[Coral-List] Response to Goreau

Mumby, Peter P.J.Mumby at exeter.ac.uk
Fri Nov 9 09:15:06 EST 2007

Dear Dr Goreau

I read with interest your message about our paper and would like to
clarify a number of misconceptions.

1) We do NOT conclude that lack of grazing is the cause of coral reef
decline nor do we conclude that parrotfish are the solution to the
problem. The model shows that reefs have a greater capacity to recover
from disturbance if total grazing and coral cover are higher.
Sensitivity analyses repeatedly showed that reefs could shift to either
a coral- or algal-dominated state when levels of grazing represented
those found today in the functional absence of Diadema. Whether the reef
will shift to either a coral or algal state depends on the initial coral
cover, the disturbance regime, the underlying dynamics of the system,
and the level of grazing. However, as grazing levels are increased by
fisheries management, the likelihood of a reef becoming entrained
towards an algal-dominated state decreases - i.e. the system becomes
more resilient. Therefore, high levels of parrotfish grazing do not
guarantee that a reef recover, but they maximise the probability that it
will. There are a number of studies from the Caribbean that show that
corals can respond when grazing levels increase, so this really isn't

2) Model outcomes are NOT an inevitable restatement of assumptions. The
model was an individual-based (or agent based) approach that assigns
empirical rates of processes to corals, categories of algae, herbivores,
and so on. In other words, we use the existing literature to build the
reef from scratch - e.g. corals recruit at the rates we see in the
field, grow at observed rates, etc, etc. This is very different from
creating an explicit mathematical relationship such that algal cover is
merely a function of coral cover. It is also important to bear in mind
that each model parameter is based on an empirical field study from the

3) Your statements about the functioning of the model are incorrect.
Corals experience both partial and whole-colony mortality at chronic
background levels observed in nature (from studies by Rolf Bak and John
Bythell). The causes of this mortality are varied and include parrotfish
predation, occasional diseases, storm damage, etc. We then estimate the
'equilibrial' dynamics of such reefs between major disturbance events -
i.e. does the reef tend to increase in coral cover, remain the same, or
decline over time after a disturbance of given size? We then add a
disturbance regime which, by way of example was hurricanes occurring
every 20 years, and assess the net outcome of disturbance and recovery.
This highlights whether the reef has the capacity to absorb these
impacts or not, which is the essence of resilience.

Other assertions on the parameterisation are also incorrect. Algae do
not grow at constant rates and they are extensively scoured (removed)
after major storms. I'll deal with nutrients separately. 

3) Nutrients. It's very important to understand that the model was
parameterised for a particular type of reef - seaward forereefs. These
reefs have low suspended sediment concentration and typically have very
high levels of primary production because high wave exposure maximises
the delivery of nutrients to algal fronds. Therefore, the algal
colonisation and growth rates are relatively high and considerably
greater than those from leeward reefs. Changes in algal growth rate will
affect the shape of the hysteresis curve (Fig. 2 of the paper). When we
carried out sensitivity analyses to different algal growth rates (as a
proxy for nutrient flux), we found that the conclusions held despite
perturbing the growth rates of algae. In short, the model does have the
capacity to vary algal growth rates and sensitivity analyses reveal that
20% fluctuations do not alter our main conclusions. Indeed, an earlier
study, published in Ecological Modelling, found that even a doubling of
algal colonisation and growth rates had limited impact on reef
trajectories. That does not mean that nutrients are unimportant -
rather, the model provides a framework to integrate the effects of
changes in both algal growth rate and grazing. There are some
combinations of coral cover and grazing for which changes in algal
growth rate are really quite important (as illustrated by your
examples). Like most things in nature, the relative importance of
nutrient flux and grazing varies from place to place. 

We must try to incorporate the best science available to parameterise
ecological models. As new data emerge, the model (and others like it)
can be improved. For example, I have not yet seen a study that
rigorously quantifies the additive effect of rising nutrient
concentrations on the community composition of Caribbean forereefs (i.e.
factoring in other changes such as exposure, grazing, water velocity,
habitat differences, etc). If such studies are available then we can
progressively refine the model accordingly. 

Lastly, I'd just like to reiterate the main take-home messages of the

A) Previous field studies have provided clear evidence of phase changes
on coral reefs, such as those that occur when the levels of urchin or
fish grazing change. What our model reveals is that phase changes can
involve at least two alternate stable states. This means that once a
threshold of say coral cover is exceeded during a sudden disturbance,
the reef can either shift towards a coral- or algal-dominated state even
if grazing levels remain unchanged. As to whether the reef moves towards
a coral or algal state depends on the initial coral cover, underlying
reef dynamics (including nutrient levels), level of grazing, and
intensity of disturbance. Alternate states occur because of feedback
processes and this means that it becomes progressively more difficult to
restore a reef as its health declines because the negative feedback
loops maintain the algal-dominated state. This process is called
hysteresis and implies that management is more likely to be successful
when applied to reefs in the early stages of degradation rather than
waiting until the reef is fully degraded. There is, therefore, a need
for urgent action. 

B) Plots of hysteresis behaviour in reefs are useful if the axes
represent major issues of importance (e.g. coral cover, which is
battered by bleaching, hurricanes, etc) and processes under management
control (e.g. grazing or nutrient flux). The shape of the hysteresis
plot is influenced by the underlying reef dynamics and therefore reefs
in different environments will have different-shaped curves. Placing all
these factors on a single graph is useful because it puts everything
into context - the type of disturbance regime and the ability of
different management interventions to alter resilience. To illustrate
this concept, we applied an example of a disturbance regime to a seaward
forereef and estimated the probability that reefs would retain an
ability to exhibit net recovery. In other words, what is the probability
that a reef of coral cover X and grazing level Y in the year 2007, would
still be able to exhibit recovery in the year 2037 under the specified
disturbance regime?
I hope that clarifies some of the issues but would be happy to continue
the discussion privately.



Professor Peter J Mumby
Marine Spatial Ecology Lab

School of BioSciences
Hatherly Laboratory
Prince of Wales Road
University of Exeter
tel: + 44 (0)1392 263798
fax: + 44 (0)1392 263700
e-mail: p.j.mumby at exeter.ac.uk

Research: http://www.ex.ac.uk/msel
Free video clips of coral reefs: http://www.reefvid.org
Coral Reef Group at Exeter: http://www.ex.ac.uk/celp
-----Original Message-----
From: coral-list-bounces at coral.aoml.noaa.gov
[mailto:coral-list-bounces at coral.aoml.noaa.gov] On Behalf Of Thomas
Sent: 04 November 2007 02:15
To: coral-list at coral.aoml.noaa.gov
Subject: [Coral-List] Parrotfish, nutrients, and control of algae

A recent paper published in Nature uses a mathematical model of coral
cover, macroalgae cover, turf algae cover, and grazing by parrotfish and
concludes that only parrotfish grazing can prevent algae from
overgrowing and killing corals. It blames fishermen for catching
parrotfish and causing algae growth, and makes the policy recommendation
that fishermen should be stopped in order to let the corals recover.
These conclusions have been widely covered in the press.

However close examination of the model reveals that these conclusions
are no more than a restatement of the original assumptions built into
the model.  As a someone with experience doing mathematical modeling in
astronomy, spatial population distributions, biogeochemistry,
atmospheric chemistry, and paleoclimatology, I am acutely aware that no
model is better than its assumptions, and if these don't adequately
describe reality, the results are simply intellectual artifacts rather
than providing insight into how nature works. If we misunderstand the
key controlling factors, the management prescriptions we make cannot
possibly work.

The model published in Nature allows corals to die only by being
overgrown by algae, and by "natural" coral mortality, which is equated
with hurricane destruction, that is to say, it does not include
mortality from heat shock or new diseases, the major causes of coral
mortality in most places in the last few decades. The model specifies
that algae grow at a constant rate, and can only die by being grazed.
There is no allowance for algae fragmentation by waves (anyone diving
after a storm knows the bottom can be covered with algae ripped loose),
nor is there any allowance for intrinsic factors that may vary the rate
of algae growth. Now it is long known that benthic algae growth can vary
by orders of magnitude depending on nutrient concentrations, but
nutrients nowhere figure in the model as a factor affecting algae
growth. Hence the model's conclusion that only grazing can limit algae
growth, as was assumed in the first place. This tautology somehow
escaped the peer reviewers.

The model predicts that the more parrotfish the less algae, but anyone
who has actually watched the long term changes in algae and parrotfish
knows that as algae populations increase, so do the numbers of
parrotfish. The model uses the misnamed "phase shift"  
interpretation of the long term changes in algae, corals, and fish in
Jamaica that attributes algae abundance to Diadema die off and
overfishing, and which blames the fishermen for eating all the
herbivorous fish. But in fact, long term observations of changes in
reefs all around Jamaica show that algae overgrowth took place at
different times in different places over a 40 year period, and every
place they followed local population growth and sewage inputs to coastal
waters, but did not follow overfishing or Diadema mortality except
coincidentally at a few places, such as Discovery Bay that went
eutrophic at the same time (T. J. Goreau, 1992, Bleaching and reef
community change in Jamaica: 1951-1991, SYMPOSIUM ON LONG TERM DYNAMICS
OF CORAL REEFS, AMERICAN ZOOLOGIST, 32: 683-695), and that algae growth
was strongly linked to excessive nutrients from land based sources (T.
J. Goreau & K. Thacker, 1994, Coral Reefs, sewage, and water quality
CARIBBEAN: 21ST CENTURY, Kingston, Jamaica, 3:98-116 and many papers by
Brian Lapointe). In fact in this period Jamaican fish populations
changed from being dominated by fish and invertebrate eating species to
near complete dominance by herbivores, the exact opposite of what the
hypothesis of top-down control of algae by herbivores, like this recent
model, predicts, but fully consistent with the bottom-up hypothesis that
algae productivity, and herbivore populations, are controlled by
nutrient inputs.

The practical management question is: how can weedy algae be controlled
before they smother coral reefs? To my knowledge there are only two
published cases of weedy algae being removed from coral reefs on a large
scale, one of them a short term success but a long term failure, while
the other has been sustained.

In Kaneohe Bay, Oahu, Hawaii, sewage was pumped onto the reef, algae
spread out from the sewage outfall and overwhelmed the reef, and since
there was no doubt that the nutrients had caused the algae, a long
sewage outfall was built to place the problem much further away.  
As the nutrients fell the algae died back, and the coral gradually
recovered. After the point source of nutrients was removed, the
suburbanization of the watershed caused uncontrollable increases in
non-point sources of nutrients from lawn fertilizers, golf courses, road
runoff, and other nutrients that were not flushed down the sewer. These
have caused the system to again go eutrophic, and the algae have again
smothered the reef. There is a large Hawaiian literature on this that is
readily available. It shows that controlling nutrients gets rid of
algae, but only if it is sustained.

A more successful long term case is a bay in Jamaica that I got cleaned
up 10 years ago by diverting all the land based sources of nutrients and
recycling them on land. Within weeks the red and green weedy algae
smothering the reef began to die back, and two months later they were
gone (T. Goreau, 2003, Waste Nutrients: Impacts on coastal coral reefs
and fisheries, and abatement via land recycling, 28p., UNITED NATIONS
Havana, Cuba). I have just revisited this site 10 years after, and the
weedy algae are still gone, with the algae now dominated by the
oligotrophic calcareous algae.

In my experience the only way to get rid of weedy algae is to starve
them of nutrients, and then they very quickly die. But all excessive
nutrients must be controlled, and they must remain controlled. This
requires adherence to the coral reef specific nutrient standards
proposed by Brian Lapointe, Mark and Diane Littler, and Peter Bell.  
We can blame the victims by stopping fishermen from eating, but this
will not work because it is based on a seriously flawed understanding of
what controls algae growth.

The Turks and Caicos Islands are the first place in the world to propose
coral reef specific water quality standards, and the only place in the
world to require that all developers build sewage treatment plants and
recycle all of their waste water on their own property. We will not see
the algae die back in eutrophic reefs until other countries follow their
example and all the sources of anthropogenic nutrients are identified
and controlled.

Thomas J. Goreau, PhD
Global Coral Reef Alliance
37 Pleasant Street, Cambridge MA 02139
goreau at bestweb.net

Coral-List mailing list
Coral-List at coral.aoml.noaa.gov

More information about the Coral-List mailing list