Climate reconstruction and human evolution, 2: Tracking climate change

8 minute read

I am examining the pathways that climate might have influenced human evolution, and as I wrote earlier, I’m focusing first on the issue of relatively short-term climate fluctuations. How could submillennial-scale climate variability plausibly impose selection on human populations? I can think of two mechanisms:

  1. Climatic zones might shift rapidly across geographic space, meaning that humans and their prey species must constantly move to track suitable habitat. Movement and periodic habitat contraction tends to increase competition; it also favors traits that expand habitat tolerance.

  2. Global climate fluctuations might decrease the spatial autocorrelation of climate and geography. In essence, local environments become less stable, so that species fine-tuned to local habitats have lower fitness. Greater habitat tolerance is favored.

In both cases I wrote “might”, because even though global climate changes had a high amplitude during the Pleistocene, only the local consequences of those changes would have mattered to human populations. It is a matter of conjecture that those local consequences were (a) of greater amplitude or rapidity than, say, Pliocene climate changes, and (b) large compared to the ordinary year-to-year fluctuation of local environments.

Submillennial Pleistocene climate changes could well have had large effects on the climate of Africa (or Asia, or Europe), and yet those effects might still have been minor compared to the ordinary variation. That is the effect claimed for the last 100 years, where the global temperature trend is large, but the ordinary year-to-year variability is much larger.

How fast does the gradient move?

Conservation biologists are very worried about climate change over the next few hundred years. Human land use has shrunk the geographic ranges of most land species. Many are now limited to small fragments of habitat, sometimes protected by governments in national parks and reserves, sometimes not. Any population that consists of a very small number of individuals is in danger of extinction, both due to casastrophes like disease and fire, or due to the intrinsic effects of strong genetic drift – a so-called “mutational meltdown”. Some abundant species are at risk of extinction if a few key habitat fragments should be lost – the Mexican wintering grounds of monarch butterflies, for instance. If the local climates of these habitat fragments should change, the local biota may shift in ways that compromise the survival of the endangered populations.

So, there is a lot of interest in modeling how global temperature changes may influence local environments. How will rainfall and temperature patterns in small habitat fragments shift as the global system changes?

The recent paper by Loarie and colleagues (2009) is a high-profile example. These researchers attempt to identify a “velocity” of climate change as applied to different terrestrial ecosystems. The idea of a velocity rests on the assumption that local environments exist along a temperature and rainfall gradient, such that getting a little warmer or cooler will shift the boundary between adjacent habitats. If so, then one might determine how fast a biotic boundary will move with a given rate of temperature increase. Plug the global world temperature model into this ecogeographic model, and you might work out the velocity of change expected in the near future.

The result is a range of velocities for different biomes:

Using temperature change calculated from 20002100 under the intermediate A1B emissions scenario, the geometric mean velocity was 0.42?km?yr-1 (0.111.46). (Throughout, we summarize uncertainty in the mean by listing upper and lower,??1 s.d., estimates in parenthesis.) See Supplementary Fig. 17 for other emissions scenarios. We summarize velocity for biomes of the globe and rank them by increasing mean velocity (Fig. 3). Doing so shows that mountainous biomes require the slowest velocities to keep pace with climate change. In contrast, flatter biomes such as flooded grasslands, mangroves and deserts require much greater velocities.

Using these projections, Loarie and colleagues focused on a very specific question. Consider small protected areas where species now find refuge from human habitat disturbance. How long does it take one of these habitat isoclines to traverse the length of such a refuge?

To explore the interaction between protected area sizes and velocities required to keep pace with climate change, we calculated residence times, defined as the diameter of each protected area divided by velocity (km/km?yr-1 = yr). Assuming that protected areas are circular and disconnected, this index can be interpreted as the time for current climate to cross a protected area. Such residence times exceed 100?years for only 8.02% (2.6716.49) of protected areas.

This is a kind of modeling that I approach with heightened skepticism. Remember the “Bigfoot biogeography” paper from this summer? The ecogeographic models will accept any data and spit out an answer. The input data in this case come from a climate model, with its own intrinsic error. The ecogeographic data have additional observation/classification error and variance. Those errors add up.

One thing working in favor of their conclusion is that the northward (and upward) extension of the range of tree species did progress around 1 km/year in many parts of the Northern Hemisphere. They point this out in their discussion, and it brings to mind a couple of very interesting historical examples in evolutionary biology, that I’ll have to review another time.

How would this mode of change affect Pleistocene humans?

Right now, though, I just want to tackle a more limited question: Would this “velocity” of habitat change really have been any challenge for Pleistocene humans?

Historic hunter-gatherers were generally very mobile people. Most of them maintained large home ranges and lived at relatively low population densities, under 4 people per square kilometer. In northern habitats, their density was substantially lower, down to as low as 1 person per 100 square kilometers. There are lots of reasons to doubt that Pleistocene hunter-gatherers could have survived in the high-latitude grassland and tundra, places where people lived recently only with sophisticated logistical strategies, at very low densities. Loarie and colleagues mention high-latitude, low-topography regions as those predicted to have high velocities of biotic change. The temperate ecologies, on the other hand, have very broad distributions of predicted velocities, with a mean around the global average, 0.4 km/year.

That adds up to around 10 km per human generation, which is, give or take, the diameter of a large home range for a hunter-gatherer band. If Pleistocene climates shifted with velocities like those predicted for the near future, a value of 10 km/generation would imply that home ranges at the edge of a habitat regime would be at risk of disappearing (or at least, substantially changing in resources) on a human timescale.

Moving to track climate would be very easily within human capabilities. These kinds of velocities are orders of magnitude smaller than those maintained by people during long-distance migrations. If we imagine a kind of Brownian motion of individuals within the matrix of hunter-gatherer groups, any one individual probably had a larger residence shift during her lifetime than would be necessary to keep up with the secular climate trend. If we assume that people couldn’t move into their neighbors’ home ranges, then the worst effect of this kind of secular climate change would be forcing them to adjust to the resources that exist 10 km away. Not so terrible-sounding.

But maybe the direct effects of climate change were not the important factor. Resource stress in a linear swath of hunter-gatherer groups might have increased social frictions, intensifying the competition for more stable parts of the preferred habitat. If, as I think likely, human groups were embedded in a source-sink metapopulation, climate change would likely have increased the fraction of long-term sink habitat. A nice-looking home range at the edge of a favored habitat would be at high risk of shifting to a different habitat type on the timescale of a few hundred years. Eventually, climate oscillation would return this edge of the range, but possibly at the cost of area on the opposite edge.

Such changes would not be fatal to human populations – after all, people moved much farther. But speaking on the scale of generations, people may have tracked habitat mainly by reproducing more in favorable regions, deciding the issue by reproduction instead of migration. If so, climate change would have reduced effective population size to some extent. How much? Depends on the granularity and sizes of long-term favorable ranges, and the intensity of territory defense by Pleistocene groups.

UPDATE (2010-01-07): For threatened species today, the question of year-to-year variation may be less critical. We assume they are already adapted to year-to-year climate fluctuation, otherwise they’d be dead already. The species that make up their habitat need to have survived that year-to-year fluctuation also. The secular trend increases the number of bad years and makes them on average slightly worse, which may stress some resident species beyond the point they can survive. All this is much worse if genetic variation is already low – then every stochastic fluctuation in population size becomes a chance of extinction.

For Pleistocene humans, adaptation to the year-to-year variation can’t be assumed constant. The extinction of particular groups over a 100-year span is only relevant to our evolution if there was differential survival determined in part by some heritable traits. One way to adapt to year-to-year climate fluctuation is to broaden the resource base – which is also a good strategy to deal with long-term secular trends in climate. But the very ability to broaden resource dependence will tend to reduce demographic stress during bad years.

In other words, I don’t see how these kinds of climate fluctuations are going to lead to runaway or autocatalytic selection. The system seems to have a built-in buffer, at least on the timescale of a few generations.


Loarie SR, Duffy PB, Hamilton H, Asner GP, Field CB, Ackerly DD. 2009. The velocity of climate change. Nature 462:1052-1055. doi:10.1038/nature08649