does tree planting improve health — ECHO: Environmental Change and Human Outcomes Lab

Perhaps surprisingly, I don’t think we actually know.

By: Marshall Burke

A recent paper suggests that increasing greenness by about 30% in cities around the world could reduce heat related deaths by a third and save over a million lives globally. Many other papers (e.g. a related recent paper in Europe) make similar claims. These and a long list of related findings have made tree planting one of the primary interventions for local climate adaptation, particularly in urban environments where “urban heat islands” exacerbate maximum temperatures. Hong Kong has planted a remarkable 100 million trees. Many cities in the US (NYC, Chicago, and others) have similar, if somewhat less ambitious, programs. My own conversations with local governments suggests this is a very common local strategy too, with (at least until recently) fairly easily accessible grant money for local tree planting efforts in the name of climate adaptation.

But does tree planting actually reduce heat related health impacts, including the additional mortality expected under a warming climate?

A range of careful scientific work, and the lived experience of just about everyone on earth, tells us that trees provide shade and that it’s generally cooler to stand under a tree than to stand in the sun. But the amount that trees reduce surface temperatures can depend on the tree species, details about tree canopy (e.g. size, shape, fragmentation), the interaction between tree canopy and other impervious surfaces, and also how hot or humid it is on a given day. There’s some recent evidence that at very hot temperatures, trees reduce their transpiration and thus their cooling benefits are substantially reduced. There is somewhat mixed evidence on trees’ effect on nighttime temperature, with many studies suggesting no nighttime benefit. Other studies suggest that temperature benefits are non-linearly increasing in canopy cover, with limited benefits to greening at low levels of initial tree cover. Here are the data from that last study, from the obviously-internationally-representitive city of Madison, Wisconsin, showing that reductions in ambient temperature get larger at higher average levels of tree coverage and that effects are larger when measured at larger spatial scales.

Overall, though, it seems clear that there is at least some benefit of increased tree cover for maximum temperatures. Plus, trees are good for all sorts of other stuff too: they provide habitat for other critters, they can filter air pollutants, and they are nice to look at. People like trees! This nice paper uses variation in tree death induced by the spread of the ash borer in Toronto to show that one additional tree in a postal code raises property values in that postal code by 0.45%. That seems like a massive effect. Or maybe Toronto just doesn’t have that many trees. In any case, humans (among other species) clearly like trees.

But is planting trees a good climate adaptation strategy? Do they reduce heat-related health impacts? How costly is it relative to other approaches? Even if we like trees for other reasons, I think it’s worth knowing the answer to these questions: it’s useful to know what specific interventions are effective for reducing specific climate impacts we care about, and extreme heat is certainly one of the main impacts we’re worried about. Some studies suggest that heat-related mortality is the single most important climate impact globally. So, we want to know what interventions can help reduce it.

My assessment of the current literature is that, somewhat surprisingly, we don’t currently have very good causal evidence that planting trees reduces heat-related morbidity or mortality. Most of the studies cited at the top do not have research designs that, in my view, plausibly isolate the effect of tree planting on heat-related mortality. One common approach is to compare heat related mortality in places with more versus less tree cover. You might call this a “heterogeneity-on-observables” design, which requires you to be convinced that the source of heterogeneity you’re interested in (here: tree cover) is not correlated with other un-controlled-for factors (hydrology, average climate, local incomes, etc) that also could affect heat related mortality. Basically, you can estimate a causal heat-mortality relationship using time-series variation in temperature and mortality, but estimates of the heterogeneity in that relationship are typically not causal, absent some strategy to find random variation in the moderating variable. Nevertheless, estimates of heterogeneity are often interpreted as causal: improving tree cover by X would reduce heat-related mortality by Y, is the claim. There’s another related concern too: if planting trees raises home prices and only healthier and wealthier people can afford to live near trees, then we could mistake this “selection” effect for the causal impact of trees on health.

Many studies are of course aware of these concerns and try to think carefully about them. This nice plot from a detailed look at US mortality data shows that locations in the top decile of “imperviousness” (think: opposite of trees) have way higher mortality on hot days, but no differential mortality on cool days, which helps rule out many (but not necessarily all) selection and confounding stories.

Fig 5 from Chakma, Colmer, Voorhies 2024.

The other common approach to measuring the benefits of trees is to take perhaps more-reliable estimates of the effect of expanded tree cover on surface temperatures, and then push these temperature changes through existing causal temperature-mortality dose response functions to estimate the change in mortality. But this of course assumes that (1) you have the temperature change correctly estimated and (2) tree planting only affects the ambient temperature you care about and doesn’t have other non-temperature effects that shape health outcomes. But alongside the above evidence on the mixed effects on surface temperatures, there is now a host of evidence that trees affect all sorts of stuff related to health – much of it good but not all of it. Quasi-experimental studies have found that urban afforestation projects in Beijing and New York City reduced air pollution (that’s good!) but increase pollen counts and respiratory ED visits (in Beijing, bad!), and a decline in trees due to the emerald ash borer increased air pollution and mortality across US cities (bad! Meaning, trees good) but did not measurably change temperature. Here is the plot from the Beijing study (Xing et al 2023) showing pollution benefits inside the regions where trees were planted, after they were planted (crappy red annotations are mine):

A third general worry that we should have with the existing evidence is the difference between the effects of existing canopy cover, and the effects of programs designed to increase that cover. The latter is, I think, the policy question of interest. These two things are not obviously equivalent: planting trees is easier than ensuring they survive, and programs have struggled with maintenance and survival in the past. In my own conversations with local policy folks, they have repeatedly emphasized that it’s much easier to get a grant to plant trees than it is to get funding to maintain them.

None of this means that we shouldn’t plant trees, or that trees don’t have all sorts of benefits. To torture a double negative, it also doesn’t mean that trees don’t reduce temperature-related mortality. We just currently don’t have strong causal evidence that tree planting programs reduce temperature-related mortality. This matters because we want to know how to reduce heat-related mortality. Figuring out whether a tree planting program can do this seems like a really important research priority.

Does planting trees reduce heat-related health impacts?

Does prescribed burning reduce wildfire smoke?