# Importing library
library(tidyverse)
library(readr)
library(ggplot2)
library(dplyr)
library(gridExtra)
library(ggalluvial)
library(scales)
library(vegan)
library(leaflet)
library(sf)
library(tigris)
tree_data <- read_csv("../2015_Street_Tree_Census_-_Tree_Data_20241120.csv", n_max = 266134)
tree_data <- na.omit(tree_data)Lets first explore the overall composition of NYC’s tree population.
# Count the number of trees for each species and select the top 15
top_species <- tree_data %>%
count(spc_common, sort = TRUE) %>%
slice_max(order_by = n, n = 15)
# Count the number of trees for each species and select the least 15
least_species <- tree_data %>%
count(spc_common, sort = TRUE) %>%
slice_min(order_by = n, n = 15)
# Create the bar plot for Top 15 Most Common Tree Species
top_species_plot <- ggplot(top_species, aes(x = reorder(spc_common, n), y = n)) +
geom_bar(stat = "identity", fill = "#4C72B0") +
coord_flip() +
labs(
title = "Top 15 Most Common Tree Species in NYC",
x = "Tree Species",
y = "Number of Trees"
) +
theme_minimal(base_size = 14) +
theme(
plot.title = element_text(hjust = 0.5, face = "bold", size = 16),
axis.text.y = element_text(face = "bold", size = 12),
axis.text.x = element_text(face = "bold", size = 12),
axis.title = element_text(face = "bold", size = 14))
# Create the bar plot for 15 Least Seen Tree Species
least_species_plot <- ggplot(least_species, aes(x = reorder(spc_common, n), y = n)) +
geom_bar(stat = "identity", fill = "#D55E00") +
coord_flip() +
labs(
title = "15 Least Common Tree Species in NYC",
x = "Tree Species",
y = "Number of Trees"
) +
theme_minimal(base_size = 14) +
theme(
plot.title = element_text(hjust = 0.5, face = "bold", size = 16),
axis.text.y = element_text(face = "bold", size = 12),
axis.text.x = element_text(face = "bold", size = 12),
axis.title = element_text(face = "bold", size = 14)
)
# Arrange the plots side-by-side using gridExtra
grid.arrange(
top_species_plot,
least_species_plot,
ncol = 2
)
The bar plot shows that the London planetree is the most common species in NYC, with over 30,000 trees, simply for its adaptability to the urban condition, such as resistance to pollution, drought, and a wide range of soil pH. While it contributes to NYC’s fall aesthetics with its dense canopies and abundant shedding, its foliage lacks the vibrant colors that species like red maple and sweetgum display, which define the iconic autumn palette. Other major contributors to NYC’s population are honeylocust, Callery pear, and pin oak. However, the high dependence on one species, the London planetree, does raise concerns of monoculture vulnerabilities as any pest or disease affecting it would result in very uneven damage. These implications indicate that further diversification could improve ecological resilience by reducing the challenges of maintenance, given the immense cleanup required for fallen debris from the London planetree.
The second bar plot shows that NYC’s most rare species, including red pine, red horse chestnut, and Douglas-fir, have fewer than 30 trees. These are indicative of the difficult conditions needed for urban adaptation, such as very specific ecological requirements and high sensitivity to pollution. Relatively rare tree species that include Osage-orange and Himalayan-cedar illustrate underutilized biodiversities that can have promise for conservation efforts. Some of these rare species could be assisted to live in an urban environment with improved soil preparation, less competition, or protection from pests. By planting a more diverse set of species, including some of these rare ones, risks from monoculture could be reduced, creating a better balance in the makeup and health of the urban forest.
We will see the prevalence of various tree problems in NYC, categorized into trunk, branch, and root issues using the faceted bar chart.
problem_columns <- c("trunk_wire", "trnk_light", "trnk_other",
"brch_light", "brch_shoe", "brch_other",
"root_stone", "root_grate", "root_other")
# Prepare data
problem_data <- tree_data %>%
select(all_of(problem_columns)) %>%
pivot_longer(cols = everything(), names_to = "problem_type", values_to = "presence") %>%
filter(!is.na(presence)) %>%
group_by(problem_type, presence) %>%
summarise(count = n(), .groups = "drop")
# Create the faceted bar chart
ggplot(problem_data, aes(x = presence, y = count, fill = presence)) +
geom_bar(stat = "identity", position = "dodge") +
scale_fill_manual(values = c("Yes" = "#FF6347", "No" = "#4682B4")) +
facet_wrap(~ problem_type, scales = "free_y") +
labs(
title = "Presence of Tree Problems by Type",
x = "Problem Presence (Yes/No)",
y = "Number of Trees"
) +
theme_minimal(base_size = 14) +
theme(
plot.title = element_text(hjust = 0.5, face = "bold", size = 16),
axis.text.x = element_text(size = 12),
axis.text.y = element_text(size = 12),
axis.title = element_text(face = "bold", size = 14)
)
Branch Problems
brch_light) have the highest frequency of any of the branch issues but still quite rare when compared to “No”.brch_shoe) and other branch problems (brch_other) are negligible.Root Problems
root_stone) are the most significant among all categories, with a notable number of “Yes” responses. This highlights the impact of paved areas on root health.root_grate) and other root problems (root_other) are nearly nonexistent.Trunk Problems
trunk_wire) and lights (trnk_light), are minimal across all trees.trnk_other) are extremely rare and show almost no significant impact.The graph highlights root problems caused by stones as the primary concern, underscoring the need for improved urban planning around tree beds. Branch and trunk problems are negligible, indicating effective maintenance or fewer external stressors.
The bar chart illustrates the Shannon Diversity Index for tree species across different boroughs in New York City.
# Calculate Shannon Diversity Index for Each Borough
diversity_data <- tree_data %>%
group_by(borough, spc_common) %>%
summarize(count = n(), .groups = "drop") %>%
group_by(borough) %>%
summarize(diversity_index = diversity(count))
# Bar Chart
ggplot(diversity_data, aes(x = reorder(borough, -diversity_index), y = diversity_index, fill = borough)) +
geom_col(color = "black", width = 0.7) +
scale_fill_brewer(palette = "Set3") +
labs(
title = "Tree Diversity Index by Borough",
x = "Borough",
y = "Diversity Index",
fill = "Borough"
) +
theme_minimal(base_size = 14) +
theme(
plot.title = element_text(face = "bold", size = 16, hjust = 0.5),
legend.position = "none"
)
The diversity index, which quantifies the variety and distribution of species, reveals some subtle but meaningful patterns:
Queens emerges with the highest diversity index, indicating that this borough has the most balanced mix of tree species. This could be due to its larger green spaces or more distributed urban forestry initiatives.
Bronx and Brooklyn follow closely, showing comparable diversity indices. These boroughs likely benefit from active reforestation programs or ecological conditions that support a variety of species.
Staten Island and Manhattan have slightly lower diversity indices. Staten Island’s lower population density might explain fewer diverse tree planting efforts. On the other hand, Manhattan’s highly urbanized landscape and limited green space could restrict opportunities for introducing diverse tree species.
This analysis suggests that urban planning, tree planting policies, and available green space play significant roles in influencing tree diversity across boroughs. While the differences are not drastic, the diversity index highlights the relative ecological richness of Queens, Bronx, and Brooklyn compared to the more urbanized Manhattan. These findings could guide future biodiversity-enhancing efforts in areas with lower diversity.
The stacked diverging bar chart provides a visual summary of tree health distribution.
# Prepare data
tree_data_filtered <- tree_data %>%
filter(!is.na(health), !is.na(borough)) %>%
group_by(borough, health) %>%
summarise(count = n(), .groups = "drop") %>%
mutate(
health = factor(health, levels = c("Poor", "Fair", "Good"))
)
# Plot data
ggplot(tree_data_filtered, aes(x = count, y = borough, fill = health)) +
geom_bar(stat = "identity", position = "stack") +
scale_fill_manual(
values = c("Poor" = "#F44336", "Fair" = "#FFC107", "Good" = "#4CAF50"),
name = "Tree Health"
) +
labs(
title = "Tree Health Distribution by Borough",
x = "Number of Trees",
y = "Borough"
) +
theme_minimal(base_size = 14) +
theme(
plot.title = element_text(hjust = 0.5, face = "bold", size = 16),
axis.text.x = element_text(size = 12),
axis.text.y = element_text(size = 12),
legend.title = element_text(face = "bold", size = 12))
Overall Tree Health
Borough Comparison
Poor Health Distribution
Brooklyn and Staten Island
Potential for Action
The Cleveland dot plot shows the average tree diameter (DBH) for various tree species, highlighting differences in growth patterns and size among species in NYC.
# Aggregate mean diameter by species
species_avg <- tree_data %>%
group_by(spc_common) %>%
summarise(mean_dbh = mean(tree_dbh, na.rm = TRUE)) %>%
arrange(desc(mean_dbh)) %>%
slice(1:50)
# Create the plot
ggplot(species_avg, aes(x = mean_dbh, y = reorder(spc_common, mean_dbh))) +
geom_point(color = "blue", size = 3) +
labs(
title = "Average Tree Diameter by Species",
x = "Average Diameter (DBH)",
y = "Tree Species"
) +
theme_minimal(base_size = 14) +
theme(
plot.title = element_text(hjust = 0.5, face = "bold", size = 16),
axis.text.y = element_text(size = 12),
axis.text.x = element_text(size = 12),
axis.title.x = element_text(face = "bold", size = 12),
axis.title.y = element_text(face = "bold", size = 12)
)
Top Species with Largest Average Diameters
Species with Smallest Average Diameters
Insights for Urban Planning
Species-Specific Growth Patterns
The pie chart illustrates the proportion of trees in NYC with adjacent sidewalk damage.
# Calculate proportions
sidewalk_proportion <- tree_data %>%
count(sidewalk) %>%
mutate(prop = n / sum(n))
# Enhanced pie chart
ggplot(sidewalk_proportion, aes(x = "", y = prop, fill = sidewalk)) +
geom_bar(stat = "identity", width = 1) +
coord_polar(theta = "y") +
geom_text(
aes(label = scales::percent(prop, accuracy = 0.1)),
position = position_stack(vjust = 0.5),
color = "white",
size = 5
) +
labs(
title = "Proportion of Trees with Sidewalk Damage",
fill = "Sidewalk Condition"
) +
theme_void() +
theme(
plot.title = element_text(hjust = 0.5, face = "bold", size = 16),
legend.position = "right",
legend.title = element_text(size = 12, face = "bold"),
legend.text = element_text(size = 10)
) +
scale_fill_manual(values = c("Damage" = "#F44336", "NoDamage" = "#4CAF50"))
Majority with No Sidewalk Damage:
Minority with Sidewalk Damage:
Urban Impact Insight:
Actionable Consideration:
The alluvial plot visualizes the relationship between curb location, sidewalk damage, and tree health.
# Prepare data for alluvial plot
alluvial_data <- tree_data %>%
group_by(curb_loc, sidewalk, health) %>%
summarise(count = n(), .groups = "drop")
# Alluvial plot with improved axis labels
ggplot(alluvial_data, aes(axis1 = curb_loc, axis2 = sidewalk, axis3 = health, y = count)) +
geom_alluvium(aes(fill = health)) +
geom_stratum() +
geom_text(stat = "stratum", aes(label = after_stat(stratum)), size = 4) +
labs(
title = "Flow of Trees from Curb Location to Sidewalk Damage to Health",
x = "Tree Characteristics",
y = "Number of Trees"
) +
scale_fill_manual(values = c("Good" = "#4CAF50", "Fair" = "#FFC107", "Poor" = "#F44336")) +
theme_minimal() +
theme(
axis.text.x = element_text(hjust = 1, size = 12),
axis.text.y = element_text(size = 12),
axis.title = element_text(size = 12, face = "bold"),
plot.title = element_text(hjust = 0.5, size = 14, face = "bold"))
Curb Location
Sidewalk Damage
Tree Health
Relationships
The plot highlights the interplay between tree placement, sidewalk integrity, and tree health. Proper placement (e.g., OffsetFromCurb) could mitigate sidewalk damage and promote better tree health.
ggplot(tree_data, aes(x = sidewalk, y = tree_dbh, fill = sidewalk)) +
geom_violin() +
labs(
title = "Tree Diameter and Sidewalk Damage",
x = "Sidewalk Condition",
y = "Tree Diameter (DBH)"
) +
scale_fill_manual(values = c("Damage" = "#F44336", "NoDamage" = "#4CAF50")) +
coord_cartesian(ylim = c(0, 100)) + # Adjust the y-axis limits to zoom in
theme_minimal()
Distribution of Tree Diameters
Potential Correlation
Tree Size and Damage
NoDamage Category
The bubble chart offers several insights into the distribution of tree species and their health status across different boroughs in New York City.
# Calculate the Top 5 Most Common Species
top_species <- tree_data %>%
count(spc_common, sort = TRUE) %>%
top_n(5, wt = n) %>%
pull(spc_common) # Extract species names as a vector
# Prepare Data for Bubble Chart
bubble_data <- tree_data %>%
group_by(borough, spc_common, health) %>%
summarize(count = n(), .groups = "drop") %>%
filter(spc_common %in% top_species) # Focus on top 5 species
# Bubble Chart
ggplot(bubble_data, aes(x = borough, y = spc_common, size = count, color = health)) +
geom_point(alpha = 0.7) +
scale_size_area(max_size = 15) +
scale_color_manual(values = c(
"Good" = "#27ae60",
"Fair" = "#f1c40f",
"Poor" = "#e74c3c"
)) +
labs(
title = "Tree Species vs. Borough by Health",
x = "Borough",
y = "Species",
size = "Tree Count",
color = "Health Status"
) +
theme_minimal(base_size = 14) +
theme(
plot.title = element_text(face = "bold", size = 16, hjust = 0.5),
axis.text.x = element_text(angle = 45, hjust = 1),
legend.position = "right"
)
# Prepare Data for Scatter Plot
scatter_data <- tree_data %>%
filter(tree_dbh > 0) %>%
pivot_longer(
cols = starts_with("root") | starts_with("trunk") | starts_with("brch"),
names_to = "problem_type",
values_to = "has_problem"
) %>%
filter(!is.na(has_problem) & has_problem == "Yes")
ggplot(scatter_data, aes(x = tree_dbh, y = health, color = borough)) +
geom_jitter(alpha = 0.3, size = 1.5) +
facet_wrap(~problem_type, ncol = 4, nrow = 2, scales = "free_y") + # 4 columns and 2 rows
scale_x_continuous(
trans = scales::log_trans(base = 2),
breaks = c(1, 2, 4, 8, 16, 32, 64), # Log2 breaks
labels = c("1", "2", "4", "8", "16", "32", "64")
) +
scale_color_brewer(palette = "Set2") +
labs(
title = "Tree Diameter vs. Health by Problem Type",
x = "Tree Diameter (DBH) (Log2 Scale)",
y = "Health Status",
color = "Borough"
) +
theme_minimal(base_size = 14) +
theme(
plot.title = element_text(face = "bold", size = 16, hjust = 0.5),
legend.position = "top",
strip.text = element_text(face = "bold"),
axis.text.x = element_text(angle = 45, hjust = 1)
)
root_grate, root_other, root_stone) display a significant decline in health as their diameter (DBH) increases.We can get intutiton that:
# Custom labels for log base 2
log2_labels <- function(x) {
paste0("2^", log2(x))
}
# Prepare Data for Stewardship Analysis
steward_health_data <- tree_data %>%
filter(!is.na(steward)) %>%
group_by(steward, health) %>%
summarize(count = n(), .groups = "drop")
# Stacked Bar Chart with Log Base 2 Transformation and Labels
ggplot(steward_health_data, aes(x = steward, y = count, fill = health)) +
geom_bar(stat = "identity", color = "black", position = "stack") +
scale_y_continuous(
trans = scales::log_trans(base = 2), # Logarithmic transformation with base 2
breaks = scales::trans_breaks("log2", function(x) 2^x), # Breaks at powers of 2
labels = log2_labels # Custom labels in the form of 2^n
) +
scale_fill_manual(values = c(
"Good" = "#27ae60",
"Fair" = "#f1c40f",
"Poor" = "#e74c3c"
)) +
labs(
title = "Tree Health by Stewardship Level (Log Base 2 Scale)",
x = "Stewardship Level",
y = "Number of Trees (Log Base 2 Labels)",
fill = "Health Status"
) +
theme_minimal(base_size = 14) +
theme(
plot.title = element_text(face = "bold", size = 16, hjust = 0.5),
axis.text.x = element_text(hjust = 1),
axis.text.y = element_text(size = 12), # Larger font size for y-axis labels
legend.position = "top"
)
The stewardship level significantly influences the health of trees, as evidenced by the data. The graph, plotted on a log base 2 scale, highlights that trees without any recorded stewardship (“None”) constitute the largest group. This category also exhibits the widest variability in health status, with a substantial proportion of trees in “Good” health, followed by “Fair” and “Poor” health. This could indicate that while trees without stewardship are at higher risk of deteriorating health, a significant number still maintain good health, potentially due to favorable environmental conditions or maintenance by other means.
Trees under stewardship levels of “1or2” and “3or4” show similar patterns, with a notable dominance of trees in “Good” health. This suggests that even minimal stewardship efforts play a crucial role in maintaining tree health. However, the diminishing trend in tree count across higher stewardship levels indicates fewer trees fall under intensive care or monitoring (“4orMore”), which might imply resource limitations or a focused approach on only select trees.
The log scale representation emphasizes the exponential difference in tree counts across stewardship levels, making it evident that the majority of trees lack any formal stewardship. This highlights an opportunity for urban forestry programs to engage more stewards or allocate resources to currently neglected trees.
Overall, the graph suggests that stewardship has a positive impact on tree health, but the lack of stewardship for a significant proportion of trees is an area of concern. This could warrant further investigation or policy recommendations to enhance stewardship participation and ensure a healthier urban forest.
The map visualization provides a detailed and intuitive representation of tree health across New York City boroughs, allowing us to draw several key observations:
invisible(
capture.output(
nyc_boroughs <- tigris::counties(state = "NY", cb = TRUE) %>%
filter(NAME %in% c("New York", "Kings", "Queens", "Bronx", "Richmond"))
)
)Retrieving data for the year 2022sampled_tree_data <- tree_data %>% sample_n(5000)
leaflet(sampled_tree_data) %>%
addProviderTiles(providers$CartoDB.Positron) %>%
addPolygons(data = nyc_boroughs,
color = "#2c3e50",
weight = 2,
fillColor = "#ecf0f1",
fillOpacity = 0.3,
label = ~NAME) %>%
# Add Circle Markers for Tree Data
addCircleMarkers(lng = ~longitude,
lat = ~latitude,
radius = ~ifelse(tree_dbh > 20, 6, 3), # Tree size based on diameter
color = ~case_when(
health == "Good" ~ "#27ae60", # Green for "Good"
health == "Fair" ~ "#f1c40f", # Yellow for "Fair"
health == "Poor" ~ "#e74c3c", # Red for "Poor"
TRUE ~ "#bdc3c7" # Gray if health is missing
),
stroke = TRUE,
weight = 0.7,
fillOpacity = 0.8,
popup = ~paste0(
"<strong>Species: </strong>", spc_common,
"<br><strong>Health: </strong>", health,
"<br><strong>Status: </strong>", status,
"<br><strong>Diameter (DBH): </strong>", tree_dbh,
"<br><strong>Borough: </strong>", borough
)) %>%
# Cluster the markers to avoid overcrowding
addMarkers(lng = ~longitude,
lat = ~latitude,
clusterOptions = markerClusterOptions()) %>%
# Borough Legend
addLegend(position = "topright",
colors = c("#2c3e50"),
labels = c("Borough Boundaries"),
title = "Map Key",
opacity = 0.8) %>%
# Health Legend
addLegend(position = "bottomright",
colors = c("#27ae60", "#f1c40f", "#e74c3c"),
labels = c("Good Health", "Fair Health", "Poor Health"),
title = "Tree Health",
opacity = 0.8)