STAT220 Final Project - Birds, Storms, and Vegetation Index in Maricopa County, AZ

Introduction

Bird populations are influenced by a myriad of environmental factors, including temperature, habitat conditions, and storm events. Storms in particular can alter bird behavior and abundance by changing food availability, habitat quality, or displacing birds temporarily from their habitats. Understanding how weather phenomena and birds are related is important to us as biologists and amateur statisticians. In exploring possible relationships, we may be able to infer how birds respond to a variable environment.

We also chose to look at vegetation index, which is the measurement of greenery in a given area. At a very broad scale, most birds prefer areas with higher vegetation to sparse vegetation as it provides shelter and food. We hoped that with the addition of vegetation indices to our data, we might be able to get a sense of how birds respond to greenery in this urban environment.

We chose to focus on the Phoenix, Arizona area because of its climate extremes and storms that have been becoming more frequent in recent years. Through this project, we aim to explore the relationship between bird community metrics (species diversity, overall abundance) and storm activity along Salt River survey locations in Maricopa County (where Phoenix is located).

Data Sources

Three datasets were joined for our analyses.

Bird Population Data

Our first dataset contains bird point count survey observations along Salt River sites since 2013. This data set was sourced from the Environmental Data Initiative (EDI), a collaboration between the University of New Mexico and UW-Madison. We deem the data trustworthy because its collection is supported by two NSF grants, and affiliated with two accredited higher education institutions. We also deem it to be accurate because each season, each survey location is visited independently by three knowledgeable birders, who count all birds seen/heard within 15 minutes. This data package is in the public domain, and was downloaded in a CSV format.

Storm Events Data

Our second dataset contains storm event data from Maricopa County, AZ, between January 2000 and November 2025. We acquired this data from the National Oceanic and Atmospheric Administration (NOAA) Storm Events Database, whose data is peer-reviewed. NOAA is a U.S. federal agency, so we believe the data to be accurate. Reporting is done by the National Weather Service (NWS) and other sources, such as media reports and trained storm spotters. One data discrepancy is that some data is missing for certain storms, such as magnitude and exact coordinates. We removed these from our analyses. This data was available to us publicly in CSV format. Each CSV file contains only 500 events, so we compiled roughly 10 different CSV files to acquire the full 2000-2025 span.

NASA MODIS NDVI

Another factor we considered was vegetation index as provided by NASA’s MODIS Normalized Difference Vegetation Index. This data is collected by satellite every 16 days for the entire earth. For our purposes, we extracted the NDVI data from each of our sites for each year of our bird data range (2014-2024), and the NDVI value was averaged per year and site.

Research Questions

1. Does bird abundance/species richness vary on storm/no storm days? This is important for understanding how weather influences urban bird communities. It is also answerable with our data, since storm days are matched to bird survey days.

2. Do certain storm types have stronger effects on bird abundance/species richness? Different storm types may affect birds differently (noise, precipitation, habitat disruption). This is answerable with our data, since joined data allows us to compare across storm categories.

3. Do storms affect some areas of Maricopa County more than others? Storm impacts may vary spatially depending on habitat type, urbanization, etc. This is answerable with our data, since storm dataset has latitude and longitude, and the birds dataset has site code.

4. Does vegetation index affect bird abundance and species richness? Birds rely on vegetation for shelter and food, and because these cites differ in their proximity to urban areas, this is answerable with the spatial data we have.

Graphics

This line plot shows the average amount of bird species observed per survey month across all sites and years (2013-2025). We see that species richness appears to fluctuate seasonally, with higher species richness in the period of spring migration (April-June) and lower values during colder, non-migratory months.

This line plot compares monthly bird species richness across discrete years. Although the overall seasonal pattern is similar across years, the vast majority of years have their lowest species richness in the winter month of January, and highest richness during migration (April-May). 2024 seems to have the lowest overall species richness of all years displayed, which may reflect changes in annual environmental conditions (increasing weather variation, habitat changes, etc.) These year-to-year fluctuations could also be a by-product of variable survey effort.

This boxplot compares bird species richness on storm days versus days with no storms. The two distributions appear similar, which may suggest that storms do not significantly alter the number of bird species detected during surveying. However, this similarity may be due to the fact that storms alter habitat most significantly the day after the event, not the day of. It could be the case that a point count survey was conducted in the morning on a specific day, with the storm occurring later that same day. This would result in a species richness reflecting a non-storm day, on a day that did have a storm(s) occurring.

This boxplot compares the total amount of birds detected during surveys on storm days versus days with no storms, displayed on a logarithmic scale due to large count differences. As the previous boxplots, these distributions overlap. However, there are many more higher values of bird abundance on non-storm days compared to storm days. This suggests that storm presence may influence bird abundance. We see that some surveys recorded large masses of birds, which may occur when birds are detected in flocks.

This boxplot displays variation in species richness among different river survey sites. There were five broad areas where surveys were conducted: Ave35, Ave67, BM, Price, Priest, Rio, and Tonto. These areas are further specified into three sections depending on latitude: low, mid, and upper. Some locations, such as the Tonto sites, display significantly higher median species richness than other sites. Tonto is non-urban and restored habitat, whereas the other sites are either urban and restored, or urban and non-restored. This suggests that non-urban sites, such as Tonto, have more favorable habitat conditions for a higher variety of species. Differences in species richness across sites may be connected to vegetation, restoration status, and surrounding land use.

This lineplot displays the amount of storm events recorded per year in Maricopa County, Arizona. We can see that the amount of storm events has been increasing incrementally, spiking in 2017. The drop in 2020 is likely due to underreporting during the onset of the COVID-19 pandemic. This general trend of increasing storm frequency aligns with our understanding of the progression of climate change, particularly the intensification of weather extremes as a result.

These boxplots compare species richness at each bird count site during storm days versus non-storm days. We see that most sites have similar ranges between these two conditions, but generally the median and range for species richness is lower for storm days. The variation between sites suggests that certain locations with certain types of habitat may be more sensitive to weather disasters than other sites.

Interactive Map of Bird Survey Sites

This interactive map displays the specific geographic locations of the bird survey sites used along the Salt River from our dataset colored by their species richness in 2024. Placing these sites on a visual map helps spatially contextualize our analyses and clarifies how the survey sites are distributed. Bodies of water are crucial corridors for birds, providing a source of water, vegetation, and a path for migration.

This dot plot demonstrates that in sites with lower NDVIs have lower species richness, as seen in the lighter shades of orange. The size of the dot corresponds to the bird count, and while this doesn’t seem to be affected as strongly, species richness appears to show a positive correlation with mean annual NDVI.