How we found out there are three trillion trees on Earth

Trees occupy much of the Earth’s land surface. Many of us interact with them on a daily basis, whether driving down a rural country road, back-country skiing, or having a picnic in a city park. Whether large and small, trees play a wide range of roles in our global ecosystem, from sequestering atmospheric carbon to pulling nutrients to the surface from deep within the soil.

Trees offer a suite of ecosystem services for humans and other species, and myriad organizations have long-standing tree planting initiatives to help offset the effects of deforestation and climate change. Tree species have been studied on every forested continent for much of the last century.

Yet, while the field of forest ecology is quite advanced, until now, policymakers and scientists have relied primarily on satellite images to provide estimates of global forest area. Such remote-sensing techniques have not addressed tree numbers, densities or timber stocks, which are important for understanding biological processes and ecosystem structure. So just how many trees are out there? One billion, ten billion, one hundred billion?

Let the counting begin

At the request of a number of organizations interested in understanding the role that tree planting initiatives play in our global ecosystem, my colleagues and I recently completed a study to estimate just how many trees cover the Earth’s surface. Over the two years we spent amassing data and generating estimates, we would often joke about just how long it might take someone to count all the trees if they took to the field with pen and paper. This may be a funny image to conjure, but it isn’t as far from the truth as you might imagine.

A map of tree density at the square-kilometer pixel scale.Crowther, et al, Author provided

Our work started by reaching out to scientists and database specialists to obtain over 420,000 ground-sampled estimates of tree density from more than 50 different countries across the globe. Many of these estimates came from databases that are associated with large-scale inventories, such as National Forest Inventory (NFI) analyses from 21 different countries. Other estimates came from established scientific organizations, such as the Smithsonian Tropical Research Institute, and still others came from peer-reviewed studies published in the last ten years.

Regardless of where the field data came from, researchers visited each and every site where they counted, measured and qualified trees within a one-hectare field plot. These plot data were central to our goals and provided a concrete way to strengthen information obtained from satellites.

Independently, we also acquired or developed a large number of datasets representing vegetative, climatic, topographic and anthropogenic information. These included measures of precipitation, landscape roughness, and photosynthetic activity, to name a few. In the context of this study, these datasets took the form of global maps. Many came from well-known databases, such as the WorldClim database of global climate information, while others were created directly, such as an Enhanced Vegetation Index, which was based on the analysis of satellite imagery.

The heart of our study was developing relationships between these different datasets and the estimates of tree density that came from our ground-sampled field plots. The guiding principle is that some combination of datasets can accurately capture the changes in tree density across the Earth, and through those relationships, we can predict tree density at locations we didn’t measure. For example, if high rainfall was associated with greater tree densities at a number of field sites, we might expect to find high tree densities at other locations that also had high rainfall, even if we didn’t have field data for those locations.

These relationships, or models, relied on multiple linear regression, a form of statistical analysis. We used powerful computing technology to create these models – ultimately developing one predictive equation for each of the world’s 14 major biomes, which are vast areas that share similar vegetative trends and climatic drivers.

How many trees? More than we thought

We exist in a spatial world. Everything has a location, or set of geographic coordinates. To use the above-mentioned models to estimate the number of trees in each of the planet’s biomes, we moved our analysis from a strictly tabular framework (think spreadsheets and tables of numbers) to a Geographic Information System, or map-based framework (think digital image). This enabled us to divide the world into one square kilometer grid cells and to estimate, or predict, a unique tree density for each of the grid cells on Earth.

The result? A map of tree density, which we then summarized and evaluated to arrive at reliable broad-scale estimates of just how many trees exist in each biome and globally.

We estimate that there are approximately 3.041 trillion trees in the world, an entire order of magnitude greater than the previous estimate of 400.25 billion. For each person on Earth there are 422 trees. Many of these trees reach their highest densities in boreal and sub-arctic environments where they can be quite small and tightly packed. However, warmer climates contain the greatest proportion of trees with over 40% of the global population. By far, the biggest factor affecting tree density was land use by people.sian monument/flickr, CC BY-NC-ND

Beyond calculating the number of trees in tropical regions, select countries or globally, our study reveals broad-scale patterns in tree densities between biomes and provides insights into the factors that may control tree density within an ecosystem. For example, tree density often increases with temperature and moisture availability across ecosystems. More startling, perhaps, has been the role that humans have played in shaping forested ecosystems. Anthropogenic impact was the single most consistent driver of tree density across all biomes. Where humans cut forests for timber, for development, and to convert land to agriculture, we have had an overwhelming effect on the planet’s tree populations.

Through our study we estimate that humans cause a net loss of around 10 billion trees annually. Drawing on data from the United Nations Environment Programme that quantifies the Earth’s forest area following the last ice age, we estimate and that the global number of trees has fallen by about 46% since the start of modern human civilization. The rate of tree loss has been highest in tropical regions, but the scale and consistency of this effect across all forested ecosystems highlight how historical land-use decisions have shaped natural ecosystems on a global scale.

A platform for inquiry

The analysis of massive, global datasets is becoming increasingly popular, and our predictions of tree density add to a growing body of geographic data. Just as we drew on a suite of datasets in constructing our predictive models, scientists are now able to take our findings and use them in better evaluating a wide variety of relationships and questions.

Whether researchers use our estimates to explore what constitutes suitable habitat for an animal species, how trees relate to biodiversity, or what role trees play in carbon cycling or other biogeochemical processes, our findings add to our fundamental knowledge of the Earth’s terrestrial ecosystems.

It’s important to remember, however, that this work is not the first, nor the last, step in better understanding the role that trees play in our world. We expect future efforts to continue to refine our estimates of tree density and to reveal all manner of fascinating findings and uncertainties related to them. After all, science is known for generating more questions than answers.

Thomas Crowther receives funding from the British Ecological Society and the Yale Climate and Energy Institute.

Henry B Glick does not work for, consult, own shares in or receive funding from any company or organization that would benefit from this article, and has disclosed no relevant affiliations beyond the academic appointment above.