Global Change Across Scales

Across the studies pooled in one 2011 synthesis, the median land species moved 11 metres uphill and about 17 kilometres toward the pole every decade. Individual species varied widely around those medians. In the same decades, forests were cleared, ships and planes carried species between continents, nitrogen fell from polluted air and carbon dioxide dissolved into the sea. This lesson sets out what each line of evidence shows, how fast a species would have to move to keep pace with warming, and how to tell a detected change from one that has been attributed to a cause.

Moving: range shifts uphill and poleward

A species’ range is the area where its populations live and reproduce. As climate and biomes showed, temperature falls with latitude and with elevation, so warming moves an isotherm, a line of equal temperature, poleward and uphill. If temperature sets the edge of a species’ tolerable conditions, part of its niche, the edge should follow.

Think of a tide line on a beach. As the tide comes in the line moves up the sand, and anything that stays just above the water has to keep moving with it.

I-Ching Chen, Chris D. Thomas and colleagues pooled published range shifts of terrestrial species in a meta-analysis, a statistical synthesis that treats each study’s result as one data point (Science, 2011). The median shift was 16.9 km per decade toward higher latitudes and 11.0 m per decade uphill, and shifts were greatest in studies whose regions had warmed most. Individual species varied greatly. A median summarises that spread.

In the BioShifts database of 30,534 range-shift records, Lenoir and colleagues (2020) found marine species moving poleward about six times faster than terrestrial ones, and they found the records biased toward the developed regions of the Northern Hemisphere.

Worked problem: climate velocity on a mountain. The rate at which an isotherm moves across the landscape is called climate velocity. On a slope it depends on how fast the air warms and on the lapse rate, how fast temperature falls with height.

isotherm shift per decade = (warming per decade / temperature drop per metre of elevation)

Suppose a mountainside warmed 0.3 C per decade, and temperature there falls 0.6 C per 100 m of elevation. Both numbers are invented round figures; only the species’ rate below comes from data.

1. Isotherm speed. 0.3 C per decade / 0.6 C per 100 m = 0.5 x 100 m = 50 m per decade uphill.
2. Species speed. Suppose a species moved at Chen and colleagues’ terrestrial median, 11.0 m per decade. It would cover 11.0 / 50 = 0.22, or 22% of the isotherm’s shift.
3. Lag over 30 years. Thirty years is 3 decades. The isotherm moves 3 x 50 = 150 m; the species moves 3 x 11.0 = 33 m. The species ends up 117 m below the elevation that now has its former temperature.
4. Assumptions. Warming and the lapse rate are constant, the species’ range is set by temperature alone, and nothing such as a road, a farm or a cliff blocks it.

A real lag of that kind could reflect slow dispersal, cues other than temperature, competitors or predators already upslope, or loss of forest. And because the median hides wide variation, a particular species could be far ahead of the isotherm or far behind it.

Timing: phenology and mismatch

Phenology is the seasonal timing of life-cycle events: leafing, flowering, migration, egg laying. Many spring events have moved earlier as springs have warmed. An earlier date is not evidence of successful adjustment.

In the field, warm springs and early laying go together, but warm springs also bring earlier food, so the pattern alone cannot say what the birds respond to. Marcel Visser, L. J. Holleman and S. P. Caro tested temperature directly. They kept breeding pairs of great tits, Parus major, in climatized aviaries for six years, with food freely available, under warm or cold temperature patterns that differed by 4 C in the period that mattered (Proceedings of the Royal Society B, 2009). The warm treatment advanced laying in five of the six years. Because food was unlimited, they interpreted temperature as likely acting as a cue, not only easing the energy cost of egg production.

Timing matters to demography when a consumer and its food shift by different amounts. This is trophic mismatch: the period of greatest demand, such as nestlings in the nest, no longer lines up with the period of greatest supply. Christiaan Both, Marcel Visser and colleagues compared nine Dutch populations of the pied flycatcher, Ficedula hypoleuca, a long-distance migrant (Nature, 2006). Where caterpillars peaked early, flycatcher populations had declined by about 90% over the preceding two decades. Where food peaked late, and early breeders were still well timed, declines were weak at most. That contrast across sites is evidence consistent with timing affecting population size. It is still an observational comparison, and the cross-site caterpillar peaks were measured in one year, 2003, with the share of great tits laying second broods in 1985 to 1990 as a second index of food timing.

Other drivers: land use, invaders, nitrogen and acidifying seas

Land use. Converting forest and grassland to farms, pasture and towns changes local communities directly. Newbold and colleagues (Nature, 2015) modelled a global database of terrestrial assemblages and estimated that land use and related pressures have reduced within-sample species richness by an average of 13.6%, and by 76.5% in the worst-affected habitats. These are model estimates from sampled sites. Land use also interacts with climate: in 2,798 elevational shifts from 43 sites, Guo, Lenoir and Bonebrake (2018) found that forest loss in warmer regions tended to accelerate upslope movement. An uphill shift is not by itself proof that climate acted alone.

Invasive species. An alien species is one moved by people beyond its native range (Blackburn and colleagues, 2014). Seebens and colleagues (2017) compiled 45,813 first records of established aliens, covering 16,926 species, and 37% of those records fell in 1970 to 2014. A first record can lag behind the real introduction, so the curve tracks detection as well as arrival. Impacts run through different mechanisms, including competition, predation and hybridisation, and impacts vary greatly among alien species, from minimal to massive.

Nitrogen deposition. Burning fuel and making fertilizer put reactive nitrogen into the air, where it falls on distant land (see nutrient cycling). On Minnesota prairie, where ambient deposition was 6 kg N per hectare per year, Christopher Clark and David Tilman (Nature, 2008) found that adding just 10 kg N per hectare per year more reduced plant species number by 17% relative to controls. In a second experiment, species number had recovered a decade after additions stopped, but species abundances had not. Why does a nutrient cut diversity? In experimental grassland plant communities, Hautier, Niklaus and Hector (Science, 2009) added light to the understory and prevented the loss caused by fertilization, which implicates competition for light. In a semi-arid grassland in China, Li and colleagues (2024) found that adding water prevented species loss under nitrogen alone or warming alone, but only partly when both were combined. Neither mechanism is universal. How species loss changes ecosystem processes is the subject of biodiversity and ecosystem function.

Ocean acidification. Carbon dioxide dissolving in seawater forms carbonic acid (the equilibria are in water and weak bonds):

CO2 + H2O ⇌ H2CO3 ⇌ H+ + HCO3-, and H+ + CO3 2- → HCO3-

The extra hydrogen ions lower pH, and by combining with carbonate ions they lower the saturation state of calcium carbonate, the measure of how readily shells and skeletons form. Surface seawater stays alkaline, above pH 7, while this happens; “acidification” names the direction of change. Organisms differ. Ries, Cohen and McCorkle (Geology, 2009) held 18 species of benthic calcifiers at controlled CO2 levels for 60 days: 10 calcified less, 7 calcified more at intermediate or high CO2, and the blue mussel did not respond.

How strong is the attribution?

Keep three kinds of claim apart.

• Detection shows that a change happened beyond what chance or natural variation would produce. Detecting a change does not by itself show what caused it.
• Attribution links the change to a cause and tests the alternatives. In the IPCC’s 2007 assessment (AR4; later assessments have updated the wording), “joint attribution” links a biological change to regional climate change and then links part of that climate change to human causes, while assessing plausible alternative explanations.
• Projection is what a model gives under a stated scenario of emissions or land use. It is a conditional result, not an observation of the future.

Rosenzweig and colleagues (Nature, 2008) found that continental-scale warming associated with observed changes in physical and biological systems could not be explained by natural climate variation alone. They also noted that the observations were concentrated in Europe and North America.

Within ecology, designs differ in strength. A manipulation assigns the driver: aviary temperatures, prairie nitrogen, seawater CO2. It shows cause in that preparation but says nothing about why the field warmed. A comparison across sites that differ in exposure, like the flycatcher forests, is stronger than a single trend but can be confounded. A covariate-adjusted association, such as a mixed model with warming and forest loss, can expose a confounder it measured and stays blind to one it did not. For any claim, ask five questions: what response was measured, what driver is proposed, what alternatives were tested, where the independent variation came from, and how large the uncertainty is. Decisions in conservation rest on those answers.

Design problem

At one site, a bird’s laying date has advanced 10 days over 30 years, while the peak of its caterpillar prey advanced 20 days. The population has fallen. What evidence would show that the growing mismatch caused the decline, and what would it take to connect that to human-caused warming?