Assessing the Relationship Between Land Cover Variables and the Diversity of Acoustic Bat Detections in Urban and Rural Areas
Nicholas Comparato1*
1Miami University, 501 E High St, Oxford, OH 45056. *Corresponding author.
Urban Naturalist, No. 67 (2023)
Abstract
Urbanization is a growing threat to native biodiversity around the world and one that
is exacerbating a host of other conservation pressures currently impacting North American bat populations. Acoustic surveys are the most accessible means of monitoring how bats are responding to these pressures. This study examines the interaction between land cover variables and the species richness of acoustic bat detections in an effort to assess how urban land cover impacts the results of acoustic surveys. Five parks in Westchester County, New York, USA, a region located directly north of New York City, were surveyed for two seasons with an acoustic bat detector. These parks represent a diverse sampling of the urban-to-rural spectrum, ranging from a small park located in an urban city to an expansive park surrounded by rural areas. The land cover surrounding these 5 parks was quantified in 4 categories: urban, low development, water, and tree cover. To account for the varying roosting and foraging ranges of different bat species, this quantification was carried out at 4 different spatial scales: 3 km, 1 km, 500 m, and 100 m. The relationship between land cover variables at each scale and the species richness of acoustic detections was then modeled using linear mixed effects models. The results indicated that urban land cover and low-development land cover both had a negative impact on the diversity of acoustic bat detections, while tree cover has a positive impact.
Download Full-text pdf
Site by Bennett Web & Design Co.
Volume 10, 2023 Urban Naturalist No. 67
Assessing the Relationship
Between Land Cover
Variables and the Diversity
of Acoustic Bat Detections
in Urban and Rural Areas
Nicholas Comparato
Urban Naturalist
The Urban Naturalist (ISSN # 2328-8965) is published by the Eagle Hill Institute, PO Box 9, 59 Eagle Hill Road, Steuben, ME 04680-
0009. Phone 207-546-2821 Ext. 4, FAX 207-546-3042. E-mail: office@eaglehill.us. Webpage: http://www.eaglehill.us/urna. Copyright
© 2023, all rights reserved. Published on an article by article basis. Special issue proposals are welcome. The Urban Naturalist is
an open access journal. Authors: Submission guidelines are available at http://www.eaglehill.us/urna. Co-published journals: The
Northeastern Naturalist, Southeastern Naturalist, Caribbean Naturalist, and Eastern Paleontologist, each with a separate Board of
Editors. The Eagle Hill Institute is a tax exempt 501(c)(3) nonprofit corporation of the State of Maine (Federal ID # 010379899).
Board of Editors
Myla Aronson, Rutgers University, New Brunswick, NJ,
USA
Joscha Beninde, University of California at Los Angeles,
CA, USA.
Sabina Caula, Universidad de Carabobo, Naguanagua,
Venezuela
Sylvio Codella, Kean University, Union New Jersey, USA
Julie Craves, University of Michigan-Dearborn, Dearborn,
MI, USA
Ana Faggi, Universidad de Flores/CONICET, Buenos
Aires, Argentina
Leonie Fischer, Technical University of Berlin, Berlin,
Germany
Chad Johnson, Arizona State University, Glendale, AZ,
USA
Erik Kiviat, Hudsonia, Bard College, Annandale-on-
Hudson, NY, USA
Sonja Knapp, Helmholtz Centre for Environmental
Research–UFZ, Halle (Saale), Germany
David Krauss, City University of New York, New York,
NY, USA
Joerg-Henner Lotze, Eagle Hill Institute, Steuben, ME.
Publisher
Kristi MacDonald, Hudsonia, Bard College, Annandale-on-
Hudson, NY, USA
Ian MacGregor-Fors, University of Helsinki, Finland.
Editor
Tibor Magura, University of Debrecen, Debrecen, Hungary
Brooke Maslo, Rutgers University, New Brunswick, NJ,
USA
Mark McDonnell, Royal Botanic Gardens Victoria and
University of Melbourne, Melbourne, Australia
Mike McKinney, University of Tennessee, Knoxville, TN,
USA
Desirée Narango, University of Massachusetts, Amherst,
MA, USA
Joseph Rachlin, Lehman College, City University of New
York, New York, NY, USA
Travis Ryan, Center for Urban Ecology, Butler University,
Indianapolis, IN, USA
Michael Strohbach, Technische Universität Braunschweig,
Institute of Geoecology, Braunschweig, Germany
Katalin Szlavecz, Johns Hopkins University, Baltimore,
MD, USA
Alan Yeakley, Portland State University, Portland, OR,
USA
♦ The Urban Naturalist is a peer-reviewed and
edited interdisciplinary natural history journal
with a global focus on urban areas (ISSN 2328-
8965 [online]).
♦ The journal features research articles, notes,
and research summaries on terrestrial, freshwater,
and marine organisms and their habitats.
♦ It offers article-by-article online publication
for prompt distribution to a global audience.
♦ It offers authors the option of publishing large
files such as data tables, and audio and video
clips as online supplemental files.
♦ Special issues - The Urban Naturalist welcomes
proposals for special issues that are based
on conference proceedings or on a series of invitational
articles. Special issue editors can rely
on the publisher’s years of experiences in efficiently
handling most details relating to the
publication of special issues.
♦ Indexing - The Urban Naturalist is a young
journal whose indexing at this time is by way of
author entries in Google Scholar and Researchgate.
Its indexing coverage is expected to become
comparable to that of the Institute's first 3 journals
(Northeastern Naturalist, Southeastern Naturalist,
and Journal of the North Atlantic). These 3
journals are included in full-text in BioOne.org
and JSTOR.org and are indexed in Web of Science
(clarivate.com) and EBSCO.com.
♦ The journal's staff is pleased to discuss ideas
for manuscripts and to assist during all stages of
manuscript preparation. The journal has a page
charge to help defray a portion of the costs of
publishing manuscripts. Instructions for Authors
are available online on the journal’s website
(http://www.eaglehill.us/urna).
♦ It is co-published with the Northeastern Naturalist,
Southeastern Naturalist, Caribbean Naturalist,
Eastern Paleontologist, Eastern Biologist,
and Journal of the North Atlantic.
♦ It is available online in full-text version on the
journal's website (http://www.eaglehill.us/urna).
Arrangements for inclusion in other databases
are being pursued.
Cover Phograph: A silver-haired bat (Lasionycteris noctivagans) seen on the Bronx River Parkway in
Bronxville, NY in the early spring of 2020. © Nic Comparato.
Urban Naturalist
N. Comparato
2023 No. 67
1
2023 Urban Naturalist 67:1–14
Assessing the Relationship Between Land Cover Variables
and the Diversity of Acoustic Bat Detections in Urban and
Rural Areas
Nicholas Comparato1*
Abstract–Urbanization is a growing threat to native biodiversity around the world and one that
is exacerbating a host of other conservation pressures currently impacting North American bat
populations. Acoustic surveys are the most accessible means of monitoring how bats are responding
to these pressures. This study examines the interaction between land cover variables and the species
richness of acoustic bat detections in an effort to assess how urban land cover impacts the results of
acoustic surveys. Five parks in Westchester County, New York, USA, a region located directly north of
New York City, were surveyed for two seasons with an acoustic bat detector. These parks represent a
diverse sampling of the urban-to-rural spectrum, ranging from a small park located in an urban city to
an expansive park surrounded by rural areas. The land cover surrounding these 5 parks was quantified
in 4 categories: urban, low development, water, and tree cover. To account for the varying roosting
and foraging ranges of different bat species, this quantification was carried out at 4 different spatial
scales: 3 km, 1 km, 500 m, and 100 m. The relationship between land cover variables at each scale and
the species richness of acoustic detections was then modeled using linear mixed effects models. The
results indicated that urban land cover and low-development land cover both had a negative impact
on the diversity of acoustic bat detections, while tree cover has a positive impact.
Introduction
Urbanization describes a process in which natural habitats are destroyed and replaced
with impervious structures, characteristic of urban environments, most notably roads and
buildings. Such environments typically offer little in terms of habitat for wildlife beyond
heavily manicured parks. Pushed by an ever-growing movement of human populations
to urban centers, the spread of urbanization has been directly linked to a loss in native
biodiversity and the emergence of monocultures of urban-tolerant species (McKinney
2002). Bats are among the many taxonomic families to have experienced shrinking diversity
across habitat ranges due to this growth of urban environments. Previous research has
suggested that the light pollution, roadways, reduced insect abundance, and reduced tree
cover associated with urbanization have all contributed to the declining diversity of global
bat populations (Moretto and Francis 2017).
In North America, urbanization is just one of a host of conservation pressures currently
threatening bat populations. White-nose syndrome, a fatal condition caused by the fungal
pathogen Pseudogymnoascus destructans Minnis and Lindner has devastated cave-hibernating
obligate species throughout the eastern half of the continent. Spread of this fungal pathogen
has caused population declines of >95% in some cases (Hoyt et al. 2021). The increased
prevalence of utility-scale wind energy developments is also contributing to declining bat
populations. Expansive fields of wind turbines constructed along the pathways of migratory
species have resulted in large numbers of bat fatalities (Arnett and Baerwald 2013). These
threats have each contributed a present state of crisis in North American bat conservation.
1Miami University, 501 E High St, Oxford, OH 45056. *Corresponding author – ncomparato@gmail.com
Associate Editor: Michael McKinney, University of Tennessee
Urban Naturalist
N. Comparato
2023 No. 67
2
Surveying and monitoring bat populations are crucial tasks in maintaining an accurate
assessment of these conservation issues and their developing impacts. The use of specialized
acoustic detection devices is the most accessible and least interruptive means of doing so.
Unlike other methods, such as netting and roost counts, acoustic detection does not require
any direct interaction or proximity between humans and bats (Szewczak and Morrison
2020). The development of low-cost detection devices, like the AudioMoth, have further
increased the accessibility of this approach by addressing the once formidable barrier of
equipment costs (Hill et al. 2019). Though acoustic detection has yet to achieve the level
of species identification accuracy afforded by directly observing and handling bats, it has
gained widespread adoption in the research literature and among large scale monitoring
efforts (Loeb et al. 2009, 2015, Parkins and Clark 2015, Gallo et al. 2018).
This study analyzes the impact of urbanization on the diversity of bat species detected
in Westchester County, New York during acoustic surveys conducted in 2020 and 2021.
Westchester County is well suited to studies of urbanization as it is located directly north of
New York City, the most densely populated urban region in the United States (US Census
Bureau 2023). While the county is heavily developed and populated along its southern
border with New York City, it gradually transitions to suburban and rural environments
located in the northern half of the county (Westchester County Department of Planning
2011). Survey data came from five county parks located in the southern, central, and
northern regions of this area (Fig. 1). Each park is surrounded by a different level of urban
development, ranging from a small park located within a heavily developed city to a ~4000
acre park surrounded by rural townships. I chose the metric of species richness to examine
how these surroundings impacted the biodiversity of local bat populations. Species richness
measures biodiversity in its simplest terms, by counting the number of species present in
the composition of a population. This metric works well with the limitations of acoustic
detection as this method can establish the likely presence of a species but not how prevalent
the species is (Loeb et al. 2009).
There is a substantial body of literature examining the relationship between urbanization
and the health of bat populations. Previous studies have established that tolerance of
urban environments varies and is based on land cover characteristics at different spatial
scales, depending on the species (Dixon 2011, Gallo et al. 2018). Some bat species are
quite tolerant of the urban environment, especially when provided with improved habitat
options like green roofs (Parkins and Clark 2015). Other species, like Myotis lucifugus Le
Conte (Little brown bat), may readily inhabit urban areas, but are actually most successful
in transitional environments between urban and rural areas (Coleman and Barclay 2011).
A general trend among studies of urban bats, however, is that species diversity is lower in
urban areas than in less developed areas (Moretto et al. 2017).
This study contributes to the literature on bats and urbanization by specifically examining
what impact urbanization can be expected to have on the number of species recorded by
an acoustic bat detector on any given night. Following the examples of Dixon (2011) and
Gallo et al., (2018), I chose to examine this potential impact at multiple spatial scales
to account for the varying foraging and roosting ranges of the bat species present in the
study area. Survey locations were placed along the urban-to-rural spectrum by quantifying
what percentage of the surrounding land cover fell into four different categories: urban,
low development, tree cover, and water. My primary hypothesis (H1) was that urban land
cover would have a negative relationship with the number of bat species detected across
all spatial scales. A related secondary hypothesis (H2) was that tree cover would have a
positive relationship with the number of species detected at all spatial scales. The low
Urban Naturalist
N. Comparato
2023 No. 67
3
development category covered pasture lands and similar areas dominated by open grass
with sparse anthropogenic structures. Previous research has suggested such areas may
benefit bat populations (Coleman and Barclay 2011), thus a tertiary hypothesis (H3) was
that low development land cover would also have a positive relationship with the diversity
of bat species detected.
Site Description
Westchester County encompasses 450 square miles (~1165 sq. km) in southern New
York. It is bordered by large bodies of brackish water on both sides, with the western border
defined by the Hudson River and the eastern border by the Long Island Sound. The county
has maintained a consistent population of ~1,000,000 people in recent years, who inhabit
a total of 45 cities and towns spread throughout the area (Westchester County Government
2022). The bat surveys generating data for this research were originally conducted on
behalf of the county’s Parks and Recreation Department with the intent of assessing local
species diversity. Site selection for these surveys was heavily influenced by the onset of
the COVID-19 pandemic and other stakeholder issues. Public health measures and park
administrative complications limited access to potential survey locations. The five parks
included were selected from a group of parks I was given permission to conduct bat surveys
in. These parks were selected based on the diversity of the surrounding environments in
terms of the rural-to-urban spectrum.
The most urban of these parks was Lenoir Preserve, a 40-acre park surrounding a
historic mansion. It is located near the southwestern border Westchester County shares
with the Bronx in the city of Yonkers. Marshlands Conservancy, located in Mamaroneck
along the Long Island Sound, offers 147 acres of parkland in southeastern Westchester
amidst an otherwise heavily developed suburban landscape. This park is unique from the
other survey locations in that it is located on the coast and contains an extensive network of
brackish wetlands. Cranberry Lake Preserve is a 190-acre park surrounding a glacial lake
in central Westchester. The area immediately surrounding the park features relatively little
development. However, it is only ~ 6.5 km from the urban center of White Plains. Muscoot
Farms is an interpretive farm surrounded by an expansive 777 acres of parkland. It is
located near the town of Katonah in northern Westchester, a region that is substantially less
developed and densely populated than the county’s southern reaches. At 4315 acres, Ward
Pound Ridge Reservation is the largest park and most rural area surveyed in this study. It is
located in northeastern Westchester along the Connecticut border (See Fig. 1).
Materials and Methods
Passive acoustic surveys
I conducted passive acoustic surveys from mid-May to mid-August of 2020 and 2021.
For each survey, a single Pettersson D500X bat detector with an external microphone
was set up at a location within one of the five parks. I selected survey locations based on
their suitability for capturing high quality recordings of bat sonar. These were locations
near forest edges in which the microphone could be aimed at an open area with no
reflective surfaces that may cause recording interference. To further reduce the risk of
interference, the microphone was elevated four meters from the ground on a telescoping
pole (Szewczak and Morisson 2020). For each deployment, I left the bat detector in place
until it recorded a minimum of five nights with optimal weather for bat activity. Ancillary
Urban Naturalist
N. Comparato
2023 No. 67
4
data collected with each deployment consisted of location coordinates, microphone
direction, and weather data for each night (precipitation, hourly wind speeds, and high
and low temperatures). This survey protocol was based on the recommendations provided
by the USGS NABat program (Loeb et al. 2015).
Bat calls recorded during surveys were analyzed using SonoBat 4.5.5, North America.
SonoBat’s automatic classification tool was used to initially classify recordings to the species
level. To ensure the accuracy of SonoBat’s classifications, I then manually vetted its suggested
species classifications. Based on NABat’s suggested protocol for data analysis, I examined
calls for each survey night until I could find at least one high quality, conclusive recording for
each species suggested to be present by the software (Reichert et al. 2018).
Land cover analysis
Land cover analyses of the five survey locations were performed using QGIS 3.28.3
and the 2019 National Land Cover Database (NLCD) map for the contiguous United
States. The NLCD map classifies land cover into a number of different categories. Using
the “Raster Calculator” tool on QGIS, I split the NLCD map into four component maps,
each aggregating several NLCD categories into one overarching category. The map of
urban land cover consisted of the NLCD categories “developed low intensity”, “developed
medium intensity”, and “developed high intensity”. A map of tree cover was created from
the NLCD categories “deciduous forest”, “evergreen forest”, and “mixed forest”. Another
map was created of bodies of water by aggregating the NLCD categories “open water”,
“woody wetlands”, and “herbaceous wetlands”. A final map was made to account for
areas that featured some development, but were not developed enough to be included on
the urban map. This map of low development areas combined the NCLD categories of
“developed open space” and “pasture/hay”. Though there are other categories included
in the NLCD map, none of them were present enough around the survey locations to
collectively account for anything close to 1% of the total land cover.
The survey locations were plotted on the four component maps and a series of fixedradius
spatial buffers of 100 m, 500 m, 1 km, and 3 km were established around each
survey location. Each buffer represents a different scope of interest for this analysis.
Figure 1. Survey locations.
Urban Naturalist
N. Comparato
2023 No. 67
5
The scales of 100 m, 500 m, and 1 km were taken from studies on multi-scale, spatial
effects by Dixon (2011), and Gallo et al., (2018). These studies selected 100 m as a “local
scale” because it was the smallest area at which substantial variation in land cover can be
observed. The “broad scale” of 1 km was selected because it captures the known foraging
ranges of multiple common bat species. An “intermediary scale” was added at the halfway
point of 500 m between the local and broad scales. I added a fourth scale of 3 km to
account for the occasional larger movements of migratory species like Lasionycteris
noctivagans (McGuire et al. 2012). The percentages of each land cover type (urban,
low development, tree cover, and water) present in the buffer zones at each scale were
calculated using the “overlap analysis” tool in QGIS.
Statistical analysis
Five nights of detection data were selected from a deployment of the Pettersson
D500X in each of the five survey locations, giving a total sample size of n=25 detector
nights. Five was chosen because it was the minimum number of clear weather nights the
detector was left in each location. For survey locations that had more than five nights of
data, the first consecutive, or closest to consecutive five, nights were selected. For each
night, the total number of verifiable species detected were recorded with the land cover
percentages of the survey location at 100 m, 500 m, 1 km, and 3 km. Seasonality was not
accounted for in the planning of the surveys, and the potential random influence of this
factor needed to be accounted for in calculating any relationship between land cover and
species diversity. To facilitate this, the month in which the survey was conducted was
also included in the data.
All statistical analyses were performed on JASP 0.17.1. I first checked the normal
distribution of all variables using Q-Q plots. Urban land cover at the 100 m scope was the
only variable that did not have a normal distribution. Even after log10 transformation,
this variable could not be normalized and it was subsequently removed from any further
analysis. Linear mixed effects models (LMMs) were used to examine the relationships
between land cover variables and species diversity. I chose this statistical approach
because it allows for greater accuracy in calculating relationships between variables
when there is potential influence from a random factor not accounted for in study design.
A conventional rule of LMMs is that the random grouping variable should have at least 5
levels, however the validity of this convention has been disputed (Gomes 2022). In this
case the grouping factor, month, only had 4 levels as the bat detection season runs from
May to August. I chose to continue with LMMs after efforts to analyze this data with
fixed-effect regressions consistently yielded high RMSE scores.
The small size of the data set required a simple LMM structure to avoid singular fit. For
most land cover variables, a random intercept LMM was used to assess the relationship
with diversity of detections. This suggests that the effect size of most variables on the
number of species detected remained consistent from one month to the next, despite
the seasonal presence or absence of different species. Tree cover at 1 km and water at
3 km both required a random slope-only LMM to avoid singular fit. This indicated that
the effect size of these two variables on the richness of species detections differed from
month to month. If the LMM for a land cover variable resulted in a singular fit even after
simplification measures, the variable was discarded from the study. Tree cover at 500 m
and 100 m, water at 1km and 100 m and low development at 3 km, 500 m and 100 m were
all discarded due to singular fit.
Urban Naturalist
N. Comparato
2023 No. 67
6
Results
Passive acoustic surveys
In total, the equipment ran for 40 monitoring nights, producing 3236 recordings. Of these,
1608 (~50%) yielded calls that could be identified to the species level. The most common
species detected was Eptesicus fuscus Beauvois (Big brown bat), with 1104 detections
across all five parks. Myotis leibii Audubon and Bachman (Eastern small-footed bat) was
the next most prominent species with 263 detections. This unique species, which prefers to
roost under lose rocks on sunny talus slopes, was detected exclusively at Ward Pound Ridge
Reservation (WPRR). Lasiurus cinereus Beauvois (Hoary bat) was another species detected
at a single location. This species was detected 96 times at Muscoot Farms. Myotis lucifugus
was detected in small numbers at all parks with the exception of Marshlands Conservancy
(MC), producing a total of 94 detections. The long-distance migratory species Lasionycteris
noctivagans Le Conte (Silver-haired bat) was detected at both MC and WPRR, however
the WPRR detections proved to most likely be misidentified calls by E. fuscus upon
vetting. Only the 36 detections from MC were included in subsequent analyses. Perimyotis
subflavus Cuvier (Tri-colored bat) was detected once at WPRR and 8 times at Cranberry
Lake Preserve (CLP). Lasiurus borealis Müller (Eastern red bat), a bat that is normally
common in southern New York (Parkins and Clark 2015), was only detected 3 times, all of
which came from Lenoir Preserve (LP, Table 1).
Land cover analysis
At the 3 km scope, MC and LP feature the most surrounding urban land cover at roughly
36% and 35% respectively. CLP was surrounded by 24% urban land cover at this scope
and MF was surrounded by 9%. WPRR has the least surrounding urban land cover at 3 km
with 2%, but the most surrounding tree cover at this scope with 76%. MF has the next most
tree cover at 3 km with 65%, followed by CLP at 33% and LP at 16%. MC has the least
surrounding tree cover at 3 km, with only 2%. A coastal park, MC also has the most water
at the 3 km scope with 42%. LP is near the coast and features 30% water at this scope. CLP
has 21% water at this scope, WPRR has 14%, and MF has 12%. CLP has the most low
development land cover at 3km with 21%, followed by MC at 20% and LP at 18%. MF is
surrounded by 13% low development land cover at 3 km and WPRR is surrounded by 8%.
At 1 km, the percentage of urban land cover surrounding LP increases to 41%. For MC,
on the other hand, urban land cover percentage decreases to 19% at 1 km. Urban land cover
percentage at this scope around CLP is 13%. MF is 3% and WPRR is 2%. WPRR continues
to have the most tree cover, being surrounded by 85% tree cover at the 1 km scope. MF is
surrounded by 58% tree cover at this scope and CLP is surrounded by 52%. LP features 19%
tree cover at 1 km and MC features 13%. MC has the most water at 1 km with 34%. MF
follows at 30%, LP at 18%, CLP at 13%, and WPRR at 6%. MC also features the most low
development land cover at the 1 km scope with 34%. LP follows at 21% and CLP at 20%.
Table 1. Species detected in each park.
Park Species Detected
Lenoir Preserve E. fuscus, M. lucifugus, L. borealis
Marshlands Conservancy E. fuscus, L. noctivagans
Cranberry Lake Preserve E. fuscus, M. lucifugus, P. subflavus
Muscoot Farms E. fuscus, M. lucifugus, L. cinereus
Ward Pound Ridge Reservation E. fuscus, M. lucifugus, P. subflavus, M. leibii
Urban Naturalist
N. Comparato
2023 No. 67
7
WPRR features 8% low development land cover at 1 km and MF feat ures 7%.
At 500 m, LP featured 37% urban land cover. MC followed at 12%, CLP and MF at
4%, and WPRR at 1%. WPRR is surrounded by 89% tree cover at 500 m. MF also features
a lot of tree cover at this scope with 78%, as does CLP with 73%. MC features 39% tree
cover at 500 m and LP features 34%. MC continues to feature the most water, with 25%
water at 500 m. MF is surrounded by 13% water at this scope, CLP at 8%, WPRR at 6%,
and LP at 2%. LP features the most low development land cover at 500 m with 25%, with
MC following closely at 23%. CLP is surrounded by 14% low development at this scope,
MF by 4%, and WPRR by 2%.
At 100 m, most parks did not have any urban land cover surrounding them. LP was the
only one, with 16% urban land cover at this scope. Tree cover percentage was relatively
high across all parks. MC features 74% tree cover at 100 m, WPRR features 71%, and CLP
features 60%. LP is surrounded by 57% at this scope and MF by 51%. MF is surrounded
by the most water at 100 m with 47%. CLP features 40% water at this scope and WPRR
features 29%. MC and LP both feature 0% water at 100 m. CLP, WPRR, and MF all
featured 0% low development land cover at this scope. LP is surrounded by 27% low
development at 100 m and MC by 26% (Fig. 2).
Significant relationships with land cover variables
Urban, low development, and water all had a negative effect on the species richness of
acoustic detections each night, while tree cover was the only land cover variable to have
a positive effect. The effect size of all land cover variables had an inverse relationship
Figure 2. Land cover composition by park and spatial scale.
Urban Naturalist
N. Comparato
2023 No. 67
8
with spatial scale. Effect sizes grew slightly as the spatial scope of consideration was
reduced. No significant relationship between any land cover variables and the number of
species detected was found at 100 m. The effect sizes and estimated marginal means for all
significant relationships are listed in Table 2.
Urban. Urban land cover was found to have a significant (p <0.05) negative relationship
with the number of species detected on each night at the 3 km, 1 km, and 500 m scales.
Spatial scale and the effect size of urban land cover had an inverse relationship, with the
effect size growing by 0.01 species per percentage of urban land cover as the spatial scale
got smaller. The estimated marginal means suggest that surveys in parks with a lower
percentage of surrounding urban land cover were more likely to detect 1 additional species
than parks surrounded by a high percentage on any given night. This estimate becomes less
reliable at the 500 m scale, however, as the estimated marginal means were calculated on a
range of 25% to − 2%, the latter of which is not an actual poss ibility.
Tree cover. Tree cover was found to have a significant (p <0.05) positive relationship
with the number of species detected on each night at 3 km and 1km. Spatial scales and the
effect size of tree cover had an inverse relationship, with effect size growing by 0.02 from
3 km to 1 km. The estimated marginal means indicate that surveys in parks surrounded by
a high percentage of tree cover were more likely to detect 1 additional species than parks
surrounded a low percentage on any given night.
Water. Water was found to have a significant (p <0.05) negative relationship with the
number of species detected on each night at 3 km and 500 m. Spatial scales and the effect
size of water had an inverse relationship, with the effect size growing by 0.08 between 3 km
and 500 m. The estimated marginal means indicate that surveys in parks surrounded by a low
percentage of water were more likely to detect 1 additional species than parks surrounded by
a high percentage on any given night. Both the effect size and marginal means suggest water
Table 2. Effect size of each land cover variable and its related estimated marginal means.
3 km (+/−) Range at
3 km 1 km (+/−) Range at
1 km 500 m (+/−) Range at
500 m
Urban −0.024 (0.008) — −0.025 (0.008) — −0.026 (0.009) —
EMM* High 2.03 (0.163) 35.00% 2 (0.164) 30.00% 2.02 (0.174) 25.00%
EMM Low 2.7 (0.164) 7.00% 2.73 (0.167) 1.00% 2.73 (0.179) −2.00%
L.D.** — — −0.026 (0.012) — — —
EMM High — — 2.09 (0.202) 28.00% — —
EMM Low — — 2.62 (0.197) 8.00% — —
Tree Cover 0.011 (0.004) 0.013 (0.005) — — —
EMM High 2.7 (0.166) 67.00% 2.72 (0.197) 72.00% — —
EMM Low 2.04 (0.166) 10.00% 2.02 (0.159) 18.00% — —
Water −0.03 (0.012) — — — −0.022 (0.017) —
EMM High 2.015 (0.213) 35.00% — — 2.170 (0.309) 19.00%
EMM Low 2.7 12.00% — — 2.533 (0.294) 3.00%
*Estimated marginal means, ** Low Development
Note: EMM High represents areas with a high percentage of the relevant land cover variable and EMM
low represents area with a low percentage. The adjacent column, “Range at x”, describes the range of
percentages the EMMs were calculated at.
Urban Naturalist
N. Comparato
2023 No. 67
9
had less impact on the number of species detected than urban land cover or tree cover.
Low development. Low development land cover was found to have a significant (p
<0.05) negative relationship with the number of species detected on each night at 1 km.
As this relationship could not be calculated at any other scale, the relationship between
scale and effect size for low development land cover cannot be described from this data.
It is interesting to note that the effect size of low development land cover at 1 km is 0.026
species per percentage of low development land cover, which is the same effect size urban
land cover had at 500 m. The estimated marginal means suggest surveys in parks surrounded
by a low percentage of low development land cover were more likely to detect 1 additional
species than parks surrounded by a high percentage (See Table 2).
Discussion
Comparing species between Westchester and New York City
There are several recent studies of bat populations within New York City (NYC) that provide
a valuable comparison for the results of acoustic surveys in Westchester. In their acoustic
study of green roofs within NYC, Parkins and Clark (2015) found Lasiurus borealis to
be the most common bat in the area, making up over 60% of the calls they were able to
identify. The prevalence of this species in NYC was more recently confirmed by Partridge,
et al. (2020) in a study that preceded my earliest surveys by only 1 year. That this species
was only detected three times during the two years of acoustic bat surveys in neighboring
Westchester County is very surprising. L. borealis is a migratory species and it may be the
case that the timing of my surveys simply missed the seasonal window in which they would
be passing through Westchester. Their prevalence in the results of these prior, season-long
acoustic surveys suggests, however, that NYC hosts a resident population. This species has
previously been found to be tolerant of urban environments and there is conflicting evidence
as to whether it prefers them for foraging and roosting (Walters et al. 2007, Dixon 2011).
It is notable that the three L.borealis detections were in the most urban location surveyed,
Lenoir Preserve. These unexpectedly low numbers may also be cause for concern as they
could be an indication that local L. borealis populations have been impacted by wind energy
developments along their migratory path (Arnett and Baerwald 2013).
Eptesicus fuscus has consistently been found to have a limited presence in NYC,
which is unexpected as this species is considered a habitat generalist tolerant of urban
environments (Walters et al. 2007, Parkins and Clark 2015, Gallo et al. 2018, Partridge
et al. 2020). A dedicated acoustic survey of the Bronx, the borough of NYC immediately
bordering Westchester to the south, yielded higher numbers of recordings for E. fuscus than
other boroughs, however L. borealis remained the dominant species (Parkins et al. 2016).
In contrast, E. fuscus has a dominant presence in Westchester County, accounting for 69%
of all recordings identified to the species level. True to its generalist nature, the presence of
E. fuscus in Westchester did not appear to be influenced by any land cover variable as it was
the only species to be present in all five parks surveyed. This result, when combined with the
results of acoustic surveys of NYC, suggests there may be some feature of the city’s more
southern boroughs, specifically Manhattan and Queens, which makes them less desirable
for E. fuscus (Parkins and Clark, 2015, Partridge et al. 2020).
Lasionycteris noctivagans has been found to have a substantial presence in NYC
during the brief period it passes through on its migratory route (Parkins and Clark 2015).
This species’ presence in neighboring Westchester follows a similar trend. L. noctivagans
appeared almost exclusively during a May survey in Marshlands Conservancy, where it
Urban Naturalist
N. Comparato
2023 No. 67
10
accounted for 60% of recordings identified to the species level. Though SonoBat suggested
this species was present in a small number of recordings from other parks surveyed in
later months, these recordings proved to most likely be misidentified calls from E. fuscus
when manually vetted. Interestingly, Parkins et al. (2016) accumulated the majority of their
L. noctivagans detections in October during their acoustic surveys of the Bronx. Taken
together, my results and those of Parkins et al. suggest L. noctivagans travels northward
through Southern New York early in the season and does not pass back through until well
past the end of the standard monitoring season for bats. The 3 km spatial scale was included
in this study to account for the large scale movements of L. noctivagans, which has been
observed to travel up to 3 km in a single night during migration (McGuire et al. 2012).
Previous studies have indicated, however, that this species can be sensitive to urban land
cover at smaller spatial scales (Dixon 2011). The area surrounding MC at the 500 m and 100
m scales feature a substantial percentage of urban and low-development land cover. These
results, in conjunction with the results from NYC surveys, suggest that impervious surfaces
in the immediate environment may not be a major deterrent to L. noctivagans in choosing
temporary habitat during migration.
Lasiurus cinereus, another migratory species, has been detected in NYC in small
numbers, especially during the mid-to-late summer months (Parkins and Clark 2015,
Parkins et al. 2016). In Westchester County, this species was only detected at Muscoot
Farms during a June survey, where it accounted for 35% of all recordings identified to the
species level. Previous comparisons between urban and rural locations hosting L. cinereus
have found the species to be tolerant of urban environments, but preferential towards areas
with high percentages of water and tree cover (Dixon 2011, Gallo et al. 2018). The area
surrounding MF matches these preferences, featuring the highest percentages of open, fresh
water at the 1 km, 500 m, and 100 m scales (See Fig. 2). It is surprising that this species
was detected at no other park in Westchester during the two seasons of surveys. The area
surrounding Cranberry Lake, for example, features similar percentages of tree cover and
open, fresh water at the 500 m and 100 m scales. Seasonal timing for surveys at MF and CLP
were also close, suggesting it is unlikely that a seasonal window of activity for L. Cinereus
was missed in the survey of the latter park.
Perimyotis subflavus has been detected in very low numbers in NYC (Parkins and
Clark 2015, Parkins et al. 2016). The results in Westchester were similar, with this species
only being detected at Cranberry Lake Preserve, where it accounted for 4% of detections
identified to the species level, and Ward Pound Ridge, where it accounted for less than 1%.
Tree cover, water, and urban land cover have all been found to have a positive relationship
with the presence of P. subflavus in previous studies, with at least one survey finding them
to be more common in urban than rural areas (Dixon 2011, Gallo et al. 2018). These prior
results suggest this species is most attracted to park locations adjacent to or within urban
areas. As a relatively large park surrounding a lake near the city of White Plains, CLP fits
this profile. The low number of detections of P. subflavus in this survey may be attributable
to the impact of white-nose syndrome, which has caused an estimated population decline
for this species of 95% in the Northeastern U.S. (Hoyt et al. 2021).
Myotis lucifugus has not been detected in any recent published results from acoustic
bat surveys of NYC, though there is unpublished data suggesting it may be present there
(Partridge et al. 2020). This species was detected in 4 of the 5 parks surveyed in Westchester,
with the exception being MC. M. lucifugus has been established as an urban-tolerant species
with a preference for transitional environments in the middle of the urban-to-rural spectrum
(Coleman and Barclay 2011). Its willingness to inhabit a variety of habitats is evident from
Urban Naturalist
N. Comparato
2023 No. 67
11
its wide distribution among areas of varying land cover compositions in Westchester. M.
lucifugus is another species that has been heavily impacted by white-nose syndrome, with
populations in the Northeastern U.S. experiencing an estimated decline of 96% (Hoyt et
al. 2021). This substantial decline may be the reason the species has not recently been
detected in NYC and is likely why M. lucifugus only accounted for a small percentage of
the detections in Westchester.
Myotis leibii is a less common species that was detected exclusively at WPRR, where
it accounted for 29% of detections identified to the species level. This species is known
to use anthropogenic structures for roosting habitat, however its tolerance level for urban
land cover is currently unknown (Harvey et al. 2011). M. leibii has never been detected by
acoustic surveys in NYC. The area surrounding WPRR is the most rural location surveyed
and the most distant from Westchester’s urban border with NYC. M. leibii’s exclusive
presence here may indicate a habitat preference for areas with low percentages of urban
land cover. That this species was present at WPRR and not MF, which features a similar
surrounding land cover composition, may be attributable to the presence of large “riprap”
embankments near WPRR. Riprap is a type of anthropogenic rock piling used for erosion
control that can be attractive roosting habitat for M. leibii (P. Moosman, VA Military
Institute, Lexington, VA, pers. comm.).
The relationship between land cover variables and species richness
The results of the LMMs estimating the relationship between the percentage of
surrounding urban land cover and the diversity of acoustic detections confirm H1. Urban
land cover negatively impacted the species richness of acoustic bat detections at all spatial
scales considered (3 km, 1 km, and 500 m). In Westchester, where species richness of
acoustic detections ranged between 1 and 3 species per night, the estimated marginal
means of the LMMs indicated acoustic surveys in areas with less surrounding urban land
cover are more likely to detect 1 additional species than areas with a high percentage.
This result was consistent across the 3 spatial scales (See Table 2). The percentage of
surrounding tree cover was found to have a positive relationship with the diversity of
acoustic detections at the 2 spatial scales considered (3 km and 1 km), confirming H2.
Estimated marginal means suggest that acoustic surveys in areas with a high percentage of
surrounding tree cover in Westchester are more likely to detect 1 additional species than
areas with a lower percentage.
Low-development land cover proved to have a negative relationship with the species
richness of acoustic detections at the one spatial scale it could be modeled, 1 km. This
refutes H3, which had proposed low-development land cover would positively impact
species richness. Low development was included as a land cover category in this study
to account for areas that fell into the NLCD categories of “developed open space” and
“pasture/hay”. The hypothesis that these areas would encourage greater species richness in
acoustic detections was based on the presumption that they would feature ample amounts
of forest edge, ideal foraging habitat for some bat species (Harvey et al. 2011). This land
cover category was also the closest match to what Coleman and Barclay (2011) describe as
“transitional” areas between urban and rural, which was found to be the preferred habitat of
Myotis lucifugus in the Canadian prairie. Estimated marginal means of the statistical models
indicated that surveys surrounded by higher percentages of low-development land cover in
Westchester were likely to detect 1 less species than areas surrounded by a low percentage.
It may be the case that roads and similar developments have made these areas less desirable
for foraging and roosting (Moretto and Francis 2017).
Urban Naturalist
N. Comparato
2023 No. 67
12
Though water was found to have a significant negative relationship with species
diversity at 3k m and 500 m, this result is compromised by the fact that the NLCD map
does not differentiate between bodies of salt, brackish and fresh water. That fresh water
is a crucial resource for bats has been shown in previous studies to encourage higher rates
of bat activity for many species (Dixon 2011). The two areas with the highest percentages
of water land cover in this study, MC and LP, are adjacent to large bodies of brackish
water, which are likely of little value to bats. This study also appears to suggest that no
land cover variable has a significant impact at the 100 m scale, contradicting previous
studies that suggest many bat species react significantly to land cover composition at
this scale (Dixon 2011, Gallo et al. 2018). In this case, the lack of significant impact
from land cover variables at 100 m most likely reflects the increasing homogeneity of
survey locations at smaller spatial scales. Though LP and MC both feature substantial
percentages of developed land cover at 100 m, the land cover composition of all 5 survey
locations is dominated by tree cover at this scale (See Fig. 2).
Acoustic surveys of Westchester detected a total of 7 species and the highest number
detected at a single park was 4 (See Table 1). The average number of species detected
among the parks was 2 with a standard deviation of 1. This relatively limited diversity
likely restricted the degree to which any land cover variable could impact the species
richness of acoustic detections. Note that the estimated marginal means for each LMM
correspond to the standard deviation for species richness. In areas with greater diversity,
the impact of land cover variables may be greater and show greater variation between
spatial scales. In the context of bat conservation and monitoring, however, the presence
of 1 additional species in acoustic detection results can be both valuable and informative.
The presence of L. cinereus and P. subflavus in Westchester, for example, was limited to
locations that featured very specific ratios of land cover variables that met these species’
preferences. While these species are both present in the highly urban environment of
NYC, this result suggests that they may avoid urban areas and form concentrations
in more preferential habitat when it is available. The exclusive presence of M. leibii
in Westchester’s most rural areas, in conjunction with its complete absence in NYC,
indicate that this species may have a strong preference for areas with a low percentage of
developed land cover.
The ability of this study to draw greater conclusions regarding the impact of urban
land cover and the preferences of individual species was limited by a small data set. This
data set was excerpted from a small survey effort constrained by lack of equipment and
technicians. Only one acoustic detector was available for surveys, of which I was the
sole operator. Multiple detectors and technicians would have made it possible to survey
urban and rural locations simultaneously, thus mitigating the influence of seasonality in
comparisons between the two. Such a research design is exemplified in Parkins and Clark
(2015). A greater variety of locations may have also brought greater heterogeneity to land
cover composition at the 100 m scale, providing a greater sample to model the potential
impact of land cover variables at this scale. Unfortunately, digital storage failure also
limited the present study, as 2020 survey data was corrupted before site-by-night reports
could be generated. Site-by-night data has previously been used to calculate speciesspecific
activity levels from acoustic surveys. This data could have been used to model the
impact of land cover variables on the activity levels of different species and the likelihood
of detecting them during acoustic surveys, as exemplified by Dixon (2011).
Urban Naturalist
N. Comparato
2023 No. 67
13
Conclusion
Urbanization is a growing issue for native biodiversity around the world and one that
is exacerbating a host of conservation pressures currently facing bats in North America.
Acoustic detection is the most accessible means of monitoring how bats are responding
to these pressures. With this study, I sought to examine the interaction between land cover
variables and the species richness of acoustic bat detections. My hypotheses were that
species richness would have a negative relationship with urban land cover and a positive
relationship with tree cover and low-development land cover. The hypotheses regarding
urban land cover and tree cover were both confirmed by statistical analyses. Estimated
marginal means from statistical models indicated that surveys in areas surrounded by a
high percentage of urban land cover were likely to detect 1 less species than less urban
areas. Similarly, surveys in areas with a high percentage of surrounding tree cover were
likely to detect 1 additional species than areas surrounded by lower percentages of tree
cover. These results were consistent across spatial scales. The hypothesis regarding lowdevelopment
land cover was refuted by statistical analysis, which revealed a negative
relationship between this variable and species richness at the spatial scale of 1 km. These
conclusions agree with several previous studies suggesting urban land development has
a negative impact on species diversity. This study also provides limited evidence that
certain bat species can have very specific preferences regarding the land cover composition
surrounding their habitat.
Acknowledgements
I would like to thank Donal Solick, Kaitlyn Parkins and Carl Herzog, who provided
valuable feedback on an earlier version of this manuscript. Further thanks go out to Sondra
Comparato and Haley Cotton for their support.
Literature Cited
Arnett, E.B., and E.F. Baerwald. 2013. Impacts of wind energy development on bats: implications for
conservation. Pp. 435–456, In R.A. Adams and S.C. Pederson (Eds.). Bat Evolution, Ecology, and
Conservation. Springer Science+Business Media, New York, NY, USA. 549 pp.
Coleman, J.L., and R.M.R. Barclay. 2011. Influence of urbanization on demography of little brown
bats (Myotis lucifugus) in the prairies of North America. PLoS ONE 6(5):e20483. Available online
at https://doi.org/10.1371/journal.pone.0020483. Accessed on 22 August 2023
Dixon, M.D. 2011. Relationship between land cover and insectivorous bat activity in an urban
landscape. Urban Ecosystems 15:683–695.
Gallo, T., E.W. Lehrer, M. Fidino, R.J. Kilgour, P.J. Wolff, and S.B. Magle. 2018. Need for
multiscale planning for conservation of urban bats. Conservation Biology 32(3):1–10.
Gomes, D.G.E. 2022. Should I use fixed effects or random effects when I have fewer than five levels
of a grouping factor in a mixed-effects model? PeerJ 10:e12794. Available online at https://doi.
org/10.7717/peerj.12794. Accessed on 22 August 2023
Harvey, M.J., J.S. Altenbach, and T.L. Best. 2011. Bats of the United States and Canada. The Johns
Hopkins University Press, Baltimore, MD, USA. 202 pp.
Hill, A.P., P. Prince, J.L. Snaddon, C.P. Doncaster, and A. Rogers. 2019. AudioMoth: A low-cost
acoustic device for monitoring biodiversity and the environment. HardwareX 6:e00073. Available
online at https://doi.org/10.1016/j.ohx.2019.e00073. Accessed on 22 August 2023
Hoyt, J.R., A.M. Kilpatrick, and K.E. Langwig. 2021. Ecology and impacts of white-nose syndrome
on bats. Nature Reviews Microbiology 19(3):196–210.
Urban Naturalist
N. Comparato
2023 No. 67
14
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. Reichard, 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.H.
Johnson. 2015. A Plan for the North American Bat Monitoring Program (NABat). United States
Department of Agriculture, Asheville, NC. 112 pp.
McGuire, L. P., C.G. Guglielmo, S.A. Mackenzie, and P. D. Taylor. 2012. Migratory stopover in the
long‐distance migrant silver‐haired bat, Lasionycteris noctivagans. Journal of Animal Ecology
81(2):377–385.
McKinney, M.L. 2002. Urbanization, biodiversity, and conservation: The impacts of urbanization on
native species are poorly studied, but educating a highly urbanized human population about these
impacts can greatly improve species conservation in all ecosystems. BioScience 52(10):883–890.
Moretto, L., and C.M. Francis. 2017. What factors limit bat abundance and diversity in temperate,
North American urban environments? Journal of Urban Ecology 3(1):1–9.
Parkins, K.L., and J.A. Clark. 2015. Green roofs provide habitat for urban bats. Global Ecology and
Conservation 4:349–357.
Parkins, K.L., M. Mathios, C. McCann, and J.A. Clark. 2016. Bats in the Bronx: Acoustic Monitoring
of Bats in New York City. Urban Naturalist 10:1–16.
Partridge, D., K.L. Parkins, S.B. Elbin, and J.A. Clark. 2020. Bat Activity Correlates with Moth
Abundance on an Urban Green Roof. Northeastern Naturalist 27(1) :77–89
Reichert, B., C. Lausen, S. Loeb, T. Weller, R. Allen, E. Britzke, T. Hohoff, J. Siemers, B. Burkholder,
C. Herzog, and M. Verant. 2018. A guide to processing bat acoustic data for the North American
bat monitoring program (NABat). United States Geological Survey, Reston, VA. 33 pp.
Szewczak, J., and M.L. Morrison. 2020. Use of bioacoustics monitoring systems in wildlife research.
Pp. 365–380. In N.J. Silvy (Ed.). The Wildlife Techniques Manual, Vol. 1: Research. Johns
Hopkins University Press, Baltimore, MD. 1400 pp.
Walters, B.L., C.M. Ritzi, D.W. Sparks, and J.O. Whitaker Jr. 2007. Foraging behavior of Eastern Red
Bats at an urban-rural interface. The American Midland Naturalist 157(2):365–373.
Westchester County Department of Planning. 2011. Population Density by 2010 Census Block.
Available online at https://planning.westchestergov.com/images/stories/MapPDFS/
popdensityblock2010.pdf. Accessed on 7 April 2023.
Westchester County Government. 2022. General Information: About Westchester County. Available
online at https://www.westchestergov.com/about-westchester. Accessed on 29 August 2023
US Census Bureau. 2023. City and town population totals: 2020–2021. Available online at https://
www.census.gov/data/tables/time-series/demo/popest/2020s-total-cities-and-towns.html.
Accessed on 7 April, 2023.