FIGURE 5 | Marine heatwave duration-intensity ( x - y ) phase space. (A) Solid lines indicate the probability distributions (95% confidence contours) of global annual mean MHW properties from the (blue) historicalNat simulation for 1850–2005, (gray) historical simulation for 1900–1950, (black) historical simulation for 1970–2000, (red) RCP8.5 simulation for 2005–2035 runs and (purple) the RCP8.5 simulation for 2050–2080. Blue overlapping dots indicate annual mean MHW properties from all historicalNat simulations. Orange shading indicates increasing probability of ecological impact (being contours of constant degree-heating-days, i.e., the product of MHW intensity and duration, every 2 ◦ C days over the range 4–20 ◦ C days). (B) Examples of ecological impacts of an organism’s exposure to extreme temperatures as a function of intensity and duration. For a short-term response, filled circles show time to 50% mortality ( M = 0.5, where M is the mortality rate) for intertidal barnacles exposed to temperatures above a threshold defined as that giving 50% mortality after 100 h exposure, with additional fitted lines giving times to expected 80% and 90% mortality ( M = 0.8 and M = 0.9, respectively) (Foster, 1969) (see Supplementary Figure 8 for details of symbols and underlying data). For a medium-duration response, filled areas show threshold degree heating weeks for coral bleaching ( > 4 ◦ C-weeks) and mortality ( > 8 ◦ C-weeks) (Liu et al., 2006; Kayanne, 2017). For the effect of MHWs on annual scales, dashed lines show total annual degree days (above 1988–1999 mean maximum temperatures) associated with low and high annual mortality rates of seagrasses (Marba and Duarte,