Patterns, processes, and impacts of abrupt climate change in a warm world: the past 11,700 years

Rapid ecological changes as large as the effects of Euro‐American land clearance in the 19th century repeatedly punctuated the Holocene in response to episodes of temperature change as large as 3–6°C per century. The maximum possible rates of temperature change (a) were calculated by assuming that hydrogen isotope changes in lake sediment records from New England20,21 are entirely attributable to temperature, which ensures that the apparent relationship between ecosystem and temperature change is conservative. (b) Rates of vegetation change were calculated by applying the squared‐chord distance metric to fossil pollen samples.22

Mid‐Holocene climate transitions and variability8,76,79 coincide in time with vegetation changes in eastern North America,6,86 including the hemlock (Tsuga) decline and possible post‐decline periods of aborted recovery (dashed lines). Gray bars mark episodes of drought in the northeast U.S.

The rate of change in June insolation25 (a) is compared with the timing of abrupt changes in (b) European elm (Ulmus) abundance, (c) eastern North American hemlock (Tsuga) abundance, and (d) multiple environmental records from northern Africa (Table 1). For each variable, both a representative individual data series86,116,182 is shown as well as a histogram of ages of abrupt events.8,97

Northern Hemisphere hydroclimate datasets show a series of Holocene abrupt changes, which contrast with (a) the typically ‘stable’ Holocene observed in Arctic ice cores.108 Abrupt changes are recorded (b) off west Africa,102,116 (c) in moisture reconstructions from Wyoming,117,118 (d) in the evidence for changes in shoreline position in a Rocky Mountain lake,117 (e) in the ages of loess deposition in the western Great Plains,119 (f) in a lake sediment record of aeolian deposition in central Minnesota,120 and (g) in a diatom‐inferred record of lake salinity from North Dakota.121 Gray bars mark periods of severe aridity in central North America.

Datasets from the Holocene show patterns similar to the hypothetical scenarios in Figure 2. Data derive from (a) a lake‐level reconstruction,74 (b) the mean of a representative series of North Atlantic sea‐surface temperature (SST) reconstructions,75–77 (c) a fossil pollen record from central Massachusetts, USA,78 and (d) a lacustrine oxygen isotope record from central New York, USA79 (thick line) and a pollen record from coastal Connecticut, USA80 (thin line).

Hypothetical changes in climate (a) and their impacts (b) were produced using an arbitrary trend and adding random variation. The ratio of the magnitudes of the noise (N) and of the trend (T) depends on the time scale viewed; compare the left and right panels at top (a). Different nonlinear responses were calculated for the two time scales. A smooth unimodal response function (b) was applied to both the trend (thin line) and the trend plus variability (gray shaded curve). On the bottom right (c), both a linear response with a threshold (dark line) and an exponential response function (thin line) were applied to the trend and variability above. Abrupt changes in the exponential response are marked A and B.