After the last post, I couldn't get the topic out of my head, and so decided to dive a little deeper into the mechanisms (climatic or otherwise) behind the projected spatial shifts in climate suitability of various kinds.
The effects of projected climate changes on crops can be fairly neatly separated into two groupings: those due to increases in temperature, and those due to increases in CO2. On the whole the former are expected to cause decreases in yields, while the latter will cause increases (though with many interactions and variations by species and region). The temperature effect, at least to me, seemed counterintuitive — though this is perhaps an artifact of hailing from a place that's always limited by too little heat rather than too much. But major crops, even ones famous for their love of heat like corn and cotton, have yields that fall off precipitously when temperatures climb above 30 or 32 deg C (see left panel below). [Not coincidentally, this maximum tolerance of about 90 deg F matches up well with the climate of the Tehuacan Valley of Mexico, where corn is believed to have been domesticated.] This susceptibility to extremes contributes to an optimal growing temperature somewhat cooler than might be expected, and one that is already surpassed for many of the world's staple crops (see right panel below, and first link).
Hidden in the below charts are a multitude of effects: for example, high temperatures contribute to a higher vapor-pressure deficit of leaves relative to air — causing water stress that reduces photosynthesis rates — as well as to more agricultural pests, more ozone, and greater direct danger from the heat itself. There are also fewer cold extremes, but this factor is apparently outweighed, at least when considering the world's current agricultural regions. Meanwhile, higher CO2 has a fertilization effect on "C3" plants (e.g. the majority which are better adapted to high-CO2 conditions, including wheat, rice, and vegetables; the C4 pathway evolved fairly recently in semi-arid climates as global CO2 levels fell). Therefore, C4 plants like corn and sorghum that pump in CO2 will see relatively less CO2 benefit and more temperature damage. The CO2 benefit will also likely be weaker in the tropics, where nutrients and fertilizer usage are both currently low, because increased CO2 will tend to decrease protein levels without additional nitrogen fertilization. Finally, changes in consistency and/or timing of precipitation will strike the tropics the hardest, both because it will require the most adjustment of practice, and because the agricultural sector there is least-equipped to adapt to the change.
Almost all climate projections, including some I cited in the previous post, suggest that the semi-arid and arid areas of the subtropics will become hotter and drier over the course of this century. But it's not self-evident that this is a major factor driving latitudinal shifts in suitability on a global scale, considering that this may be a minor factor of importance only in sparsely populated regions, and that water is routinely moved hundreds of miles from mountain or lake to city. A recent article from a member of my research group succinctly dispels that notion by first observing that about 1.9 billion people live in areas that depend to some extent on snowmelt runoff to meet their water needs, and then mapping the projected change in runoff by 2080 (see figure below) — the decreases coincide in a number of locations with likely demand growth due to temperature and population increases. The large-scale spatial correlation between basins means that in many cases solutions won't be as simple as pumping water from a neighboring watershed. The impressive water-conservation figures coming out of California fortunately show that this projection is not destiny for those who dream of life in warm sunny locales, and technology from pipelines to desalination always has a role to play, but it does help illustrate the flat or increasing suitability of the high latitudes juxtaposed with the decreasing overall suitability of the expanding subtropics.
One may wonder why cold extremes are expected to retreat at a faster rate than warm ones will expand (see figure below) — effectively increasing the latitudinal band spared from conditions outside of the range of human comfort. For one thing, atmospheric moisture increases at about 7%/K, meaning that in the already moist deep tropics it will even more strongly limit temperature extremes, leading to only modest gains in wet-bulb temperature. The strong high-latitude warming on the coldest night of the year is generally thought to be related to a combination of weakening inversions and the albedo effect of retreating snow cover/sea ice. That study found the primary cause of "Arctic amplification" to be the near-constant high-latitude inversion, meaning that warming is distributed over a shallower layer than in the tropics where heat is quickly mixed up to the tropopause. Summertime humidity and cold meltwater in the North Atlantic will further help hold down the warming of the hottest day at high latitudes. This warming, while more spatially uniform, will be strongest in subtropical and mid-latitude grassland regions that presently have just enough moisture to moderate potential high extremes vis-à-vis those in deserts at the same latitudes, but for which that will not be the case 75 years hence. At the forefront of the disparity between trends in cold and warm extremes, and consequently the increasing frequency of the latter relative to the former, are urban areas, where heat waves are presently more intense, and cold waves less intense, than in nearby rural areas, with a widening gap going into the future. This widening can be further ascribed, in part, to decreases in windiness in urban areas, which tends to exacerbate the urban-heat-island effect while simultaneously making high-latitude cities more livable by lessening the windchill.
So perhaps certain cities that are currently quite inhospitable and quiet should begin preparing to gain new cachet as the poleward fringe of the zone for pleasant human habitation accelerates their way. These benefits, of course, are far outweighed by the negatives of habitat loss, ecosystem shifts, and the accompanying extinction of local traditions. The air may feel mild in Nunavut circa 2100, but what would be the point of visiting if there were no polar bears to see, or sealskins to decorate, or walrus tusks to carve?
We've all heard by now about animals shifting their ranges poleward and/or uphill, which has been seen not just in traipsing large mammals but also in more parochial species like insects and fish. Over the last few decades, this progression has been on the order of 1.7 km poleward and 1.1 m uphill per year. [Using a standard estimation of the lapse rate as about 6.5 deg C per 1000 m, this implies an altitude adjustment rate of 0.72 deg C per century, which is quite similar to the observed rate of warming in the atmosphere since 1900 (using GISS data). Estimating the annual-mean equator-pole temperature difference as 50 deg C (equivalent to about 5 deg C per 1000 km) implies a latitude adjustment rate of 0.85 deg C per century. Always gratifying when these back-of-the-envelope calculations imply something reasonable — in this case, that species are shifting their ranges uphill and poleward at the same rate with respect to temperature, and that this leading edge is keeping pace with the warming.] The complete picture is not quite that neat, though: there is also evidence that the equatorial bounds of ranges are shifting less than the poleward ones, if at all. The figure below shows some evidence that the hot end of a climatic range is better tolerated than we would theoretically expect (whereas from the same study the cold end is, empirically, a harder limit).
A comparison of the lab-measured thermal tolerance of a variety of species (determining their placement on the y-axis, and with their latitudinal ranges shown in green) against the highest temperature recently observed in each latitudinal band. While a gross simplification, it makes the point that most species (if not individuals) have likely experienced temperatures in their native habitats exceeding those they would theoretically be expected to not survive. Source: Sunday et al. 2012, supplement.
Given the intriguing findings above, I was curious about whether these concepts had any applicability to humans. After all, despite all our trappings of civilization and thus limited exposure to the unadulterated elements, humans resemble other animals in key ways: we are thermally sensitive, and we acclimate to the climate around us, so that where it's cold we die more from heat and where it's hot we die more from cold (also see top two figures in the panel below). In the U.S., heat is usually considered deadlier than cold as tallied up by the National Weather Service, but in fact a meta-analysis of 74 million deaths published in The Lancet earlier this year found exactly the opposite: for the regions considered (primarily also in the mid-latitudes), cold has killed many more people than heat in the past 30 years. Another interesting finding was that much of this excess mortality was associated not with the (rare) most-extreme temperatures, but rather with the more-common yet still anomalous ones. This pattern held up geographically as well: it wasn't Canada that had the most deaths from cold, or Thailand that had the most deaths from heat, but rather (of the 13 countries examined) China and Italy respectively.
Together, these findings go against many of our presuppositions about the effect of extreme temperatures on society-wide health. Italy is generally considered an epitome of a benign and temperate climate! One hypothesis that occurs to me is that it is the rate of variability that matters most, more than the absolute temperature: too fast and animals' body temperatures stay low, they can burrow into the soil, avoid activity, etc; too slow and acclimatization cuts into mortality. This could be a partial explanation for why the animal species in the figure above were evidently able to survive extreme heat when they 'shouldn't have.' A major caveat on the human side is that the limited data from tropical areas makes it unclear if the pattern in Thailand, Taiwan, and Brazil — more deaths from cold than heat! — is accurate, and if so, if it's global. Because of the much greater amount of moisture, the maximum temperatures are lower in tropical environments than in the arid subtropics, though the wet-bulb temperatures are typically higher... More empirical studies are needed because intuition doesn't provide much of a guide here. It's certainly a complex story — not surprising when finely tuned and highly responsive biological and climatic systems are thrown into opposition. But again, the human-mortality data back up the idea that under CO2-driven warming the tropics stay reasonably habitable even as the high latitudes become less uninhabitable.
Percentage of all recorded deaths attributable to temperatures that were 'moderately anomalous' (simply warmer or colder than optimal) and to extremes, for the period 1985-2012. Note that overall about 7.5% of all deaths were attributable to temperature as the immediate cause. Source: Gasparrini et al. 2015.
Considering the above, all other things being equal we would expect the line of minimum mortality to shift poleward in a warming climate (with many complications for possible changes in temperature variability, storm tracks, moisture levels during the hottest events, etc). U.S. cities with the lowest vulnerability to climate-change impacts are generally in the North, even after taking out the heat-stress component, due in large part to anticipated water shortages. Together with the findings of the Gasparrini et al. study, and factoring in things like the wicked heat observed in parts of the Persian Gulf, maybe the (expanding) subtropics are the most-marginal places currently inhabited, and in that case will be the toughest to remain in. The overall uninhabitability crown will surely be kept by the Antarctic interior, though, which even with exceptional warming will stay bitterly cold.
Going back to the very first sentence of this post, another one of the truisms about climate change is that it is in effect more of a redistribution — of heat, of moisture; in short, of suitability for life — than anything else. This is of course appropriate given that the same could be said of the entire climate system itself. Without redistribution, there'd be local ('radiative') equilibrium, which would look a lot different than what's observed (in the linked figure, E is for Earth observations). The projections for arable land resemble those for mortality, the only difference being that they are based on physiological stresses on plants rather than animals. Again, increased temperature and its corollary, increased aridity (via the vapor-pressure deficit), are the main culprits in losses of arable land in South America and Africa, but globally arable land will likely increase moderately due to gains in the highest latitudes. In the absence of large-scale latitudinal migrations much larger and unlikelier than those discussed in the previous post, and barring any major technological breakthroughs, this shift in regions of agricultural and hydrological suitability over the 21st century away from regions of high population growth — and toward ones of slow or negative growth — means that the cities of tomorrow will have to be supplied by another resource redistribution of similarly massive magnitude.