Plain Language Summary
Periods of extremely high ocean temperatures that persist for days to
months, known as Marine Heatwaves (MHWs), can cause the loss of marine
life and impact coastal communities and economies. Climate change is
expected to drive substantial increases in the length, strength and
frequency of MHWs this century. There has been less analysis, however,
of the characteristics of individual MHWs, like the rate at which they
develop. In this research, we examine how well climate models can
simulate these characteristics and the implication for future
projections. We find considerable biases in the simulation of some key
MHW characteristics in parts of the ocean due to model limitations in
capturing physical processes like surface winds along the equator. Most
MHW characteristics like duration and total heat stress are projected to
increase sharply this century, particularly for coral reefs and kelp
forests, although the increases in some regions are likely overestimated
due to model biases. Conversely, we project decreases in “priming” –
periods of sub-lethal heat stress that help marine life prepare for heat
waves. These findings identify regional errors to consider when
interpreting MHW projections and can help researchers identify areas for
improving model performance.
1. Introduction
Over the past few decades, marine heatwaves (MHWs) have become longer,
stronger and more frequent (Frölicher et al., 2018; X. Li & Donner,
n.d.; Oliver et al., 2018). These periods of anomalously high sea
surface temperatures (SSTs) have severely affected marine ecosystems
including changes in species distributions, mass mortality, loss of
biomass, degradation of ecosystem function and decline in ecosystem
services (Arias-Ortiz et al., 2018; Cheung et al., 2021; Smale et al.,
2019). MHWs during the summer or warm-season, when temperatures are more
likely to exceed organisms’ upper thermal tolerance, are a particular
threat to habitat-forming systems in which the foundational species are
vulnerable to heat stress. Warmwater coral reefs are susceptible to heat
stress of as little as 1-2 °C above long-term average summer
temperature, which can interrupt the symbiont relationship between coral
and microalgae living in coral tissue, leading to the phenomenon known
as coral bleaching. For example, more than 75% of warmwater coral reefs
experienced some bleaching during 2014 and 2017, which caused mass loss
of living coral and cascading effects on reef ecosystems (Hughes et al.,
2017; W. J. Skirving et al., 2019; Sully et al., 2019). Kelp forests are
also severely threatened by MHWs, which can cause mass mortality,
changes in the food web and phase shifts to urchin-dominated systems
(Arafeh-Dalmau et al., 2019; Filbee-Dexter et al., 2020; Rogers-Bennett
& Catton, 2019; Smale, 2020).
It has been well documented that MHWs are likely to become more
frequent, intensive and longer-lasting under climate change throughout
the 21st century (Frölicher et al., 2018; Oliver et al., 2019). Most
studies of the projected impacts of MHWs on marine ecosystems have
focused on the frequency and intensity of MHWs, and not considered other
properties which can affect marine ecosystems and organisms. For
example, most projections of the effects of MHWs on coral reefs employ
accumulated heat stress, a metric measuring the combination of duration
and magnitude of heat stress as the indicator of coral bleaching
conditions (Skirving et al., 2020), while the rate of heat stress
development, which can influence mortality of coral reef fish (Genin et
al., 2020), has not been assessed. In addition, there has been limited
analysis of the duration of pre-MHW “priming” – a period of
sub-lethal heat stress in advance of warm-season MHW development which
can influence the response of corals and other marine organisms to
severe heat stress (Ainsworth et al., 2016; Hilker et al., 2016).
Evaluating these fine-scale MHW properties could help better understand
and project how MHWs affect marine ecosystems.
Projections of MHW properties and their effects on marine ecosystems
depend on the ability of models to represent the atmospheric and oceanic
processes that influence MHW development and dissolution. While previous
studies evaluated MHW projections with outputs from ensembles of General
Circulation Models (GCMs) and Earth System Models (Frölicher et al.,
2018; Oliver et al., 2019; Plecha et al., 2021), there has been less
analysis of model biases in simulating the baseline characteristics of
MHWs, and how such biases may affect future projections. Challenges in
simulating air-sea interactions, the periodicity and diversity of El
Niño / Southern Oscillation (ENSO) dynamics, and other key climate
phenomena due to limits of model resolution and other factors can lead
to regional biases in mean and seasonal SST (Brown et al., 2020; Guo et
al., 2022; Jiang et al., 2021; G. Li & Xie, 2012; Toniazzo &
Woolnough, 2014; Wang et al., 2014). These model biases could reduce
accuracy of the projected MHW thermal properties and their ecological
impacts (Hoeke et al., 2011; van Hooidonk & Huber, 2012). Though a
large ensemble of models might present a more accurate representation of
mean SSTs (Frölicher et al., 2016; Weigel et al., 2010), some of the
process-derived biases in individual models cannot cancel each other out
(Frölicher et al., 2018; Oliver et al., 2019). Evaluating the model
biases can indicate key processes to target in model development, and
identify biases to be considered when projecting local or regional
ecological impacts of warm-season MHWs.
In this study, we aim to improve our understanding of the future thermal
properties of warm-season MHWs by assessing their projected changes in
light of historical model biases, using three CMIP6 models. First, we
compare historical model simulations against observations to identify
the regional biases in warm-season MHW properties, including the
duration, peak intensity, accumulated heat stress, heating rate and
duration of the priming period. Second, we evaluate future projections
of warm-season MHW properties under three Shared Socio-Economic Pathways
(SSPs). Third, we examine the MHW projections for coral reef and kelp
systems worldwide considering the role of the regional model biases. We
then discuss the possible physical drivers of regional model biases and
disagreement between model projections.
2. Methods
2.1 Definition of warm-season MHW and the metrics characterizing its
thermal properties
A warm-season MHW is defined here as a period of positive anomalies of
daily SSTs or HotSpots (HS), relative to the thermal threshold known as
the Maximum Monthly Mean (MMM), that represents the climatological
warm-season SST and is commonly used for predicting coral bleaching. The
MMM in each grid cell is calculated as the maximum from a 1985-2014
monthly mean SST climatology. To test the effects of theoretical
acclimation or adaptation to warming by marine ecosystems, we repeat the
analysis using a rolling climatology (Logan et al., 2014), in which the
MMM is calculated based on the previous sixty year period.
We define a set of metrics for characterizing warm-season MHWs in terms
of magnitude, duration, accumulated heat stress and heating rate (Table
1), following Li & Donner (2022). The duration of heat stress is
described by Dc, the duration of continuous positive HS,
and Dtot, the total number of days with positive HS. The
“priming” period (Dp), a period of sub-lethal heat
stress that might train marine organisms’ thermal tolerance (Hilker et
al., 2016), is computed as the number of days from the first positive HS
in a year to the onset of Dc. The accumulated heat
stress over the continuous period (Dc) and for the
annual total (Dtot) are described by the metrics
Ac and Atot, respectively. As the total
number of positive HS days (Dtot) is longer than or at
minimum equal to the duration of continuous heat stress
(Dc), the accumulated heat stress over all HS days is
greater than or at minimum equal to that over the period of continuous
heat stress. Finally, the heating rate (HRc) is the rate
of warming from the start of the continuous heat stress period to the
date of peak HS.