use a robust sampling design
to investigate the distribution and habitat use of the Florida bonneted bat and offers our
first real insight into the habitat selection patterns of this endangered species.
42
Background
Knowledge of a species’ distributions and environmental associations is critical
for conservation and management of wildlife (Reid et al. 2008). Once important
environmental features are identified, proper conservation actions can be implemented
and the effects of potential disturbances can be assessed (Chamberlain et al. 1999,
Milne et al. 2006). This knowledge also makes it possible to designate scientifically
rigorous and tractable recovery criteria for federally listed species (Doak et al. 2015).
The Florida bonneted bat (Eumops floridanus) is one of just 9 species of bats
listed as endangered under the Endangered Species Act (ESA) in the United States
(U.S. Fish and Wildlife Service [USFWS] 2013). The bat was federally listed in 2013,
and is listed as critically endangered by the International Union for Conservation of
Nature (IUCN) (USFWS 2013, IUCN 2015). Surprisingly little is known about the Florida
bonneted bat (hereafter, bonneted bat). This species is endemic to southern Florida,
and is believed to have one of the most limited geographic ranges (approximately
12,000 km2 or less) of any bat species in North America (Belwood 1992, Florida Fish
and Wildlife Conservation Commission [FWC] 2011). Existing information on distribution
and environmental associations of the bonneted bat comes mostly from acoustic
surveys (Figure 3-1); (Brian Scofield, Avon Park Air Force Range, pers.comm., Marks
and Marks 2012). However, previous acoustic surveys were limited because of variable
methodologies and equipment, and were biased in favor of areas presumed to be
“good” habitat (e.g., mostly pine dominated natural areas). Accordingly, the current
portion of bonneted bats’ occupied range is assumed to be small, but remains largely
unknown (Marks and Marks 2012, FWC 2013).
43
The lack of reliable information on the distribution and environmental
associations of the bonneted bat hinders the development of effective conservation and
management plans for this species. While the species has been detected in a wide
variety of habitats, including urban, agriculture, uplands, and wetlands (Marks and
Marks 2008, Marks and Marks 2012, USFWS 2013), there is currently no information on
whether they use some areas more commonly than others. Based on records of
colonies roosting in slash and longleaf pine (Pinus spp., Belwood 1992) and foraging in
pine flatwoods in Charlotte County, Florida, it is generally believed that bonneted bats
select for pineland forests (USFWS 2013). However, there is no quantitative evidence
suggesting that pinelands are used more than any other land cover type.
It is possible that the unique climate of southern Florida influences the
distribution of the bonneted bat. This area boasts the only tropical climate in the
continental United States (Kottek et al. 2006). Up until 2008, the Florida bonneted bat
was considered a subspecies of Wagner’s bonneted bat (E. glaucinus), a Neotropical
species that is distributed across Central and South America north of Argentina (Best et
al. 1997, IUCN 2015). There is anecdotal evidence that bonneted bats may be cold-
sensitive. In 2010, a cold spell resulted in the permanent disappearance and presumed
mortality of half of the bonneted bats using a bat house in North Fort Myers (USFWS
2013). These factors may indicate that bonneted bats in Florida may be at the northern
edge of their range.
The distributions of other insectivorous species of bats have been found to be
influenced by prey availability, with higher occurrence probabilities in areas where insect
productivity is expected to be higher (e.g., Rodhouse et al. 2015). We would therefore
44
expect to find bonneted bats in areas that have high insect abundance that the bats
could successfully exploit. Insect abundance and diversity has been linked with
increased amounts of precipitation and tree canopy cover (Williams 1951, Silva et al.
2011), thus, these factors may influence the distribution of the bonneted bat.
The major goal of this study was to investigate the distribution and environmental
associations of the bonneted bat. Because the bonneted bat is found in such a small
area, we have the unique opportunity to sample across their entire geographic range.
By developing hierarchical models, we can also attain information on what influences
detection rates of this species, allowing for the development of effective monitoring
programs that can be used in the future. The specific objectives of our study were to (1)
investigate the distribution of the bonneted bat throughout southern Florida, (2) identify
potential environmental associations that influence the distribution of the bonneted bat,
and (3) determine factors influencing detection of the bonneted bat, which can be used
to make recommendations for future monitoring efforts.
Methods
Study Area
Our research was conducted in all counties of known or suspected occurrence of
the bonneted bat. This included 16 counties in southern Florida: Polk, Osceola,
Highlands, Okeechobee, Sarasota, Desoto, Charlotte, Glades, Martin, Lee, Hendry,
Palm Beach, Collier, Broward, Monroe and Miami-Dade. South Florida has a subtropical
to tropical climate (Figure 3-2), with a wet season during the summer months and a dry
season that extends from mid-fall through late spring (Duever et al. 1994). The major
source of rainfall is thunderstorms, with the average annual precipitation being nearly
constant for the past 100 years (Duever et al. 1994). The highest precipitation amounts
45
are found along the east coast, with Hialeah recording an average of 178.77
centimeters of precipitation each year compared to an average of 128.22 centimeters in
Venice, along the west coast (FSU Climate Center 2014).
The average temperatures are warm all year, only rarely dropping below
freezing; the northern counties (Polk and Osceola) have an average of 1 – 2 freeze
days annually, while the rest of the study area averages 0 days where the minimum
temperature drops below freezing (Southeast Regional Climate Center 2015). The
warmest month on average over the entire study area is August, with average
temperatures around 27°C in the northern part of the study area and between 28.5°C
and 29°C to the south of Lake Okeechobee. The coolest month is February; the coolest
temperatures (about 15.5°C) occur in the interior, northern part of the study area. The
Gulf coast is slightly warmer, and the east coast is the warmest, with average
temperatures around 19°C.
The study area encompasses a number of protected areas. The protected areas
include the nationally-managed Everglades National Park and Big Cypress National
Preserve, and a number of smaller, state-managed parks and wildlife management
areas. Major vegetation types include pine forest, cypress forest, mangrove swamps,
sloughs, and marsh-prairies (Webb 1999). In the northern portion of the study regions
lies the Lake Wales Ridge, a series of sandy ridges that supports relatively unique, xeric
communities of scrub, pine, and scrubby flatwoods. The southwest coast of Florida is
dominated by mesic flatwoods, which supports the third highest species richness of any
vegetative communities in Florida (Beever and Dryden 1998). The area is also
characterized by expansive dry prairies and mangroves along the coast (Webb 1999).
46
The Atlantic Coastal Ridge runs 160 km down the heavily urbanized east coast of
southern Florida (Webb 1999). This limestone ridge supports a number of ecological
communities, including pine rocklands, pine flatwoods, wet prairie, coastal strand, and
maritime and tropical hardwood hammocks (Webb 1999). The southern portion of the
study area is dominated by the cypress, pine and hardwood forests of the Big Cypress
subregion and the Everglades, the vast ‘river of grass’ that historically stretched across
much of the Florida peninsula south of Lake Okeechobee (Webb 1999). The natural
communities of the Everglades include forested wetlands, marshes and upland rockland
communities (Gunderson 1994, Webb 1999).
There are a wide variety of agricultural commodities produced throughout
southern Florida. The Everglades Agricultural Area encompasses nearly 300,000 ha of
land to the south of Lake Okeechobee (Rice et al. 2002). The dominant crop in this area
is sugarcane, with rice and vegetables also grown (Whalen and Whalen 1994). The
northern shores of Lake Okeechobee are dominated by the cattle industry (Webb 1999).
Approximately half of agricultural land in southern Florida consists of rangeland for
cattle (Webb 1999), and many wildlife management areas serve as improved and
unimproved cattle pastures (FWC 2013). Polk, Osceola, and Highlands counties
produce the most cattle in the state (Fresh from Florida 2013). Over half of all of
Florida’s citrus is produced in southern Florida (USDA 2014). Polk, Desoto, Hendry,
Highlands, and Hardee counties are the top citrus producers in the state (USDA 2014).
Miami-Dade and Collier counties are among the leading producers of tomatoes in the
country (USDA 2014).
47
Both the east and west coasts of the study area are dominated by urban areas. 4
of the top 10 largest metropolitan statistical areas in Florida are within this area: Miami-
Fort Lauderdale-West Palm Beach, North Port-Bradenton-Sarasota, Cape Coral-Fort
Myers and Lakeland-Winter Haven. As of 2014, there were close to 8 million people
living within these metropolitan areas, with the Miami-Fort Lauderdale-West Palm
Beach area accounting for nearly 6 million (U.S. Census Bureau 2015). Our sampling
included areas across the urban matrix from rural to suburban to high density urban
areas.
Site Selection
We established a grid system comprised of 5 km by 5 km cells that we overlaid
across southern Florida (Ormsbee 2010, Loeb 2015), similar to the Bat Grid and North
American Bat Monitoring Program (NABAT) survey protocols (Hayes et al. 2009, Loeb
2015). We overlaid this grid across our study area using ArcMap 10.1 (ESRI; Redlands,
CA). We selected cells using a stratified random sampling design to ensure a
representative sample of bat activity across all major land cover types in southern
Florida. We stratified land uses using the cooperative land cover map developed by the
Florida Natural Areas Inventory [FNAI] (FNAI 2012). We simplified the FNAI
classifications into four major categories: agriculture, developed, upland and wetland
(Table 3-1). We used the ‘Tabulate Area’ feature in the Spatial Analyst toolbox in
ArcMap to determine the dominant land cover type used to classify each grid cell.
We randomly selected 17 grid cells of each of the 4 land cover types for sampling
during two field seasons. After all grid cells were selected, we used the ‘Create
Random Points’ feature in the Data Management toolbox in ArcMap 10.1 to place 5
48
random sampling points in each cell to reduce biases from spatial variation (Hayes
2000). We placed each point at least 400 meters from other points.
Acoustic Surveys
We used acoustic recording equipment to survey for bonneted bats. Acoustic
methods are especially effective for surveying bat species with distinguishable
echolocation calls (Hayes 2000; Hayes et al. 2009). Bonneted bats have a low
frequency call that is easily distinguishable from that of other species of bats in south
Florida (Belwood 1992) and can be detected at ranges of up to approximately 30 meters
(Marks and Marks 2012).
We used SM2BAT+ detectors with SMX-US microphones (Wildlife Acoustics,
Maynard, MA) for acoustic surveys. All detectors were set to record continuously from
15 minutes before sunset to 15 minutes after sunrise. We mounted the microphone of
each detector on a fiberglass painter’s pole that extended to a height of 3.4 meters. All
microphones were mounted horizontally with a slight downward tilt to minimize the risk
of damage from weather (Wildlife Acoustics 2013). We used an ultrasonic calibrator
(Wildlife Acoustics, Maynard, MA) to test all microphones once per week. We replaced
microphones when the calibrator read -36 dB (as per the recommendations provided by
Wildlife Acoustics).
We surveyed from 20 January to 13 June 2014 and from 13 January to 12 May
2015. We visited each cell 3 times per year; separating each visit by at least 3 weeks.
During each visit, the detectors recorded bat activity for 2 to 3 nights depending on
logistics. We placed a detector within a 100 meter buffer of each of the 5 random points
generated in each cell during site selection. We placed detectors at locations within
each buffer that would maximize the probability of detecting bats (e.g. more open areas
49
where calls could propagate farther; Ormsbee 2010). The majority of points in
developed and agricultural cells were located on private land. If access permission was
denied at the original point, we moved to the next closest property in a sequential
manner until permission was granted to set the equipment.
During 2014, we set out a detector at each of the 5 points per cell during the first
visit. Equipment and logistical issues resulted in only 4 points in each cell being
sampled during the second and third visits. When only 4 detectors were deployed in
each cell, we randomly rotated the 1 point that was not sampled, ensuring that we
sampled each point for a minimum of 4 nights over the course of the year. In 2015, we
followed the same protocol as in 2014, but sampled all 5 points during all visits to each
cell.
Covariate Sampling
Detection covariates
Many bat species are more active as temperatures increase (Rodhouse et al.
2015). To quantify the influence of temperature on bonneted bat detection, we
programed the SM2+ detectors to record the external temperature every 15 minutes
while deployed and recorded the minimum temperature from each survey night
(surveytemp). To account for potential changes in detectability resulting from seasonal
movements, phenology and increases in detection as juveniles became volant during
the study period, we converted the date of each survey to Julian date (Jdate) and used
it as a covariate.
Occupancy Covariates
We used ArcGIS 10.1 to derive a number of metrics to reflect patterns of land
cover and land use. We estimated the average canopy cover (cc) in each grid cell using
50
the National Land Cover Dataset (NLCD) developed by the U.S. Geological Survey
(NLCD 2011). We used the Spatial Statistics toolset in ArcGIS 10.1 to measure the
average canopy cover within each grid cell sampled. We also used the ‘Tabulate Area’
tool in the Spatial Analyst toolset to determine the area of each grid cell dominated by
pine (Pinus spp.), as identified by FNAI.
We broke the 4 categories into 8 land cover classes because the land cover
categories used during site selection were broad and had considerable variation within
categories. Agricultural lands included improved pasture and rangeland that maintained
many characteristics of more natural environments and also heavily managed crop-
fields. To account for this variation we divided the agricultural land cover into rangeland
and crop-dominated agriculture. We selected all cells that had previously been identified
during site selection as dominated by ‘agriculture’ based on FNAI categories. We
identified all patches that were classified as having a site-level land cover of ‘improved
pasture’ or ‘unimproved/woodland pasture’ as being rangeland (FNAI 2012). We then
separated uplands into non-forested uplands (i.e., dry prairies, palmetto prairies, scrub,
shrub and brushland and barren and outcrop communities identified by FNAI) and
forested uplands (see Table 3-1). We also separated wetlands into 3 different
categories: open water (communities identified by FNAI as lacustrine, palustrine,
riverine and estuarine), non-forested wetlands (communities identified by FNAI as
freshwater non-forested wetlands, freshwater marshes, and non-vegetated wetlands),
and forested wetlands (see Table 3-1). We used the ‘Tabulate Area’ tool in the Spatial
Analyst toolset to calculate the percent of each cell that was covered by forested
wetlands (forest.wetland), non-forested wetlands (non.wetland), forested uplands
51
(forest.upland), non-forested uplands (non.upland), crop-dominated agricultural areas
(ag), range-lands (range), developed areas (developed), and open water (water).
We used the National Oceanic and Atmospheric Administration’s 1981 – 2010
climate normals available from the Southeast Regional Climate Center (SERCC) to get
historical climate data for the areas sampled. We determined the average annual
minimum temperature (mintemp), average annual precipitation amounts (precip),
average spring (January – May) minimum temperatures (spring.mintemp), and average
spring precipitation amounts (spring.precip) from the climate center nearest each survey
point.
Data Analysis
We analyzed all recorded files in Kaleidoscope Pro 3.1.0 using the ‘Bats of
Florida 3.1.0’ classifier (Wildlife Acoustics; Maynard, MA). Kaleidoscope Pro compares
recorded calls to a known call library that contains between 1,631 and 130,435 calls for
each species (Wildlife Acoustics 2015). Using a set of trained Hidden Markov Models
(HMMs) for each species, Kaleidoscope Pro calculates the prior probability that the call
in question was generated by each species (Agranat 2012). Kaleidoscope Pro then
calculates a Fisher score for each call sequence by looking at HMM-generated
identification of each call in the sequence. The Fisher score identifies what call
parameters are most significant for identification, and then multiple pairwise
comparisons are performed. These comparisons provide an output including what bat
species the software believes produced each call sequence and an associated
confidence factor. If the normalized confidence factor exceeds a certain threshold, the
file is classified to species; if the confidence factors do not exceed the threshold or if the
52
most likely species class is the unidentifiable class, the file is labeled as ‘NoID’ (Agranat
2012).
We programmed Kaleidoscope Pro to identify to species each call sequence (a
series of at least 2 calls with less than a 5 second gap between them; Britzke et al.
2002). In addition, we manually looked at all calls identified by Kaleidoscope as either
bonneted bat or NoID to ensure that no bonneted bat calls were missed (Mona Doss,
Wildlife Acoustics, pers. comm.). We classified calls with a minimum frequency between
10 and 18 kHz and a maximum frequency between 16 and 22 kHz as bonneted bat
calls, as per recommendations by experts (Bruce Miller, Bat Sound Services, pers.
comm.).
We used Bayesian hierarchical occupancy models to investigate the distribution
of the bonneted bat and elucidate the influence of environmental factors on its
occurrence throughout southern Florida. We used a Bayesian approach to integrate
hierarchical effects that address potential spatial autocorrelation between points in each
grid cell. We also accounted for imperfect detection when estimating the state of
occupancy (Royle and Dorazio 2008). We modeled the observation process as
conditional on the latent occupancy state, yj(i) | z(i,k) ~ Bern(z[i]*pij) where pij is the
probability of detection during survey j given presence at point i, under a Bernoulli
distribution. yj(i) represents the detection history, with each taking a value of 1 or 0
representing “detection” or “non-detection” for survey j at site i. In order to speed
Markov chain Monte Carlo (MCMC) convergence and improve interpretability, we
standardized all covariates (Gelman and Hill 2007). We did not include correlated
covariates (r2 > 0.60, Rodhouse et al. 2015) in interactive models.
53
We created models by first holding occupancy constant at the latent occupancy
state and modeled the effects of survey.temp and Jdate on detection. We used JAGS v
3.4.0 (Plummer 2003) launched from RStudio v.0.98 with the R2jags library (Su and
Yajima 2015) to implement Bayesian estimation of model parameters via MCMC
samples of posterior distributions. We input each covariate into the model as a random
effect using vague, normally distributed [N(0, 0.01)] priors on all logit-scale parameters
(Kery and Royle 2015). Posterior summaries were based on 10000 MCMC samples of
the posterior distributions from 3 chains run simultaneously, thinned by a factor of 10,
following an initial burn-in of 2000. We assessed convergence of MCMC chains with
trace plots and the Gelman-Rubin diagnostic (R̂); convergence was reached for all
parameters according to the criteria | R̂ – 1| < 0.1 (Ntzoufras 2009).
We evaluated models using a stepwise comparison predictive performance of
each model, removing covariates where the 95% credible intervals crossed zero and re-
running the model after each covariate was removed (Kery and Schaub 2012). This
approach was used rather than the deviance information criterion (DIC) because it is
more robust when evaluating predictive performance of models (Gelman et al. 2013).
Once a final model was selected for detection, we evaluated occupancy with pine,
mintemp, spring.mintemp, precip, spring.precip, developed, forest.upland, non.upland,
forest.wetland, non.wetland, range, ag and water as covariates. Again, we used a
stepwise approach, holding detection constant as a function of the best-fitting model
found above. We reported all beta estimates and credible intervals from the final
selected model and graphed the effects of all covariates that were included in the final
model.
54
Results
We sampled a total of 330 points (66 cells) during 2014 and 2015 (Figure 3-3).
We recorded over 500 bonneted bat call sequences at 60 of these points. Bonneted
bats were detected within 10 counties in southern Florida (Broward, Charlotte, Collier,
Desoto, Glades, Hendry, Highlands, Lee, Miami-Dade and Polk). The best model for
detection included surveytemp and Jdate as covariates (Table 3-3). Julian date (Jdate)
had the strongest effect on detection, with bonneted bats being more likely to be
detected later in the year [β = 0.70 (95% credible interval: 0.42 – 0.80)] (Figure 3-4).
Minimum temperature of each survey night (surveytemp) had a positive effect (Figure 3-
5) on detection probabilities of bonneted bats [β = 0.26 (95% credible intervals: 0.03 to
0.50)]. Overall nightly detection probability of bonneted bats was 0.29 (0.23 to 0.35).
Based on the estimated detection probabilities, it would take between 9 and 10 survey
nights to determine with 95% certainty whether bonneted bats are present at a site
(Figure 3-6).
The best model for occupancy included developed, precip, spring.mintemp,
spring.precip, and ag as covariates that influenced bonneted bat occupancy
probabilities (Table 3-3). Developed areas (developed) had the largest effect [β = -1.20
(-1.92 to -0.63)] (Table 3-3), with occupancy probability decreasing with increasing
amount of development in the grid cell (Figure 3-7, Figure 3-8). Precipitation (precip)
had the second largest effect [β = 0.87 (0.42 to 1.32)] (Table 3-3), with occupancy
probability of bonneted bats increasing with increasing average annual precipitation
levels recorded between 1982 and 2010 (Figure 3-9). Average spring precipitation
levels (spring.precip) had a negative effect (Figure 3-10) on bonneted bat occupancy
probability [β = -0.69 (-1.10 to -0.32)] (Table 3-3). Average spring minimum temperature
55
(spring.mintemp) had a relatively large effect [β = 0.74 (0.26 to 1.25)] (Table 3-3), with
occupancy probability increasing with average spring minimum temperature recorded
between 1982 and 2010 (Figure 3-12, Figure 3-13). Ag had a positive effect on
bonneted bat occupancy [β = 0.52 (0.21 – 0.85)] (Table 3-3), with occupancy
probabilities increasing with the amount of crop-based agriculture in the grid cell (Figure
3-14, Figure 3-15). The final model estimated that bonneted bats were present at about
77 (66 – 91) sites, compared to our observed 60 sites (Figure 3-16).
Discussion
This study is the first to use a systematic and rigorous sampling design (Hayes
1997, USFWS 2013) to investigate the distribution and habitat use of the Florida
bonneted bat. We found that the range of the bonneted bat is larger than previously
believed. Although distribution cannot be directly linked to abundance, it is considered a
reliable method for assessing population status at broad spatial scales (Holt et al. 2002,
MacKenzie et al. 2005, Jones 2011).Overall, bonneted bats were estimated to be
present in > 20% of our study area, indicating that these bats may be more common
than originally thought (Marks and Marks 2012, USFWS 2013).
Bonneted bats were found using all land cover types investigated in this study;
however, increased development in the grid cell appeared to negatively influence
occupancy probabilities. Other studies have suggested that the high wing-loading of
Molossids has resulted in them being pre-adapted for foraging in urban areas, where
they can successfully exploit concentrations of insects that occur around street-lamps
(Scanlon and Petit 2008, Threlfall et al. 2011). Despite this, Avila-Flores and Fenton
(2005) noticed a similar trend to us with the western mastiff bat (E. perotis) in Mexico
City, Mexico. Although the western mastiff bat had previously been captured in the city,
56
the authors only recorded the species acoustically in natural forest surrounding the city
(Avila-Flores and Fenton 2005). They hypothesized that the larger Molossids, such as
those in the genus Eumops, are not maneuverable enough to exploit insect prey amid
the structural clutter that comes with development. This could explain why bonneted
bats were negatively associated with urbanization, even while other Molossids have
shown positive trends with increasing urbanization (Jung and Threlfall 2016).
In contrast to development, crop-based agriculture is largely void of structural
clutter (Williams-Guillen et al. 2016). In addition, while insect abundance in agricultural
areas varies based on use of pesticides, well-irrigated crop fields often foster an
abundance and diversity of insects, particularly in areas that frequently experience
droughts similar to the dry season of southern Florida (Noer et al. 2012). Increased
occupancy probabilities of bonneted bats in agricultural areas is likely a result of the
combination of insect abundance in these areas and the ability of bats to successfully
forage in these open areas and exploit these populations. Similar trends have been
observed in other species of Molossids, including Brazilian free-tailed bats (Tadarida
brasiliensis) (Cleveland et al. 2006), little free-tailed bats (Chaerephon pumilus) and
Angolan free-tailed bats (Mops condylurus) (Noer et al. 2012).
The tropical climate of southern Florida allows the area to support a large
number of Caribbean and Neotropical species (Webb 1999). The steep decline in
occupancy probabilities of the bonneted bat when historical average spring minimum
temperatures dropped below 15°C provides evidence that the bonneted bat is limited to
this area as a result of the climate. This is not surprising, considering that the genus
57
Eumops is primarily distributed throughout the Neotropics (Best et al. 1997, IUCN
2015).
Interestingly, bonneted bats were more common in areas with higher average
annual rainfall amounts, but seemed to prefer areas with less spring rainfall. Productivity
is positively correlated with precipitation (Williams 1951, Silva et al. 2011), and thus it is
not surprising that bonneted bats were found more commonly in areas with higher
precipitation levels. The negative association between bonneted bat occurrence and
high spring precipitation levels may have to do with the timing of precipitation events.
The vast majority of rainfall in southern Florida occurs during the summer wet season
(Webb 1999), when maximum rainfall amounts occur during the early to mid-afternoon
(Schwartz and Bosart 1979). Maximum rainfall during the dry season (winter and spring)
occurs in the early to late evening (Schwartz and Bosart 1979). This peak roughly
corresponds with the peak insect activity (Kunz 1973). Nighttime precipitation has been
shown to decrease the activity levels of species of bats (Fenton 1977, Kunz 1973), likely
as a result of increased thermoregulatory costs and decreased prey (Burles et al. 2009).
However, when we ran a similar analysis on other species of bats recorded in southern
Florida, we did not see the same trends in any other species (Appendix 3-1). It is likely
that bonneted bats are responding to some other variable that we did not explicitly
measure, which is correlated with spring precipitation levels.
Southern Florida is an area that is expected to undergo dramatic changes in the
future. The population of Florida is expected to grow by nearly 20,000,000 people by
2060, and the landscape of southern Florida is expected to become mostly urbanized to
account for the growing population (Zwick and Carr 2006). Southwest Florida and
58
south-central Florida are expected to undergo the most extreme land cover changes in
the state, with nearly all agricultural and natural lands predicted to be converted to
development (Zwick and Carr 2006). If these land use changes occur, bonneted bats
could see a dramatic contraction of suitable foraging areas.
These future scenarios highlight the importance of continued monitoring of the
bonneted bat to document potential shifts in the range and status of this endangered
species. Future monitoring efforts should leave detectors in an area for at least 10
nights to have a 95% chance of detecting bonneted bats if they are present. Our results
suggest that future monitoring efforts should be focused on warm nights later in the
spring to maximize detection probabilities. The higher detection probabilities observed
in the later months of both field seasons are likely a result of increased insect
abundance due to increased temperatures, humidity and precipitation (Williams 1951)
influencing the activity levels of bonneted bats. Unfortunately, we cannot reliably
extrapolate our results beyond May. While it is possible that bonneted bat detection
probabilities are higher during the summer months, this also corresponds to the wet
season in southern Florida. During this time of year, many areas are inaccessible due to
high water levels (Webb 1999), limiting the utility of sampling during this period.
This study helps to fill some of the data gaps in our knowledge of the bonneted
bat. By having a better understanding of the species distribution, we can now start to
investigate more fine-scale patterns influencing their occurrence and habitat selection
patterns. In addition, we can more accurately assess how certain management activities
will influence the bonneted bat. This study provides a solid foundation upon which to
build future research on this endangered species.
59
Figure 3-1. Tentative range map of the Florida bonneted bat, as developed by acoustic surveys conducted from 2006 – 2012. Adapted from Marks and Marks (2012).
60
Table 3-1. Florida Natural Areas Inventory (FNAI) state-level land cover types grouped into broad land covers used during site selection for acoustic surveys conducted for bonneted bats throughout southern Florida.
Broad Land Cover Type
Agriculture Developed Upland Wetland
Agriculture Low intensity urban Upland hardwood forest* Freshwater non-forested wetlands
Row crops High intensity urban Mesic hammock* Freshwater marshes
Field crops Rural Rockland hammock* Cypress/tupelo*
Improved pasture1 Transportation Scrub Cypress*
Unimproved/woodland pasture1 Communication Upland mixed woodland* Other coniferous wetlands*
Citrus Extractive Upland coniferous* Hardwood wetlands*
Fruit orchards Utilities Sandhill* Non-vegetated wetlands
Vineyard and nurseries Dry flatwoods* Wet coniferous plantation*
Pine rockland* Lacustrine2
Dry prairie Riverine2
Palmetto prairie Estuarine2
Hardwood forest* Exotic wetland hardwoods*
Shrub and brushland
Coastal uplands*
Barren and outcrop
Coniferous plantations*
*forested upland or wetland 1 rangeland 2 open water
61
Figure 3-2. Koppen-Geiger Climate Classification map. Aw (pink) is an equatorial desert climate, Am (red) is an equatorial monsoonal climate, Af (dark red) is an equatorial humid climate and Cfb (green) is a temperate humid climate (Kottek et al. 2006).
62
Table 3-2. Description of covariates used in occupancy models for the Florida bonneted bat in southern Florida.
Variable Description
Non.upland Percent of grid cell covered by non-forested uplands.
Forest.upland Percent of grid cell covered by forested uplands.
Non.wetland Percent of grid cell covered by non-forested wetlands.
Forest.wetland Percent of grid cell covered by forested wetlands.
Ag Percent of grid cell covered by agriculture (row crops, citrus).
Range Percent of grid cell covered by rangeland.
Developed Percent of grid cell covered by developed areas.
Water Percent of grid cell covered by open water.
CC Canopy cover at each site.
Pine Distance of each site from the nearest pine-dominated area.
Spring.precip Average precipitation from January - May (from NOAA 1981 - 2010 climate normals).
Precip Average annual precipitation (from NOAA 1981 - 2010 climate normals).
Mintemp Average annual minimum temperature (from NOAA 1981 - 2010 climate normals).
Spring.mintemp Average minimum temperature from January - May (NOAA 1981 - 2010 climate normals).
63
Figure 3-3. Land cover map of our study area in southern Florida. Black squares represent grid cells that were acoustically sampled for bonneted bats from January – June 2014 and January – May 2015.
64
Table 3-3. Beta estimates of final model predicting bonneted bat occupancy, including variables affecting detection (p) and occupancy (psi). Bonneted bat occurrence was modeled from acoustic survey data collected in southern Florida during January - June of 2014 and January – May 2015.
Variable Estimate Standard Deviation
95% Credible Interval
Lower Upper
Julian date (p) 0.70 0.15 0.42 0.80
Survey temperature (p) 0.26 0.12 0.03 0.50
Developed (psi) -1.20 0.30 -1.82 -0.63
Agriculture (psi) 0.52 0.16 0.21 0.85
Precipitation (psi) 0.87 0.24 0.42 1.35
Spring precipitation (psi) -0.69 0.20 -1.10 -0.32
Spring min temperature (psi) 0.74 0.25 0.26 1.25
65
Figure 3-4. Effect of Julian date on detection probability of bonneted bats acoustically
surveyed in southern Florida from January – June 2014 and January – May 2015. Black lines show the posterior mean, and gray lines show the relationships based on a random posterior sample of size 200 to visualize estimation uncertainty.
66
Figure 3-5. Effect of minimum temperature during each survey night on detection
probability of bonneted bats acoustically surveyed in southern Florida in January – June 2014 and January – May 2015. Black lines show the posterior mean, and gray lines show the relationships based on a random posterior sample of size 200 to visualize estimation uncertainty.
67
Figure 3-6. The relationship between P*, the probability to detect bonneted bats at a site
acoustically at least once during n surveys. The dashed line indicates 95% certainty to detect the species when present.
68
Figure 3-7. Effect of the percent of grid cell classified as developed on the occupancy
probability of bonneted bats acoustically surveyed in southern Florida in January - June 2014 and January – May 2015. Black lines show the posterior mean, and gray lines show the relationships based on a random posterior sample of size 200 to visualize estimation uncertainty.
69
Figure 3-8. Map of percent of each grid cell in the study area covered by development.
Sampled cells represent cells surveyed acoustically for the bonneted bat between January – June 2014 and January – May 2015.
70
Figure 3-9. Effect of average annual precipitation on the occupancy probability of
bonneted bats acoustically surveyed in southern Florida in January - June 2014 and January – May 2015. Black lines show the posterior mean, and gray lines show the relationships based on a random posterior sample of size 200 to visualize estimation uncertainty.
71
Figure 3-10. Effect of average annual spring precipitation on the occupancy probability
of bonneted bats acoustically surveyed in southern Florida in January - June 2014 and January – May 2015. Black lines show the posterior mean, and gray lines show the relationships based on a random posterior sample of size 200 to visualize estimation uncertainty.
72
Figure 3-11. Map of average annual precipitation from 1981 – 2010 of each grid cell
within our study area. Sampled cells represent cells surveyed acoustically for the bonneted bat between January – June 2014 and January – May 2015.
73
Figure 3-12. Effect of the average spring minimum temperature on the occupancy
probability of bonneted bats acoustically surveyed in southern Florida in January - June 2014 and January – May 2015. Black lines show the posterior mean, and gray lines show the relationships based on a random posterior sample of size 200 to visualize estimation uncertainty.
74
Figure 3-13. Map of average spring minimum temperature from 1981 – 2010 of each
grid cell in the study area. Sampled cells represent cells surveyed acoustically for the bonneted bat between January – June 2014 and January – May 2015.
75
Figure 3-14. Effect of the percent of grid cell classified as crop-dominated agriculture on
the occupancy probability of bonneted bats acoustically surveyed in southern Florida in January - June 2014 and January – May 2015. Black lines show the posterior mean, and gray lines show the relationships based on a random posterior sample of size 200 to visualize estimation uncertainty.
76
Figure 3-15. Map of percent of each grid cell in the study area covered by crop-
dominated agriculture. Sampled cells represent cells surveyed acoustically for the bonneted bat between January – June 2014 and January – May 2015.
77
Figure 3-16. Posterior distribution of the number of sites occupied by the bonneted bat based on acoustic surveys conducted in southern Florida in January – June 2014 and January – May 2015. Vertical line indicates the observed number of 60 occupied sites.
78
CHAPTER 4 ECOLOGY OF BATS IN HUMAN-DOMINATED HABITATS IN SOUTHERN FLORIDA
Synopsis
A growing global human population has converted large amounts of lands for
agricultural production and development. Changes in land cover has had drastic effects
on wildlife communities across the globe. The threat of land use change leading to
alteration in wildlife communities is pronounced in southern Florida, which has one of
the fastest growing human populations in the United States. We investigated community
and species-specific effects of human-dominated landscapes on bats across southern
Florida. We predicted that bat species richness would decline with increasing levels of
human influence, and that species with broad wings and low wing-loading would be
most vulnerable to anthropogenic land uses. Our study found significant effects of
agricultural intensification and urban development on all species of bats across the
landscape. Occupancy probabilities of all species were negatively correlated with
increasing amounts of human influence; this trend was highly variable between different
species. Within human-dominated areas, the loss of tree canopy cover and distance
from undeveloped areas appeared to have the greatest effects on the bat community.
Background
Humans have drastically altered natural processes of our planet, creating a new
geological epoch, the ‘Anthropocene’, or human epoch (Crutzen and Stoermer 2000).
As the human population has grown, few natural ecosystems have remained
untouched. Agricultural land uses cover nearly 40% of our planet (Food and Agriculture
Association [FAO] 2011, Robertson and Swinton 2005, Power 2010). While urban areas
do not cover as much surface area, they support the vast majority of the world’s human
79
population, and have a disproportionally large and negative impact on the surrounding
environment (Grimm et al. 2008).
In the United States, habitat conversion resulting from the development of urban
and agricultural areas has been considered the number one threat to wildlife (Wilcove et
al. 1998). This threat is particularly pronounced in southern Florida, which has
previously been identified as the most endangered landscape in the United States
(Noss et al. 1995). Since 1920, the human population in southern Florida has risen
exponentially (Solecki 2001). As of 2014, there were close to 8 million people living
within 4 metropolitan areas in the region, with the Miami-Fort Lauderdale-West Palm
Beach area accounting for nearly 6 million (U.S. Census Bureau 2015). This growth has
caused a decline in natural vegetation, with approximately 90% of Florida’s forests
altered or destroyed (Wear and Greis 2002). Wear and Greis (2011) predicted that
peninsular Florida will lose more forested land than any other area in the southern
United States by 2060.
The mild climate of the region also lends itself to a year-round growing season.
Over two-thirds of the nation’s winter and spring vegetables are grown in southern
Florida (Kranzer et al. 1995) and this area produces more sugarcane than anywhere
else in the U.S. (Baucum et al. 2006). Nearly 90% of Florida’s citrus is grown in
southern Florida (Webb 1999). As a result of predicted population growth, southern
Florida is expected to lose most of its agriculture and natural areas and become mostly
urbanized (Zwick and Carr 2006).
Increasing urbanization and agricultural intensification typically leads to
decreased richness and homogenization of wildlife and plant species (McKinney 2008).
80
A large body of research has focused on the effects of anthropogenic habitat conversion
on birds (Garden et al. 2006), but there is still a paucity of information for many other
taxa (Garden et al. 2006, McKinney 2008). Bats make a good model to study the effects
of human-dominated landscapes on wildlife communities. Their slow life histories make
them good bioindicators (Fenton 2003), and they provide key ecosystem services such
as the control of insect populations (Kunz and Fenton 2003).
The responses of bats to urbanization and agricultural intensification appear to
be mostly negative, with most studies documenting decreased activity and species
richness of bats in areas with increasing human influence (Gaisler et al. 1998, Legakis
et al. 2000, Lesinski et al. 2000, Jung and Threlfall 2016). Development is often
considered to be the more severe of the 2, often resulting in complete alterations of the
landscape (McIntyre and Hobbs 1999, Jung and Threlfall 2016). Responses of bats to
urbanization are largely species-specific, with some species avoiding all urban areas
and others potentially benefiting from urbanization (Avila-Flores and Fenton 2005, Jung
and Kalko 2011). Mobility explains most of the variation in responses of different
species of bats (Bader et al. 2015). In bats, mobility and maneuverability is directly
related to wing shape and the ratio between bat body size and wing area (Norberg and
Rayner 1987). Species with low wing-loading are often slower but more maneuverable,
while species with high wing-loading are less maneuverable but capable of flying fast
(Norberg and Rayner 1987). In general, broad-winged species with a low aspect ratio
are most sensitive to urbanization, with some species disappearing altogether from
areas with even low density development (Gonsalves et al. 2013, Luck et al. 2013). In
contrast to this, species with high wing-loading that are adapted for fast flight in open
81
areas typically have a weaker response to urbanization (Avila-Flores and Fenton 2005,
Jung and Kalko 2011). Some studies have even found that certain species with high
wing-loading, most often species in the family Molossidae, appear to benefit from
urbanization (Avila-Flores and Fenton 2005, Jung and Kalko 2011). These species are
well-adapted for foraging in sparsely vegetated, urban landscapes, and have been
observed exploiting concentrations of insects around streetlights (see reviews in Jung
and Threlfall 2016 and Rowse et al. 2016).
The responses of bats to agriculture are generally considered to be less severe
than responses to urbanization (Williams-Guillen et al. 2016), but the effect is largely
dependent on the type of agriculture. It has been suggested that open agricultural areas
with few natural features can be as impermeable to some species of bats as open water
(Ekman and de Jong 1996). However, when some canopy cover is retained, the bat
community within agricultural areas can be indistinguishable from surrounding natural
areas (Williams-Guillen et al. 2016). Similar to urbanization, species with broad-wings
that are adapted for foraging in cluttered environments tend to be the most sensitive to
agricultural intensification. Species with higher wing-loading that forage along edges
and in open areas are often less affected, with some species even preferentially
selecting to forage over agriculture (Noer et al. 2012).
The mechanisms behind bat species responses to urbanization and agricultural
intensification are poorly understood. Natural areas often have higher insect diversity
and abundance compared to altered landscapes (Avila-Flores et al. 2005), and provide
more abundant roosting habitat for species that roost in tree cavities and foliage
(Duchamp et al. 2004). The loss of canopy cover and connectivity to natural areas that
82
often occurs with urbanization and agricultural intensification presumably affects many
species of bats (Fuentes-Montemayor et al. 2011). These effects are thought to be most
severe for species with low wing-loading with limited ability to cross developed and
agricultural landscapes during foraging bouts (Norberg and Rayner 1987, Jung and
Threlfall 2016). For many different species of mammals, roads are considered a major
barrier to movement (Forman and Alexander 1998, Berthinussen and Altringham 2012).
The volant nature of bats presumably makes them less susceptible to this particular
threat, but there is still some evidence that roads with high traffic volume may be
avoided by certain bat species (Kerth and Melber 2009, Berthinussen and Altringham
2012). Additionally, slower bats with low wing-loading that forage near the ground may
be more susceptible to collisions with vehicles (Lesinski 2007). Increased nighttime light
levels in urban areas may also influence bats in human-dominated landscapes. Certain
species of bats are known to avoid light, presumably to reduce predation risk (see
review in Rowse et al. 2016). However, it has been suggested that the ability of
streetlights to concentrate insects is one of the main reasons why certain species
appear to benefit from urbanization (see review in Rowse et al. 2016).
The majority of studies to date have focused primarily on responses of bats to
either agriculture or urbanization, with few studies examining the responses of bats to a
landscape mosaic (Duchamp et al. 2004). Additionally, most prior research has focused
on one metropolitan or agricultural area rather than over the broad landscape. Since
bats are capable of traveling large distances while foraging (Kerth et al. 2001,
Bontadina et al. 2002), landscape-scale effects are likely the most telling when looking
at bat distributions in human-dominated areas (Gehrt and Chelsvig 2003). An additional
83
flaw of many previous studies is the fact that they often focused on overall bat activity or
on one species of bat, rather than on species-specific responses across the bat
community. It is clear that different bat species respond to agriculture and development
in drastically different ways (see reviews in Williams-Guillen et al. 2016, Russo and
Ancillotto 2014), and trends can be lost when species are pooled. Coleman and Barclay
(2011) found higher bat activity in urban areas than in surrounding rural and natural
areas; however, when they looked at species-level responses, they found that only little
brown bats (Myotis lucifugus) activity increased in urban areas, while all other species
declined. While we would expect to be able to explain some general trends based on
bat mobility, even species with similar ecomorphology have been found to have varying
responses to agricultural intensification and levels of development (Luck et al. 2013,
McConville et al. 2014), further highlighting a need for species-specific investigations on
the effects of human-dominated landscapes on bats.
The goal of this study was to investigate the responses of the bat community and
of individual species to broad-scale patterns of development and agriculture across the
south Florida landscape. We also investigate potential mechanisms that may influence
distribution trends within human-dominated landscapes. Our objectives were to (1)
compare bat species richness among crop-dominated agriculture, rangeland, developed
areas, and undeveloped ecosystems (wetlands and uplands) in southern Florida, (2)
investigate species-specific use of human-dominated landscapes throughout southern
Florida, and (3) examine species-specific responses to potential mechanisms affecting
bat distribution in human-dominated landscapes (fine-scale land use, canopy cover,
distance to undeveloped habitat, distance to roads, and levels of nighttime light). We
84
predict that species richness will be lower in agricultural and developed areas compared
to undeveloped areas, such as wetlands and uplands. In particular, we hypothesize that
species with lower wing-loading will be negatively affected by the amount of
development and agriculture across the landscape, while Molossids with high wing-
loading will be least affected. We predict that species with average to high wing-loading
that typically forage along edges will have neutral or slightly negative responses to
agricultural intensity and development. Within human-dominated landscapes, we predict
that canopy cover will be especially important to clutter-adapted species with lower
wing-loading. We also predict that all species will increase with proximity to
undeveloped areas and as distance to roads increases. Finally, we predict that
Molossids will be positively impacted by increased levels of nighttime light, while
species with low wing-loading will be negatively impacted.
Methods
Study Species
There are 10 species of bats that are resident to southern Florida. These include
2 species in the Molossidae family and 8 in the Vespertilionidae. The 2 Molossids are
the Florida bonneted bat (Eumops floridanus, EUFL) and the Brazilian free-tailed bat
(Tadarida brasiliensis, TABR). These species have long, narrow wings and high wing-
loading; they are well-adapted for foraging in open environments (Norberg and Rayner
1987). The Florida bonneted bat is endemic to southern Florida. The species is listed
as federally endangered under the Endangered Species Act (ESA) and is considered
critically endangered by the International Union for the Conservation of Nature (IUCN)
(FWS 2013, IUCN 2015).The Brazilian free-tailed bat is distributed throughout the state
and is considered a species of least concern by the IUCN (IUCN 2015).
85
All of the Vespertilionids found in southern Florida are considered species of
greatest conservation need (SGCN) in Florida (FWC 2005). These species include the
Northern yellow bat (Lasiurus intermedius, LAIN), Seminole bat (L. seminolus, LASE),
red bat (L. borealis, LABO), big brown bat (Eptesicus fuscus, EPFU), Rafinesque’s big-
eared bat (Corynorhinus rafinesquii, CORA), southeastern myotis (Myotis
austroriparius, MYAU), tri-colored bat (Perimyotis subflavus, PESU) and evening bat
(Nycticeius humeralis, NYHU). The southeastern myotis, red bat, Rafinesque’s big-
eared bat and big brown bat are considered rare in southern Florida, with the other
species found throughout the state. Rafinesque’s big-eared bat and the tri-colored bat
both have broad wings and low wing-loading, allowing them to forage in cluttered
environments (Norberg and Rayner 1987). The yellow bat, Seminole bat, big brown bat
and evening bat have relatively long wings and are built for foraging along edges and in
more open environments (Norberg and Rayner 1987).
Study Area
We surveyed for bats in 16 counties in southern Florida: Polk, Osceola,
Highlands, Okeechobee, Sarasota, Desoto, Charlotte, Glades, Martin, Lee, Hendry,
Palm Beach, Collier, Broward, Monroe and Miami-Dade. South Florida has a subtropical
to tropical climate, with a wet season during the summer months and a dry season that
extends from mid-fall through late spring (Duever et al. 1994). The major source of
rainfall is thunderstorms, with the average annual precipitation being nearly constant for
the past 100 years (Duever et al. 1994). The highest precipitation amounts are found
along the east coast, with Hialeah recording an average of 178.77 centimeters of
precipitation each year compared to an average of 128.22 centimeters in Venice, which
is along the west coast (FSU Climate Center 2014).
86
The average temperatures are warm all year, only rarely dropping below
freezing; the northern counties (Polk and Osceola) have an average of 1 – 2 freeze
days annually, while the rest of the study area averages 0 days where the minimum
temperature drops below freezing (Southeast Regional Climate Center 2015). The
warmest month on average over the entire study area is August, with average
temperatures around 28°C in the northern part of the study area and between 28.5°C
and 29°C to the south of Lake Okeechobee. The coolest month is February; the coolest
temperatures (about 15.5°C) occurring in the interior, northern part of the study area.
Land-uses in our study area ranged from heavily urbanized to remote natural
areas. Major vegetation types within natural areas included pine forest, cypress forest,
mangrove swamps, sloughs, and marsh-prairies (Webb 1999). The southern portion of
the study area was dominated by the cypress (Taxodium distichum), pine (Pinus spp.)
and hardwood forests of the Big Cypress subregion and the Everglades, the vast
shallow wetland that historically stretched across much of the Florida peninsula south of
Lake Okeechobee (Webb 1999). Some of the interior natural communities within the
Everglades included forested wetlands, freshwater marshes and upland rockland
communities (Gunderson 1994, Webb 1999).
There were a wide variety of agricultural land use throughout our study area
(Figure 4-1). The Everglades Agricultural Area, dominated by sugarcane production
(Whalen and Whalen 1996), encompassed nearly 300,000 ha of land to the south of
Lake Okeechobee (Rice et al. 2002). The northern shores of Lake Okeechobee have
been predominantly used for cattle ranching (Webb 1999). Approximately half of
agricultural land in southern Florida consists of rangeland for cattle (Webb 1999), and
87
many wildlife management areas serve as improved and unimproved cattle pastures
(FWC 2013). The areas surrounding Lake Okeechobee produced the most citrus in the
state (US Department of Agriculture [USDA] 2014). Miami-Dade and Collier counties in
the southern portion of our study area were among the leading producers of tomatoes in
the country (USDA 2014).
Both the east and west coasts of the study area were dominated by urban areas,
with the highest intensity development occurring on the east coast (Figure 4-2). Housing
densities decreased as you moved away from the coast, becoming mostly rural in the
interior portion of the state (Mulkey 2007, FWC 2013). Our sampling included areas
across the urban matrix from rural to suburban to high density urban areas. Nighttime
light levels were highly correlated with amount of development, with highest ambient
light levels occurring in developed areas along the coasts. Roads crisscross most of the
southern Florida landscape. Road densities were highest in developed areas along the
coast, but many roads in the interior portion of the state (eg. Alligator Alley) had a high
volume of traffic (Florida Department of Transportation [DOT] 2015).
Site Selection
We established a grid system comprised of 5 km by 5 km cells that we overlaid
across southern Florida (Ormsbee 2010, Loeb 2015), similar to the Bat Grid and North
American Bat Monitoring Program (NABAT) survey protocols (Hayes et al. 2009, Loeb
2015). We overlaid this grid across our study area using ArcMap 10.1 (ESRI; Redlands,
CA). We selected cells using a stratified random sampling design to ensure a
representative sample of bat activity across all major land cover types in southern
Florida. We stratified land uses using the cooperative land cover map developed by the
Florida Natural Areas Inventory [FNAI] (FNAI 2012). We simplified the FNAI
88
classifications into four major categories: agriculture, developed, upland and wetland
(Table 4-1). We used the ‘Tabulate Area’ feature in the Spatial Analyst toolbox in
ArcMap to determine the dominant land cover type used to classify each grid cell.
We randomly selected 17 grid cells of each of the 4 land cover types for sampling
during two field seasons. After all grid cells were selected, we used the ‘Create Random
Points’ feature in the Data Management toolbox in ArcMap 10.1 to place 5 random
sampling points in each cell to reduce biases from spatial variation (Hayes 2000). We
placed each point at least 400 meters from other points.
Acoustic Surveys
We used acoustic recording equipment to survey for bats: SM2BAT+ detectors
with SMX-US microphones (Wildlife Acoustics, Maynard, MA). We set all detectors to
record continuously from 15 minutes before sunset to 15 minutes after sunrise. We
mounted the microphone of each detector on a fiberglass painter’s pole that extended to
a height of 3.4 meters. We tested microphone sensitivity once per week using an
ultrasonic calibrator (Wildlife Acoustics, Maynard, MA); we replaced microphones when
the calibrator read -36 dB (as per the recommendations provided by Wildlife Acoustics).
We surveyed from 20 January to 13 June 2014 and from 13 January to 12 May
2015. We visited each cell 3 times per year; separating each visit by at least 3 weeks.
During each visit, the detectors recorded bat activity for 2 to 3 nights depending on
logistics. We placed a detector within a 100 meter buffer of each of the 5 random points
generated in each cell during site selection. We placed detectors at locations within
each buffer that would maximize the probability of detecting bats (eg. more open areas
where calls could propagate farther; Ormsbee 2010). The majority of points in
developed and agricultural cells were located on private land. If permission was denied
89
at the original point, we moved to the next closest property in a sequential manner until
permission was granted to set the equipment. During 2014, equipment and logistical
issues resulted in 2 of 5 points being sampled for only 4 instead of 6 nights. In 2015, we
followed the same protocol as in 2014, all 5 points were surveyed for 6 nights.
Detection Covariates
Many bat species are more active as temperatures increase (Rodhouse et al.
2015). In order to account for potential changes in detection as a result of temperature,
we recorded minimum temperature for each survey night from the SM2+ detectors
(surveytemp). To account for potential changes in detectability of species resulting from
seasonal movements, phenology and increases in detection as juveniles became volant
later in our study period, we converted the date of each survey to the number of days
after January 1st and used it as a covariate (date).
We used the number of noise files recorded by detectors as a proxy for the
amount of background noise during each site visit. All recorded .WAV files were
scanned through Wildlife Acoustic’s Kaleidoscope Pro software (Maynard, MA). This
initial scan separated files that contained bat calls from files that contained noise by
looking for the distinctive frequency modulated sweep of a bat call (Agranat 2012). By
scanning for the sweeps, the software is able to separate bat calls from false triggers
caused by factors such as rain, wind, insect and frog calls, and anthropogenic noise.
We counted the number of files that were classified as ‘noise’ by Kaleidoscope Pro for
each survey at each site and used this as a covariate (noise).
90
Occupancy Covariates
Broad-scale analysis
We used ArcGIS 10.1 to estimate variables related to human-dominated
landscapes that we thought would potentially influence occurrence of bat species
across the southern Florida landscape. We broke our 4 land stratifications into 8 land
cover classes. The agricultural lands strata included improved pasture and rangeland
that maintained many characteristics of more natural environments and also heavily
managed crop agriculture. To account for this, we divided the agricultural land cover
into rangeland and crop-dominated agriculture. We selected all cells that had previously
been identified during site selection as dominated by ‘agriculture’ based on FNAI
categories. We classified land covers of ‘improved pasture’ or ‘unimproved/woodland
pasture’ as being rangeland (FNAI 2012). We then separated uplands into non-forested
uplands (ie. dry prairies, palmetto prairies, scrub, shrub and brushland and barren and
outcrop communities identified by FNAI) and forested uplands (see Table 4-1). We also
separated wetlands into 3 different categories: open water (communities identified by
FNAI as lacustrine, palustrine, riverine and estuarine), non-forested wetlands
(communities identified by FNAI as freshwater non-forested wetlands, freshwater
marshes, and non-vegetated wetlands), and forested wetlands (see Table 4-1). We
used the ‘Tabulate Area’ tool in the Spatial Analyst toolset to calculate the percent of
each cell that was covered by forested wetlands (forest.wetland), non-forested wetlands
(non.wetland), forested uplands (forest.upland), non-forested uplands (non.upland),
crop-dominated agricultural areas (ag), range-lands (range), developed areas
(developed), and open water (water).
91
Fine-scale analysis
To investigate potential mechanisms influencing the species-specific distributions
of bats in human-dominated landscapes, we calculated a number of additional variables
for each sampling point within all cells that were dominated by either crop-based
agriculture or development. We calculated the influence of vehicular traffic (traffic) by
importing an average annual daily traffic (AADT) layer (FL DOT 2013); we used the
‘Near’ feature in the Proximity toolbox to estimate the distance each point was from a
road with AADT of 3000 or higher. We selected 3000 as the cut-off, as this was the level
at which avoidance behavior of roads was observed in little brown bats (Myotis
lucifugus; Altringham and Kerth 2016). We also used the ‘Near’ feature in the Proximity
toolbox to estimate the distance each point was from non-developed upland and
wetland areas (natural). Increases in occupancy probability as proximity to undeveloped
areas increases has been previously documented in bat species (Williams-Guillen et al.
2016).
To further investigate a number of variables potentially influencing bat occupancy
within human-dominated landscapes, we placed a buffer of 400 meters around each
point using the ‘Buffer’ tool. To investigate potential effects of anthropogenic night
lighting on bats, we imported a layer that provided ambient nighttime light levels at 1 km
x 1 km throughout North America (CLED 2011). We then calculated the average
ambient nighttime light level (light) within the 400 m buffer surrounding each point using
the Spatial Analyst toolbox. We estimated the average canopy cover (cc) in each buffer
using the National Land Cover Dataset (NLCD) developed by the U.S. Geological
Survey (NLCD 2011). We used the Spatial Statistics toolset in ArcGIS 10.1 to measure
the average canopy cover within each grid cell sampled, as it has been suggested that
92
the loss of canopy cover in human-dominated landscapes plays a major role in the loss
of bat species in these areas (Coleman and Barclay 2012).
Species responses to development and agriculture have been shown to vary
based on the intensity of the land cover (see reviews in Jung and Threlfall 2016 and
Williams-Guillen et al. 2016). To investigate the potential that species would respond
differently to high intensity development vs. rural areas vs. crop-dominated agriculture,
we further separated human-dominated sites at the fine-scale. Within all human-
dominated grid cells, we placed a 400 meter buffer around each point. We used the
‘Tabulate Area’ tool to calculate the total area of each buffer that was either high density
urban (high), low density urban (low), rural (rural), orchards (tree.ag), row and field
crops (crop) or extractive (extractive) using the Florida Natural Areas Inventory (FNAI
2012, Table 4-2).
Data Analysis
We analyzed all files in Kaleidoscope Pro 3.1.0 using the Bats of Florida 3.1.0
classifiers (Wildlife Acoustics; Maynard, MA). Kaleidoscope Pro compares recorded
calls to a known call library that contains between 1,631 and 130,435 calls for each
species (Wildlife Acoustics 2015). Using a set of trained Hidden Markov Models (HMMs)
for each species, Kaleidoscope Pro calculates the prior probability that the call in
question was generated by each species’ HMM (Agranat 2012). Kaleidoscope Pro then
calculates a Fisher score for each call sequence by looking at HMM-generated
identification of each sequence. The Fisher score identifies what call parameters are
most significant for identification, and then multiple pairwise comparisons are
performed. These comparisons provide an output including what species the software
believes the sequence came from and a confidence factor associated with this. If the
93
normalized confidence factor exceeds a certain threshold, the file is considered
classified to species; if the confidence factors do not exceed the threshold or if the most
likely species class is the unidentifiable class, the file is labeled as ‘NoID’ (Agranat
2012).
We used an identification filter that identifies a bat call sequence as a series of at
least 5 calls with less than a 5 second gap between them (Britzke et al. 2002). Because
EUFL has an extremely distinctive call, sequences consisting of at least 2 calls were
used to identify this species. High quality calls were identified by Kaleidoscope Pro as
one of eight species or species classes: CORA, EUFL, LAIN/EPFU, LASE/LABO,
MYAU, NYHU, PESU or TABR. We grouped LAIN and EPFU into one species class
because their calls are very similar, and Kaleidoscope Pro could not reliably separate
the two species. We did the same with LASE and LABO. Previous acoustic and mist-
netting studies conducted in southern Florida had never documented LABO in the area
(Marks and Marks 2006); however, mist-netting in Florida Panther National Wildlife
Refuge in Collier County recorded several LABO in the spring of 2015 (E. Braun de
Torrez, University of Florida, pers. comm). We included LABO in the possible species
groupings to account for uncertainties in the range of the species.
The output of Kaleidoscope Pro provides a maximum likelihood estimate (MLE)
associated with species presence for each survey. A key factor in the development of
the MLE probability values is the confusion matrix; this matrix is unique for each
classifier in Kaleidoscope (Wildlife Acoustics 2015). The confusion matrix was
developed by running a set of known calls of each species through the classifier, and
determining the number of sequences that were correctly identified and how many were
94
identified as a different species (Wildlife Acoustics 2015). These misclassifications are
then taken into consideration when estimating whether a species was present during
each survey at a site, with a probability value of 0 meaning that a species is likely
present and a probability value of 1 meaning that a species is likely absent (Wildlife
Acoustics 2015). If the p-values calculated by Kaleidoscope Pro were 0.5 or less, we
considered a species to be present at a site.
The only species treated differently was EUFL. In order to ensure that all EUFL
calls were identified, we manually looked at all calls identified by Kaleidoscope as either
EUFL or NoID (Mona Doss, Wildlife Acoustics, pers. comm.). We identified calls with a
minimum frequency between 10 – 18 kHz and a maximum frequency between 16 – 22
kHz as EUFL (Bruce Miller, Bat Sound Services, pers. comm.).
We used multi-species Bayesian hierarchical occupancy models to investigate
the effects of human-dominated landscapes on bat species. We used a Bayesian
approach to integrate a hierarchical effect that addresses potential spatial
autocorrelation between points in each grid cell. We also accounted for imperfect
detection when estimating the state of occupancy (Royle and Dorazio 2008). We
modeled the observation process as conditional on the latent occupancy state, yj(i,k) |
z(i,k) ~ Bern(z[i,k]*pijk) where pijk is the probability of detection during survey j given
presence of species k at point i under a Bernoulli distribution. yj(i,k) represents the
detection history, with each taking a value of 1 or 0 representing “detection” or “non-
detection” for species k during survey j at site i. In order to speed Markov chain Monte
Carlo (MCMC) convergence and improve interpretability, we standardized all covariates
(Gelman and Hill 2007). We did not include correlated covariates (r2 > 0.60, Rodhouse
95
et al. 2015) in interactive models. We built all models at the community level and
analyzed the responses of each species within the community (Kery and Royle 2016)
Broad-scale model
We created models by first holding occupancy constant at the latent occupancy
state and modeled the effects of surveytemp, date and noise on detection. We used
JAGS v 3.4.0 (Plummer 2003) launched from RStudio v.0.98 with the R2jags library (Su
and Yajima 2015) to implement Bayesian estimation of model parameters via MCMC
samples of posterior distributions. We input each covariate into the model as a random
effect using vague, normally distributed [N(0, 0.01)] hyperpriors on all logit-scale
parameters (Kery and Royle 2016). Posterior summaries were based on 10000 MCMC
samples of the posterior distributions from 3 chains run simultaneously, thinned by a
factor of 10, following an initial burn-in of 2000. We assessed convergence of MCMC
chains with trace plots and the Gelman-Rubin diagnostic (R̂); convergence was reached
for all parameters according to the criteria | R̂ – 1| < 0.1 (Ntzoufras 2009). We evaluated
models using a stepwise comparison predictive performance of each model, removing
covariates where the 95% credible intervals crossed 0 and re-running the model after
each covariate was removed (Kery and Schaub 2012). We used this approach rather
than the deviance information criterion (DIC) because it is more robust when evaluating
predictive performance of models (Gelman et al. 2013). Our final model only included
variables where the 95% credible intervals did not cross 0 for at least one species (Kery
and Royle 2016).
Once we had a final model for detection, we evaluated occupancy of bat species
across the southern Florida landscape with covariates developed, ag, range,
96
non.wetland, forest.wetland, non.upland, forest.upland and water. Again, we used a
stepwise approach technique, holding detection constant as a function of the best-fitting
model found above. We reported all beta estimates and credible intervals from the final
selected model at the community and species level and graphed the species level
effects of all covariates that were included in the final model. We calculated species
richness of each of the 8 major land cover types investigated during our acoustic
surveys using the Nsite values provided by the final model. We graphed average
species richness between the major land cover types.
Fine-scale model
We further investigated finer-scaled effects of features within human-dominated
landscapes on bats. For this analysis, we only used grid cells that were originally
classified as dominated by either development or agriculture during site selection. We
modeled detection as a function of surveytemp, date, and noise using the same
technique described for the broad-scale analysis. We used a stepwise approach
technique to select variables included in the final model; removing variables where the
95% credible intervals crossed 0 for every species and re-running the analysis with the
variable removed (Kery and Schuab 2012). Once we selected a final model for
detection, we evaluated occupancy with the continuous variables high, low, rural, crop,
tree.ag, extractive, traffic, light, natural and cc as covariates. We again used a stepwise
approach, holding detection constant as a function of the best-fitting model found
above. Our final model included only variables where the 95% credible intervals did not
cross 0 for at least one species. We reported all beta estimates and credible intervals
from the final selected model and graphed the species-level effects of all covariates that
were included in the final model.
97
Results
Broad-scale model
Both surveytemp and date influenced detection probabilities of multiple species
(Table 4-3). All species except for MYAU had higher detection probabilities later in the
survey season [95% credible intervals (CRIs) did not cross 0 for LASE/LABO: β = 0.24
(CRI: 0.13 to 0.35), NYHU: β =0.2 (0.05 to 0.35), PESU: β = 0.25 (0.11 to 0.40), and
TABR: β = 0.33 (0.2 to 0.45)]. All species except for MYAU had significantly higher
detection probabilities on nights with higher minimum temperatures [CRIs did not cross
0 for CORA: β = 0.58 (0.07 to 1.11), EUFL: β = 0.47 (0.12 to 0.82), LAIN/EPFU: β =
0.88 (0.70 to 1.06), LASE/LABO: β = 0.37 (0.21 to 0.52), NYHU: β = 0.84 (0.61 to 1.08),
PESU: β = 0.18 (0.01 to 0.37) and TABR: β = 0.35 (0.18 to 0.52)]. The covariate noise
was not included in the final model, as the CRIs overlapped 0 for every species.
At the community level, species richness did not differ significantly between
major land cover types (Figure 4-3). CRIs crossed 0 for all major land cover types,
including development [β = - 0.22 (-0.42 to 0.10)] and agriculture [β = -0.19 (-0.51 to
0.14)]. However, at the species level, occupancy probabilities of all bat species were
lower in cells dominated by development. CRIs did not cross 0 for CORA [β = -0.81 (-
1.67 to -0.06)] EUFL [β = -0.70 (-1.31 to -0.14)], NYHU [β = -0.59 (-1.057 to -0.07)] and
PESU [β = -0.96 (-1.57 to -0.75)] (Figure 4-4, Table 4-4). Occupancy probabilities of
most species were lower in crop-based agriculture (Figure 4-5, Table 4-4); CRIs did not
cross 0 for LASE/LABO [β = -1.86 (-3.51 to -0.35), LAIN/EPFU [β = -1.07 (-1.96 to -
0.17)], MYAU [β = -1.94 (-5.01 to -0.07)] and PESU [β = -1.22 (-2.13 to -0.38)].
Occupancy probabilities of EUFL [β = 0.81 (-0.13 to 1.70)] and TABR [β = 0.18 (-1.81 to
98
3.14)] were positively influenced by crop-based agriculture, though the CRIs crossed 0
for both species (Figure 4-5, Table 4-4).
Both forested wetlands and forested uplands were positively correlated with
occupancy probabilities for most species of bats (Table 4-4). The amount of forested
wetlands in a cell was positively correlated with occupancy probabilities for all species
of bats; CRIs did not cross 0 for LASE/LABO [β = 0.587 (0.037 – 1.705)] and NYHU [β =
0.475 (0.053 – 1.022)] (Figure 4-6, Table 4-4). The amount of forested uplands also was
positively correlated with occupancy probabilities of most bat species; CRIs did not
cross 0 for LASE/LABO [β = 2.494 (1.300 – 4.163)], NYHU [β = 1.233 (0.683 – 1.856)]
and PESU [β = 1.277 (0.741 – 1.925)] (Figure 4-7).
Fine-scale model
Within human-dominated landscapes, we saw the same trends in detection as in
the broad-scale analysis, with detection probabilities for most species being higher later
in the survey season and on warmer nights. The best model at the community level
included distance to undeveloped habitats (natural) and canopy cover (cc) as covariates
on occupancy. The overall bat community showed a decline in occupancy probabilities
as distance to undeveloped habitats increased [β = -0.79 (-1.52 to -0.16)] (Figure 4-9).
In contrast, occupancy probabilities at the community level were positively correlated
with canopy cover [β = 0.72 (0.26 – 1.40)] (Figure 4-11).
At the species level, CRIs for all land cover types for all species crossed 0, so
these variables were not included in the final model. CRIs for all species crossed 0 for
the covariates light and traffic; these covariates were also dropped from analysis.
Occupancy probabilities of all species of bats increased as the distance to undeveloped
habitats (natural) decreased (Table 4-5). CRIs did not cross 0 for CORA [β = -1.31 (-
99
2.77 to -0.66)], LAIN/EPFU [β = -1.14 (-1.80 to -0.93)], LASE/LABO [β = -0.80 (-1.44 to -
0.61)], NYHU [β = -0.95 (-1.75 to -0.70)] and PESU [β = -1.22 (-2.20 to -0.88)] (Table 4-
5, Figure 4-8). Occupancy probabilities of nearly all species were positively influenced
by increasing canopy cover (cc) (Table 4-5). CRIs did not cross 0 for LASE/LABO [β =
0.66 (0.06 to 1.37)], MYAU [β = 0.88 (0.38 to 1.46)], NYHU [β = 1.21 (0.31 to 2.73)], and
PESU [β = 0.85 (0.33 to 1.51)] (Table 4-5, Figure 4-10).
Discussion
Human modifications to the landscape were associated with changes in
occupancy probabilities for all bat species across southern Florida. An overall negative
correlation was most prevalent in developed areas, where all species appeared to have
decreased occupancy probabilities in areas with higher amounts of development,
though the effects were of varying strengths. As expected, there was a relatively strong
negative correlation between the amount of development in a cell and occupancy
probabilities of tri-colored bats (PESU), Rafinesque’s big-eared bats (CORA) and
southeastern myotis (MYAU), the species with the lowest wing-loading. The association
was not as strong as expected for southeastern myotis, but this is likely a result of the
small number of detections (detected at 8% of sites). Of all the species with average
wing-loading, evening bats (NYHU) had the strongest negative apparent response to
development. Although evening bats have relatively long wings relative to other species
within the Vespertilionidae family (Norberg and Rayner 1987), they are known to have
slow, erratic flight (Walker 1964) and have been documented avoiding urban areas
when foraging (Duchamp et al. 2004). Their avoidance of developed areas is likely also
affected by their roosting behavior. While evening bats have occasionally been
documented roosting in buildings, they usually roost in tree cavities (Whitaker et al.
100
2002, Whitaker and Gummer 2003). Even in developed areas with canopy cover, there
are often few cavities available for bats (Duschamp et al. 2004). Trees in these areas
are often pruned, eliminating the formation of cavities from dead branches (Duschamp
et al. 2004). Bats that primarily roost in cavities have been found to be more affected by
urbanization than foliage-roosting bats, as dead foliage is often not removed from trees
in developed areas (Duschamp et al. 2004).
In contrast to other studies, we found that occupancy probabilities of both
Molossids were negatively correlated with the amount of development. The vast
majority of prior studies have been focused in temperate and tropical locales, where
urbanization implied deforestation and a loss of vertical structure. In southern Florida,
the majority of development occurs near the coast (Webb 1999). On the east coast,
development has infringed upon the vast sawgrass plains that comprise the Greater
Everglades Ecosystem (Webb 1999). On the west coast, heavy development is
surrounded by treeless dry prairies and open canopy mesic pine flatwoods (Webb
1999). In contrast to the majority of prior studies, it is likely that development in southern
Florida results in an increase in vertical structure; thus it may be more efficient for
Molossids to forage in the relatively open natural areas surrounding development. It is
possible that this apparent effect was especially pronounced in the bonneted bat
(EUFL), as this species is over twice the size of the Brazilian free-tailed bat (TABR)
(FWC 2011), and likely even less maneuverable (Avila-Flores and Fenton 2005).
The apparent responses of bat species to crop-dominated agriculture were more
varied than the responses to development. The broad-winged tri-colored bat and
southeastern myotis both showed strong negative responses to crop-dominated
101
agriculture. Surprisingly, Rafinesque’s big-eared bat showed a neutral response to crop-
dominated agriculture. This species is specialized to glean Lepidoptera from vegetation;
Johnson and Lacki (2013) suggested that their nocturnal movements are regulated by
the abundance and diversity of moths. Moth species abundance and diversity have
been found to be highest along forest-agricultural edges (Ricketts et al. 2001, Johnson
and Lacki 2013). It is possible that Rafinesque’s big-eared bats foraging along
agricultural edges or on Lepidopteran pest species that specialize on crops (eg. Pena et
al. 1996) help to mitigate potential avoidance of large expanses of agriculture (Johnson
and Lacki 2013).
Occupancy probabilities of Seminole/red bats (LASE/LABO) and yellow/big
brown bats (LAIN/EPFU) also showed a relatively strong negative correlation with the
amount of crop-based agriculture in a cell. These species are well-adapted for foraging
along edges and in open areas, so it was surprising that they showed such a strong
apparent response. However, it is likely that this response is related to roosting habits of
these species. Seminole, red, and yellow bats are all foliage roosters (Menzel et al.
1998), with yellow bats roosting primarily in Spanish moss (Barbour and Davis 1969).
Big brown bats are known to roost in trees and also in buildings (Barbour and Davis
1969). In southern Florida, most intensively managed crop-based agriculture have few
trees (Webb 1999). The one exception would be citrus groves (Webb 1999). These
groves are often intensively pruned and managed, and dead foliage that could provide
roosting habitat for foliage-roosting bats would likely be removed. Similar to other
studies, the Molossidae in our study were least affected by crop-dominated agriculture;
both bonneted bats and Brazilian free-tailed bats had positive but not strong (CRIs
102
included 0) associations with this land cover (Noer et al. 2012). The high-wing loading
and low, narrowband calls of these species make them well-adapted for exploiting
insects in these open areas (Norberg and Rayner 1987).
The findings from our broad-scale analysis have significant implications in the
context of southern Florida, where anthropogenic alteration of landscapes is inevitable.
The population of Florida is expected to grow by nearly 20,000,000 people by 2060, and
even more land cover changes will occur (Zwick and Carr 2006). Southwest Florida and
south-central Florida are supposed to undergo the most extreme land cover changes in
the state, with nearly all agricultural and natural lands expected to become urbanized
(Zwick and Carr 2006). Nearly all unprotected land throughout southern Florida is
expected to become heavily urbanized by 2060, with the exception of some agricultural
land surrounding Lake Okeechobee (Zwick and Carr 2006). The results from our study
show that the conversion of natural and agricultural land could have drastic effects on
the bat community of this area.
The results from our fine-scale analysis provide some clues as to how we can
potentially mitigate negative responses of bats to increasing urbanization. Similar to
other studies (Jung and Threfall 2016, Altringham and Kerth 2016), we found that
proximity of human-dominated landscapes to undeveloped areas increases occupancy
probabilities of all species of bats found in southern Florida. In addition, we found that
increased canopy cover within human-dominated landscapes benefitted all species
except for Brazilian free-tails and Rafinesque’s big-eared bats. While distance to roads
and levels of night-time lighting have been found to affect bats in other studies (see
Altringham and Kerth 2016 and Rowse et al. 2016 for reviews), we saw no such trends
103
during this study. However, it is important to note that our study was not specifically
designed to look at relationships between nighttime lighting and roads, and more
focused research should be done in the area to reliably examine potential relationships.
Even so, our results suggest that bat species persist in human-dominated landscapes,
including high-density development, if canopy cover and connectivity to undeveloped
areas is maintained. Developers and managers should take these factors into
consideration in order to protect these important species.
104
Figure 4-1. Map of percent of each grid cell in the study area covered by crop-dominated agriculture. Sampled cells represent cells surveyed acoustically for bats between January – June 2014 and January – May 2015.
105
Figure 4-2. Map of percent of each grid cell in the study area covered by development. Sampled cells represent cells surveyed acoustically for bats between January – June 2014 and January – May 2015.
106
Table 4-1. Florida Natural Areas Inventory (FNAI) state-level land cover types grouped into broad land covers used during
site selection for acoustic surveys conducted for bats throughout southern Florida.
Broad Land Cover Type
Agriculture Developed Upland Wetland
Agriculture Low intensity urban Upland hardwood forest* Freshwater non-forested wetlands
Row crops High intensity urban Mesic hammock* Freshwater marshes
Field crops Rural Rockland hammock* Cypress/tupelo*
Improved pasture1 Transportation Scrub Cypress*
Unimproved/woodland pasture1 Communication Upland mixed woodland* Other coniferous wetlands*
Citrus Extractive Upland coniferous* Hardwood wetlands*
Fruit orchards Utilities Sandhill* Non-vegetated wetlands
Vineyard and nurseries Dry flatwoods* Wet coniferous plantation*
Pine rockland* Lacustrine2
Dry prairie Riverine2
Palmetto prairie Estuarine2
Hardwood forest* Exotic wetland hardwoods*
Shrub and brushland
Coastal uplands*
Barren and outcrop
Coniferous plantations*
*forested upland or wetland 1 rangeland 2 open water
107
Table 4-2. Covariates used in the fine-scale Bayesian hierarchical model for each bat species recorded in southern Florida from January – June 2014 and January – May 2015.
Variable Description
High intensity urban > 2 dwellings per acre Low intensity urban Less than 2 dwellings per acre Rural Less than 1 dwelling per 5 acres Crops Agricultural land which is managed for the production of row or field crops Tree ag Agricultural land managed for citrus, fruit orchards, or tree nurseries Extractive Surface and subsurface mining operations and industrial complexes
108
Table 4-3. Beta estimates and credible intervals for detection covariates included in the final broad-scale Bayesian hierarchical model for each bat species recorded in southern Florida from January – June 2014 and January – May 2015.
Species
Survey date Minimum temperature
survey night
Beta estimate
Lower bound of CI
Upper bound of CI
Beta estimate
Lower bound of CI
Upper bound of CI
CORA 0.02 -0.38 0.32 0.58 0.07 1.11
EUFL 0.10 -0.14 0.30 0.47 0.12 0.82
LASE/LABO 0.11 -0.02 0.23 0.88 0.70 1.06
LAIN/EPFU 0.24 0.13 0.35 0.37 0.21 0.52
MYAU -0.04 -0.47 0.26 0.13 -0.43 0.66
NYHU 0.20 0.05 0.36 0.84 0.61 1.08
PESU 0.25 0.11 0.40 0.18 0.01 0.37
TABR 0.33 0.20 0.45 0.35 0.18 0.52
109
Table 4-4. Beta estimates and credible intervals for occupancy covariates included in the final broad-scale Bayesian hierarchical model for each bat species recorded in southern Florida from January – June 2014 and January – May 2015.
Species
Development Crop-based ag Forested wetlands Forested Upland
Beta estimate
Lower bound of
CI
Upper bound of
CI
Beta
estimate
Lower bound of CI
Upper bound of CI
Beta
estimate
Lower bound of CI
Upper bound of CI
Beta estimate
Lower bound of CI
Upper bound of CI
CORA -0.24 -0.59 0.08 -0.17 -0.64 0.26 0.288 -0.407 0.967 0.520 -0.198 1.412
EUFL -0.18 -0.42 0.07 0.28 0.03 0.56 0.376 -0.025 0.767 -0.447 -1.057 0.110
LASE/LABO -0.16 -0.39 0.10 -0.26 -0.51 -0.02 0.587 0.037 1.705 2.494 1.300 4.163
LAIN/EPFU -0.17 -0.52 0.30 -0.61 -1.04 -0.20 0.236 -0.597 0.879 1.006 -0.149 2.750
MYAU -0.10 -0.37 0.25 -0.24 -0.69 0.14 0.875 -0.045 3.774 0.330 -0.386 1.037
NYHU -0.23 -0.43 -0.02 -0.10 -0.34 0.15 0.475 0.053 1.022 1.233 0.683 1.856
PESU -0.39 -0.66 -0.16 -0.32 -0.58 -0.10 0.392 -0.050 1.030 1.277 0.741 1.925
TABR -0.22 -0.60 0.18 -0.07 -0.54 0.54 0.060 -1.026 0.718 0.752 -0.404 2.363
110
Figure 4-3. Bat species richness in grid cells of each major land cover type surveyed from January – June 2014 and January – May 2015. Southern Florida, USA.
111
Figure 4-4. Effect of developed land cover on occupancy probabilities of species of bats sampled acoustically from January – June 2014 and January – May 2015. Effects calculated during broad-scale analysis. Southern Florida, USA.
TABR
PESU
NYHU
MYAU
LAIN/EPFU
LASE/LABO
EUFL
CORA
112
Figure 4-5. Effect of crop-based agriculture land cover on occupancy probabilities of species of bats sampled acoustically from January – June 2014 and January – May 2015. Effects calculated during broad-scale analysis. Southern Florida, USA.
TABR
PESU
NYHU
MYAU
LAIN/EPFU
LASE/LABO
EUFL
CORA
113
Figure 4-6. Effect of amount of forested wetlands within sampled grid cells on
occupancy probabilities of species of bats sampled acoustically from January – June 2014 and January – May 2015. Effects calculated during broad-scale analysis. Southern Florida, USA.
114
Figure 4-7. Effect of amount of forested uplands within sampled grid cells on occupancy
probabilities of species of bats sampled acoustically from January – June 2014 and January – May 2015. Effects calculated during broad-scale analysis. Southern Florida, USA.
115
Table 4-5. Beta estimates and credible intervals for covariates included in the final fine-scale Bayesian hierarchical model
for each bat species recorded in human-dominated landscapes in southern Florida from January – June 2014 and January – May 2015.
Species
Distance to undeveloped areas Canopy cover
Beta estimate
Lower bound of CI
Upper bound of CI
Beta estimate
Lower bound of CI
Upper bound of CI
CORA -1.31 -2.77 -0.66 -0.23 -1.38 0.86 EUFL 0.12 -0.38 0.27 0.18 -0.26 0.58 LASE/LABO -1.14 -1.80 -0.93 0.66 0.06 1.37 LAIN/EPFU -0.80 -1.44 -0.61 0.48 -0.62 2.37 MYAU -0.60 -1.56 -0.28 0.88 0.38 1.46 NYHU -0.95 -1.75 -0.7 1.21 0.31 2.73 PESU -1.22 -2.20 -0.88 0.85 0.33 1.51 TABR -0.09 -0.73 0.12 -0.41 -1.07 0.32
116
Figure 4-8. Effect of distance to undeveloped habitats in human-dominated landscapes
on occupancy probabilities of bats acoustically sampled from January – June 2014 and January – May 2015. Southern Florida, USA.
117
Figure 4-9. Response of bat species to distance to undeveloped habitat in human-dominated landscapes. The black line represents the community-level response, with colored lines representing the responses of individual species. All bat species were acoustically sampled in January – June 2014 and January – May 2015. Southern Florida, USA.
118
Figure 4-10. Effect of canopy cover in human-dominated landscapes on occupancy
probabilities of bat species acoustically sampled in January – June 2014 and January – May 2015. Southern Florida, USA.
119
Figure 4-11. Response of bat species to canopy cover in human-dominated landscapes. The black line represents the community-level response, with colored lines representing the responses of individual species. All bat species were acoustically sampled in January – June 2014 and January – May 2015. Southern Florida, USA.
120
CHAPTER 5 CONCLUSION
The major objective of this project was to investigate bat ecology in the rapidly
changing southern Florida landscape. We focused on closing data gaps for the endemic
Florida bonneted bat by investigating demographic rates and environmental
associations of this endangered species. We investigated apparent survival, recruitment
and population growth rate of a population of bonneted bats using a number of bat
houses on Babcock-Webb Wildlife Management Area (BWWMA). In addition, we
investigated how well a number of environmental associations explained the current
distribution of the bonneted bat throughout southern Florida. Further, we analyzed the
impact of human-dominated landscapes on the bonneted bat and other bat species
found in this region, and investigated fine-scale associations of bat species within
human-dominated landscapes to better understand the responses we may expect as
urbanization pressure increases in the coming years.
Our study provided the first estimates of demographic rates for the Florida
bonneted bat. We found that bonneted bats had relatively low apparent survival rates
compared to other species of bats (O’Shea et al. 2004), although the low survival rates
observed in the study may have been partially a result of tag loss. Like other species of
tropical Molossids (eg. Gager et al. 2016), the bonneted bat had an extended
reproductive season, with recruitment remaining constant throughout our study period.
The population of bonneted bats using bat houses on BWWMA was relatively stable
throughout our 2 year study period, although the estimated population growth rates for
adult females were suggestive of a downward trend. The potential loss of females in this
population is concerning, as it could represent a loss in reproductive potential. Although
121
our estimates only apply to a single population of bonneted bats using bat houses on
BWWMA, this information will be useful in the development of recovery and
management plans.
The bonneted bat is believed to have one of the most limited geographic ranges
of any bat species in the United States; however, there have been no previous
investigations into environmental associations of this species. Climactic factors
appeared to influence the bonneted bat distribution in southern Florida. Bonneted bats
were found more often in areas with high average spring minimum temperatures and
high average annual precipitation levels. The preference of bonneted bats for warm,
humid environments may reflect the typical Neotropic range of other species of
Eumops, including the closely related E. glaucinus. We found bonneted bats in every
major land cover type investigated, although occupancy probabilities of bonneted bats
decreased in grid cells with high amounts of development. In contrast, bonneted bats
appeared to prefer grid cells with high levels of crop-dominated agriculture.
The threat of land use change leading to alteration in wildlife communities is
pronounced in southern Florida, which has one of the fastest growing human
populations in the United States. We investigated community and species-specific
effects of human-dominated landscapes on bats across southern Florida. Our study
found the bat community shifted in response to agricultural intensification and urban
development. Although increasing levels of development did not lead to significant
changes in occupancy rates of the bat community, all species showed a negative trend
in occupancy with increasing levels of development. The magnitude of this trend was
highly variable between different species. Five of the 8 species investigated responded
122
negatively to agricultural intensification; whereas the bonneted bat and Brazilian free-
tailed bat responded positively to the amount of agriculture within grid cells. Within
human-dominated areas, the loss of canopy cover and distance from undeveloped
areas appeared to have the greatest effects on the bat community.
As a result of a rising human population, leading to pressure from development
and intensive agriculture and a loss of natural areas, the southern Florida landscape is
considered one of the most threatened landscapes in the United States (Noss and
Peters 1995). Our research shows that resulting land cover changes are shaping the
distribution of bats in southern Florida, including the federally endangered Florida
bonneted bat. Future research should determine if survival rates and population growth
rates of bonneted bats observed in this study occur throughout the range of the species,
or solely in the bat houses on BWWMA. In addition, we recommend additional attention
be focused on environmental associations of the bonneted bat and other species of bats
in southern Florida to elucidate additional threats, particularly those relating to the rising
human population and subsequent land cover changes predicted to occur in southern
Florida in the future.
123
APPENDIX A RESULTS OF SIMULATIONS IN CHAPTER 2
Results of simulations when tag retention = 0.90 between capture events.
Confidence Interval
Parameter Actual Estimate Lower Upper
φ 0.7 0.731 0.629 0.812
p 0.6 0.563 0.446 0.673
λ 1.25 1.218 1.124 1.321
Results of simulations when tag retention = 0.50 between capture events.
Confidence Interval
Parameter Actual Estimate Lower Upper
φ 0.7 0.315 0.211 0.442
p 0.6 0.653 0.356 0.865
λ 1.25 1.219 1.119 1.329
Results of simulations when φ = 0.90 between capture events.
Confidence Interval
Parameter Actual Estimate Lower Upper
φ 0.9 0.848 0.755 0.909
p 0.6 0.552 0.454 0.646
λ 1.25 1.255 1.163 1.354
Results of simulations when φ = 0.70 between capture events.
Confidence Interval
Parameter Actual Estimate Lower Upper
φ 0.7 0.759 0.654 0.84 p 0.6 0.507 0.393 0.62 λ 1.25 1.187 1.094 1.288
124
Results of simulations when φ = 0.50 between capture events.
Confidence Interval
Parameter Actual Estimate Lower Upper
φ 0.5 0.504 0.351 0.657 p 0.6 0.427 0.241 0.635 λ 1.25 1.097 0.992 1.214
125
APPENDIX B SPECIES MAPS IN RELATION TO DEVELOPMENT AND CROP-BASED AGRICULTURE
126
127
128
129
130
131
132
133
LIST OF REFERENCES
Agranat, I. 2012. Bat species identification from zero crossing and full spectrum echolocation calls using HMMs, Fisher scores, unsupervised clustering and balanced Winnow pairwise classifiers. Wildlife Acoustics, Inc. Concord, Massachusetts, USA.
Altringham, J.D. 1996. Bats: Biology and Behavior. Oxford University Press, Inc., New
York, New York, USA. Altringham, J., and G. Kerth. 2016. Bats and roads. Pages 35 – 60 in C.C. Voigt and T.
Kingston, editors. Bats in the Anthropocene: Conservation of Bats in a Changing World. Springer Press, New York City, New York, USA.
Avila-Flores, R., and M.B. Fenton. 2005. Use of spatial features by foraging
insectivorous bats in a large urban landscape. Journal of Mammalogy 86(6): 1193 – 1204.
Bader, E., K. Jung, E.K.V. Kalko, R.A. Page, R. Rodriquez, and T. Sattler. 2015.
Mobility explains the response of aerial insectivorous bats to anthropogenic habitat change in the Neotropics. Biological Conservation 186: 97 – 106.
Bakker, V.J., D.F. Doak, G.W. Roemer, D.K. Garcelon, T.J. Coonan, S.A. Morrison, C.
Lynch, K. Ralls, R. Shaw. 2009. Incorporating ecological drivers and uncertainty into a demographic population viability analysis for the island fox. 79(1): 77 – 108.
Barbour, R. W. and W. H. Davis. 1969. Bats of America. The University Press of
Kentucky, Lexington, KY. Barclay, R.M.R. and L.D. Harder 2003. Life histories of bats: Life in the slow lane.
Pages 209 – 253 in T.H. Kunz and M.B. Fenton, editors. Bat Ecology, University of Chicago Press, Chicago, Illinois, USA.
Barnosky, A.D. et al. 2011. Has the Earth’s sixth mass extinction already arrived?
Nature 471(7336): 51 – 57. Bartlett, S.N., M.M. McDonough, and L.K. Ammerman. 2013. Molecular systematics of
bonneted bats (Molossidae: Eumops) based on mitochondrial and nuclear DNA sequences. Journal of Mammalogy 94(4): 867 – 880.
Baucum, L.E., R.W. Rice and T.J. Schueneman. 2006. An overview of Florida
sugarcane. Agronomy Department, Florida Cooperative Extensioin Service, Institute of Food and Agricultural Sciences, University of Florida. Publication #SS-AGR-232.
134
Beever, J.W. III, and K.A. Dryden. 1998. The hydric pine flatwoods of southwest Florida: a community profile update. Office of Environmental Services, Florida Game and Fresh Water Fish Commission.
Bellamy, C., C. Scott, and J. Altringham. 2013. Multiscale, presence-only habitat
suitability models: fine-resolution maps for eight bat species. Journal of Applied Ecology 50: 892 – 901.
Belwood, J. J. 1981. Wagner’s mastiff bat, Eumops glaucinus floridanus (Molossidae) in
southwestern Florida. Journal of Mammalogy 62: 411 – 413. Belwood, J. J. 1992. Florida mastiff bat Eumops glaucinus floridanus. Pp. 216 – 223 In
S. R. Humphrey (ed.), Rare and endangered biota of Florida. Vol. I. Mammals. University Press of Florida, Gainesville, FL.
Berthinussen, A., and J. Altringham. 2012. The effect of a major road on bat activity and
diversity. Journal of Applied Ecology 49: 82 – 89. Best, T.L., W.M. Kiser, and J.C. Rainey. 1997. Eumops glaucinus. Mammalian Species
551: 1 – 6. Bielby, J., G.M. Mace, O.R.P. Bininda-Edmonds, M. Cardillo, J.L. Gittleman, K.E. Jones,
C.D.L. Orme, and A. Purvis. 2007. The fast-slow continuum in mammalian life history: an empirical reevaluation. The American Naturalist 169(6): 748 – 757.
Bontadina, F., H. Schofield, and B. Naef-Daenzer. 2002. Radio-tracking reveals that
lesser horseshoe bats (Rhinolophus hipposideros) forage in woodland. Journal of Zoology 258(3): 281 – 290.
Britzke, E. R. 2003. Use of ultrasonic detectors for acoustic identification and study of
bat ecology in the eastern United States. PhD. Dissertation. Tennessee Technological University, Cookeville, TN.
Britzke, E. R., K.L. Murray, J.S. Heywood, and L.W. Robbins. 2002. Acoustic
identification. Pages. 220 – 224 in A. Kurta and J. Kennedy, editors. The Indiana bat: biology and management of an endangered species. Bat Conservation International, Austin, Texas, USA.
Burles, D.W., R.M. Brigham, R.A. Ring, and T.E. Reimchen. 2009. Influence of weather
on two insectivorous bats in a temperate Pacific Northwest rainforest. Canadian Journal of Zoology 87: 132 – 138.
Campbell, S.P., J.A. Clark, L.H. Crampton, A.D. Guerry, L.T. Hatch, P.R. Hosseini, J.J.
Lawler, and R.J. O’Connor. 2002. An assessment of monitoring efforts in endangered species recovery plans. Ecological Applications 12(3): 674 – 681.
135
Chamberlain, D.E., A.M. Wilson, S.J. Browne, and J.A. Vickery. 1999. Effects of habitat type and management on the abundance of skylarks in the breeding season. Journal of Applied Ecology 36: 856 – 870.
Cleveland, C.J., et al. 2006. Economic value of the pest control service provided by
Brazilian free-tailed bats in south-central Texas. Frontiers in Ecology and the Environment 4(5): 238 – 243.
Cockrum, E.L. 1960. Distribution, habitat and habits of the mastiff bat, Eumops perotis,
in North America. Journal of the Arizona Academy of Science 1(3): 79 – 84. Coleman, J.L., and R.M.R. Barclay. 2012. Urbanization and the abundance and
diversity of Prairie bats. Urban Ecosystems 15: 87 – 102. Coleman, J.L., and R.M.R. Barclay. 2013. Prey availability and foraging activity of
grassland bats in relation to urbanization. Journal of Mammalogy 94(5): 1111 – 1122.
Crutzen, P.J., and E.F. Stoermer. 2000. The Anthropocene. Global Change News 41:
17 – 18. Davis, W.H. and H.B. Hitchcock. 1965. Biology and migration of the bat, Myotis
lucifugus, in New England. Journal of Mammalogy 46(2): 296 – 313. Dixon, M.D. 2012. Relationship between land cover and insectivorous bat activity in an
urban landscape. Urban Ecosystems 15: 683 – 695. Duchamp, J.E., and R.K. Swihart. 2008. Shifts in bat community structure related to
evolved traits and features of human-altered landscapes. Landscape Ecology 23: 849 – 860.
Duchamp, J.E., D.W. Sparks and J.O. Whitaker, Jr. 2004. Foraging-habitat selection by
bats at an urban-rural interface: comparison between a successful and a less successful species. Canadian Journal of Zoology 82: 1157 – 1164.
Duever, M. J., J. F. Meeder, L. C. Meeder and J. M. McCollom. 1994. The climate of
South Florida and its role in shaping the Everglades Ecosystem. Pp. 225 - 248 In S. Davis and J. C. Ogden [eds.], Everglades: The Ecosystem and its Restoration. CRC Press, Boca Raton, Florida.
Duchamp, J.E., M. Yates, R.M. Muzika, and R.K. Swihart. 2006. Estimating probabilities
of detection for bat echolocation calls: an application of the double-observer method. Wildlife Society Bulletin 34(2): 408 – 412.
Ehrlich, P.R. 1994. Energy use and biodiversity loss. Philosophical Transactions of the
Royal Society of London B. 344: 99 – 104.
136
Ekman, M., and J. de Jong. 1996. Local patterns of distribution and resource utilization
of four bat species (Myotis brandti, Eptesicus nilssoni, Plecotus auritus and Pipistrellus pipistrellus) in patchy and continuous environments. Journal of Zoolgy 238: 571 – 580.
Ellison, L.E., T.J. O’Shea, D.J. Neubaum, and R.A. Bowen. 2007. Factors influencing
movement probabilities of big brown bats (Eptesicus fuscus) in buildings. Ecological Applications 17(2): 620 – 627.
Estrada, A., R. Coates-Estrada and D. Meritt, Jr. 1993. Bat species richness and
abundance in tropical rain forest fragments and in agricultural habitats at Los Tuxtlas, Mexico. Ecography 16: 309 – 318.
Evans, K.L., P.H. Warren, and K.J. Gaston. 2005. Species-energy relationships at the
macroecological scale: a review of the mechanisms. Biological Reviews 80: 1 – 25.
Farrow, L.J., and H.G. Broders. 2011. Loss of forest cover impacts the distribution of the
forest-dwelling tri-colored bat (Perimyotis subflavus). Mammalian Biology 76: 172 – 179.
Fenton, M. B. 2003. Science and the conservation of bats: where to go next? Wildlife
Society Bulletin 31: 6 – 15. Fenton, M.B., N.G.H. Boyle, T.M. Harrison, and D.J. Oxley. 1977. Activity patterns,
habitat use, and prey selection by some African insectivorous bats. Biotropica 9(2): 73 – 85.
Fieberg, J., and S.P. Ellner. 2001. Stochastic matrix models for conservation and
management: a comparative review of methods. Ecology Letters 4(3): 244 – 266. Florida Bat Conservancy. 2006. Bats of Florida. Bay Pines, Florida
<http://www.floridabats.org/FloridaBats.htm> Florida Fish and Wildlife Conservation Commission [FWC]. 2003. A conceptual
management plan for Fred C. Babcock – Cecil M. Webb Wildlife Management Area. Florida Fish and Wildlife Conservation Commission, Tallahassee, Florida, USA.
Florida Fish and Wildlife Conservation Commission [FWC]. 2005. Florida’s Wildlife
Legacy Initiative. Florida’s Comprehensive Wildlife Conservation Strategy. Tallahassee, Florida, USA.
137
Florida Fish and Wildlife Conservation Commission [FWC]. 2011. Biological status review report for the Florida bonneted bat (Eumops floridanus). Florida Fish and Wildlife Conservation Commission, Tallahassee, Florida, USA.
Florida Fish and Wildlife Conservation Commission [FWC]. 2013. DRAFT: A Species
Action Plan for the Florida Bonneted Bat Eumops floridanus. Florida Fish and Wildlife Conservation Commission, Tallahassee, Florida, USA.
Florida Natural Areas Inventory [FNAI]. 2012. Guide to the natural communities of
Florida. Florida Natural Areas Inventory, Tallahassee, Florida, USA. Food and Agriculture Association [FAO]. 2011. The state of the world’s land and water
resources for food and agriculture: manageing systems at risk. 308 pages. Ford, W.M., M.A. Menzel, J.L. Rodrigue, J.M. Menzel, and J.B. Johnson. 2005. Relating
bat species presence to simple habitat measures in a central Appalachian forest. Biological Conservation 126: 528 – 539.
Foreman, R.T.T. and L.E. Alexander. 1998. Roads and their major ecological effects.
Annual Review of Ecological Systems 8: 629 – 644. Frick. W.F., D.S. Reynolds, and T.H. Kunz. 2010. Influence of climate and reproductive
timing on demography of little brown myotis Myotis lucifugus. Journal of Animal Ecology 79(1): 128 – 136.
Fuentes-Montemayor, E., D. Goulson, and K.J. Park. 2011. Pipistrelle bats and their
prey do not benefit from four widely applied agri-environment management prescriptions. Biological Conservation 144: 2233 – 2246.
Gager, Y., O. Gimenez, M.T. O’Mara, and D.K.N. Dechmann. 2016. Group size, survival
and surprisingly short lifespan in socially foraging bats. BMC Ecology 16(1): 1 – 12.
Gaisler, J., J. Zukal, Z. Rehak, and M. Homolka. 1998. Habitat preference and flight
activity of bats in a city. Journal of Zoology 244: 439 – 445. Garden, J., C. McAlpine, A.N.N. Peterson, D. Jones, and H. Possingham. 2006. Review
of the ecology of Australian urban fauna: a focus on spatially explicit processes. Austral Ecology 31: 126 – 148.
Gehrt, S.D., and J.E. Chelsvig. 2003. Bat activity in an urban landscape: patterns at the
landscape and microhabitat scale. Ecological Applications 13(4): 939 – 950. Gehrt, S.D., and J.E. Chelsvig. 2004. Species-specific patterns of bat activity in an
urban landscape. Ecological Applications 14(2): 625 – 635.
138
Gelman, A., and J. Hill. 2007. Data Analysis Using Regression and Multilevel/Hierarchical Models. Cambridge University Press, New York, New York, USA.
Gelman, A., J.B. Carlin, H.S. Stern, D.B. Dunson, A. Vehtari, D.B. Rubin. 2013.
Bayesian Data Analysis, Third Edition. Chapman and Hall/CRC Press, Boca Raton, Florida, USA.
Gonsalves, L., S. Lamb, C. Webb, B. Law, and V. Monamy. 2013. Do mosquitoes
influence bat activity in coastal habitats? Wildlife Research 40(1): 10 – 24. Gorresen, P.M., A.C. Miles, C.M. Todd, F.J. Bonaccorso, and T.J. Weller. 2008.
Assessing bat detectability and occupancy with multiple automated echolocation detectors. Journal of Mammalogy 89(1): 11 – 17.
Greaves, G.J., R. Mathieu, and P.J. Seddon 2005. Predictive modelling and ground
validation of the spatial distribution of the New Zealand long-tailed bat (Chalinolobus tuberculatus). Biological Conservation 132: 211 – 221.
Grimm, N.B., S.H. Feath, N.E. Golubiewski, C.L. Redman, J. Wu, X. Bai and J.M.
Briggs. 2008. Global change and the ecology of cities. Science 319(5864): 756 – 760.
Gunderson, L.H. and J.R. Snyder. 1994. Fire patterns in the southern Everglades.
Pages 291 – 305 in S.M. Davis, and J.C. Ogden, editors. Everglades – the Ecosystems and its Restorations. St. Lucie Press, Delray Beach, Florida, USA.
Hayes, J. P. 1997. Temporal variation in activity of bats and the design of echolocation-
monitoring studies. Journal of Mammalogy 78: 514 – 524. Hayes, J. P. 2000. Assumptions and practical considerations in the design and
interpretation of echolocation-monitoring studies. Acta Chiropterologica 2: 225 – 236.
Hayes, J.P., H.K. Ober, and R.E. Sherwin. 2009. Survey and monitoring of bats. Pages
112 – 129 in T.H. Kunz and S. Parsons, editors, Ecological and Behavioral Methods for the Study of Bats, 2nd Edition. Johns Hopkins University Press, Baltimore, Maryland, USA.
Hayman, D.T.S., R. McCrea, O. Restif, R. Suu-Ire, A.R. Fooks, J.L.N. Wood, A.A.
Cunningham, J.M. Rowcliffe. 2012. Demography of straw-colored fruit bats in Ghana. Journal of Mammalogy 93(5): 1393 – 1404.
Hines, J. E. and J. D. Nichols. 2002. Investigations of potential bias in the estimation of
lambda using Pradel’s (1996) model for capture-recapture data. Journal of Applied Statistics 29: 573 – 587.
139
Holt, A.R., K.J. Gaston, and F. He. 2002. Occupancy-abundance relationships and
spatial distribution: a review. Basic and Applied Ecology 3: 1 – 13. Humphrey, S.R. 1975. Nursery roosts and community diversity of Nearctic bats. Journal
of Mammalogy 56: 321 – 346. Humphries, M.M., D.W. Thomas, and J.R. Speakman. 2002. Climate mediated
energetic constraints on the distribution of hibernating mammals. Nature 418: 313 – 316.
IUCN. 2015. IUCN red list of threatened species. Vienna: International Union for
Conservation of Nature and Natural Resources (IUCN). Jaberg, C., and A. Guisan. 2001. Modelling the distribution of bats in relation to
landscape structure in a temperate mountain environment. Journal of Applied Ecology 38: 1169 – 1181.
Johnson, J.B., J.E. Gates, and W.M. Ford. 2008. Distribution and activity of bats at local
and landscape scales within a rural-urban gradient. Urban Ecosystems 11: 227 – 242.
Johnson, J.B., M.A. Menzel, J.W. Edwards, and W.M. Ford. 2002. A comparison of 2
acoustical bat survey techniques. Wildlife Society Bulletin 30(3): 931 – 936. Johnson, J.S., and M.J. Lacki. 2013. Habitat associations of Rafinesque’s big-eared
bats (Corynorhinus rafinesquii) and their lepidopteran prey in bottomland hardwood forests. Canadian Journal of Zoology 91: 94 – 101.
Johnson, W.E. et al. 2010. Genetic restoration of the Florida panther. Science 329:
1641 – 1644. Jones, J.P.G. 2011. Monitoring species abundance and distribution at the landscape
scale. Journal of Applied Ecology 48: 9 – 13. Jones, K.E., A. Purvis, and J.L. Gittleman. 2003. Biological correlates of extinction risk
in bats. The American Naturalist 161: 601 – 614. Jung, K., and C.G. Threlfall. 2016. Urbanisation and its effects on bats – a global meta-
analysis. Pages 13 – 33 in C.C. Voigt and T. Kingston, editors. Bats in the Anthropocene: Conservation of Bats in a Changing World. Springer Press, New York City, New York, USA.
Jung, K., and E.K.V. Kalko. 2010. Where forest meets urbanization: foraging plasticity of
aerial insectivorous bats in an anthropogenically altered environment. Journal of Mammalogy 91(1): 144 – 153.
140
Jung, K., and E.K.V. Kalko. 2011. Adaptability and vulnerability of high flying
Neotropical aerial insectivorous bats to urbanization. Diversity and Distributions 17(2): 262 – 274.
Kerth, G., and M. Melber. 2009. Species-specific barrier effects of a motorway on the
habitat use of two threatened forest-living bat species. Biological Conservation 142: 270 – 279.
Kerth, G., M. Wagner, and B. Konig. 2001. Roosting together, foraging apart:
information transfer about food is unlikely to explain sociality in female Bechstein’s bats (Myotis bechsteinii). Behavioral Ecology and Sociobiology 50(3): 283 – 291.
Kery, M. 2011. Towards the modelling of true species distributions. Journal of
Biogeography 38: 617 – 618. Kery, M., and M. Schaub. 2012. Bayesian Population Analysis Using WinBUGS: a
Hierarchical Perspective. Academic Press, Waltham, Massachusetts, USA. Kery, M., and J.A. Royle. 2016. Applied Hierarchical Modeling in Ecology: Analysis of
distribution, abundance and species richness in R and BUGS. Academic Press, London, UK.
Kottek, M., J. Grieser, C. Beck, B. Rudolf, and F. Rubel. 2006. World map of the
Koppen-Geiger climate classification updated. Meteorologische Zeitschrift 15(3): 259 – 263.
Kranzer, B. et al. 1995. The Governor’s Commission for a Sustainable South Florida:
Initial Report. Florida Department of Community Affairs, Coral Gables, Florida, USA.
Kunz, T.H. 1973. Resource utilization: temporal and spatial components of bat activity in
central Iowa. Journal of Mammalogy 54: 14 – 32. Kunz, T.H. and M.B. Fenton. 2003. Bat Ecology. University of Chicago Press, Chicago,
Illinois, USA. Legakis, A., C. Papdimitrou, M. Gaetglich, and D. Lazaris. 2000. Survey of bats of the
Athens metropolitan area. Myotis 38: 41 – 46. Lesinski, G., E.B. Fuszara, and M. Kowalski. 2000. Foraging areas and relative density
of bats (Chiroptera) in differently human transformed landscapes. International Journal of Mammalian Biology 65: 129 – 137.
141
Lewis, S.E. 1995. Roost fidelity of bats: a review. Journal of Mammalogy 76(2): 481 – 496.
Li, H., and K.T. Wilkins. 2014. Patch or mosaic: bat activity responds to fine-scale urban
heterogeneity in a medium-sized city in the United States. Urban Ecosystems 17: 1013 – 1031.
Loeb, S.C., C.J. Post, and S.T. Hall. 2009. Relationship between urbanization and bat
community structure in national parks of the southeastern U.S. Urban Ecosystems 12: 197 – 214.
Loeb, S.C., T.J. Rodhouse, L.E. Ellison, C.L. Lausen, J.D. Richard, K.M. Irvine, T.E.
Ingersoll, J.T.H. Coleman, W.E. Thogmartin, J.R. Sauer, C.M. Francis, M.L. Bayless, T.R. Stanley, and D.M. Johnson. 2015. A plan for the North American bat monitoring program (NABat). General Technical Report SRS-208. United States Department of Agriculture, Forest Service, Southern Research Station, Asheville, North Carolina, USA.
Luck, G.W., L. Smallbone, C. Threlfall, and B. Law. 2013. Patterns in bat functional
guilds across multiple urban centres in south-eastern Australia. Landscape Ecology 28: 455 – 469.
MacKenzie, D.I., and J.A. Royle. 20015. Designing occupancy studies: general advice
and allocating survey effort. Journal of Applied Ecology 42: 1105 – 1114. Marks, G. E. and C. S. Marks. 2008a. Status of the Florida bonneted bat (Eumops
floridanus). Final report submitted to the U.S. Fish and Wildlife Service by the Florida Bat Conservancy under grant agreement number 401815G192. Florida Bat Conservancy, Bay Pines, Florida.
Marks, G. E. and C. S. Marks. 2008b. Bat conservation and land management
Kissimmee River WMA. Report submitted to the Florida Fish and Wildlife Conservation Commission by the Florida Bat Conservancy, Bay Pines, Florida.
Marks, G. E. and C. S. Marks. 2012. Status of the Florida bonneted bat (Eumops
floridanus). Final report to U.S. Fish and Wildlife Service by the Florida Bat Conservancy under grant agreement number 40181AG121. Florida Bat Conservancy, Bay Pines, Florida.
McCleery, R., M.K. Oli, J.A. Hostetler, B. Karmacharya, D. Greene, C. Winchester, J.
Gore, S. Sneckenberger, S.B. Castleberry, and M.T. Mengak. 2013. Are declines of an endangered mammal predation-driven, and can a captive-breeding and release program aid their recovery? Journal of Zoology 291: 59 – 68.
142
McConville, A., B. Law, T. Penman, and M. Mahony. 2014. Contrasting habitat use of morphologically similar bat species with differing conservation status in south-eastern Australia. Austral Ecology 39: 83 – 94.
McCracken, G.F. 2010. Social dispersion and genetic variation in two species of
emballonurid bats. Ethology 66(1): 55 – 69. McIntyre, N.E. 2000. Ecology of urban arthropods: a review and a call to action. Annals
of the Entomological Society of America 93(4): 825 – 835. McIntyre, S., and R. Hobbs. 1999. A framework for conceptualizing human effects on
landscapes and its relevance to management and research models. Conservation Biology 13: 1282 – 1292.
McKinney, M.L. 2002. Urbanization, biodiversity and conservation. BioScience 52(10):
883 – 890. McKinney, M.L. 2008. Effects of urbanization on species richness: a review of plants
and animals. Urban Ecosystems 11(2): 161 – 176. Menzel, J. M., W. M. Ford, M. A. Menzel, T. C. Carter, J. E. Gardner, J. D. Garner and
J. E. Hofmann. 2005. Summer habitat use and home-range analysis of the endangered Indiana bat. Journal of Wildlife Management 69(1): 430 – 436.
Menzel, M.A., T.C. Carter, B.R. Chapman, and J. Laerm. 1998. Quantitative comparison
of tree roosts used by red bats (Lasiurus borealis) and Seminole bats (L. seminolus). Canadian Journal of Zoology 76(4): 630 – 634.
Mickleburgh, S. P., A. M. Hutson, and P. A. Racey. 2002. A review of the global
conservation status of bats. Oryx 36: 18 – 34. Milne, D.J., A. Fisher, and C.R. Pavey. 2006. Models of the habitat associations and
distributions of insectivorous bats of the Top End of the Northern Territory, Australia. Biological Conservation 130: 370 – 385.
Moore, J., and R. Ali. 1984. Are dispersal and inbreeding avoidance related? Animal
Behaviour 32(1): 94 – 112. Moore, J.E., and R. K Swihart. 2005. Modeling patch occupancy by forest rodents:
incorporating detectability and spatial autocorrelation with hierarchically structured data. Journal of Wildlife Management 69(3): 933 – 949.
Morris, A.D., D.A. Miller, and L.M. Conner. 2011. Home-range size of evening bats
(Mycticeius humeralis) in southwestern Georgia. Southeastern Naturalist 10(1): 85 – 94.
143
Morris, W.F., and D.F. Doak. 2002. Quantitative conservation biology: theory and practice of population viability analysis. Sinauer Associations, Sunderland, Massachusetts, USA.
Mulkey, S. 2007. Climate change and land use in Florida: interdependencies and
opportunities. A report prepared for the Century Commission for a Sustainable Florida, 43 pages.
Myers, N. 1996. Two key challenges for biodiversity: discontinuities and synergisms.
Biodiversity Conservation 5: 1025 – 1034. Myers, N., R.A. Mittermeier, C.G. Mittermeier, G.A.B. da Fonesca, and J. Kent. 2000.
Biodiversity hotspots for conservation priorities. Nature 403: 853 – 858. Nagy, M., L. Gunther, M. Knornschild, F. Mayer. 2013. Female-biased dispersal in a bat
with a female-defence mating strategy. Molecular Ecology 3(22): 1733 – 1745. Noer, C.L., T. Dabelsteen, K. Bohmann, and A. Monadjem. 2012. Molossid bats in an
African agro-ecosystem select sugarcane fields as foraging habitat. African Zoology 47(1): 1 – 11.
Norberg, U.M., and J.M.V. Rayner. 1987. Ecological morphology and flight in bats
(Mammalia; Chiroptera): wing adaptations, flight performance, foraging strategy and echolocation. Philosophical Transactions of the Royal Society of London, B, Biological Sciences 316: 335 – 427.
Noss, R.F., E.T. LaRoe III, and J.M. Scott. 1995. Endangered ecosystems in the United
States: a preliminary assessment of loss and degradation. Biological Report 28. National Biological Service, Washington, DC, USA.
Ntzoufras, I. 2009. Bayesian Modeling Using WinBUGS. Wiley Series in Computational
Statistics, Hoboken, New Jersey, USA. O’Shea, T.J., M.A. Bogan, and L.E. Ellison. 2003. Monitoring trends in bat populations
of the United States and territories: status of the science and recommendations for the future. Wildlife Society Bulletin 31(1): 16 – 29.
O’Shea, T.J., L.E. Ellison, and T.R. Stanley. 2004. Survival estimation in bats: historical
overview, critical appraisal, and suggestions for new approaches. Pages 297 – 336 in W. Thompson, editor. Sampling Rare or Elusive Species: Concepts, Designs, and Techniques for Estimating Population Parameters. Island Press, Washington, DC, USA.
Ormsbee, P. 2010. The Bat Grid inventory and monitoring project: a regional approach
to inventorying and monitoring bat populations, 2008 inventory report. Report submitted to the Department of Defense: Legacy Resource Management. 39 pp.
144
Pena, J.E., R. Duncan, and H. Browning. 1996. Seasonal abundance of Phyllocnistis
citrella (Lepidoptera: Gracillariidae) and its parasitoids in South Florida citrus. Biological Control 25(3): 698 – 702.
Plummer, M. 2003. JAGS: a program for analysis of Bayesian graphical models using
Gibbs sampling. Proceedings of the 3rd International Workshop on Distributed Statistical Computing, March 20 – 22, Vienna, Austria.
Power, A.G. 2010. Ecosystem services and agriculture: tradeoffs and synergies.
Philosophical Transactions of the Royal Society B 365: 2959 – 2971. Pradel, R. 1996. Utilization of capture-mark-recapture for the study of recruitment and
population growth rate. Biometrics 52: 703 – 709. Pradel, R., A.R. Johnson, A. Viallefront, R.G. Nager, and F. Cezilly. 1997. Local
recruitment in the greater flamingo: a new approach using capture-mark-recapture data. Ecology 78: 1431 – 1445.
Pryde, M.A., C.F.J. O’Donnell, and R.J. Barker. 2005. Factors influencing survival and
long-term population viability of New Zealand long-tailed bats Chalinolobus tuberculatus: implications for conservation. Biological Conservation 126: 175 – 185.
Purvis, A., J.L. Gittleman, G. Cowlishaw, G.M. Mace. 2000. Predicting extinction risk in
declining species. Proceedings of the Royal Society B 267(1456): 1947 – 1952. Razgour, O., J. Hammer, and G. Jones. 2011. Using multi-scale modelling to predict
habitat suitability for species of conservation concern: the grey long-eared bat as a case study. Biological Conservation 144 (12): 2922 – 2930.
Reid, J.M., E.M. Bignal, S. Bignal, D.I. McCracken, M.I. Bogdanova, and P. Monaghan.
2008. Investigating patterns and processes of demographic variation: environmental correlates of pre-breeding survival in red-billed choughs Pyrrhocorax pyrrhocorax. Journal of Animal Ecology 77: 777 – 788.
Ricketts, T.H., G.C. Daily, P.R. Ehrlich, and J.P. Fay. 2001. Countryside biogeography
of moths in a fragmented landscape: biodiversity in native and agricultural habitats. Conservation Biology 15(2): 378 – 388.
Ritchie, J.C., J.H. Everitt, D.E. Escobar, T.J. Jackson, and M.R. Davis. 1992. Airborne
laser measurements of rangeland canopy cover and distribution. Journal of Range Management 1: 189 – 193.
145
Rhodes, M., and C. Catterall. 2008. Spatial foraging behavior and use of an urban landscape by a fast-flying bat, the Molossid Tadarida australis. Journal of Mammalogy 89(1): 34 – 42.
Rice, R.W., F.T. Izuno, and R.M. Garcia. 2002. Phosphorus load reductions under best
management practices for sugarcane cropping systems in the Everglades Agricultural Area. Agricultural Water Management 56(1): 17 – 39.
Robertson, G., and S. Swinton. 2005. Reconciling agricultural productivity and
environmental integrity: a grand challenge for agriculture. Frontiers in Ecology and the Environment 3: 38 – 46.
Rodhouse, T.J., P.C. Ormsbee, K.M. Irvine, L.A. Vierling, J.M. Szewczak, and K.T.
Vierling.2012. Assessing the status and trend of bat populations across broad geographic regions with dynamic distribution models. Ecological Applications 22: 1098 – 1113.
Rodhouse, T.J., P.C. Ormsbee, K.M. Irvine, L.A. Vierling, J.M. Szewczak, and K.T.
Vierling. 2015. Establishing conservation baselines with dynamic distribution models for bat populations facing imminent decline. Diversity and Distributions 21: 1401 – 1413.
Romesburg, H.C. 1981. Wildlife science: gaining reliable knowledge. Journal of Wildlife
Management 45(2): 293 – 313. Rowse, E.G., D. Lewanzik, E.L. Stone, S. Harris, and G. Jones. 2016. Dark matters: the
effects of artificial lighting on bats. Pages 187 – 213 in C.C. Voigt and T. Kingston, editors. Bats in the Anthropocene: Conservation of Bats in a Changing World. Springer Press, New York City, New York, USA.
Royle, J.A., and R.M. Dorazio. 2008. Hierarchical modeling and inference in ecology.
Elsevier, London, UK. Russo, D., and L. Ancillotto. 2014. Sensitivity of bats to urbanization: a review.
Mammalian Biology 80: 205 – 212. Safi, K., and G. Kerth. 2004. A comparative analysis of specialization and extinction risk
in temperate-zone bats. Conservation Biology 18: 1293 – 1303. Scanlon, A.T., and S. Petit. 2008. Biomass and biodiversity of nocturnal aerial insects in
an Adelaide City park and implications for bats (Microchiroptera). Urban Ecosystems 11: 91 – 106.
Schorcht, W., F. Bontadina, and M. Schaub. 2009. Variation of adult survival drives
population dynamics in a migrating forest bat. Journal of Animal Ecology 78(6): 1182 – 1190.
146
Schwartz, B.E., and L.F. Bosart. 1979. The diurnal variability of Florida rainfall. Monthly
Weather Review 107: 1535 – 1545. Sexton, J.P., P.M. McIntyre, A.L. Angert and K.J. Rice. 2009. Evolution and ecology of
species range limits. Annual Review of Ecology, Evolution, and Systematics 40: 415 – 436.
Silva, P.S.D., C.M. Knoechelmann, M. Tabarelli, and J.S. Almeida-Cortez. 2011.
Richness of gall morphospecies along a secondary successional gradient of Atlantic Forest in northeastern Brazil. Revista Brasileira de Biociencias 9(3): 270 – 277.
Solecki, W. D. 2001. The role of global-to-local linkages in land use/land cover change
in south Florida. Special section – south Florida: the reality of change and the prospects for sustainability. Ecological Economics 37: 339 – 356.
Spiegelhalter, D. J. et al. 2003. WinBUGS User Manual (Version 1.4). Cambridge:
MrcBiostatistics Unit. <http://www.mrc-bsu.cam.ac.uk/bugs/>. Su, Y., and M. Yajima. 2011. R2jags: a package for running jags from R.
<http://CRAN.R-project.org/package=R2jags>. R package version 002-17. Tear, T.H., J.M. Scott, P.H. Hayward, and B. Griffith. 2002. Recovery plans and the
Endangered Species Act: are criticisms supported by data? Conservation Biology 9(1): 182 – 195.
Threlfall, C., B. Law, T. Penman and P.B. Banks. 2011. Ecological processes in urban
landscapes: mechanisms influencing the distribution and activity of insectivorous bats. Ecography 34: 814 – 826.
Timm, R. M. and H. H. Genoways. 2004. The Florida bonneted bat, Eumops floridanus
(Chiroptera: Molossidae): distribution, morphometrics, systematics, and ecology. Journal of Mammalogy 85: 852 – 865.
Turbill, C., C. Bieber, and T. Ruf. 2011. Hibernation is associated with increased
survival and the evolution of slow life histories among mammals. Proceedings of the Royal Society of London B: Biological Sciences 278(1723): 3355 – 3363.
Tuttle, M.D., and D. Stevenson. 1982. Growth and survival of bats. Pages 105 – 150 in
T.H. Kunz, editor. Ecology of Bats, Springer US, New York, New York, USA. Tyre, A.J., B. Tenhumberg, S.A. Field, D. Niejalke, and K. Parris. 2003. Improving
precision and reducing bias in biological surveys: estimating false-negative error rates. Ecological Applications 13(6): 1790 – 1801.
147
U.S. Fish and Wildlife Service [USFWS]. 2011. U.S. Fish and Wildlife Service Species Assessment and Listing Priority Assignment. <http://pdadupws.nrc.gov/docs/ML1219/ML12193A530.pdf>.
U.S. Fish and Wildlife Service [USFWS]. 2012. Propose endangered species status for
the Florida bonneted bat. Federal Register 77(193): 60750 – 60776. U.S. Fish and Wildlife Service [USFWS]. 2013. Endangered species status for the
Florida bonneted bat. Federal Register 78(191): 61004 – 61043. Valiente-Banuet, A. et al. 2015. Beyond species loss: the extinction of ecological
interactions in a changing world. Functional Ecology 29: 299 – 307. Voigt, C.C., and T. Kingston. 2016. Bats in the Anthropocene. Pages 1 – 9 in C.C. Voigt
and T. Kingston, editors. Bats in the Anthropocene: Conservation of Bats in a Changing World. Springer Press, New York City, New York, USA.
Walker, E.P. 1964. Mammals of the World. Johns Hopkins Press, Baltimore, Maryland,
USA. Walters, B.L., C.M. Ritzi, D.W. Sparks, J.O. Whitaker. 2007. Foraging behavior of
eastern red bats (Lasiurus borealis) at an urban-rural interface. The American Midland Naturalist 157(2): 365 – 373.
Wear, D.N., and J.G. Greis. 2002. Southern forest resource assessment – technical
report. General technical report SRS-53. U.S. Department of Agriculture, Forest Service, Southern Research Station, Asheville, North Carolina. <http://www.srs.fs.usda.gov/sustain/report/index.htm>.
Wear, D. N. and J. G. Greis. 2011. The Southern Forest Futures Project: summary
report. Draft report. U.S. Department of Agriculture, Forest Service, Southern Research Station. Research Triangle Park, North Carolina. <http://www.srs.fs.usda.gov/futures/>.
Webb, S. D. 1999. The South Florida Ecosystem. Pages 1 – 84 in U.S. Fish and Wildlife
Service, Multi-Species Recovery Plan for South Florida. Fish & Wildlife Service Reference Service, Bethesda, Maryland, USA.
Weller, T.J. 2008. Using occupancy estimation to assess the effectiveness of a regional
multiple-species conservation plan: bats in the Pacific Northwest. Biological Conservation 141: 2279 – 2289.
Wilcove, D.S., D. Rothstein, J. Dubow, A. Phillips, and E. Losos. 1998. Quantifying
threats to imperiled species in the United States. BioScience 48(8): 607 – 615. Wilkins, K.T. 1989. Tadarida brasiliensis. Mammalian Species 331: 1 – 10.
148
Wilkinson, G.S., and J.M. South. 2002. Life history, ecology and longevity in bats. Aging
Cell 1(2): 124 – 131. Williams, C.M. 1951. Biochemical mechanisms in insect growth and metamorphosis.
Federation Proceedings 10(2): 546 – 552. Williams-Guillen, K., E. Olimpi, B. Maas, P.J. Taylor and R. Arlettaz. 2016. Bats in the
anthropogenic matrix: challenges and opportunities for the conservation of Chiroptera and their ecosystem services in agricultural landscapes. Pages 151 – 186 in C.C. Voigt and T. Kingston, editors. Bats in the Anthropocene: Conservation of Bats in a Changing World. Springer Press, New York City, New York, USA.
Whalen, P.J., and B.M. Whalen. 1994. Nonpoint source regulatory program for the
Everglades Agricultural Area. Paper FL94-101. Florida Section of the American Society of Agricultural Engineers.
White, G.C., and K.P. Burnham. 1999. Program MARK: survival estimation from
populations of marked animals. Bird Study 46 Supplement, 120 – 138. Whitaker, J.O. Jr., and S.L. Gummer. 2003. Current status of the evening bat,
Nycticeius humeralis, in Indiana. Proceedings of the Indiana Academy of Science 112(1): 55 – 60.
Whitaker, J.O. Jr., V. Brack Jr., and J.B. Cope. 2002. Are bats in Indiana declining?
Proceedings of the Indiana Academy of Science 105: 87 – 94. Wright, D.H. 1983. Species-energy theory: an extension of species-area theory. Oikos
41: 496 – 506. Zwick, P.D., and M.H. Carr. 2006. Florida 2060. A population distribution scenario for
the State of Florida. A research project prepared for 1000 Friends of Florida. Prepared by the Geoplan Center at the University of Florida, Gainesville, Florida.
149
BIOGRAPHICAL SKETCH
Mandy graduated summa cum laude from the University of Maine in May 2012
with a Bachelor of Science in wildlife ecology. Mandy went on to pursue a Master of
Science in wildlife ecology and conservation at the University of Florida, which she
completed in the spring of 2016.
