[Coral-List] HadISST 1.1 data: Bleaching is not due to Warming

[Martin P ê cheux]

Dear Coral-listers,

I know that I will shock the common belief on this very important topic. But
the facts are the facts. I was myself surprised. So I decided to send this
draft to the list before submission to a publication, hoping for comments or
critics. Thanks a lot.
A long version is available at request.

HadISST 1.1 data : Coral Reef Mass Bleaching is not due to Global Warming.

Martin Pêcheux, June 2004
Nice University, 94260 Fresnes, France
Email : martin-pecheux at wanadoo.fr

ABSTRACT
Coral reef mass bleaching is the main threat on an Earth ecosystem. There is
a consensus among reef researchers that Global Warming is the cause. Here I
analyse HadISST 1.1 data, a new release of Sea Surface Temperature set, from
1950 to 2000, with 1° degree resolution, monthly averaged, at 50 reef sites
with known bleaching record.
First, the Global Warming is all but global. Whereas in Indian ocean the
warming of yearly maximum SST is significative, as well as in the Pacific,
the Caribbean has seen a general cooling from 1950 to about 1986, then a
warming. Secondly, I determine for the studied sites the relevant parameter
for bleaching ÆT, i.e. the minimum temperature at which bleaching has
occurred, minus the maximum temperature at which bleaching was absent during
the 1950-1978¹s. In fact, only 16% of the sites have positive ÆT : there was
generally higher temperature in 1950-1978 without bleaching than now with
bleaching eliciting. Bleaching threshold is also analysed.
I must conclude that, although temperature is an obvious parameter (as
irradiance), Global Warming is not the primary cause of bleaching. This lets
a direct bleaching action of CO2 rise and seawater acidification as the only
possible other global cause of coral reef mass bleaching.

INTRODUCTION
Coral reef mass bleaching begin in 1979 (reviews in Williams and
Bunkley-Williams, 1990, Smith and Buddemeier, 1992, Glynn, 1996, Brown,
1997, Pêcheux, 1997 - unpublished/online, Hoegh-Guldberg, 1999, Coles and
Brown, 2003). Reef researchers were at first destabilised about its cause.
But with repetitive reports of bleaching with "above average" temperatures
in summer, and the following of "hot spot" formation from satellite (Goreau
and Hayes, 1994) in particular in real time (NOAA indices, Strong et al.,
1997, Lui and Strong, 2003,
orbit-net.nesdis.noaa.gov/orad/coral_bleaching_index), an overwhelming
consensus emerges as to pinpoint Global Warming as the cause. But, by
definition, "above average" temperatures and "hot spots" are a normal
phenomenon, which have occurred also before mass bleaching began. My goal
was to examine such phenomenon, i. e. what is the difference ÆT between
present temperatures with bleaching and past maximum without bleaching.
Mainly due to lack of historical data, there have been only one serious
attempt to check this feature. Coffroth et al. (1989) examine the 1982 event
of the (well studied) Great Barrier Reef. They conclude unambiguously that
the temperature was elevated but not exceptional. They also discard any
conjunction with other physical factors, such as luminosity, water
transparency, winds, rainfall or tide. Elms (1992, discussed in Atwood et
al., 1992) analyses the GOSTA data from 1950 to 1989 over fifteen 5°x5°
zones in Caribbean and found cooling of the yearly mean in fourteen of them,
with a mean of ­1.00°C/century. Some other less relevant works are analysed
in Pêcheux (1997).
Lough (2000) studies the GISST data (1903-1994) and IGOSS data (1982-1999)
at 47 world-wide coral reef sites, in an approach rather similar to mine.
She focuses on global trends and on the unprecedented 1997-1998 event.
Nonetheless, she pinpoints that 1958 was more extreme for pooled Caribbean
sites. Last but not least, it is known that bleaching of large foraminifers
peaks at summer solstice and not during the maximum temperature a few months
later (Williams et al., 1997).

METHODS
Here I used the HadISST 1.1 data, newly released by the British Atmospheric
Data Centre, Met Office Hadley Centre, UK, an improved reconstructed Sea
Surface Temperature record (Rayner et al., 2003;
badc.nerc.ac.uk/home/index.html) and descendant of GOSTA-GISST data. I
restricted my analysis to the years 1950-2000. SST are averaged on a 1°
longitude x 1° latitude grid for each month.
Bleaching records at 50 world wide sites are taken from the literature or
from the Reef Base (www.reefbase.org) and the GCRMN reports
(www.aims.gov.au). The bleaching temperature was taken as the temperature of
the month at which the bleaching began if it was known, or (rarely) of the
month before if the temperature was superior, otherwise it was used the
maximum temperature of the year (Table 1).

RESULTS (Table 2)
Variance of yearly maximum SST.
At each site, yearly maxima appear as normal distributed (stochastic) (Fig.
2). There is a slight excess of "abnormal" standard score greater than 3
standard deviations (return time >741 years - Gauss integral). They
correspond to the El Niño 1983 (Panama, Galapagos) and the 1998 cycle
(Bahamas, Galapagos, Phuket, Seychelles, Maldives and Kenya), and also
Lizard Island, GBR, January 1970.
For the 50 sites, the standard deviation of the yearly maximum is
Poisson-distributed, being 0.384°C±0.137 (range 0.240 to 0.966). A
geographical differentiation exists: Caribbean is less variable,
0.298°C±0.030, than Pacific, 0.461°C±0.173, and Indian ocean, 0.350°C±0.040.
Warming of the yearly maximum SST.
A simple analysis of warming, not of the year data, but only of the maximum
each year, is informative. In Indian ocean (9 sites, without Persian Gulf),
the warming is regular, being +1.055°C/century, p<0.0001. In Pacific (20
sites), the warming is not so great, being +0.707°C/century, p=0.008. At
contrast, in Caribbean (18 sites), there is a cooling during a first period,
1950-1986 (determined by maximising slope) of -1.075°C/century, p<0.0001,
then a strong recent warming, 1987-2000, +1.803°C/century, p=0.008.
ÆT, minimum SST with bleaching minus maximum SST 1950-1978 without.
ÆT is generally negative. Only 9 sites have a positive value. The mean value
is -0.668°C±0.685 (-2.25 to 1.01), normally distributed. Values are all
negative for Caribbean (-0.821°C±0.466, -1.68 to ­0.02), negative for Indian
ocean (-0.331°C±0.493, -1.15 to 0.41). It is flat distributed for Pacific
(-0.672°C±0.883, -2.25 to 1.01), where positive value are encountered mainly
in the East Pacific with the El Niño 1983 (Coco Islands and Galapagos), and
in the West Pacific. The 1950-1978 maxima are rather evenly distributed,
apart 13 world wide occurrences of the year 1969, as already noted by Elms
(1992) in Caribbean.
Bleaching threshold from maximum mean.
I search for a bleaching threshold by subtracting the mean of yearly maximum
SST 1950-2000 from the minimum bleaching SST. The value is, without high
values of the El Niño 1983 at Galapagos and Coco Island, -0.0057°C±0.540
(-1.184 to 1.157). The threshold is well correlated with ÆT (r2=O.785,
p<0.0001). The discrepancy with NOAA indices (bleaching at 1°C above maximum
mean) does not come from different maximum means.

DISCUSSION
Bleaching record
.It is certainly incomplete. But a more precise record could only lower the
minimum temperatures at which bleaching begin. There is some inverse
correlation between ÆT and the number of event per site (r2=0.274,
p=0.0001), but whatever the sites are well monitored or not. The ÆT is based
on the assumption that no "unexplained" mass bleaching took place before
1979 (in Bonaire, Hot in Williams and Bunkley-Williams, 1990). This
certainly holds since 1950. There was enough coverage of reefs by
scientists, at least in enough places. There are several reports of mass
bleaching before 1979, always with obvious local causes (heated reef flat,
thermal pollution, fresh or turbid water, cold event, sedimentation (see
ref. in Pêcheux, 1997). And, at contrast to corals which stop calcification,
foraminifers show frequent spectacular shell abnormalities in association
with bleaching, which were unknown in previous time (Talge et al., 1997,
Pêcheux, 1997) apart at the Cretaceous/Tertiary boundary (Pêcheux, 1999).
Relevance of HadISST data for coral reef mass bleaching.
Its accuracy might not be perfect, but I used only the most recent data. In
any case, the magnitude of the mean ÆT, 0.68°C, is far above any inaccuracy.
The average over one month may not capture the exact value of SST, but the
time dynamics are slow and regular. The 1°x1° grid (~110 km square) may
obscure slightly the formation of relevant "hot spots", and be less
significative for shore reefs. But this surely does not hold for fore reefs
well bathed by oceanic waters. And I am confident that the physical
relationship between mesoscale open ocean temperatures and local shore ones
could not have changed so dramatically since 1979.
HadISST data can be well compared with the NOAA satellite bleaching-warning
indices in real time, which are quite successful. They have a slight higher
spatial resolution (50 km) but bleaching events show extended zone of warm
waters. Time resolution is twice a week, but the most precise index, the
Degree Heating Weeks, must reach 4 on its scale before bleaching,
corresponding to 1 degree above maximum mean during 4 weeks.
Bleaching threshold.
It is null because I used the minimum bleaching SST, at contrast to Goreau
and Hayes (1994) and NOAA indices, which established empirically a +1°C
threshold from all records. The value of the mean standard deviation,
0.384°C, means that a SST with "1°C above mean maximum", i.e. 2.6 time the
standard deviation, occurs each 217 years (range 63500 to 6.65 years), which
is obviously false. The standard deviation has not increase recently (mean
SD=0.362°C for 1990-2000). I guess that NOAA indices capture important
events but not all, in particular in Caribbean with low variance.
A null minimum bleaching threshold means that bleaching is in the yearly
maximum mean, said, it has a 50% probability, or can occurs every two years,
which is also obviously false. This does not mean that the data are false,
of course. Temperature is a factor but not the only one. There are also
other triggering factors, the firsts in mind being clear sky (full light and
UV) and dolldrum time (reduced cooling evaporation, more water transparency,
less circulation, and, more important in my opinion, impaired exchanges at
organism interface, i.e. the "unstirred layer effect"). Whatever the case,
the small variances mean a high sensitivity to a few tenth of °C.
A very crude calculation of the true bleaching threshold can be done using
the frequency of the bleaching events and the variance of the maximum
temperature. The mean frequency of bleaching is 14.2% (mean of event number
2.98/21 years) (return time 7.05 years), corresponding to 1.076 standard
deviation of the maximum mean (0.384°C), thus giving a bleaching threshold
of 0.413°C.

CONCLUSION
The cooling of the maximum SST over the Caribbean till 1986 and, more
definitive, the negative values of most ÆT, almost twice the standard
deviation, conduce to the disturbing but inescapable conclusion that the
increasing temperature is not the primary cause of coral reef mass
bleaching. It appears when maximum summer stress, of course, but my result
call for other synergetic factor(s). And this is not contradictory with the
fact that mass bleaching may occur also more and more frequently with future
Global Warming.
Ultra-violets, at first often considered, are not in cause, as they have not
increase in tropics, and there is no relationship between ozone drawdowns
and bleaching events (Pêcheux, 1996a). Moisture deficit increases over warm
pools in tropics (Flohn and Kapala, 1989, Graham, 1995), but it is of low
probability that it induces in the 80¹s an hydrological pattern change great
enough to be responsible of local new maxima 0.7°C above previous one. It
would also rather increase nebulosity over warm waters ("cirrus anvil
thermostat"). And of course there were summer dolldrum times with clear sky
before 1979.
This lets a direct effect of CO2 rise as the last possible global cause.
There are many arguments in favor of this hypothesis (Pêcheux, 1997). CO2
is, like many other parameters, a bleaching factor (Pêcheux, 1994, 1996b).
CO2 rise from 280ppm to 360ppm CO2 is already equivalent to a warming of
0.4°C, more probably 1.2°C (Pêcheux, 2002), which fits well with ÆT. Often
forgotten, the fact that bleaching affects not only corals but also all
other cnidarians, molluscs, sponges, ascidians, and large foraminifers in
symbioses with either dinoflagelates, diatoms or cyanobacteria implies a
fundamental limitation of their photosynthesis, the two main of which being
photoinhibition and photorespiration, both under strong influence of
symbiose-limited CO2.
Given future expected CO2 level, and with the addition of Global Warming,
coral reefs must be considered in great danger. The belief in adaptation to
CO2 and temperature rises 100 faster than at the interglacial termination is
rather a matter of opinion, with very few facts (discussion in Coles and
Brown, 2003, Hughes et al., 2003). I think that coral reefs will disappear
in the next decades if CO2 rise is unabated. If so, fight for strong
mitigation of anthropogenic CO2 rise must be our top priority.