Tree Diversity and Urban Forests:
New York City Tree Census Analysis
By: Alexandra Stassinopoulos
13 December, 2020
Lots of Trees Grow in NYC
Although New York City is perceived primarily as an urban landscape, the map below shows that this is in fact not strictly true; New York City is home to a hidden urban forest. Even when we exclude trees found in parks and other specifically designated green spaces, the "forest" in the City is dense enough that it is possible to create a street map using the positions of trees. The ability to outline streets simply by tree position is an indication of how influential trees and canopy cover are in urban environments — even if their presence typically blends into the background.
Quick Note: Below, the majority of the areas "without" trees are in fact parks, especially the blank spots in Staten Island and Manhatten. Tree data from parks (including Central Park) was not included in this data set.
The map above plots each tree in the same way. Doing this demonstrates the sheer size of the the city's tree population. In reality, however, the New York City tree population if far from homogeneous; one of the world's larges cities, New York City is home to 132 distinct tree species, which belong to 68 different genuses. Some tree species are more common than others; the most common tree in New York City is the London Planetree, shown below. This species accounts for just over 13.3% of the total tree population. Of the 132 species, however, only the top 15 most common species account for more than 10,000 trees.
*An example of a London Planetree, the most common type of tree in New York City.* Photo Courtesy of NetPS Plant Finder
The second map (below) shows the city's trees mapped by the top most common genuses (species --> genus --> family -->). This maps allows one to understand the scope of the tree population's diversity.
On the surface, there appears to be high biodiversity, even though this map has separate categories for only the the top ten different genuses. In fact, the city's tree population is even more diverse than it appears below. Click on individual points to learn each tree's species and genus.
How Diverse is Diverse Enough?
For any forest, urban or otherwise, it is important to maintain a diverse population of tree species in order to prevent a single invasive insect species or tree disease from wiping out all the trees in a single area (Guyot et al., 2015). This phenomenon, called catastrophic deforestation, has both negative economic and quality of life effects for affected communities; in an urban area, tree canopy can regulate ground temperature (Plumer and Popovich, 2020) and help protect privacy from remote mapping technologies. In natural forests, a diverse tree population promotes biodiversity through the entire ecosystem.
Unlike natural forests, however, urban forests, such as New York City's, are planned. To ensure tree diversity urban planners commonly try to use the "10-20-30 rule," which states that a population should comprise no more than 10% of one tree species, 20% of one tree genus, or 30% of one tree family (Bassuk and Cowett, 2017). In New York City, the percent of the forest population occupied by the London Planetree breaks this rule by a small amount; it accounts for just over 13% of the total population. An important thing to keep in mind, however, is that both the second and third most common trees species each account for 10% of remaining total population. Overall, around 30% of the city's tree population is composed of just 3 separate tree types.
These statistics, however, look at the diversity of the tree population as a whole. In this analysis, I will compare these population statistics with those at the zip code level. Although zip codes are an artificially created level of classification, it is useful to use this level of granularity for my analysis. Zip code districts group New York City into small enough sections to have enough data points, without going so small that one would not expect a reasonable amount of tree diversity in that area.
Note: To see the code I used to generate the population statistics and the maps, see the annotated code at the very end of this page.
Total and Sick Tree Density Across New York City
Below, I've mapped the density of all trees and the density of sick trees by zip code. To account for the fact that zip code districts can vary in size, the density values are normalized by the area of the zip code in 1,000,000 square miles.
The overall density data (green layer) shows that certain areas have a much higher density of trees than others; overall, zip code districts in Queens and areas near Central Park in Manhattan seem to have the most trees. This is important because it means that trees are by no means equally distributed across the city; therefore, the statistics I mentioned in the earlier section only generalize the city as a whole. They are not accurate when speaking about a specific borough or zip code.
Below the map of the overall density, there is a layer showing the density of sick trees in the city. (Toggle the top layer off to access.) Sick trees are those classified as being in "Poor" health in the census data set. A comparison between the green and pink layers show that the areas with the most trees are also the ones with the most sick trees. In terms of overall urban forestry health, this is a good thing; it means that the density of sick trees in certain zip codes is correlated to the fact that these zip codes have more trees in general, rather than to a devastating disease.
Tree Diversity by Zip Code
As established above, the most common tree in New York City, the London Planetree composes 13.3% of the total tree population. Already, that percentage violates the 10-20-30 rule. As mentioned above, the chance of such a tree disease or invasive insect species causing catastrophic canopy loss is increased if a certain area's tree population is much higher than 10%.
The map beneath this text shows what percentage a single tree species makes up in each zip code area. The areas in green show zip codes where the most common tree takes up 10% of the tree population in that district. The yellow, orange, and red districts have more than 10% of a single tree species, with red areas having the most percentage of a single tree species.
Click on individual zip code districts to see their most common tree species and what percent that species makes up in the district's population.
Note: This map does not include the two zip codes which had only one type of tree species; these zip codes were very small, encompassing only one block. Because they were extreme outliers compared to the other values, I filtered them out so that the cloropleth styling would accurately display the rest of the data.
The map above displays a startling result; although New York City as a whole does not violate the 10-20-30 rule by a large amount, 113 out of 185 zip code districts have one tree species accounting for more than 15% of that district's total population. Out of those, 20 zip codes have one species comprising over 20% of the population, and 5 of those have a single species counting for more than 50% of their population. Furthermore, from the map, we can see that some of the districts with the highest percentages are large zip codes in Staten Island and Queens. That is to say, it is not only tiny Manhattan zip codes that have a biodiversity problem.
What Does That Mean for New York City?
Overall, New York City roughly follows the 10-20-30 rule for its tree population. If a particularly aggressive disease affecting London Planetrees were to wipe out the whole species in the City, it would affect less than the 15% of the tree canopy currently provided by the urban forest. However, based on the previous map, the majority of individual zip code districts are at risk of widespread deforestration if their area gets hit by a deadly disease affecting that affects the zip code's most common tree (Guo et al., 2019). Obviously, zip code boundaries are not absolute; they merely indicate a geographic area and nothing more. It is likely that areas neighboring those with low biodiversity are also at risk of possible catastrophic deforestation.
Are these Zip Codes Currently Facing a Deadly Disease?
No. At the moment, most of the zip codes with a disproportionate population of one tree species are not battling a disease with the potential to wipe out a large chunk of its population.
However, the Northeast United States has faced deadly tree diseases in the past, a notable example being the Dutch Elm disease. Inadvertantly brought to the U.S. by boat in the 1930s, this disease single handedly devastated populations of American Elms in Northeastern states (Bassuk and Cowett, 2017; TreeSpeak).
Although it is not widespread, New York City is battling a similarly devastating disease, Oak Wilt. As its name implies, this disease affects oak tree species (Finch, 2017; Showalter and Fanelli, 2018). In New York City, there are over 70,000 different oak trees of varying species. Oak Wilt is deadly to oak trees, and, even though different species have varying levels of susceptibility, a tree can die 4-6 weeks after infection.
Although I do not have enough data to know what is causing the poor health of the sick trees in the census data set, there is a surface level correlation between oak tree species and which trees are sick in New York City. Below, I have created a table which counts the number of zip code districts for which each tree species accounts for the highest number of sick trees. (In other words, the species with the most sick trees in 55 different zip codes is Willow Oak.)
If this data is sorted by the number of zip codes in which a certain species is the sickest tree, two of the top three tree species most likely to be sick in a zip ocde are oak: Willow Oak and White Oak. This result adds an extra dimension to the map of the density of sick trees by zip code. Although sick trees, in general, are more common in areas which have more trees to begin with, certain trees species are more likely to be sick. And, more importantly, the types of species most likely to be sick are not the species which are the most widespread. Both Willow and White Oak species are relatively uncommon in New York City. The popularity rank column in the table below ranks tree species by how common they are in the tree census. This columns shows that Willow and White Oak trees are respectively the 30th and 44th most popular tree species in the city (indexing for all tables starts at 0).
While this data set is from 2015, and it is unclear what caused the sick trees' "poor" health, it is certainly sugestive that the two species accounting for most sick trees in the most zip code districts are both oak.
Although New York City as a whole has a relatively diverse tree population, the extent of this diversity breaks down when the city is looked at through a smaller level of classification. At the zip code level, the majority of individual districts break the 10-20-30 rule, a commonly accepted best practice for ensuring tree diversity in urban forests. The clustering of tree species in this manner puts certain neighborhoods at risk for catastrophic deforestration if a deadly, infectious tree disease were to hit their districts. In the past, the Northeast United States has had certain tree populations, most notabely the American Elm, decimated by such a disease.
At the moment, there are not any deadly tree diseases threatening specific regions of New York City in such a manner. However, there is still reason to be worried. At the moment New York City, state, and surrounding regions are battling Oak Wilt. Even though this disease is not currently widespread, it is possible to tell that there is something affecting the health of oak trees in New York City by analyzing which species are most commonly accounting for the most sick trees in a district. Oak Wilt is especially deadly because it can kill certain oak species in a matter of weeks; if a similar disease were to threaten, for example, the London Planetree and become widespread, the results could be devasting. There are certain zip code districts where London Planetrees account for more than 50% of their tree population; if hit by such a disease, these areas could lose more than half of their tree canopy at great economic and quality of life costs.
This next section showed how I worked with the 2015 NYC Tree Census data, NYC shape files and American Community Surveys data to create the maps and figures above.
In addition to other techniques, I analyzed the data through numerous table joins (i.e. merging dataframes), aggregations, and filters. I also created new columns through expressions. I chose to do this analysis in Python on Deepnote, and not in Carto or Excel, due to the size of my data set. Because Carto locks data after one analysis, it would have been impossible to download and re-upload my data set (which ultimately had to be split into six parts for Carto to accept).
Preliminary Data Cleaning of NYC 2015 Tree Census Data
This part of the analysis will clean and condense the main data set I used for my project, the 2015 NYC Tree Census. In addition to removing unnecessary columns, I added columns that associated categorical datas with integer values so that they could mapped by category in Carto.