Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Urban areas and avian diversity: using citizen collected data to explore green spaces
(USC Thesis Other)
Urban areas and avian diversity: using citizen collected data to explore green spaces
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
Urban Areas and Avian Diversity:
Using Citizen Collected Data to Explore Green Spaces
by
Ephriam Joseph Daniels
A Thesis Presented to the
Faculty of the USC Graduate School
University of Southern California
In Partial Fulfillment of the
Requirements for the Degree
Master of Science
(Geographic Information Science and Technology)
August 2019
Copyright © 2019 by Ephriam Joseph Daniels
This thesis is dedicated to Yi-Hsuan Chu who encouraged me to pursue my dreams and to my
family for their loving support
iv
Table of Contents
List of Figures.................................................................................................................................vi
List of Tables.................................................................................................................................vii
List of Equations ..........................................................................................................................viii
Acknowledgements ........................................................................................................................ix
List of Abbreviations .......................................................................................................................x
Abstract...........................................................................................................................................xi
Chapter 1 Introduction.....................................................................................................................1
1.1. Urbanization and Green Spaces..........................................................................................1
1.2. Avian Species in the Green Spaces.....................................................................................2
1.3. GIS and VGI Data within Taipei ........................................................................................3
1.4. Objectives ...........................................................................................................................5
Chapter 2 Background .....................................................................................................................8
2.1. Urban Shift and the UHI Phenomenon ...............................................................................8
2.2. Green Spaces and Urban Vegetation ..................................................................................9
2.3. Taipei Green Space and Urban Vegetation Endeavors.....................................................11
2.4. Citizen Collected Data and GIS Monitoring.....................................................................12
Chapter 3 Methods.........................................................................................................................15
3.1. Data Exploration and Management ..................................................................................16
3.2. Data Preparation................................................................................................................22
3.2.1. Determining All Green Spaces ................................................................................23
3.2.2. EBird Modifications.................................................................................................28
3.3. Data Analysis....................................................................................................................28
Chapter 4 Results...........................................................................................................................34
4.1. Rarefaction and Richness..................................................................................................34
v
4.2. Dates Comparison.............................................................................................................35
4.3. Biodiversity and Residency ..............................................................................................38
4.4. Jaccard Clustering.............................................................................................................39
Chapter 5 Discussion .....................................................................................................................44
5.1. Citizen-collected Datasets.................................................................................................44
5.2. Different Variables............................................................................................................45
5.3. Species Composition.........................................................................................................46
5.4. Limitations ........................................................................................................................47
Chapter 6 Conclusion ....................................................................................................................48
6.1. Major Findings..................................................................................................................48
6.2. Future Work......................................................................................................................49
6.3. Overall Conclusions..........................................................................................................50
Reference .......................................................................................................................................52
Appendix .......................................................................................................................................56
vi
List of Figures
Figure 1: Administrative boundary of the municipality of Taipei...................................................4
Figure 2: The workflow diagram for the methods used in this study............................................15
Figure 3: The explorative analysis of the eBird dataset ................................................................18
Figure 4: NLSC datasets representing Landcover of Taipei of 2018 ............................................20
Figure 5: Landsat 8 NDVI of Taipei Area.....................................................................................21
Figure 6: The Avian Database .......................................................................................................22
Figure 7: Urban areas within Taipei with 50m buffer ...................................................................24
Figure 8: All possible green space found within Taipei from landcover dataset ..........................25
Figure 9: All 25 green spaces selected in Taipei ...........................................................................27
Figure 10: Refraction of eBird dataset from 2016 to 2018 using R ..............................................35
Figure 11: Averages of biodiversity of all sites between 2016 and 2018......................................36
Figure 12: Biodiversity of each site over the years of 2016 to 2018.............................................37
Figure 13: Collective Richness and Biodiversity ..........................................................................38
Figure 14: Migratory and resident species in all selected sites .....................................................39
Figure 15: All species composition of each site. ...........................................................................40
Figure 16: Resident species composition of each site. ..................................................................41
Figure 17: Migratory species composition of each site ................................................................42
Figure 18: Observations recorded of eBird dataset between the dates of 2016 and 2018.............45
vii
List of Tables
Table 1: Datasets used in study .....................................................................................................16
viii
List of Equations
Equation 1 Rarefraction.................................................................................................................30
Equation 2 Shannon-Weiner Index................................................................................................31
Equation 2 Jaccard Similarity Index..............................................................................................31
ix
Acknowledgements
I am grateful to USC faculty for the direction I needed and my other faculty who assisted me
when I needed it. This is especially true with my thesis committee members; Dr. Bernstein, for
insight in developing my thesis, Dr. Wu in organizing and manipulation my database and Dr.
Marx for biological insight when working with GIS. I am also grateful for Dr. Quang’s lab at
Academia Sinica for teaching me R-studio and statistical applications.
x
List of Abbreviations
BLI Birdlife International
GIS Geographic Information System
JSI Jaccard Similarity Index
MIT Ministry of Interior of Taiwan
NSLC National Survey and Land Center
NDVI Normalized Difference Vegetation Index
ODBI Open Database Connectivity
SSI Spatial Sciences Institute
USC University of Southern California
UHI Urban Heat Island
VGI Volunteer Generated Information
xi
Abstract
Urban development is expanding in today's world, and the impacts on humans and the
environment are strained within these modern cityscapes. The Urban Heat Island phenomena and
habitat loss have raised concerns about the future of many the ecological health of many
metropolitan areas. Due to these concerns, cities have taken steps to reduce the negative impacts
on the urban environment with the use of green spaces. On the Island of Taiwan, the
municipality of Taipei is one metropolitan area that has experienced dramatic urban growth.
While multiple studies have investigated the avian diversity of Taipei's green spaces, most
studies have used traditional data collecting methods. These surveys are financially taxing and
time-consuming, which can limit the volume of recorded events. In this study, Volunteer
Generated Information (VGI) and Geospatial Information Systems (GIS) was used to determine
the biodiversity, richness, and species composition of 25 green spaces selected by data-driven
selection process within Taipei from 2016 to 2018. The eBird dataset and multiple indexes
served as indicators of ecological health allowed for monitoring of green spaces. This study
determined that there are relationships between biodiversity, richness and species composition of
green spaces within Taipei. However, specific site’s species compositions in VGI showed weak
links between the richness of the green space. VGI datasets and GIS could enable a cost-effective
way to monitor a city's green spaces effectively in the future.
1
Chapter 1 Introduction
Human migration from rural regions into urban areas has produced multiple negative
implications to both humans and the animals that inhabit these urban areas. Adverse outcomes
are caused by two large factors; urban shift and the Urban Heat Island (UHI) phenomena. To
combat these factors, many cities have created or improved designated green spaces. Studies
have investigated specific green spaces in the city of Taipei Taiwan. However, there were large
constraints on the data that was collected (e.g., time, locations, etc.). Citizen-collected data can
be used to explore multiple areas that are less limited by these constraints while still having
higher avian counts than previous studies. Therefore, using these types of datasets can increase
the number of selected sites based on the dataset and not the pre-selected sites.
1.1. Urbanization and Green Spaces
The pressures of urbanization will increase with time and will have significant impacts on
urban environments. Currently, 54% of the human population lives in urban areas; this number is
predicted to grow to 60% in the next 30 years (The World Bank, 2018; United Nations, 2011).
This urban migration will need to be managed by city planners to promote a healthy city. These
city planners must focus not only on social aspects when designing and improving urban areas,
but also environmental ones. Urban migrations impact local ecosystems and displace several
native species (Barbosa et al. 2007). That said, the urban shift is not the only factor that causes
environmental displacement.
As urbanization increases globally, concerns arise due to the shift from rural regions to
more densely populated human-made environments. One concern with this urban shift is the UHI
phenomenon. One of the phenomenon’s traits is an increase in temperature within urban areas,
which can have adverse effects on both humans and the environment. The UHI phenomenon
2
increases health risks and wildlife displacement due to the increased temperature (Akbari &
Kolokotsa, 2016). Awareness and concerns about this phenomenon have led to the improved
urban design and designation of green spaces.
Green spaces and designated vegetated areas have been used by city planners to address
the effects of urban shift and the UHI phenomenon. Green spaces and the diverse vegetation
provide benefits both to humans and the environment (Wolch, Byrne, & Newell, 2014). But not
all green spaces are created equal. Many studies have found that the spatial distribution of green
spaces (islands vs. networks) and the overall site quality, can impact positively the biodiversity
of the green space (Aida et al., 2016; Chamberlain et al., 2007; Imai & Nakashizuka, 2010).
Because of this effect of green spaces, many cities have started implantation more green spaces
or improving existing ones within their administrative boundaries.
1.2. Avian Species in the Green Spaces
Many studies have used avian species as indicators of an ecosystem’s health (Melles,
Glenn, & Martin, 2003). One of the advantages of using avian species versus other organisms is
the ability to differentiate species by phenotypical traits and calls easily. Understanding species
richness and diversity within a specific region allows researchers and city planners to interpret
the level biological of stability and success of the green space post-implementation. Multiple
studies have linked the ecological health of green spaces to the richness and biodiversity of avian
species (Shih, 2010; Tajima, 2003). They have also noted that the area of green spaces may play
a smaller role compared to the biodiversity that is within its boundaries.
In recent years, research has been conducted on avian species and their distribution in
urban areas. Findings concluded that diversity in green spaces within urban areas dramatically
increases the biodiversity of the municipality and its surroundings (Chamberlain et al., 2007;
3
Shih, 2018; Strohbach, Lerman, & Warren, 2013). This increase of avian biodiversity in green
spaces has been studied in the city of Taipei by Dr. Shih (Shih, 2018). However, the Shih study
had low recorded sightings and a limitation of the study period. Because of these limitations, a
misrepresentation of the green spaces could be observed. By using geospatial information
systems (GIS) selection tools and citizen collection datasets, this study explored the relationship
between biodiversity and green spaces of Taipei. Because of the dataset used and the method for
site selection allowed for a more robust observational dataset for a more extended period.
1.3. GIS and VGI Data within Taipei
With the increased adoption of geospatial technologies, researchers can more easily
identify and explore green spaces. The ability allowed by GIS applications allows for more
reliable and quicker collected datasets than traditional monitoring methods (Silvertown 2009).
This improvement in methods is partially due to the increase in handheld devices and intuitive
collection applications. When compared with previous studies, the use of volunteer-generated
information (VGI) data has proved useful when working with conservation and/or monitoring
efforts (Chandler et al., 2017). These tools can be used in cities around the world.
The municipality of Taipei is located within the Taipei Basin which is surrounded by
mountains in the Island of Taiwan in East Asia (Figure 1). Most populated parts of the
municipality are alongside the Tamsui River. While the center of the city is at low elevation,
much of the municipality is within a steep mountain range. The municipality of Taipei is in the
subtropical climate. The temperature in Taipei is warm, with precipitation moderate to high all
year round (Chang, Li, and Chang 2007). From June to October Taiwan experience typhoons,
tropical cyclones that occur in the Pacific Ocean. Because of the high levels of precipitation and
4
being in a subtropical region, Taipei is conducive to having a high amount of vegetation. Even
with high amounts of vegetation, the municipality has changed over the years.
Figure 1: Administrative boundary of the municipality of Taipei.
Dating back to the 18th century, Taipei has been the capital of Taiwan and is currently
home to over 2.7 million people with over 5 million people that commute between its
5
administrative borders (Shih 2010). The dense population has resulted in multiple governmental
concerns. One of these concerns is the effects population density on the ecological health. The
loss of native ecological habitats has led to limited resources for animals within cities (Goddard,
Dougill, and Benton 2010). This limitation is compounded with the increase of multi-story
buildings which have been linked to increasing the average city temperature which leads to the
displacement of multiple native species (Tratalos et al. 2007). Because of these concerns, the
municipality of Taipei has taken steps to improve the health and wellbeing of its ecosystems with
the use of green spaces. Green spaces have been used to promote biodiversity in Taipei, yet few
studies focus on showing their effectiveness.
There have been significant changes in improved urban planning policies over the years
in the municipality of Taipei that have been beneficial for native species. These policies have
been used to increase the construction of green spaces as well as increase the diversification of
vegetation of existing green space (Weng, Lu, and Schubring 2004). This increase in green
spaces can improve multiple native faunae. Taiwan is home to a diverse number of avian species.
According to Birdlife International (BLI), Taipei is listed as having 342 species within its
municipality (BirdLife 2018). While there are a large number of species that can be found within
Taipei, the biodiversity of any single location can be high depending on the quality of the site
(Goddard, Dougill, and Benton 2010). This high species count in part is due to Taipei location
which is found at a low altitude with mountainous and aquatic features (Huang and Chan 2014).
1.4. Objectives
The objective of this study was to use geographic information science and volunteer-
generated information datasets to monitor biodiversity and ecological richness in green spaces in
the city of Taipei, Taiwan. While past studies about the relationships between green spaces and
6
avian distribution have been conducted in Taipei, none have used VGI data. The VGI dataset
consists of the higher volume of avian sightings and allows for a more supported conclusion of
the ecological health of Taipei’s green spaces. This high volume of records would support the
study to understand the ecological health of the data selected green spaces within the
municipality of Taipei through multiple years.
There are two specific research objectives in this study. The first objective is to examine
how biodiversity and richness varies across green spaces in Taipei. Biodiversity is defined here
as the abundance of species in selected areas and not by the dominant species. For this both the
Shannon-Weiner Index and rarefication analysis on each of the selected sites. The study
compared the levels of biodiversity and richness between different sites. The study also explored
the relationship the green spaces have over multiple years. From January 2016 to December
2018 each green space’s average biodiversity was determined. The rarefication analysis will
further determine the most optimal sample size needed to be collected for green space. These
measurements will allow an understanding of the ecological health of each location and optimal
sample size for VGI datasets.
The second objective of this study is to determine if there are links between selected
green space’s biodiversity, richness, and species sites composition. This was further
differentiated between resident vs. migratory species, as many sites may have a stronger
abundancy of species depending on the species residency status. The understanding of the
species residency statuses and how they are distributed between the green spaces better defines a
link between specific green spaces to species composition. To further support this link, species
site composition of all selected green spaces was analyzed using the Jaccard Similarity Index
7
(JSI). By exploring these links, this study will provide an understanding and further support the
final objective.
The results of the study can be used to not only identify the quality of differing green
spaces based on their avian biodiversity but will also allow a deeper understanding of which
types of avian species compositions are affecting specific green spaces locations inside Taipei.
Both sub-questions will support the finding in the primary objective of the study.
8
Chapter 2 Background
This chapter describes the background information related to this study and focuses on
four subjects: Urban shift and the UHI phenomenon, green space, and urban vegetation, Taipei
green space and their vegetation endeavors and finally citizen-collected data and GIS
monitoring.
2.1. Urban Shift and the UHI Phenomenon
Much of the global population is found in urban centers. The U.N. predicts that over 60%
of the earth’s population will live in cities by 2050 (United Nations 2011). This endeavors to
create urban center needs land and large amounts of resources in concentrated areas. Urban
development influences human and non-human life within the population center.
The human population is migrating from rural areas to highly populated urban areas. This
change in population placement has had a pervasive ecological cost. While most rural housing is
spread out over a broader region, cities condense their housing area. This urban shift effects
various species as it narrows the ecosystems they can have and limits the available natural
resources while displaces multiple species that were originally native to a formerly rural area.
This phenomenon will degrade the quality and function of an ecosystem. This degradation is
only compounded by other effects caused by the UHI phenomenon (Shiflett et al. 2017).
The construction in urban development due to population migration significantly
increases the amount of human-made material in a concentrated area, and many of these
materials (asphalt, cement, metals) are quick to collect solar radiation (Chen et al. 2006).
Because of the low specific heat capacity that is found in human-made materials, many urban
constructions can collect so much solar radiation they increase the surrounding temperature of
the area. The UHI has led to multiple issues including a decrease in biodiversity and multiple
9
health concerns within larger cities (Guo et al. 2015; Tan et al. 2010; Akbari and Kolokotsa
2016). Because urban areas are prone to multiple factors related to UHI and urban development,
city planners have taken steps to reduce the UHI effects.
2.2. Green Spaces and Urban Vegetation
The United States Environmental Protection Agency (US EPA) (2018) defines open
spaces as a piece of land that is undeveloped and accessible to the public. Of all the categories
open spaces, green spaces have been shown to have a cooling effect on both local and
surrounding regions which has been documented to reducing the effect of UHI (Estoque,
Murayama, and Myint 2017). Greens spaces defined in this study are open spaces designed to be
vegetated for human, and non-humans use. Much research has been done on the health of
humans and the environment with the creations of these spaces (Barbosa et al. 2007; Fuller et al.
2007; Wolch, Byrne, and Newell 2014). One of the findings is that green spaces because of their
vegetation produce shade, which reduces stored thermal energy in the surrounding areas (Li et al.
2011). Because of this finding, remote sensing has been used to locate areas of interest. Datasets
that use thermal sensors have been linked to areas of vegetation (Tan et al. 2010). The
Normalized Difference Vegetation Index (NDVI) has been used in multiple studies to locate
areas suspected of reducing temperatures because of thermal storage (Li et al. 2011). This
cooling effect has been seen to reduce the power needed to cool electronics and reduce the
amount to ozone, which impart reduces cardiovascular disease (Akbari and Kolokotsa 2016).
While there have been direct links between human health and financial benefits of green spaces,
there are also links connecting green spaces to ecological health.
Richness and biodiversity are two defined terms that have been used to determine a
location's ecological health. Richness is defined as the number of individual species in a specific
10
ecological location (James and Rathbun 1981). This term is calculated by the sum of all species
in a specific location. While there is a single method for determining the richness, there are
multiple indexes used for biodiversity. Biodiversity is defined as the number of individuals and a
variety of species in an ecological community (Chao et al. 2006). While there is a single
definition, the term is often broken into two different methods depending on a study’s focus:
Simpson Index and Shannon-Wiener Index. The Simpson Index is weighted by dominant
species, which is used to measure the degree of concentration when individuals are selected for
specific groups (Oksanen 2016). The Shannon-Wiener Index focuses on the abundance which
quantifies the uncertainty of the predicted species in an area (Tramer 1969). This index was used
in this study as it works on the abundance of a species in a specific site. Understanding both the
richness and biodiversity of the term allows for insight into green spaces.
Green space has been shown to improve their effectiveness to an environment’s
biodiversity, as they enable multiple biological niches that were not previously present (Goddard,
Dougill, and Benton 2010). Greens paces have been linked to improved health in both people
and the biophysical environment (Tan et al. 2010; Wolch, Byrne, and Newell 2014). Green
spaces with high levels of biodiversity have affected multiple species as these organisms fill new
ecological niches (Matsuba, Nishijima, and Katoh 2016). While green spaces to increase the
amount of biodiversity in an area, there is a limit to that diversity. Researchers determined that
with the rarefaction test, different sites have a maximum richness threshold in avian diversity
depending on the variables of the site (James and Rathbun 1981). This finding gives support to
the idea that each site’s features determine different environmental health as levels of both
richness and biodiversity are not shown to be equal.
11
One standard indicator used for environmental health studies is the diversity and
abundance of avian species. Studies have used avian species to monitor the health of specific
environments (Blair 1999). Avian species are apex organisms that are relatively easily to identify
compared to other organisms and have been linked to the biodiversity of the surrounding areas
(Kong et al. 2010). This link was further studied when examining green space relations to the
residency of differing avian species (Matsuba, Nishijima, Katoh 2016). As different species need
different requirements in their habitat, trends start to emerge depending on the species residency.
While there are overlapping areas, migratory and resident avian species have been more
conducive to specific areas (Fontana et al. 2011). This spatial pattern is further limited in a
cityscape’s green spaces where natural vegetation confined and vary significantly (Shih 2010).
Because of a green space’s limitation, multiple species compositions have been shown to
highlight areas that are conducive for higher levels of biodiversity. The JSI has been used to not
only to determine these sites but show sites relation to varying levels of richness (James and
Rathbun 1981). Because of avian and their biological status are classified as higher-level
organism, it is easy to discern and link them to green spaces.
2.3. Taipei Green Space and Urban Vegetation Endeavors
Taipei is in a sub-tropical climate that is conducive to yearlong vegetation growth. With
most of the city inside the Taipei basin, the location has allowed the city to have a relatively
large amount of vegetation and a diversity of animal species. When Taiwan started to
industrialize in the mid-1970s, the number of green spaces was reduced (Taipei DUD, 2009). In
recent years, Taipei has seen a rise in its urban population and has become one of the most
densely populated cities in the world (Shih 2018). Started in 1992 with the Taipei
12
Comprehensive Urban Plan, Taipei city planners first focused on urban development with little
emphasis on the biophysical environment (Shih 2010).
With concerns about UHI phenomena and intent on improving the city’s image, Taipei
began using planning infrastructure changes to improve urban ecosystems in the early 1990s
(Shih 2010). It was in 1999 that development policies included the increase in green spaces. The
emphasis in urban vegetation was further expanded with the creation of the Green Master Plan
for Taipei (Huang and Chan 2014). After the establishment of the Green Master Plan, Taipei
implemented the Beautiful Taipei Policies to set specifics policies for land improvement that
included increasing the number of green spaces and urban vegetation locations in 2010 (EPA,
ROC 2018). Because of these policies, the city increased and improved vegetative diversity
many green spaces within its administrative borders. Currently, the city of Taipei’s Department
of Urban Development contains 281 units of designated green spaces and over 80,000 trees
planted on roadsides (Taipei DUD 2009). This number does not account for how the city, as it
breaks larger units into multiple parcels of green spaces. While studies have been used to predict
the best placements for green spaces, there is little work on analyzing their efficacy with respect
to biodiversity after they have been created. This exploration can be done with the use of avian
species and their compositions as seen in earlier studies (Shih 2010). With this information, one
can investigate how effective green spaces are to the ecological health to where they have been
implemented in Taipei.
2.4. Citizen Collected Data and GIS Monitoring
Citizen-collected data has been gathered and used for hundreds of years. Charles Darwin
was an unpaid companion to Captain Robert Fitzroy when he traveled on the Beagle collecting
13
data on wildlife (Silvertown 2009). With the emergence of new technologies, data collection has
become much easier and more accessible. As a result, the volume of citizen-collected data has
grown immensely (Silvertown 2009). Many organizations have committed to storing and
maintaining large datasets including citizen-collected data and making them accessible to the
general public.
New technologies allow users to achieve a clearer understanding of ecological
information. GIS has been used in avian studies to find patterns of migration and habitat ranges
(Bouten et al. 2013; Fink et al. 2011). Researchers have also been able to use geospatial analysis
to understand urban green spaces and the correlations it has with species biodiversity (Matsuba,
Nishijima, and Katoh 2016; Wiens et al. 2009). In recent years, there has been an increasing
amount of research using GIS with open-source databases. The availability of datasets has only
increased with the inclusion of VGI.
There has been an increased use of VGI databases. The accessibility and increased usage
of VGI datasets have been shown improved their reliability (Silvertown 2009; Conrad and
Hilchey 2011; McKinley et al. 2017). This reliability is primarily supported by the number of
recorded events that VGI databases can produce. VGI database allows for a larger number of
moderators, which in turn gives more opportunities to verify large dataset quickly. These
datasets give public access to record and increase their ability to analyze geospatial data. They
also encourage users to add more data to datasets as many VGI databases are simplified for quick
data collection. The dataset grows and improves with every additional record. This increase in
shared data also improves the accuracy of a dataset. These datasets increase on both the temporal
and spatial scale of a study which could not be conducted by traditional means, which requires a
14
team of researchers that could only collect data for a limited amount of time (McKinley et al.
2017).
15
Chapter 3 Methods
The study was conducted within the urban areas of the city of Taipei and focused on the
distributions of avian diversity and richness within green spaces. The study was conducted in
three analytical phases; (1) primary data collection/exploration, (2) data preparation, and (3)
analysis (Figure 2). Each phase was vital to the study as each was dependent on the prior stage.
Figure 2: The workflow diagram for the methods used in this study.
16
3.1. Data Exploration and Management
In the first phase, datasets were collected from various sources (Table 1). Both
QuantumGIS (QGIS) and PostgreSQL applications were used to explore and store different
datasets. Because of availability and the user-friendly design, ArcGIS Pro was used for exporting
final map/figures. Five datasets were collected for this study; eBird, BirdLife International,
Taipei administrative parcel, Taipei landcover, and Landsat 8 imagery.
Dataset Type Description
EBird Dataset Vector – Point
Original Txt.
EBird dataset collected from
March 2010 to 2018
https://ebird.org/
Birdlife International Table Information on species names in
municipality of Taipei
Updated 2018
http://datazone.birdlife.org/home
Taipei Administrative
Parcel
Vector – Polygon
Original Shp.
Administrative borders of Taipei
region.
Updated 2018
https://data.gov.tw/
Taipei Landcover Vector - Polygon Landcover of Taipei City
Updated 2018
http://www.nlsc.gov.tw
Landsat 8 image GeoTIFF Landsat NDVI 30-meter cells of
2018 from Climate Engine
https://app.climateengine.org/
Table 1: Datasets used in study
This study utilized the eBird datasets, which is a collection of citizen-collected
information on avian species around the world. This original dataset utilized eight years data
which was recorded from January 2010 to December 2018. The dataset was selected for the area
within the municipality of Taipei using the eBird portal. The original dataset was a text file and
was first added into QGIS to make the dataset spatial by converting each record into points using
the XY coordinates. The data was then converted from a geographic coordinate system to a
17
projected coordinate system in Taiwan Datum 1997 (TWD97). The projection used throughout
the study was TWD97. The dataset was transferred into PostgreSQL for pre-analysis.
EBird datasets have one of the most extensive collections of VGI on avian species
worldwide. While the dataset is large, there are still variables that need to be understood to be
used accurately for this study. One of the concerns with the citizen-collected dataset was the
quality control that may vary between users. The dataset allows users to investigate the collector,
collection method, and if the specific recorded had been peer reviewed. This process to
investigate the eBird dataset was done through PostgreSQL, as the application was able to handle
the entire dataset efficiently. SQL script was made to find the sum of yearly observations, the
number of protocols used throughout the datasets (Appendix A, B). The reduction of anomalies
of the eBird dataset and yearly diversity were counted and refined using PostgreSQL (Table 2).
The data used in the study was from 2016 to 2018, as earlier years did not have sufficient
recorded events. The final data selection of the study was further supported, as the richness of
avian inside Taipei were starting to level from 2016. While the eBird dataset was suitable for this
study, there are limiting factors.
18
0
20000
40000
60000
80000
2010 2011 2012 2013 2014 2015 2016 2017 2018
Observations By Year
Year
Observations
eBird Dataset
Total Counts Richness Year
462 103 2010
1915 143 2011
2743 164 2012
4691 182 2013
5100 206 2014
13721 253 2015
29129 294 2016
43383 324 2017
72228 311 2018
0
100
200
300
400
2010 2011 2012 2013 2014 2015 2016 2017 2018
Overall Richness By Year
Year
Species
Figure 3: The explorative analysis of the eBird dataset. (Left) The recorded diversity of
avian species in Taipei from 2010-2018 (Top Right) Avian species recorded of Taipei from
eBird dataset (Bottom Right) Richness of avian species found in eBird dataset
While the eBird dataset has a high number of recorded events, it has some drawbacks.
There was potential bias in how humans recorded specific areas unrelated to avian distribution.
Multiple factors can cause these biases. The first factor is that the point of the collection is not
always the location of the bird, but rather the position of the collector. With this designation of
location, the distance between the collector and the bird were typically ranged between 0 to 30
meters. This protocol is marked in eBird data as P25. The second factor is with high counts but
with low time expanded for observations. Some of these cases were a result of different methods
of collection and not collector’s error. Because of this multimethod dataset, records that were
significantly different for the rest of the dataset were omitted unless the appropriate time for the
collection was given. The final factor that can affect distribution is errors in collectors recording
19
species names. EBird dataset allows users to add genius names without recording the species,
because of this richness and biodiversity may be artificially heightened because of the lack of
user precision. This correction of nomenclature was corrected using the Birdlife International
(BLI) dataset as it would have an up-to-date and standard species checklist.
The Birdlife International dataset verifies of scientific names, residency, species status.
BLI is an organization that categorized avian distribution throughout the world. The dataset also
contained threatened status which was used to explore if there are any anomalies with the eBird
dataset. The BLI dataset was selected to represented avian species presence in the municipality
of Taipei. Because of this BLI dataset, only 117 resident and 225 migratory species were
selected for this study.
To define a study area, an administrative parcel dataset was used. This dataset was
collected from the Ministry of Interior of Taiwan (MIT) online portal. This dataset gave the
administrative boundaries of all the counties and municipalities within the Island of Taiwan. This
dataset was suitable when defining the study area as it contains the municipality of Taipei’s
boundary. A text file contained parcel’s information and was obtained through the MIT web
portal. A join between the parcel shapefile and the text file was be done in QGIS. The dataset
was then be added to the geodatabase in PostgreSQL for future processing.
To define the study sites, an administrative landcover dataset was used. This dataset was
collected from the Taipei’s National Land and Survey Center (Figure 4). This landcover dataset
used was a collection of 38 individual tiles of Taipei and the surrounding areas taken from 2018
with the resolution of 5:000. Because of inconsistencies in dataset polygons, only 37 tiles were
selected for the study.
20
Figure 4: NLSC datasets representing Landcover of Taipei of 2018
21
Landsat 8 Level 1 data was used in the study to determine the extent of urban areas. The
dataset was be collected from Climate Engine. Climate Engine is a web-based application that
allows for the procurement of opensource high-resolution global remote sensing datasets. The
dataset has a 30-meter spatial resolution NDVI. As the parcel dataset was last updated in 2018,
the Landsat image was used a mosaic average from January 2018 to December 2018. The dates
of capture ensure quality control for defining the study area as it ensured the latest images of
urban expansion.
Figure 5: Landsat 8 NDVI of Taipei Area
22
3.2. Data Preparation
The second phase is the data preparation where datasets were created and manipulated. In
this phase, there was an emphasis on organizing the newly created datasets (e.g., sites table,
urban areas) in a geodatabase (Figure 6). A geodatabase was created, connected and updated
between QGIS, PostgreSQL, and R-Studio for the ease. EBird and site selected tables will be
created so there is no many to many relationships within the database. Landsat, Taiwan
administration, selected sites, and eBird datasets will be connected with their geometry under the
TWD97 projection. Much of the data processing was completed in both Terrset and PostgreSQL.
QGIS was used as an intermediate software. Final selected sites created in QGIS was compiled
into the geodatabase which was stored in PostgreSQL for ease of data manipulation and
organization.
Figure 6: The Avian Database
The process of defining a study area was conducted with the Taipei administrative parcel
dataset. Selecting only the Taipei municipality out of the entire country of Taiwan allowed for a
clear definition of the municipality’s borders. The urban area in which the selected sites will be
selected will be extracted using Terrset and QGIS.
23
3.2.1. Determining All Green Spaces
The selected sites were inclusive to only the urban areas of the city of Taipei to
represent the populated areas of Taipei. Urban areas were defined using Terrset and QGIS. This
process was conducted using a Landsat 8 Level I NDVI and the Taipei parcel dataset. The
Landsat 8 Level I’s NDVI was used to find the N-value that were between .082959 and .31047.
This value was determined with the use of the histogram tool within Terrset. The N-value peaked
at three separate occasions: vegetated areas, urban areas, and water (Appendix C). Using the
Reclass tool in Terrset a new image was created, where the values -1 to .082959 were classified
as water, value .082959 to .31047 were urban areas, and values greater the .31047 were
vegetative areas. Before exiting Terrset, the new image was converted into a vector file. This
new vector dataset was imported into QGIS.
In QGIS the newly created urban area dataset was used to determine all the green spaces
within the municipality. The urban dataset had a 50-meter buffer added to incorporating any
green spaces that could have been on a border area (Figure 7). This buffer was selected as some
larger parks were not originally selected even if they were found in the center of the city. This
new dataset was used to initially define the urban areas by creating a new border shapefile within
QGIS. The parcels that are within the new border shapefile will be qualified for the new study
area. The study area covers a total of 271 square kilometers and has 3,100 parcels of landcover
were designated green spaces that were applicable for the study area (Figure 7). The
undeveloped land was not designated as possible green space as their attributes varied
significantly.
24
Figure 7: Urban areas within Taipei with 50m buffer
25
Figure 8: All possible green space found within Taipei from landcover dataset
26
Within the municipality of Taipei, there was a total of 3,100 designated green spaces
identified from Section 3.2.2. Using PostgreSQL, 25 specific sites within the identified green
spaces were identified for further analysis. This site selection procedure was completed in
multiple steps using a QGIS’s database manger and QGIS editing tools. However, all SQL script
was tested in PostgreSQL before use in QGIS (Appendix B) .
The first step was to identify the identified green parcels that are also classified as green
spaces within the Taipei landcover dataset. The attributes of the landcover datasets, values with
an ID value between 700 series were classified as green spaces by the NLSC. The use of SQL
statements enabled the selection of the attribute series which corresponds with green areas in the
dataset. As avian species can have a broader range than the exact location in which the bird was
identified, a 300-meter buffer was created around selected sites for the avian count (Shih 2018).
This script was combined with an intersect SQL statement to determine the number of recorded
avian counts inside the newly acquired parcel zones. Only green spaces that had equal to or
greater than 100 avian observations inside their boundaries were selected. The was exported as a
vector dataset in QGIS through the database management tool in QGIS.
Initially, the top 300 were selected using the SQL script. From those 300 parcels, the
merge commands were conducted in QGIS to connect the parcels that belong to a single site. The
Open Street Map background was used to determine its borders. As the landcover dataset
separates green spaces by NLSC’s designated parcel, it can mispresent a larger site. This
misrepresentation is because some of the sites are a combination of multiple parcels. This
process changed 300 parcels into 257 potential sites. This new dataset was uploaded into
PostgreSQL under the table labeled sites. Using the SQL script to find the 300 parcels was
27
modified, selecting from the new sites table with a limitation of 50. This selection process
allowed the identification of green spaces that were suitable for analysis.
Only 25 sites that had the most significant number of records and were over 100 observed
difference from its adjacent sites were selected. The 25 sites were selected for two reasons. The
first was for computing processing limitations. The second was that lower-ranking sites had
larger gaps and temporal consistency in their datasets. Finally, previous studies in Taipei green
spaces only an equal number of sites. Because of these factors, the process used eliminates the
chance of sites with insufficient data for future analysis.
Figure 9: All 25 green spaces selected in Taipei
28
3.2.2. EBird Modifications
The avian dataset was modified using buffers for accounting for collection protocols. The
standard protocol used in eBird is P25, where an observer collects data 30 meters from the point
of collection. The new dataset was reviewed and edited before being used for further processes.
Historical records and large numbers over an extensive period of time spent was omitted. This
omission was completed using the SQL’s Group By command and selecting one of the following
columns found within eBird dataset: protocol, and time_expired. Protocols that were historical
records that had over 100 observations in a single sitting were omitted. For time_exipred, data
collection for a duration of +10 hours was omitted. This procedure will ensure quality in the
dataset for future analysis.
A new dataset was created inside of PostgreSQL for ease of comparative analysis of the
eBird dataset. This process was used as a junction table between eBird and BLI dataset. New
values were used to connect the eBird dataset, and newly created junction table key, where
values less than 700 were found only in BLI species checklist and values greater than 900
represent values only found in eBird’s that were redundant nomenclature or unclear species
names. The final database was connected with foreign keys to ensure the validating of values
before analyzing with R (Figure 11).
3.3. Data Analysis
The final phase focused on analyzing datasets the modified and refined datasets. Using R,
PostgreSQL and QGIS, richness and biodiversity were identified in the selected 25 sites. Three
methods were to be used to determine the links between green spaces and avian diversity inside
Taipei’s green areas: rarefaction analysis, Shannon-Weiner biodiversity index and Jaccard’s
similarity index. Two libraries were used when working with R: Open Database Connectivity
29
(ODBC) and vegan libraries. The ODBI library was used for data exchange between R and
PostgreSQL. All analyses used the vegan library imported into R. This library contained the R
script to run the analyses. The rarefaction analysis allowed for a greater understanding of the
optimal sample size. This is not needed with the Shannon’s Index as its large variations of
sample sizes to find comparative biodiversity of each site. The JSI allowed for a more in-depth
understanding of each of the site to one another. Two variables were used with three different
analysis: residency and temporal. Both these variables were explored were possible.
3.3.1 Rarefaction Analysis
The rarefaction analysis is a method to determine the richness of a specific area where
one controls the amount of sample being randomly counted. This process finds the most common
denominator of species richness through all the sites. The rarefaction also possesses a rarefaction
curve. This curve is the average of all options of species accumulation in of a single site. When
combined with multiple curves, it then can be used to find the most optimal sample size for
collecting a multiple site’s richness. When comparing multiple sites, the recorded counts may
vary. Because of this disparity, some selected sites have the possibility of showing a higher
richness because they have lower numbers to choose from, while sites with a larger recorded
account could show a lower amount of richness. The rarefaction equation solves this discrepancy
problem (Equation 1). The rarefaction takes a specified number of samples, the most common in
all datasets, and compares the species richness at all locations. This allows for the most optimal
pool size when comparing multiple sites that contain a larger difference in sample sizes.
30
Eqaution 1:E(S) is rarefaction, S is the number of species, N
i
the count of species i, and
(N/n) is the binomial coefficient and N
i
gives the probabilities that species i does not
occur in a sample of size. (Gotelli and Colwell 2009).
PostgreSQL and R-Studio were both used in order to find a standardized richness with all
sites. R-studio was connected to PostgreSQL using the ODBI library in R. This package allowed
both applications to communicate and save information in a single database. Under the vegan
library, the rarefy function allowed for the quick rarefaction analysis of eBird data of each of the
sites (Oksanen, 2016). The output dataset was added to a newly created table for sites analysis.
3.3.2 Shannon–Weiner Biodiversity Index
The Shannon-Weiner biodiversity index determined the diversity based on abundance to
support the notion concept of biodiversity in specific areas (Equation 2). This index, unlike the
rarefaction analysis, assumes that all recorded species are the total species in the location. The pi
in the equation can further be broken down to the amount of individuals of each species over the
total number of individuals for a site. Because of this variable, the analysis makes use of the full
dataset. This index also showed the evenness of the specific site, which was comprised of the
species number and the abundance of the species. Since a bigger dataset could be at a
disadvantage when using the rarefaction analysis as it will have sample size limitations, the
Shannon-Weiner Index can be used to support the rarefaction analysis of richness in a specific
area.
31
Equation 2: The H’ is the Shannon-Weiner Index, p
i
is the relative abundance of species i, S is
the total number of species present and ln is the natural log (Oksanen 2016).
R-Studio and PostgreSQL were used for completion of the Shannon-Wiener Diversity
analysis. For this analysis, the vegan package diversity function in R allowed the ability to
compare multiple green spaces to find each of their avian diversity. The analysis results were
stored in PostgreSQL producing a new dataset for the Shannon-Weiner biodiversity index.
3.3.3 Jaccard Similarity Index
The Jaccard similarity index was used to explore the similarity and diversity of a
community (Equation 3). This index allowed for a clear understanding of multiple sites and the
community composition of the avian species found within each site. The JSI equation focus on
the relationship between intersections and unions of different groups or datasets. Because of this
relationship, the X and Y repersents sets, where X∩Y is the intersect and XUY is the union of
the sets. Understanding the intersections of multiple sets allows the ability to find composition
similarities between different groups. This analysis was used with migratory avian species,
resident species, as well as the combination of both avian groups to determine species if there
were visible community guilds that have formed within the city of Taipei in specific green
spaces.
Equation 3: J (X, Y) is the Jaccard similarity matrix. (Chao et al. 2006).
32
Data separated from PostgreSQL and running analysis on R was used to find JSI. The use
of R’s vegan library allowed for the calculation of JSI. The JSI was plotted using a custom
command to make a dendrogram. This dendrogram determined the cluster of sites that have
similar species composition. This plot was used to create a visual representation on a map using
ArcGIS Pro.
3.3.4 Migration VS. Resident Avian Diversity
The ability to identify if these species are residents could give a greater understanding of
the biodiversity of the individual site. Differentiation between migratory and resident species
could affect the avian diversity in each of the green spaces as there could be dramatic population
fluctuations during different periods. These fluctuations affect could both analysis methods and
misrepresent the site. Because of the possibility of migratory avian species affecting the analysis,
both migratory and resident avian species biodiversity analysis will be conducted.
PostgreSQL and SQL script were used to separate resident avian species from migratory
and using the BLI created the dataset. This allowed for analysis in R-Studio between the two
different groups. These were then transferred into R using the OBDC library where they were
further analyzed (Figure 11).
3.3.5 Time Variable
Early analyses were used to determine the influence of diversity and richness on all the
selected sites. However, significant temporal variations in the diversity count could be an
indicator for inaccuracies in data. While biodiversity was determined in all the sites during all
three years together, the study of the site data annually was taken. This allowed the study to
explore temporal changes in selected areas. An overall analysis of recorded sightings completed
of all the sites that contained enough recorded events. This showed the general avian biodiversity
33
found through Taipei green spaces, yet this method does not give the full picture of these green
spaces. Further examination of the dataset will be processed in R-Studio then saved in
PostgreSQL to finally be explored in QGIS.
In the eBird dataset, only three years of data (2016, 2017 and 2018) were analyzed. Using
the Shannon-Weiner diversity index and JSI methods described in Section 3.3.1 and 3.3.2,
changes in biodiversity over time were analyzed. When working with the annual dataset, it must
be noted that there were changes in the amount of observational data collected for each year. All
results from the sites that were analyzed were saved in PostgreSQL and presented in QGIS.
34
Chapter 4 Results
A total of 112,871 events were recorded between 2016 and 2018. However, only 65,171
entries were selected due to multiple recordings having inaccurate species names. From the total
of the selected entries, 237 avian species were found within the sites. Only 152 of the avian
species were classified as migratory, and 95 were classified as resident species. Ten of the avian
species were listed as threatened or endangered.
4.1. Rarefaction and Richness
Using the rarefaction analysis, there are two significant findings of the overall citizen-
collected eBird dataset at the selected green space sites in Taipei. Rarefaction was able to
determine the optimal point collection of all 25 of the sites (Figure 12). The optimal point is
represented by the vertical line and was determined to be at 257 sample size for all sites. This
means that if the sample size of each site is collected randomly at a designated observation
amount, all sites have the most optimal levels for comparison when 257 observations were used.
The second significant finding is the rarefaction curve. The curve shows that even with larger
sample sizes there is a limitation to the richness. There is a clustering of richness that levels off
around the sites that had only 50 to 100 species present. This supports the proposal that selected
green spaces and species richness do not increase after this range.
35
Rarefaction eBird Dataset
Figure 10: Refraction of eBird dataset from 2016 to 2018 using R
4.2. Dates Comparison
When using the Shannon-Weiner biodiversity index, the overall green spaces found
within the study area showed a steady increase in biodiversity in all the sites (Figure 13). While
there was an increase, there also seemed to be a limit of biodiversity. This limit is around 3.5 in
the Shannon index. Between 2016 and 2018, there was a ten percent increase in the overall
biodiversity throughout all sites, with the largest increase being from 2017 to 2018. The amount
of biodiversity variation between sites decreased significantly in 2018. This could be an artifact
of an increased amount of observations in these sites, which produced a more accurate
biodiversity reading.
36
Figure 11: Averages of biodiversity of all sites between 2016 and 2018
Biodiversity had no significance by either location or sample size. The highest average
diversity locations were found at sites with a large difference in the number of observations
(Figure 14). Daan forest park, Shuagxi Riverside, and Mucha park had the highest numbers of
observations; however, Daan was found to have a biodiversity index around 2.5. This is in
contrast to the Shuagxi and Mucha park which averaged around 3. Unique green space attributes
may be the largest factor affecting green space diversity. Sites with lower biodiversity tended to
be more erratic in their yearly biodiversity. However, on average biodiversity increased in later
dates of collection.
37
1
1.5
2
2.5
3
3.5
Mucha Park
Huajiangyanya Natural Park
Nangang Park
Hu Shih Park
Shezidao Wetlands
Shuangxi Riverside Park
Zhongyang Park
Hengguang Bridge Riverside
Park
SanYan Park
Academia Park
Meiti Riverside Park
Zhangxin Park
Nangangquxinfulifunan
Park
Hushan Forested Area
Daan Forest Park
Donghua Park
Baoyi Park
Guanshan Riverside Park
Dajia Riverside Park
Yuqianlou
Minzu Park
Nanxi Park
Rongxing Garden
Middle Mountian 417
Gexin Park
2016 2017 2018
Biodiversity of Sites Over Time
Figure 12: Biodiversity of each site over the years of 2016 to 2018
Patterns emerged over time. These patterns may not show an actual increase over the
years, but instead stable biodiversity within the green space areas within Taipei. It is premature
to say whether positive shifts in biodiversity are affecting many of these sites or if there are other
variables relating to the site’s attributes. While the most stable sites show the highest averages of
biodiversity, this may be because of a temporal biased. Figure 13 showed that most sites in 2018
have higher biodiversity then previous years. It can be noted that some sites that had a lower
biodiversity average had higher in biodiversity for the year 2018 then sites that had higher
averages. Some of the mid-range sites found in the overall averages have stabilized with respect
to their yearly biodiversity levels. Another result was that most of the dates in 2018 tended to be
higher in the mid and lower biodiversity averages.
38
4.3. Biodiversity and Residency
The comparison between biodiversity and species richness is worth discussion
(Figure14). Sites located nearer to the city center tended to have lower biodiversity compared to
green spaces located closer to the edges of the administrative borders. This pattern implies that
the different site attributes have an influence on their biodiversity. When comparing species
richness to biodiversity, there is tends to be higher biodiversity when compared with higher
richness. However, where species richness is below 50, this trend is weaker.
Figure 13: Collective Richness and Biodiversity
39
Biodiversity between residence and migratory avian species showed differences when
viewed across the three years (Figure 15). The migratory species have higher biodiversity in
most of the sites as compared with resident species. However, where there was a lack of water
features, there tended to be an increase in resident species. This is different from migratory
species, which tended to species decrease. This was seen with all separate dates noted from 2016
to 2018.
1.2
1.4
1.6
1.8
2
2.2
2.4
2.6
2.8
3
Mucha Park
Hengguang Bridge Riverside
Park
Huajiangyanya Natural Park
Academia Park
Shezidao Wetlands
Nangangquxinfulifunan
Park
Nangang Park
Zhongyang Park
Hu Shih Park
Shuangxi Riverside Park
Minzu Park
Meiti Riverside Park
Zhangxin Park
Guanshan Riverside Park
Hushan Forested Area
Baoyi Park
SanYan Park
Dajia Riverside Park
Yuqianlou
Donghua Park
Daan Forest Park
Middle Mountian 417
Nanxi Park
Gexin Park
Rongxing Garden
migration resident
Residency of Species to Biodiversity
Figure 14: Migratory and resident species in all selected sites
4.4. Jaccard Clustering
There was a noticeable difference when comparing the migratory and residential avian
species using the Jaccard clustering tool. The dendrograms showed variations between all three
variables (Appendix D,E,F). However, the largest difference was with respect to migratory
species as it was broken into four main groups. Compared to the other variables, the dendrogram
40
only was divided into only three clusters. The species compositions were represented from blue
to yellow depending on the similarities of each site composition (Figures 16, 17 18).
When exploring all the species using JSI, there were species composition similarities that
become apparent (Figure 16). Sites closer to the city center had a similar species composition in
comparison to sites located closer to the outskirts of the city. Both the south and the eastern sites
have similar species composition. When richness and species composition was examined, there
is a weak trend between sites species compositions and site richness.
Figure 15: All species composition of each site. Each marker represents a site defined with size
as richness and color as closeness of species composition of each site of all species between 2016
to 2018
41
The residential species between different sites had similar clustering with respect to the
eBird dataset (Figure 17). However, individual sites within the city had closer species
composition to species more often found outside of the city. Both the southern and eastern
compositions were relatively similar when compared to the combined dataset.
Figure 16: Resident species composition of each site. Each marker represents a site defined with
size as richness and color as closeness of species composition of each site of all species between
2016 to 2018
42
Finally, the migratory species between different sites had similarities with residential
species (Figure 18). All plotted maps within the city map had strong species composition
similarities. Analogous to the resident species, there are clusters of similar migratory species
within the city.
Figure 17: Migratory species composition of each site. Each marker represents a site defined
with size as richness and color as closeness of species composition of each site of all species
between 2016 to 2018
43
While all three plotted species compositions had similarities and differences, there was
not enough information pointing to a strong trend. This outcome may be a result of the more
considerable abundance of the resident avian species in comparison to the migratory species that
were observed in the selected green spaces. While this appears to be a pattern among the selected
sites, the sample size may be too small to generalize to the city of Taipei.
44
Chapter 5 Discussion
Monitoring the ecological health of green spaces within cities has consistently been a challenge
for city planners and government officials. However, VGI datasets have allowed for a more
accessible method for such monitoring compared to traditional means of data collection. It was
determined with the use of eBird datasets that avian diversity in selected sites can be monitored
with varying degrees of success.
5.1. Citizen-collected Datasets
The study indicated that citizen-collected data could monitor avian species in green
spaces in the city of Taipei efficiently. Exploration of various factors inside the dataset has
allowed for a more complete understanding of overall biodiversity and health of the city Taipei.
A distinct advantage in using citizen-collected datasets is the finding appropriate
sampling size for a study. Compared to Shih’s research, the citizen-collected dataset contained
significantly more records where the study’s sites overlapped (Shih 2010). This is seen when
Shih’s overall sample size for entire study was under 1000, while the citizen-collected from the
eBird dataset had a minimum of 800 recorded events per site. These results also show a higher
biodiversity average with respect to the citizen-collected datasets as compared to prior studies.
When using the rarefaction analysis, it was determined that the optimal sample size was 257 for
an individual site. This volume of data is due to the constant recordings through the three years
the VGI dataset (Figure 19). This high volume of records would explain why the biodiversity
averages for most of the sites were significantly higher in sites that of earlier studies. Because
most sites were not able to achieve optimal sampling size using traditional collection methods,
they may be prone to underestimate as site’s avian biodiversity and community’s composition.
45
0
200
400
600
800
1000
1200
1/1/2016
2/1/2016
3/1/2016
4/1/2016
5/1/2016
6/1/2016
7/1/2016
8/1/2016
9/1/2016
10/1/2016
11/1/2016
12/1/2016
1/1/2017
2/1/2017
3/1/2017
4/1/2017
5/1/2017
6/1/2017
7/1/2017
8/1/2017
9/1/2017
10/1/2017
11/1/2017
12/1/2017
1/1/2018
2/1/2018
3/1/2018
4/1/2018
5/1/2018
6/1/2018
7/1/2018
8/1/2018
All Site Observations
Figure 18: Observations recorded of eBird dataset between the dates of 2016 and 2018
It can be noted, while the observation count of the eBird dataset is high, there is a pattern.
As Figure 16 showed, while biodiversity of all the selected sites is becoming higher over time,
the observations have noticeable peaks and valleys. Most studies in Taipei where avian
observations were used as an ecological health indicator were collected from July to November
(Shih 2018; Shih 2010). If a VGI dataset used the same collection dates as previous studies, it
would include some of the lowest recording frequencies. Because of these findings, the study
supports the idea that VGI datasets allow for a higher chance of observing a selected green
space’s richness and biodiversity.
5.2. Different Variables
Analysis of the eBird dataset over three years allowed for a better understanding of green
spaces within Taipei. Of all 25 selected green spaces, all but two of sites (Gexin and Middle
46
Mountain 417 Park) had readable levels of biodiversity for all three years. These sites reported
low levels of average biodiversity and richness due to have significantly lower observer counts
from a single individual yearly selection. However, the sites that had the overall highest average
biodiversity tended to have the most stable year to year (Figure 13). Even with less stable years,
2017 and 2018 tended to be closer in biodiversity compared to earlier dates. Because of this
trend, it can be supported that, VGI's eBird dataset showed levels of reliability when monitoring
both richness and biodiversity of a selected green space as time progresses.
The analysis of resident and migratory species confirmed the there were some unique
differences. While there was a higher number of migratory species, these species have a weak
influence on the overall avian biodiversity of the selected sites. This might be due to many of the
resident avian species being counted in higher number throughout the year while migratory
species are only recorded when passing through on their yearly routes.
5.3. Species Composition
Species composition allowed for more in-depth exploration into both the migratory and
resident avian species. There were no strong links between a site’s species compositions and
their richness. However, when visualizing the species compositions on the map (Figures 16, 17,
18), the sites located further from the urban centers and closer to the mountainous areas tended to
have similar compositions. This clustering could be due to the attributes of the surrounding area.
Geographical attributes such as elevation and proximity to structures could be affecting these
compositions. This was most strongly seen in both the eastern and southern sites where the urban
areas are less dense and closest to the mountainsides. Hillsides and heavily forest areas
demonstrated to similar species clustering effect.
47
5.4. Limitations
The study was able to show trends using citizen-collected data, though there were notable
limitations. The first limitation was the temporal dispersion of recorded events. Unlike traditional
methods of avian data collection that determines a standard recording dataset, producing
temporal evenness, citizen-collected datasets are collected on an observer’s temporal bias.
Because these datasets had peaks and valleys throughout the year and there was temporal
unevenness through the varying sites (Figure 17). This is different compared to traditional
studies that use the breeding seasons for all data entry. However, if used on this dataset it would
have had a much lower Shannon’s Index in most sites. This limitation could be addressed by
either increasing the amount of years or having a standard time based on peaks and valleys. The
second limitation is the multiple attributes that are inherent to individual selected sites. While the
study did address the practical use of citizen-collected data, it did not examine the different
factors that could be affecting their distribution. Attributes for both the locations and the species
were limited in this study. Due to this lack of selected attributes, it would be premature to define
as to the reason these trends are emerging.
48
Chapter 6 Conclusion
The objective of the thesis was to determine if citizen-collected datasets could be used as a
reliable means of determining green space health in Taipei. Other studies have used avian
surveys to conduct this process for the Taipei area in association with green spaces; none have
used citizen-collected data. While the citizen-collected dataset allowed for high numbers of
events, several of these were limited due to inaccuracy in species labels. This section summarizes
the findings of the individual elements and elaborates on further exploration of future endeavors
that could be taken.
6.1. Major Findings
The study confirmed that citizen-collected data can be used for determining a green
space’s ecological health using biodiversity and richness. The study determined a strong
relationship between avian biodiversity and richness of green spaces in Taipei. However, there
was only weak support between a site’s species composition and richness. While there were
limitations, the study was able to find the eBird’s avian datasets can be used in monitoring
multiple sites by a data-driven selection process throughout the municipality of Taipei.
Compared to traditional data collection, eBird has the benefit of simultaneously
collecting data from various sites throughout a specific region. There were originally 458 green
space parcels that correlated with the original criteria of this study. Only the 25 green spaces
selected in Taipei were selected and included for this study. While there were a large number of
green spaces within Taipei, many of these sites had too low of a count of observations to be
accurately accessed. These irregularities can be caused by observational biases for specific
locations. Even using sites with different sample sizes, there was a threshold to biodiversity and
richness found in a specific green space. Identifying the biodiversity and richness thresholds
49
found in these areas can be used in subsequent models as a baseline of overall environmental
health.
The eBird dataset was shown to be reliable from 2016 to 2018 and shows promise of
growth in the future. Many of the sites with the most entries tended to have steadily increasing
biodiversity indexes for all three years. Reviewing how the eBird database has grown over the
past ten years for Taipei, there is the expectation of the increase of usable data for more precise
analysis in the future.
When comparing the migratory and resident species, migratory species tended to have
higher species diversity and were found in more areas. This finding may because of a higher
number of possible migratory species documented under the BLI checklist. The larger list could
have shown greater bias for migratory species. When exploring community species composition,
it was determined that many of the sites have a similar trend between the two specific categories.
This was supported by the species found in the South and Southwestern green spaces.
6.2. Future Work
Although this study was able to examine the avian biodiversity of 25 sites found within
urban areas of Taipei, there were many limitations, as discussed in Chapter 5, as to what could be
ascertained from this individual study. With the ever-increasing amount of accessible data, some
of these studies’ limitation could be overcome.
With the more detailed data acquisition, an in-depth exploration of green space’s
ecological health could provide a richer understanding of the relationship of Taipei’s green
spaces and their biodiversity. All the sites within this study had a basic level of information
designated to them (e.g. area, location, and type) much was not explored. Progressing from this
study, multiple types of attributes will be essential to developing a more complete understanding
50
of specific patterns. Collecting and examining information on types of vegetation, water features,
and human factors will give a clearer picture of what could be affecting various species.
Having more data about different species allows for a better understanding of why
different species inhabit specific green spaces. Knowing nesting and foraging habits can further
understand the niches that these avian species occupy. This information would allow for a better
understanding of a Taipei green space ecological health criteria as it would understand specific
species limitations based on their different attributes. These new datasets could be updated inside
PostgreSQL database table under avian_information. The current scope of the thesis did not
require this information. If explored further, there could be more precise answers on why species
compositions were presents and what is needed for them. This information could help create
more suitable green spaces for multiple native species.
6.3. Overall Conclusions
This thesis was able to show a low-cost method to monitor the ecological health of 25
green spaces of Taipei. Using more accessible datasets, both city planners and citizens can have
a greater understanding of their environment and use this information to develop urban policy
and make planning decisions.
The objective of the study was to determine if the citizen collected dataset could be used
to monitor the health of green spaces. A baseline of avian biodiversity was established during the
exploration of various green space sites within the urban areas of the municipality of Taipei,
Taiwan. Continuous data acquisition from the VGI datasets will allow the observation of trends
in avian population and in the effectiveness of the green spaces in the urban areas over more
extended periods of time. There is an understanding that in the future the trend of urbanization of
the human population will require cities to plan for specific areas. While green spaces allow for a
51
healthier citizenry and support of other organisms within metropolitan areas, they need to be
monitored and better understood so that these limited areas are used to the utmost extent.
Because of the ease and accessibility of VGI datasets, the potential they offer to monitor a city's
valuable land resources will become ever increasingly important as more of the human
population moves to the urban environment.
52
Reference
Aida, Nurul, Selvadurai Sasidhran, Norizah Kamarudin, Najjib Aziz, Chong Leong Puan, and
Badrul Azhar. 2016. “Woody Trees, Green Space and Park Size Improve Avian
Biodiversity in Urban Landscapes of Peninsular Malaysia.” Ecological Indicators 69: 176–
83.
Akbari, Hashem, and Dionysia Kolokotsa. 2016. “Three Decades of Urban Heat Islands and
Mitigation Technologies Research.” Energy and Buildings 133: 834–42.
Barbosa, Olga, Jamie A Tratalos, Paul R Armsworth, Richard G Davies, Richard A Fuller, Pat
Johnson, and Kevin J Gaston. 2007. “Who Benefits from Access to Green Space? A Case
Study from Sheffield, UK.” Landscape and Urban Planning 83: 187–95.
“BirdLife.” BirdLife. Accessed January 2, 2019. http://www.birdlife.org/
http://www.birdlife.org/.
Blair, Robert B. 1999. “Birds and Butterflies Along an Urban Gradient : Surrogate Taxa for
Assessing Biodiversity ?” Ecological Applications 9 (1): 164–70.
Bouten, Willem, Edwin W. Baaij, Judy Shamoun-Baranes, and Kees C. J. Camphuysen. 2013.
“A Flexible GPS Tracking System for Studying Bird Behaviour at Multiple Scales.” J
Ornithol 154: 571–80.
Chamberlain, D. E., S. Gough, H. Vaughan, J. A. Vickery, and G. F. Appleton. 2007.
“Determinants of Bird Species Richness in Public Green Spaces.” Bird Study 54 (1): 87–97.
Chandler, Mark, Linda See, Kyle Copas, Astrid M Z Bonde, Bernat Claramunt López, Finn
Danielsen, Jan Kristoffer Legind, et al. 2017. “Contribution of Citizen Science towards
International Biodiversity Monitoring.”
Chang, Chi Ru, Ming Huang Li, and Shyh Dean Chang. 2007. “A Preliminary Study on the
Local Cool-Island Intensity of Taipei City Parks.” Landscape and Urban Planning 80 (4):
386–95.
Chao, Anne, Robin L. Chazdon, Robert K. Colwell, and Tsung Jen Shen. 2006. “Abundance-
Based Similarity Indices and Their Estimation When There Are Unseen Species in
Samples.” Biometrics 62 (2): 361–71.
Chen, Xiao Ling, Hong Mei Zhao, Ping Xiang Li, and Zhi Yong Yin. 2006. “Remote Sensing
Image-Based Analysis of the Relationship between Urban Heat Island and Land Use/Cover
Changes.” Remote Sensing of Environment 104 (2): 133–46.
Conrad, Cathy C, and Krista G Hilchey. 2011. “A Review of Citizen Science and Community-
Based Environmental Monitoring: Issues and Opportunities.” Environmental Monitoring
and Assessment 176 (1–4): 273–91.
Estoque, Ronald C, Yuji Murayama, and Soe W Myint. 2017. “Effects of Landscape
Composition and Pattern on Land Surface Temperature: An Urban Heat Island Study in the
Megacities of Southeast Asia.” Science of the Total Environment 577: 349–59.
53
Fink, Daniel, Wesley M Hochack, Benjamin Zuckerberg, and Steve T Kelling. 2011. “Modeling
Species Distribution Dynamics with Spatiotemoral Exploratory Models: Discovering
Patterns and Processes of Broad-Scale Avian Migrations.” In Procedia Environmental
Sciences, 7:50–55.
Fontana, Simone, Thomas Sattler, Fabio Bontadina, and Marco Moretti. 2011. “How to Manage
the Urban Green to Improve Bird Diversity and Community Structure.” Landscape and
Urban Planning 101 (3): 278–85.
Fuller, Richard A, Katherine N Irvine, Patrick Devine-Wright, Philip H Warren, and Kevin J
Gaston. 2007. “Psychological Benefits of Greenspace Increase with Biodiversity.” Biol. Lett
3: 390–94.
Goddard, Mark A., Andrew J. Dougill, and Tim G. Benton. 2010. “Scaling up from Gardens:
Biodiversity Conservation in Urban Environments.” Trends in Ecology and Evolution 25
(2): 90–98.
Gotelli, Nicholas J, and Robert K Colwell. 2009. “Estimating Species Richness.” In Biological
Diversity, 39–54.
Guo, Guanhua, Zhifeng Wu, Rongbo Xiao, Yingbiao Chen, Xiaonan Liu, and Xiaoshi Zhang.
2015. “Impacts of Urban Biophysical Composition on Land Surface Temperature in Urban
Heat Island Clusters.” Landscape and Urban Planning 135: 1–10.
Huang, Kuo Ching, and Shih Laing Chan. 2014. “A Study on the Land-Cover Change Indicators
of Taipei Metropolitan Areas.” In IOP Conference Series: Earth and Environmental
Science. Vol. 18.
Imai, Haruka, and Tohru Nakashizuka. 2010. “Environmental Factors Affecting the Composition
and Diversity of Avian Community in Mid- to Late Breeding Season in Urban Parks and
Green Spaces.” Landscape and Urban Planning 96 (3): 183–94.
James, Fraces C, and Stephen Rathbun. 1981. “Rarefaction, Relative Abundance, and Diversity
of Avian Communities.” The Auk 98 (October): 785–800.
Kong, Fanhua, Haiwei Yin, Nobukazu Nakagoshi, and Yueguang Zong. 2010. “Urban Green
Space Network Development for Biodiversity Conservation: Identification Based on Graph
Theory and Gravity Modeling.” Landscape and Urban Planning 95 (1–2): 16–27.
Li, Junxiang, Conghe Song, Lu Cao, Feige Zhu, Xianlei Meng, and Jianguo Wu. 2011. “Impacts
of Landscape Structure on Surface Urban Heat Islands: A Case Study of Shanghai, China.”
Remote Sensing of Environment 115 (12): 3249–63.
Matsuba, Misako, Shota Nishijima, and Kazuhiro Katoh. 2016. “Effectiveness of Corridor
Vegetation Depends on Urbanization Tolerance of Forest Birds in Central Tokyo, Japan.”
Urban Forestry and Urban Greening 18: 173–81.
McKinley, Duncan C., Abe J Miller-Rushing, Heidi L Ballard, Rick Bonney, Hutch Brown,
Susan C Cook-Patton, Daniel M Evans, et al. 2017. “Citizen Science Can Improve
Conservation Science, Natural Resource Management, and Environmental Protection.”
Biological Conservation 208: 15–28.
54
Melles, Stephanie, Susan M. Glenn, and Kathy Martin. 2003. “Urban Bird Diversity and
Landscape Complexity: Species-Environment Associations Along a Multiscale Habitat
Gradient.” Conservation Ecology. Vol. 7. The Resilience Alliance.
“NLSC.” National Land Surveying and Mapping Center. Last Modified 2004.
https://www.nlsc.gov.tw/En.
Oksanen, Jari. 2016. “Vegan : Ecological Diversity.” Cran.R-Project.
https://doi.org/10.1029/2006JF000545.
Shiflett, Sheri A, Liyin L Liang, Steven M Crum, Gudina L Feyisa, Jun Wang, and G Darrel
Jenerette. 2017. “Variation in the Urban Vegetation, Surface Temperature, Air Temperature
Nexus.” Science of the Total Environment 579: 495–505.
Shih, Wan-Yu. 2010. “Optimising Urban Green Networks in Taipei City: Linking Ecological
and Social Functions in Urban Green Space Systems.” University of Machester.
Shih, Wan Yu. 2018. “Bird Diversity of Greenspaces in the Densely Developed City Centre of
Taipei.” Urban Ecosystems 21 (2): 379–93.
Silvertown, Jonathan. 2009. “A New Dawn for Citizen Science.” Trends in Ecology and
Evolution 24 (9): 467–71..
Strohbach, Michael W, Susannah B Lerman, and Paige S Warren. 2013. “Are Small Greening
Areas Enhancing Bird Diversity? Insights from Community-Driven Greening Projects in
Boston.” Landscape and Urban Planning 114: 69–79.
“Taipei DUD.” 2009. Department of Urban Development. Department of Urban Development.
July 26, 2009. https://english.udd.gov.taipei/Default.aspx.
Tajima, Kayo. 2003. “New Estimates of the Demand for Urban Green Space: Implications for
Valuing the Environmental Benefits of Boston’s Big Dig Project.” Journal of Urban Affairs
25 (5): 641–55.
Tan, Jianguo, Youfei Zheng, Xu Tang, Changyi Guo, Liping Li, Guixiang Song, Xinrong Zhen,
et al. 2010. “The Urban Heat Island and Its Impact on Heat Waves and Human Health in
Shanghai.” International Journal of Biometeorology 54 (1): 75–84.
“The World Bank.” 2018. World Bank Group. 2018. http://www.worldbank.org/.
Tramer, Elliot J. 1969. “Bird Species Diversity: Components of Shannon’s Formula.” Ecology
50 (5): 927–29.
Tratalos, Jamie, Richard A Fuller, Philip H Warren, Richard G Davies, and Kevin J Gaston.
2007. “Urban Form, Biodiversity Potential and Ecosystem Services.” Landscape and Urban
Planning 83 (4): 308–17.
United Nations. 2011. Population Distribution, Urbanization, Internal Migration and
Development: An International Perspective. United Nations.
http://www.un.org/esa/population/publications/PopDistribUrbanization/PopulationDistributi
onUrbanization.pdf.
US EPA. 2018. “What Is Open Space/Green Space?” Accessed April 20, 2018.
55
https://www3.epa.gov/region1/eco/uep/openspace.html.
Weng, Qihao, Dengsheng Lu, and Jacquelyn Schubring. 2004. “Estimation of Land Surface
Temperature–Vegetation Abundance Relationship for Urban Heat Island Studies.”
Accessed April 17, 2018. https://doi.org/10.1016/j.rse.2003.11.005.
Wiens, John, Robert Sutter, Mark Anderson, Jon Blanchard, Analie Barnett, Naikoa Aguilar-
Amuchastegui, Chadwick Avery, and Stephen Laine. 2009. “Selecting and Conserving
Lands for Biodiversity: The Role of Remote Sensing.” Remote Sensing of Environment 113
(7): 1370–81.
Wolch, Jennifer R, Jason Byrne, and Joshua P Newell. 2014. “Urban Green Space, Public
Health, and Environmental Justice: The Challenge of Making Cities ‘Just Green Enough.’”
Landscape and Urban Planning 125: 234–44.
56
Appendix
SELECT lan.id, lan.geom,
count(st_intersects(St_buffer(St_Transform(eb.geom,3826),30),
St_Buffer(St_Transform(lan.geom,3826),300))) AS amount
FROM landcover AS lan
JOIN ebird AS eb
ON St_Dwithin(lan.geom, eb.geom, 300)
GROUP BY lan.id, lan.geom
ORDER BY amount DESC
LIMITS (N);
Appendix A: SQL query to determine suitable sites using modified landcover and eBird
dataset. N represent then number of limitations. Both 300 and 50 were used for the value N.
WITH situation AS (SELECT * FROM ebird
JOIN aviebird
ON aviebird.aviannum = ebird.aviannum
WHERE residency = 'M'
AND ebird.aviannum < 900
AND observat_1 BETWEEN '2016-01-01' AND '2018-12-31'),
site1 AS (SELECT SUM(observatio), scientific
FROM situation AS su
JOIN sites AS s
ON st_Dwithin(s.geom, su.geom, 300)
WHERE s.id = 1
GROUP BY su.scientific),
site(N) AS (SELECT SUM(observatio), scientific
FROM situation AS su
JOIN sites AS s
ON st_Dwithin(s.geom, su.geom, 300)
WHERE s.id = (N)
GROUP BY su.scientific)
SELECT DISTINCT situation.scientific, COALESCE(site1.sum, 0) AS site1,
COALESCE(site2.sum, 0) AS site(N),
From situation
LEFT JOIN site1 ON (situation.scientific = site1.scientific)
LEFT JOIN site2 ON (situation.scientific = site(N).scientific)
ORDER BY situation.scientific;
Appendix B: SQL script for transferring variables into R from PostgreSQL
57
Appendix C: Histogram of NDVI dataset
Appendix D: Dendrogram JSI of all species from all selected site
58
Appendix E: Dendrogram JSI of resident species from all selected site
Appendix F: Dendrogram JSI of migratory species from all selected site
Abstract (if available)
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Urban green space accessibility and environmental justice: a GIS-based analysis in the city of Phoenix, Arizona
PDF
A spatial investigation of New York City's historical shoreline
PDF
Utilizing existing museum collections and GIS for paleontological site assessment and management
PDF
Generating trail conditions using user contributed data through a web application
PDF
The movement of Mexican migration and its impact based on a GIS geospatial database
PDF
Risk assessment to wildlife from Ohio on-shore wind farm development: a landscape model approach
PDF
Using aerial imagery to assess tropical forest cover surrounding restoration sites in Costa Rica
PDF
Geographic object based image analysis for utility scale photovoltaic site suitability studies
PDF
An accessibility analysis of the homeless populations' potential access to healthcare facilities in the Los Angeles Continuum of Care
PDF
Building a spatial database of biochar research and practice with Web-GIS
PDF
Comparative 3D geographic web server development: visualizing point clouds in the web
PDF
Collecting and managing VGI infrastructure assessments in support of stability operations
PDF
GeoBAT: crowdsourcing dynamic perception of safety data through the integration of mobile GIS and ecological momentary assessments
PDF
Utilizing advanced spatial collection and monitoring technologies: surveying topographical datasets with unmanned aerial systems
PDF
Creating a Web GIS to support field operations and enhance data collection for the Animal and Plant Health Inspection Service (APHIS)
PDF
Finding the green in greenspace: an examination of geospatial measures of greenspace for use in exposure studies
PDF
Classifying the alpines: developing a methodology to track environmental changes in the alpines utilizing remote sensing and ecozone vegetation patterns
PDF
Exploring global natural disaster and climate migration data: a Web GIS application
PDF
Web-based relational spatiotemporal geodatabase of glaciers in California
PDF
Assessing woody plant encroachment in Marin County, California, 1952-2018
Asset Metadata
Creator
Daniels, Ephriam Joseph
(author)
Core Title
Urban areas and avian diversity: using citizen collected data to explore green spaces
School
College of Letters, Arts and Sciences
Degree
Master of Science
Degree Program
Geographic Information Science and Technology
Publication Date
06/27/2019
Defense Date
04/25/2019
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
avian,geographic information system,green spaces,OAI-PMH Harvest,Taipei,volunteer generated information
Format
application/pdf
(imt)
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Bernstein, Jennifer (
committee chair
), Marx, Andrew (
committee member
), Wu, An-Min (
committee member
)
Creator Email
edani061@gmail.com,ejdaniel@usc.edu
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-c89-179271
Unique identifier
UC11659990
Identifier
etd-DanielsEph-7519.pdf (filename),usctheses-c89-179271 (legacy record id)
Legacy Identifier
etd-DanielsEph-7519-0.pdf
Dmrecord
179271
Document Type
Thesis
Format
application/pdf (imt)
Rights
Daniels, Ephriam Joseph
Type
texts
Source
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Access Conditions
The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...
Repository Name
University of Southern California Digital Library
Repository Location
USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA
Tags
avian
geographic information system
green spaces
volunteer generated information