Ecosystem Collapse and Extreme Weather Events — in That Order

A Link Persistently Missing from the Climate Change Narrative

There are places on Earth where wild nature rules. But where most humans now find themselves, local ecosystems have been weakened to a morbid state, and the power of nature has faded from people’s worldview.

The two photos below were taken just ten days apart in autumn 2025. I love the bright fall colors in both, yet the worlds they depict are as far apart in natural complexity as one could imagine.

The first photo shows an untamed tributary of a tributary of the Yenisey river in Siberia. The reddish layer on the water is a drift of tree seeds which my great friend Antonio Nobre refers to as the most sophisticated technology on Earth. These tiny vessels have been carrying life forward for millions of years.

The second photo shows local people practicing their national sport on the degrading pastures in Central Asia. According to Professor Emil Shukurov, a prominent Kyrgyz ecologist, it was somewhere in these highlands that, hundreds of years ago, the heroes of the national epos used to lose their way for days as they wandered through the forests. Now this landscape, dominated by large animals, has been pared down to a stark form, radically depleted in complexity.

What are the climatic consequences of this simplification?

We were discussing this question at the United Nations Development Program office in Bishkek, the capital of Kyrgyzstan. The conversation began with the tragedy of the commons: local businesses rarely consider the long-term ecological, environmental, let alone climatic damage caused by overgrazing. But while preparing for the talk, I checked what major global companies at the World Economic Forum list as their key risks for the next ten years. Ecosystem collapse was right at the top, second only to extreme weather events.

https://www.weforum.org/publications/global-risks-report-2025/

Could this explicit risk recognition by major concentrations of global wealth translate into more extensive and effective nature-protection policies? To increase this likelihood, I thought it might help to show the broader public that the top risk, extreme weather events, is also directly tied to the loss of ecosystem wellbeing.

To this end, we recall that the familiar narrative links extreme weather events directly to global warming. (We discussed this recently in “The Rabbit–Duck Illusion in Climate Messaging: An Example from Wildfire Policy.”) Yet while the qualitative link for the global warming appears straightforward—more CO₂, a greenhouse gas, leads to a higher global mean surface temperature—the quantitative picture is less certain. The best global climate models still differ by about a factor of three in their projections of warming for a doubling of CO₂ (as discussed in another post here).

The situation with local temperature extremes is even more challenging, because what matters is not just the mean temperature but the entire temperature probability distribution. And it turns out that the global climate models do not properly capture the trends in extreme temperatures.

Below is a figure from a recent study by Kornhuber et al. 2024, “Global emergence of regional heatwave hotspots outpaces climate model simulations”, showing the difference in the rates of mean and extreme local warming in observations (red) versus models (black), along with their ratio (purple, right axis).

Here, the 99th percentile represents the median day among the hottest 2% of days in a year in a given location on land (as resolved by the ERA5 dataset), while the 87.5th percentile approximates an average summer day. If the hottest days warm faster (or more slowly) than the average summer day, the values on the horizontal axis are positive (or negative), respectively. The analysis covers the period from 1958 to 2022. That both distributions peak at zero means that in most cases the 87.5th percentile warms at the same rate as the 99th percentile (which can result from a particular choice of the percentile values).

The figure above shows that models greatly underestimate both the cases where extreme temperatures rise faster than an average summer day and the cases where they rise more slowly (that is, where extreme trends are buffered relative to the mean trend). Instead, the models tend to produce more locations where mean and extreme temperatures increase at similar rates.

So what could be the reason? Kornhuber et al. 2024 point to several factors, with an explicit mention of the hydrological cycle and vegetation:

Simpson et al. (40) found that trends in humidity, which are strongly dependent on the accurate depiction of rainfall patterns (38), evaporation (which is partially controlled by vegetation), and hydrological characteristics of the land surface, including vegetation are still not accurately reproduced, which could in part explain the discrepancies reported here.

Nevertheless, the overall conclusion, announced as usual already in the abstract (see “Why it is important to read scientific papers beyond their abstracts”), is

Our results highlight the need to better understand and model the drivers of extreme heat and to rapidly mitigate greenhouse gas emissions to avoid further harm from unexpected weather events.

Natural ecosystems as a temperature buffer

The buffering effect of ecosystems on temperature is tied to how they handle water — both locally through transpiration and at larger scales through the regulation of atmospheric moisture transport (the biotic pump). Yet water seems to be a prohibited word when discussing the reciprocal links between climate and biodiversity.

For example, in the recently released 10 New Climate Insights for 2025–2026, Insight 4 acknowledges the interdependence between climate change and biodiversity loss, using formulations like “growing evidence suggests that further loss of biodiversity can contribute to climate change, creating a destabilising feedback.” (Consider that a quarter into the 21st century, this is still presented as a new insight.) However, the supporting text is entirely about carbon, with no mention of the W-word. It almost looks as if the researchers are afraid to walk on the liquid-water surface, preferring the solid carbon ground. Or is it a taboo waiting to be released, in the algorithm that is ruling the process? Yet without embracing how ecosystems move water, there is not the slightest hope of understanding Earth’s climate stability.

There is another hidden caveat in the biodiversity–climate narrative. If biodiversity is understood simply as the number of species in a given area, that number alone does not characterize an ecosystem’s functionality or resilience. For example, early-successional habitats can maximize species counts, yet, being focused on self-recovery, they have limited capacity to regulate regional climatic conditions such as moisture transport. As a result, creating many early-successional habitats may temporarily boost local species numbers but weaken the forest’s overall ability to regulate climate and, hence, its own well-being.

How much space the early successional habitats (that are also home to large animals whom humans like to hunt, hence, for example, widespread complaints of German hunters that forest owners do not cut enough trees) occupy in a healthy forest has been determined by evolution from the condition of maximising the forest climate-regulating capacity rather than the number of species at each point. This climate-regulating function is rarely considered in biodiversity narratives weakening the appeal for conservation or even promotes policies that harm forest resilience (see “Forest-clearing to create early-successional habitats: Questionable benefits, significant costs”).

Returning to vegetation, water and temperature, destroying ecosystems leads to destabilization of the temperature regime, while their recovery helps buffer global temperature trends.

Baker and Spracklen 2019 compared temperature changes that occurred from 2000 to 2013 in nearby locations with intact Amazonian forest and with slight (non-intact), moderate, and severe disturbance. They found, as shown in the third histogram below, that severely disturbed forests warmed by almost an order of magnitude more than undisturbed forests.

Compare this to how global climate models handle temperature changes related to the removal of vegetation. Below is Fig. 7 from Lejene et al. 2017, which shows how models (LUCID and CMIP5) represent changes in maximum and minimum daily temperatures following vegetation removal. OBS refers to the observed differences in Tmax and Tmin between open land and forest, averaged over 22 paired sites in North America. We can see that the models do not reproduce even the sign of the effect.

While vegetation clearing increases temperature extremes, the natural regrowth of forests can literally buffer a large region against global trends. This is what happened with the natural recovery of forests in the Eastern United States.

Below is Figure 1 from Barnes et al. 2024,“A Century of Reforestation Reduced Anthropogenic Warming in the Eastern United States”, where panel C shows the temperature trends in 1900-2010. One can see the “warming hole” in the East, with the region cooling when other regions were warming.

This process cannot continue forever, since transpiration cannot increase without limit. However, with large-scale forest recovery, regional hydrology becomes more stable.

Of course, now that forest biomass has increased, local businesses may want to profit from the accumulated natural capital and return the landscape to a depauperate state (see the right bottom panel above), through measures such as the “Fix Our Forests Act,” which opens forests to logging and burning.

This is the curse of resource abundance. When a resource is abundant, destroying is always cheaper than protecting. Once nothing is left, the recoverers arrive to rebuild the resource for the future destroyers. Stopping this vicious cycle requires intellect and will from human society.

The bigger influence

While the above temperature-related examples that show the role of vegetation are very vivid and often used in ecorestoration narratives, they are really only the tip of the iceberg. The main weather extremes come from changes in atmospheric circulation.

To understand why this is so, let us look at this picture of forest mediated ocean to land moisture transport.

When the biotic pump works normally, moist air rises over the continent. Rain and clouds help cool the land surface. Meanwhile, over the ocean, where the air descends, there are no clouds and there is full sunshine. The area of descent warms. In the Amazon region, this creates what Antonio Nobre called the cold Amazon paradox, in which the air rises over the colder land surface and descends over the warmer ocean surface, something you do not see in textbooks.

Now imagine that we disturb the forest by burning and logging, such that its hydrological power weakens. Since the ocean never has a shortage of water, the circulation pattern readily flips, and now there is descending air motion over the Amazon. This descending air motion, persistent as it was in 2023, caused extra heat due to less clouds and more sunshine as well as because air warms rapidly as it descends adiabatically (Fernández-Alvarez et al. 2025).

Let me digress from ecology for a moment and suggest that you pause to memorize this important fact. I bet most readers have been taught, or have heard, that warm air rises. Yet the strongest heatwaves and the highest temperature extremes are associated with persistent descending air. This is not only what happened in the Amazon in 2023; you can read about other similar events for example in the study by Hotz et al. 2024.

Since it is mechanically costly to push warm air downward, this means that the work that makes it possible is done somewhere else, namely where condensation occurs and rain falls. It is the atmospheric steam engine that produces the power needed to push warm air down, a topic for another post.

Therefore, disturbing vegetation, especially on a large scale, disrupts transpiration and atmospheric moistening and can lead to a situation in which the ocean wins the tug of war with land for moisture, causing positive temperature anomalies on land.

Furthermore, natural forests not only facilitate ocean to land moisture import, but they also make it more stable by not allowing local condensation bursts. Forest trees can be compared to control rods in a nuclear reactor that protect the plant from explosion and ensure the peaceful use of enormous nuclear power. Likewise, forests tame the power of condensation, unlike hurricanes, which are more comparable to nuclear bombs than to power stations. When their control lessens, or is nonexistent, persistent local condensation wetspots can form, locking the atmosphere in this state for a prolonged period of time. This creates floods in one place and droughts and heatwaves in another.

When admitting that climate science is on uncharted territory, it was suggested that the cause of unpredicted changes is the long-term correlations in atmospheric and oceanic currents. These changes, and the role of vegetation in them, are not adequately described by climate models.

Fighting climate change

Now, if we are concerned about weather extremes, as the people at the World Economic Forum clearly are, reducing CO₂ levels does not seem to be a straightforward way to reduce these extremes, since they are not reproduced by climate models that describe CO₂-driven climate change.

Furthermore, while we still do not understand how air circulation changes, applying geoengineering methods to cool the planet, for example by spreading aerosols, is unlikely to bring about the desired climate stabilization. For air circulation, what matters are temperature and humidity gradients and their temporal evolution, and these depend on circulation itself. Even from a mechanical point of view, aerosol-based cooling might help lower the global mean temperature, but in theory it can lead to an even more destabilized circulation and more chaotic weather patterns. Adding aerosols is not thermodynamically equivalent to removing CO₂ or restoring disappearing cloud cover.

Let me conclude with an example that I think is also very vivid. Suppose we continue weakening the biotic pump on land by destroying more and more forests. At the same time, ocean temperatures rise, and the amount of water vapor over the ocean increases as a result. Therefore, even if the land to ocean circulation weakens, this additional water vapor can partially compensate for the declining power of the biotic pump.

Now suppose geoengineers succeed in reducing the global mean surface temperature. What happens then? Forests and ocean-to-land moisture transport continue to decline under human pressure, but now there is less water vapor in the air. In this situation, we can reasonably expect even more drastic disruptions of the water cycle across the planet.

Let us stand up for our ecosystems and their complexity, and refuse to slide into the ecological illiteracy and one-dimensional thinking that our degrading environments can easily breed in us.