How Forests Will Save Us

Introduction

Just over a week ago, on May 12 2024, NPR (National Public Radio) released a neat web app.[1] It compares the 2012 USDA Plant Hardiness Map to the newest version, which came out in 2023[2], to help gardeners across the nation understand how changing temperatures might affect what they can or cannot grow in their backyards. It’s a useful tool, with good visualizations and facts & figures to help Americans understand what they should or shouldn’t plant. When playing around with the new app, I put in a town I’d been to before: Moab, in Southwestern Utah, a hub for visitors to the incredible national parks and monuments in the dramatic scenery nearby.

On the 2012 Plant Hardiness Map, Moab was in Zone 7a. This meant that, on average, the coldest Moab got in the winter was somewhere between 0 and 5 degrees F (or -17.8 and -15 degrees C for those more scientifically inclined). The USDA calculates these zones via a thirty year rolling window; for the 2012 map, that meant looking between at winter temperatures between 1976 to 2005. For 2023, the window moved forward: 1991 to 2020.[1]

Today, in 2023, Moab is now in Zone 7b. Now, the lowest winter temperature gardeners should prepare for is predicted at between 5 and 10 degrees F. The new average coldest it will get in winter, according to the NPR web app, for Moab has gone up by 1.6 degrees F from a decade prior. Now, the USDA is very careful to say[3] that this map and it’s accompanying data should not be used as evidence for or dismissal of climate change, as the window of observance for climate change needs to be on a larger scale than 30 years (although it should be noted that the NPR web app has no such compunction making exactly those claims). But thankfully, there is data out in the world that can assist us in some well thought out and data-backed conjecture.

For this article, we’ll be discussing historical and projected weather data created by the U.S. Geological Survey (USGS) around various geological sites scattered across Natural Bridges National Monument, a beautiful slice of scenic south-eastern Utah, just a bit south of Moab itself. Each site surveyed came with static demographic information (location, elevation, soil composition, and plant makeup) as well as historical and projected weather information specific to each site. For more information on how the researchers created these predictions, I encourage you to visit their website; for the code scripts that cleaned and occasionally averaged their data to create the visualization below, please visit my GitHub, or inspect this document in any RMarkdown viewer.

The Sites

Our dataset covers 113 different sites, spread all across Natural Bridges National Monument. Here we can see a map showing where each site is located, along with the biome classification for each site:

We can see from this map that forests are relatively scarce outside of the north-east region of the park, with a few at the furthest edges of the south-western park boundary. The south-east appears to be dominated by woodland sites, with grasslands prominent in the west and northern areas, and shrublands scattered throughout.

These classifications came direct from the data, and have common sense definitions. Grassland and shrubland biomes are classified, as the name might suggest, by the dominance of the plant life at the site: grass and shrubs, respectively. The difference between forests and woodland sites is a bit more subtle: woodlands and forests both have trees as the dominant plant life at the site, but woodlands are characterized by an open canopy that lets light filter through to lower plants, whereas forests are dominated by closed canopy trees that block sunlight more than woodland trees. Let’s dive into these biomes in a bit more detail.

Biome

There is a roughly equal number of forest, grassland, and shrubland sites in the data (25 each to be exact), but woodland sites are slightly over-represented with 38 sites.

To paint a bit more of a mental picture here, let’s take a look at the relationship between tree canopy coverage, biome, and plant coverage. Each of the 113 sites is plotted in the four graphs below, separated out by site biome. The height of each bar corresponds with the percentage of each site that is covered by tree canopy, with the color of the bar indicating how covered the ground below and around the trees is by plants. Let’s see if we spot any relationships.

Two things may jump out at you immediately. One, forested sites with tree canopy coverage tend to be fairly covered by plants, with none showing truly sparse plant coverage of 25% or below. Oddly though, about a third of the sites don’t seem to have any tree canopy coverage at all! That seems counter intuitive to the definition of forests being defined as dominated by closed canopy trees. There are a few explanations. The simplest could be human error (a mis-measurement or a mis-record in the field), but let’s try out some other theories. I suspect that these sites are dominated by trees that are either dead or dormant; that is, they would have a closed canopy, if they had any leaves to create one.

Switching gears entirely, let’s take a look at the grassland sites:

Grasslands, as perhaps expected, are covered by less tree canopy than their forested cousin-sites. For those sites with tree canopy coverage, it also seems that there is more bare earth as well, given the paleness of many of the bars. This paints a picture of wide open spaces, with the odd tree here or there, and tufts of grass smattered between them. Oddly though, fewer of the grassland sites have 0% tree canopy coverage than forested sites!

Now we come to the shrubland sites, those sites where the predominated plantly is, appropriately, shrubs:

Very few trees to be found here! The shrubland sites with tree canopy coverage are vastly outnumbered by their tree-bare (or at least leafed-tree-bare) sister sites in the same biome. Additionally, plants on the ground are more scarce, with more bare earth showing through. This could be due to rockier terrain, or slopes that offer little protection from extreme weather for all but the hardiest plants.

And lastly, let’s take a look at the woodland sites:

Woodland sites abound! And seem to contain mostly well-tree’d areas, although with less tree canopy coverage than some of the forested sites. Very few woodland sites have no tree canopy coverage at all. Perhaps surprisingly, there seems to be a bit more variation in the plant coverage underneath the leaves of these woodland sites. Some bars are very pale indeed, indicating few plants on the ground and lots of gaps for soil to be seen through.

But tree canopy, as we’ve seen with the shrubland and grassland sites, is hardly everything! Let’s take a look at the relationship between elevation, biome, and the make-up of the plants on the ground.

Elevation

First up we have a box plot of showing the distribution of each biome. If you’ve never encountered a box plot before, there’s a trick to reading them. Given a median (the central most data point in a set), any data set can be chopped up into four quantiles which, broadly speaking, seperate the data into four roughly equal parts from one end to the other. Inside each box is a black bar for the median, and the size of each box above and below that median indicates the spread of points in the quantiles around it. A thin black line protrudes from each end of the box, travelling to the ends of the last two quantiles. Any dots seen outside that line-box-line structure are outliers, at least as far as common wisdom goes. With that established, let’s take a look at the distribution of biomes and elevation:

Forests are found almost exclusively in the middle of the parks range, between 1800 and 1950 meters above see level. Grasslands are typically found a bit lower, but some outlier sites can be found both above and below the typical range. Shrublands cover a wider dispersal than either forests or grasslands, with a median elevation similar to forests but not confined to that area. And woodlands are found at higher elevations, with a median elevation higher than any other biome.

Exposed vs Unexposed Earth

Next, let’s take a look and see if there are any trends concerning biome, elevation, and the percentage of bare earth at a site (that is, terrain not covered by plants of any kind).

Some clustering is obvious here, supporting the conclusions of the box plot above. Forests tend to lie at middling elevation and have little bare exposed earth; woodlands occupy a wide range of elevation but tend to have equal or more bare earth present than forested sites. Grassland is clustered around low elevation and more than 50% bare earth, with shrubs dominating areas with little to no other plant life. This suggests that the conclusions drawn from our earlier bar graphs about tree canopies can be applied more broadly!

Leaf Litter

Dead plant life is not to be dicounted either! Leaf litter can provide an ecological niche for insects and small rodents to shelter from extreme temperatures, as well returning extracted nutrients to the soil as the leaves rot. Let’s see if there’s a trend in leaf litter and elevation by each biome:

It seems like yes, there is! Forests have higher percentages of leaf litter, especially when viewed on average, where the differentiation between grass-, shrub-, and woodlands is not as obvious. Grasslands and woodlands tend to have about the same leaf litter regardless of elevation, with some light clustering, whereas shrublands seem to have a wide spread of leaf litter values.

Plants: Shrubs & Herbaceous Non-Annuals

Let’s take a look at the next four graphs in rapid succession, to see if we can find any trends in the relationship between elevation, biome, and non-tree plant life.

No clear trend jumps out at us here. Each of the averages for shrubs and herbaceous plants seems to be clustered fairly well together, and we don’t see a strong relationship between elevation and plant coverage in any class. This suggests that, despite the clear differences in distribution of biome and elevation, we cannot draw any clear relationship between plant coverage makeup and the elevation in question. Well, if we can’t see anything interesting in the height of the ground, let’s look at the ground itself!

Soil

Soil composition is an important factor in plant growth; after all, it’s what they live in and feed on! Each site’s soil was tested and assigned percentages of what it was made up of; let’s take a look at the relationship between sand and clay, the two most predominant soil categories.

Here we see a clear trend, which makes sense; if the percentage of clay was to rise, then we should see a corresponding drop in the percentage of sand, just to keep the laws of math working. Nothing mind blowing there! But, we can see some mild clustering within the trend; namely, that grasslands tend to be sandier than forests, with forests either favoring more clay-rich soil.

Let’s take a look at the geographic distribution and see if anything jumps out:

Generally speaker, the more eastward we move in the park, the less sandy the soil and the more clay we find in the ground. It’s unclear if this is a cause of or result of the biome distribution, or if there is another force at work on these sites that may be shaping the biome makeup. To that end, let’s move on to the next section of our project: looking at the past to understand the future.

The Past, Present, & Future

Climate change as an existential threat is hardly new; it’s been a constant refrain my entire life. But that does not soften the blow from looking at predictive models.

In addition to creating the site data discussed above, the original researchers on this project reached both backwards and forwards in time to create historical and projected weather & climate data, unique to each site. The gist of these projections took two forms via two representative concentration pathways (RCPs), RCP4.5 and RCP8.5. For those unfamiliar, RCP4.5 represents the “medium mitigation and stabilization scenario” - where we as a planet work to reduce our emissions, and mitigate some of the worst effects of climate change. RCP8.5 is not so optimistic, but instead is the projected outcomes if our current emission rates and energy policies continue into the future. Keep that in mind as we look at the next few graphs.

This is not a very heartening graph. The historical data, in that dark green on the lefthand side of the graph, shows sharp peaks and valleys year by year, but the smoothed trendline shows perhaps a multi-year cycle of warming and cooling (although without reaching farther back into climate data we cannot know for sure). The RCP8.5 scenario in red and the RCP4.5 scenario in blue are stark and inarguable. Even if we as a society make drastic changes over the next few years, we are still in for a period of dramatic warning. If we do nothing, it will be even worse.

This is not limited to annual change from increasingly hot summers: winter temperatures are projected to rise as well.

Although less extreme than the annual change, winters are projected to warm as well. This can mean less snow on the ground in winter, with more melting earlier and earlier, causing unseasonal floods. And summers will continue to get hotter and hotter:

More of the same. Heatstroke is already a danger to visitors of the myriad national parks in the desert; this will become a more and more common risk as temperatures rise.

So - what can we do! Well, first let’s take a look at some historic data broken down by biome to see if we can draw any conclusions from nature’s example.

Here we see the historic summer temperatures and their changes given a bit more room to breathe, with each biome trending relatively similarly to one another. Forests are a tad cooler than woodlands, which are about the same as grasslands (did you spot the peeks of light green on the first line plot above?), with grasslands trending the warmest. But overall, they are fairly well clumped together. Contrast that with the following plot of projected data:

Well now! Once again our unfriendly 45 degree line towards extinction has returned, but there’s a notable difference in the distribution of the biomes - specifically, forests stand apart from their biological biome brothers. While yes, forests were undoubtedly cooler in the historic data, they were not very far off from the other three biomes. Whereas now, forested areas rise in temperature at a similar rate to the rest of the climate, but they seem to lag behind by about half a degree every year.

But that’s just the average over a summer. Perhaps we will see a different trend in the true maximum.

Once again, we see a tight similar historical pattern between all four biomes, with forests as the coolest by about a quarter of a degree. But that half degree difference returns when using the projected data! This suggests there is something about forests that means they can stay cooler than their sibling biomes in the same geographic area, despite rising global temperatures. If we can find out why that is, we may glean some insight that will help us weather rising temperatures ourselves as our climate changes around us.

A Closer Look at Water

We know some difference between forests and other biomes in Natural Bridges National Monument. They tend to occur at middling elevations, but other biomes occur at those elevations frequently as well. They have more tree canopy coverage than shrubland or grassland sites, but woodlands have significant tree coverage as well and share their projected temperature rises with their less-leaf’d sister sites. They tend to have a higher percentage of plant cover, especially shrubs, but not extremely different than other biomes. But they do have a dramatically higher percentage of the ground covered in leaf litter. Damp, decomposing leaves slowly rotting and returning their cellular contents to the soil. This led me to consider a key factor in plant growth: water!

Water from the Sky

Before look at other sources of water, let’s start with one of the big ones: precipitation.

Precipitation patterns seems to vary greatly by season, and see significant change year to year. The declining rainfall in spring over the last 20 years is somewhat concerning, but doesn’t help us understand forests and their differences from other biomes. Let’s take a look at distribution of annual rainfall by biome:

This boxplot indicates that biomes do not see a marked difference in total rainfall from one another. Forests tend to be slightly damper, but only marginally, and all four biomes have almost identical minimums and maximums, with little outlier. This makes logical sense: we are looking at smaller geographic pinpricks in a massive park, which would receive similar amount of precipiation as storms rolled through. So, sheer rainfall cannot be our explanation for why forests weather climate change better. Good for us, as making it rain is hardly an easy thing to do for us mere mortals, but not helpful in aiding climate change strategies. Let’s keep looking!

Water in the Soil

Let’s take a look at water in the ground next. Volumetric water content (VWC) here represents the median amount of water in the soil, measured as a ratio of \(meters^3\) of water over \(meters^3\) of soil. In essence, how much water there will be in the soil across a season and across a year. Let’s take a look at the next five plots together: the axes have been held the same so it’s easy to visualize at a glance the difference between seasons and biomes.

Aha! And here we see a significant difference. In every season, the median of the VWC is higher in forest biomes than any other. Although they do not always have the largest maximum, or even the highest minimum, when viewed as a whole forests hold onto the water in their soil better than other biomes, despite receiving almost identical rainfall as we saw in the previous section. This provides our first clue, that biomes that can retain moisture longer may be more resistant at the micro level to resisting climate change on the macro level. But why the difference? There are two major culprits: soil makeup, and plant makeup. Let’s test both.

Soil & Plants

We know that forests tend to be found it regions with a tad more clay in the soil, namely, the eastern side of the park. But does that yield any insights into why forests can retain more moisture than other biomes in the same climate?

From this box plot we can draw a few conclusions. One jumps out at us right away: regardless of biome, the bulk of the soil is made up of sand across the park. There are no sites where less than half of the soil is sandy; most are more than 60% sand. However, when looking at forests versus the other three biomes, we do note that the median of percentage of sand in the soil is quite a bit lower. While some forest sites are pushing our viewed max of 80% soil, they are possibly outliers that are skewing the results. We see that the median of clay is clustered around the lower end, but that rocky soil is also a component in the uniqueness of soil in the forest. This may suggest that soil with good but not excessive drainage helps retain moisture, and in turn may keep local temperatures lower as global ones rise.

Let’s take a look now at the other potential half of the equation, and returning now to some of what we covered in the first section: plants.

The next two graphs are interactive, so please hover over each box to see the values of the quartiles, as well as minimums and maximums.

Forests are the only biome that have a median plant coverage of above 50%, and that tend to be made up of more plants than bare earth. This again suggests that it’s not only the composition of the soil, but what (or how much) is growing on it that assists in that water retention that cools them off.

Breaking the plants up further by type, we can look at the percentage of herbaceous (meaning low-lying, non woody stemmed) plants versus shrubs. Sadly we cannot look at the number of trees, but we know from previous analysis that forested areas have significant canopy coverage. Before we draw any conclusions, let’s look at our next and final graph:

As we move through the years, the relationship between the water in the soil (as measured by VWC) and water loss from evaporation seems highly seasonal and biome dependent. Summer and Fall move in clumps, and we see no differentiation between loss and water content in those seasons. We also see a slightly positive trend - the more water in the soil, the more lost through evaporation. Spring has some differentiation, but not much. However, in the winter, another trend emerges. Forests, despite having a higher water concentration in the soil, lose less in evaporation than the other biomes. This suggests that closed canopy coverage (as opposed to the open canopy of woodlands), combined with less sandy soil and more plants, especially more shrubs, can lead to more water retention and less moisture loss from evaporation, which is turn may account for that lower trend line in both scenarios.

Conclusion

So what can we learn from this? What can we take from this analysis into our daily lives?

For one thing, we can plant more trees.

That may sound flippant, but we know that planting more trees in city spaces “has offset an average of 0.13ºC of surface warming per decade in European cities.”[4]. Although small, offset is in line with the lower projected temperatures of our forests in Natural Bridges National Monument. De-paving can also help us increase the water retention of our soil, helping our local areas stay cool, which can have knock-on effects as cooler cities mean less energy spent on artificial cooling like AC. Changes like this cannot reverse the trend of global climate change, but perhaps they can buy us a little time as we work on more concrete strategies like emissions reduction, greenifying our electrical grids, and carbon capture.

References

In text citations:

[1] Daniel Wood, Connie Hanzhang Jin, Brent Jones, and Jeff Brady. “The USDA’s gardening zones shifted. This map shows you what’s changed in vivid detail.” National Public Radio, 2024. Accessed May 16 2024 from https://apps.npr.org/plant-hardiness-garden-map

[2] USDA Plant Hardiness Zone Map, 2023. Agricultural Research Service, U.S. Department of Agriculture. Publication date unknown. Accessed May 18 2024 from https://planthardiness.ars.usda.gov/

[3] “Map Creation”. Agricultural Research Service, U.S. Department of Agriculture. Publication date unknown. Accessed May 18 2024 from https://planthardiness.ars.usda.gov/pages/map-creation

[4] Author unknown. “Why using small green spaces can help cities to cool down”. Accessed May 20 from https://www.vectorenewables.com/en/blog/why-using-small-green-spaces-can-help-cities-to-cool-down

General references:

Hartsell, J.A., Schlaepfer, D.R., and Bradford, J.B., 2022, Climate and drought adaptation: historical and projected future exposure metrics for Southeastern Utah Group National Parks: U.S. Geological Survey data release. Accessed May 12 from https://www.sciencebase.gov/catalog/item/61a6952fd34eb622f6978d9f