Download EFFECTS OF CHANGING TEMPERATURES ON CORAL REEF

EFFECTS OF CHANGING TEMPERATURES ON CORAL REEF HEALTH:
IMPLICATIONS FOR MANAGEMENT
Elizabeth Rose Selig
A dissertation submitted to the faculty of the University of North Carolina at Chapel Hill in
partial fulfillment of the requirements for the degree of Doctor of Philosophy in the
Curriculum in Ecology
Chapel Hill
2008
Approved by:
John F. Bruno
Judith C. Lang
Charles E. Mitchell
Aaron Moody
Peter S. White
ABSTRACT
Elizabeth Rose Selig
Effects of changing temperatures on coral reef health: implications for management
(Under the direction of John Bruno)
Human-induced climate change has already led to substantial changes in a variety of
ecosystems. Coral reefs are particularly vulnerable to rises in ocean temperature as a result
of climate change because they already live near their thermal limits. However, we know
little about the spatial patterns of temperature anomalies, areas of greater than usual
temperature, which cause coral mortality and increased rates of coral disease. These gaps in
knowledge make it difficult to design effective management strategies for mitigating the
effects of ocean warming. My dissertation research uses a combination of a new satellite
ocean temperature dataset, field surveys on coral health, and data on marine protected area
(MPA) boundaries to analyze how ocean temperatures are affecting coral reef health at
regional and global scales. I discovered that temperature anomalies are spatially and
temporally variable from 1985-2005 even during El Niño events. They are also typically less
than 50 km2, smaller than the resolution of many climate models. In addition, I found a
strong relationship on the Great Barrier Reef between the number of temperature anomalies
and the number of cases of white syndrome, a prevalent coral disease. Results from this
study suggest that temperature anomalies are playing a major role in the observed decline of
coral reefs over the last 30-40 years. This decline highlights the importance of determining
whether MPAs, one of the most common management tools are effective in restoring coral
ii
cover. My analyses demonstrated that MPAs can confer some ecosystem resilience through
fisheries management and land management practices at regional scales. Coral cover on
reefs inside of MPAs did not change over time, while unprotected reefs experienced declines
in coral cover. However, MPAs do not moderate the effect of thermal stress on corals or
reduce coral decline at rates that can offset losses from thermal stress and other major natural
and human-caused disturbances. MPAs are clearly a key tool in the management of fisheries
and coral reef health. My dissertation research underscores the need for both MPAs and
additional measures aimed at reducing the anthropogenic activities driving climate change.
iii
DEDICATION
To my grandfather, my dear Papa,
who was with me every step believing in me
iv
ACKNOWLEDGEMENTS
To my advisor, John Bruno, whose enthusiasm, optimism, and passion for science has
inspired me to do my best and grow as a scientist.
To my committee, I have benefited greatly from your support and constructive criticism over
the years. To Judy Lang for traveling so far and always being willing to engage in a good
debate, Peter White for helping me see life through the music of Bob Dylan and the poetry of
Robert Frost, Charles Mitchell for always having an open door and mind, and Aaron Moody
for helping me see a different view of the world.
I am particularly grateful to the EPA STAR Fellowship program, whose support enabled me
to focus on my research for three years. I am also extremely thankful for the generosity of
Dr. M. Ross and Mrs. Charlotte Johnson for endowing my Royster Society of Fellows
Dissertation Completion Fellowship. Both of these fellowships gave me opportunities to
experience things I otherwise would have missed.
To the countless volunteers and scientists who surveyed reefs over the last 40 years, I owe a
great debt of gratitude to for all of your hard work, which has made my analyses possible.
v
I owe a huge gratitude of debt to my labmates who were there for the duration – Mary
O'Connor and Sarah Lee. They provided a rich intellectual environment that helped me
become a better scientist. They were great friends and always made me see the humor in
sometimes difficult times. I was continually awed by their strength, determination,
creativity, and incredible science. And to the Bruno Lab of the future: Pamela Reynolds and
Andrea Anton, thank you for your help and support.
To the incredible technicians and friends who have supported me and the Bruno Lab over the
years: Johanna Kertesz (the best running partner at 6 am), Meredith Kintzing, Laura Ladwig,
and Jessica Wall. A special thanks to Virginia Schutte for all of her help and her fantastic
sense of humor.
To Ken Casey at NOAA, who made collaboration a joy. I am grateful for your friendship
and all that you have taught me.
To the Department of Marine Sciences for giving me a home these last several years and
especially to Nadera Malika-Salaam and Mary Campbell for always being willing to provide
a helping hand, many thanks for your support.
To Robert Peet, Denise Kenney, and Jack Weiss, who were great resources for me and help
to make the Curriculum in Ecology a wonderful environment for graduate students.
vi
To my dear friends in Marine Sciences and the Curriculum in Ecology for their support and
friendship: Alfredo Lopez de Aretxabaleta, Catherine Edwards, Patrick Gibson, Karen
Lloyd, David Luther, Matthew McKown, Meghan McKnight, Howard Mendlovitz, Andrew
Steen, and Christy Violin.
To my family outside of Chapel Hill, particularly my mother Meg Selig, my "aunts" Ellen
and Beth Rashbaum, and my aunt Kate Kimelman, who have supplied me with care packages
and love through the duration of this process. I lost two very important family members
during these years, my grandfather Philip Rashbaum and my great aunt Dorothy Helfeld.
They helped me become who I am and their memories are a blessing.
I feel especially thankful to my family in Chapel Hill, my rocks during tough times, my great
aunt Eleanor Sorgman, my amazing and inspiring sister Heather Ragan, my wonderful niece
Ariel Kwayke and the best bike mechanic of all time, Jason Merrill. To Trond Kristiansen,
whose encouragement and support helped me through to the finish.
To Shirin Ardakani, the greatest of friends, my second sister, your friendship means the
world to me.
To my family of friends scattered near and far: Eliza Beardslee, Linda Cotton, Rachel
Galper, Amanda Marcus, Susan Minnemeyer, Hannah Purdy, and Kay Sterner. They have
been my other family and provided places of refuge, unflagging love, and support.
vii
TABLE OF CONTENTS
LIST OF TABLES.................................................................................................................... x
LIST OF FIGURES ................................................................................................................. xi
INTRODUCTION .................................................................................................................... 1
CHAPTER 1: Mapping the variability in frequency, magnitude and
spatial extent of coral reef temperature anomalies between 1985 and 2005 .......................... 16
Abstract ....................................................................................................................... 16
Introduction................................................................................................................. 17
Materials and Methods................................................................................................ 20
Results......................................................................................................................... 25
Discussion ................................................................................................................... 28
References................................................................................................................... 46
CHAPTER 2: Analyzing the relationship between ocean temperature anomalies
coral disease outbreaks at broad spatial scales ....................................................................... 52
Abstract ....................................................................................................................... 52
Introduction................................................................................................................. 53
Using remote sensing data to explore the climate warming disease outbreak
hypothesis ................................................................................................................... 56
Case study on the Great Barrier Reef ......................................................................... 59
Future research directions ........................................................................................... 69
Conclusions................................................................................................................. 70
viii
References................................................................................................................... 82
CHAPTER 3: Evidence of increased resilience to corals from
marine protected areas ............................................................................................................ 89
Abstract ....................................................................................................................... 89
Introduction................................................................................................................. 90
Methods....................................................................................................................... 93
Results....................................................................................................................... 100
Discussion ................................................................................................................. 103
References................................................................................................................. 118
CHAPTER 4: Temperature-driven coral decline: do marine
protected areas have a role? .................................................................................................. 124
Abstract ..................................................................................................................... 124
Introduction............................................................................................................... 125
Methods..................................................................................................................... 129
Results....................................................................................................................... 136
Discussion ................................................................................................................. 139
References................................................................................................................. 151
APPENDIX 1: Supplementary material for Chapter 1 ........................................................ 158
APPENDIX 2: Supplementary material for Chapter 3 ........................................................ 165
2.1 Levels of management ....................................................................................... 165
2.2 Logistic-normal distribution .............................................................................. 165
2.3 Distance pairing ................................................................................................. 166
2.4 Bayesian fitting .................................................................................................. 167
References................................................................................................................. 181
ix
LIST OF TABLES
Table 1.1 Metrics of thermal stress........................................................................................ 44
Table 2.1 Relationship between weekly averaged satellite and in situ
temperatures at nine reefs on the GBR reefs at 5-9 m depth .................................................. 79
Table 2.2. Akaike Information Criterion for thermal stress metrics...................................... 80
2
Table 3.1 R for MPA versus non-MPA model.................................................................. 115
2
Table 3.2 R for MPA-only models.................................................................................... 116
Table S1.1 Validations of the CoRTAD data using in situ data loggers
at different depths and locations. .......................................................................................... 163
Table S2.1 Percent of reef area by IUCN category ............................................................. 177
Table S2.2 List of models analyzed..................................................................................... 178
Table S2.3 Distribution of IUCN categories across oceans by surveys. ............................. 179
Table S2.4 Distribution of IUCN categories across oceans by reefs................................... 180
x
LIST OF FIGURES
Figure 1.1 Comparison of the 4 km Pathfinder with other satellite datasets ......................... 33
Figure 1.2 Region delineations for the CoRTAD analysis .................................................... 34
Figure 1.3 Cumulative number of TSAs (1985-2005)........................................................... 35
Figure 1.4 Cumulative number of WSSTAs (1985-2005)..................................................... 36
Figure 1.5 Mean frequency, duration, and magnitude of TSAs and WSSTAs...................... 37
Figure 1.6 Year of the maximum number of TSAs in the Caribbean.................................... 39
Figure 1.7 Year of maximum number of TSAs in the Pacific ............................................... 40
Figure 1.8 Severity of the 1998 El Niño event ...................................................................... 41
Figure 1.9 Severity of the 2005 warm event.......................................................................... 42
Figure 1.10 Frequency distribution of anomaly areas for TSAs
and WSSTAs for 1985-2005................................................................................................... 43
Figure 2.1 Developing the 4 km temperature dataset ............................................................ 72
Figure 2.2 Mean sea surface temperature (ºC) from 1985-2003............................................ 73
Figure 2.3 Number of white syndrome cases in 2002-2003 .................................................. 74
Figure 2.4 Example of white syndrome spreading across Acropora cytherea. ..................... 75
Figure 2.5 Relationships between the number of white syndrome cases and
total percent coral cover and the number of WSSTAs. .......................................................... 76
Figure 2.6 The effect of total coral cover and the number of WSSTAs
on the number of white syndrome cases ................................................................................. 77
Figure 2.7 Relationship between mean thermal stress (WSSTAs) and
the mean number of white syndrome cases at the sector scale ............................................... 78
xi
Figure 3.1 Location of unprotected and protected coral cover surveys............................... 109
Figure 3.2 Coefficient estimates for the MPA versus non-MPA model.............................. 110
Figure 3.3 The change in percent coral cover from 2004 to 2005
inside and outside of MPAs .................................................................................................. 111
Figure 3.4 The effect of the number of years of protection on the
1-year change in coral cover ................................................................................................. 112
Figure 3.5 Comparisons of the average coral cover per year as
predicted by the models ........................................................................................................ 113
Figure 4.1 Region delineations for the analysis................................................................... 144
Figure 4.2 Coefficient estimates for the multilevel model .................................................. 145
Figure 4.3 The effect of TSA on logit coral cover for each spatial grouping
unit split by MPA and non-MPA .......................................................................................... 146
Figure 4.4 Multilevel model predictions of the effect of thermal stress
anomalies (TSA) on the change in coral cover..................................................................... 147
Figure 4.5 Comparison of anomaly frequency within MPAs to
anomaly frequency of all reefs.............................................................................................. 148
Figure S1.1 Mean duration (weeks) of TSAs, 1985-2005 ................................................... 158
Figure S1.2 Mean duration (weeks) of WSSTAs, 1985-2005 ............................................. 160
Figure S1.3 Mean duration (weeks) of TSA in 1998........................................................... 161
Figure S1.4 Mean duration (weeks) of WSSTAs in 1998 ................................................... 162
Figure S2.1 The relationship between the MPA effect on slope and
the distance of non-MPAs surveys from MPAs. .................................................................. 169
Figure S2.2 AIC values for all models examined ................................................................ 170
Figure S2.3 Coral cover slope and intercept values from the MPA
versus non-MPA model ........................................................................................................ 171
Figure S2.4 Caterpillar plots for the slope random effects and intercept
random effects for the level-3 residuals................................................................................ 172
xii
Figure S2.5 The number of reefs categorized by the year when their
respective MPAs were established........................................................................................ 173
Figure S2.6 Times series of the number of years of protection by reef............................... 174
Figure S2.7 Distribution of years protected by ocean.......................................................... 175
Figure S2.8 Generalized additive models ............................................................................ 176
xiii
INTRODUCTION
Studies in a wide range of ecosystems have found that climate change has already had
or is predicted to have far-reaching effects (Vitousek et al. 1997; Walther et al. 2002).
Documented biological responses to climate change include shifts in species ranges (Barry et
al. 1995; Parmesan & Yohe 2003; Precht & Aronson 2004), phenological patterns (Menzel &
Fabian 1999; Parmesan & Yohe 2003; Root et al. 2003), and community structure due to
varied climate effects at different trophic levels (Schmitz et al. 2003; Voigt & al. 2003). Our
ability to document or forecast these biological responses to climate change is impeded by
several gaps in knowledge. The spatial and temporal variability of key climatic parameters
like temperature, which drive ecosystem changes, remain poorly understood, particularly in
the oceans at ecologically relevant scales. In addition, physical data are rarely linked with
biological data to build an integrated framework for analyzing how climate change is
affecting ecosystems regionally (but see Beaugrand et al. 2003; Behrenfeld et al. 2006).
Understanding these relationships is particularly important for designing and evaluating
appropriate management interventions. My dissertation uses spatially explicit databases to
examine how tropical ocean temperature variability affects coral health, and whether
management strategies like marine protected areas can mitigate the effects of climate change.
Reef-building or scleractinian corals are widely recognized as particularly vulnerable
to climate warming because of their thermal sensitivity (Coles et al. 1976; Hughes et al.
2003; Sheppard 2003). Scleractinian corals are the foundation species of coral reefs (Bruno
& Bertness 2001), one of the most diverse ecosystems in the world (Reaka-Kudla 1997;
Roberts et al. 2002). Globally, these ecosystems have undergone rapid, extensive changes
during the last 20-30 years. On some reefs, these changes are unprecedented in the Holocene
paleontological record (Aronson & Precht 2001; Wapnick et al. 2004; but see Hubbard et al.
2005). Declines have been extensive, affecting a wide range of taxa in both protected and
unprotected areas (Rogers & Beets 2001; Edmunds 2002; Bellwood et al. 2004). One of the
most commonly used metrics for evaluating coral health is the percentage of live coral tissue
covering the hard reef substratum. In the Caribbean, average hard coral cover has decreased
from approximately 50% in the late 1970s (Gardner et al. 2003) to 17% in 2005 (Schutte et
al. in prep). Similar declines have occurred in the Indo-Pacific, where regional averages in
2003 were 22% (Bruno & Selig 2007). Even on well-managed reefs like the Great Barrier
Reef, average coral cover has decreased from 40% in the 1960s to 20% in 2000 (Bellwood et
al. 2004).
The global and indiscriminate nature of the decline in coral cover is at least partially
attributable to increases in ocean temperature as a result of climate change (Hoegh-Guldberg
1999). Climate change is expected to continue to have devastating effects on coral reef
ecosystems due to rising sea level (Blanchon & Shaw 1995), ocean circulation changes
(Harley et al. 2006), further increases in ocean temperature (Sheppard 2003; McWilliams et
al. 2005), and ocean acidification (Hoegh-Guldberg et al. 2007). In the short-term, increases
in ocean temperature have had some of the most visible and dramatic effects because they are
drivers of two major causes of coral mortality: coral disease and coral bleaching (Arceo et al.
2001; Aronson & Precht 2006; Graham et al. 2006).
2
Coral bleaching is perhaps the most widely known and documented effect of ocean
warming. Bleaching describes the separation of the algal symbionts, known as
zooxanthellae, from the coral host. Typically, bleaching occurs when temperatures exceed
the local average summertime maximum temperature by approximately 1°C (Glynn 1993).
Bleaching is a natural stress response (Glynn 1993) and corals can recover from bleaching
episodes (Glynn 1993; Brown 1997). However, when temperature magnitudes are too high
or prolonged, bleaching can cause coral mortality (Glynn 1993). Mass or large-scale
bleaching events due to severe El Niño events in 1982-1983 and 1997-1998 were coincident
with major regional coral mortality (Glynn et al. 1988; Wilkinson 2000; Bruno et al. 2001;
Glynn et al. 2001). Models predict that warming ocean temperatures as a result of climate
change will increase the number and severity of these mass bleaching events (HoeghGuldberg 1999; Sheppard 2003; McWilliams et al. 2005; Sheppard & Rioja-Nieto 2005).
The relationship between ocean temperature and coral disease is less well understood.
Coral diseases have caused significant mortality in the last 30 years, particularly in the
Caribbean, where white band disease decimated acroporid populations (Aronson & Precht
2001). Recent work suggests that rises in ocean temperatures may be partially responsible
for the emergence of new diseases and increases in disease reports (Sutherland et al. 2004;
Ward & Lafferty 2004; Bruno et al. 2007). Warmer temperatures have been associated with
faster pathogen development for several coral diseases (Alker et al. 2001; Richardson & Kuta
2003). In addition, field studies have documented summertime increases in the number of
cases for a variety of coral diseases (Kuta & Richardson 2002; Patterson et al. 2002; Jones et
al. 2004; Willis et al. 2004; Boyett et al. 2007).
3
Mitigating the effects of coral decline as a result of coral disease and bleaching
represents a major management challenge. A substantial success in marine ecosystem
management has been the use of marine protected areas (MPAs) to restore the abundance of
targeted populations in both temperate and tropical ecosystems (Babcock et al. 1999;
Murawski et al. 2000; Halpern 2003; Russ et al. 2004). Several studies in coral reef
ecosystems have documented increases in fish abundance inside MPAs and spillover effects
to neighboring fisheries (McClanahan & Mangi 2000; Roberts et al. 2001; Evans & Russ
2004). Successful fisheries management and the subsequent restoration of a more complete
trophic structure through protection in MPAs is now proposed as a possible mechanism for
indirectly increasing resilience to climate change (Hughes et al. 2003; Mumby et al. 2006;
Mumby et al. 2007a). Although MPAs cannot directly affect temperature, they are
hypothesized to increase resilience to thermal events by reducing other stressors like
overfishing or terrestrial run-off (Grimsditch & Salm 2005; Wolanski & De'ath 2005;
Hughes et al. 2007). After bleaching or disease events, bare substrate can be colonized by
macroalgae, which prevents coral recruitment (Diaz-Pulido & McCook 2002; Birrell et al.
2005). Increased herbivory on reefs can prevent a shift to macroalgal dominance by
maintaining space for coral recruitment and growth (Hughes et al. 2007; Mumby et al.
2007a; Mumby et al. 2007b). In spite of these potential benefits, the large-scale effects of
MPAs on corals are not well understood. A lack of awareness about their limitations could
undermine future efforts to establish MPAs, especially in the face of some failures. In
addition, the intense focus on MPAs could fail to catalyze political will around other
necessary policy efforts.
4
Although climatic changes affect ecosystems at local scales, we need to understand
variability in temperature at regional and global scales to make predictions about future
ecosystem changes and design effective management strategies. The cornerstone of my
dissertation was the development of a 4-km coral reef temperature anomaly database
(CoRTAD), which calculated anomalies globally from 1985-2005. I used these data to
explore the spatial and temporal variability in the temperature anomalies driving coral
decline through bleaching and disease. By coupling these temperature data with field survey
data on white syndrome, a coral disease causing mortality of key coral taxa on the Great
Barrier Reef, I was able to determine that the number of temperature anomalies was
positively related to disease frequency. The results of these analyses highlighted the need to
explore whether MPAs could be an effective tool for reducing coral decline. Initially, I
addressed this question with spatial data on MPA boundaries and surveys of live coral cover
from around the world. Then I integrated spatial data on MPA boundaries with the
temperature anomaly database to analyze whether management through MPAs could be an
effective strategy for reducing the effects of temperature stress. Together this work
described how warmer temperatures are affecting coral reef health at broad spatial scales and
what strategies may be necessary for reversing coral decline.
Chapter 1: Mapping the variability in frequency, magnitude and spatial extent of coral reef
temperature anomalies between 1985 and 2005
In spite of the evidence that temperature anomalies play a fundamental role in coral
reef health, their spatial and temporal patterns have not been analyzed. I used the CoRTAD
5
to quantify key anomaly characteristics including anomaly frequency, magnitude, duration,
and area across years and regions. I found substantial fine-scale regional and local variability
in these parameters, which suggests that temperature-related disease and bleaching is likely
to be highly patchy. I calculated that 48% of bleaching-related and 44% of disease-related
anomalies were less than 50 km2, smaller than the resolution of most models used to forecast
climate changes. Understanding the link between the scale of temperature data and
biological processes will facilitate the design of appropriate management strategies and
accurate predictions about the impacts of climate change on coral reef ecosystems.
Chapter 2: Analyzing the relationship between ocean temperature anomalies and coral
disease outbreaks at broad spatial scales
Rising terrestrial and ocean temperatures have been hypothesized to cause the
emergence of new diseases and increase the frequency and severity of known diseases
(Harvell et al. 1999; Harvell et al. 2002). However, the temperature thresholds relevant to
most coral disease outbreaks are not known. In Chapter 2, I used the temperature anomaly
database created in Chapter 1 and a long-term disease database from the Australian Institute
for Marine Science (AIMS) to determine which temperature thresholds best describe changes
in disease frequency. In addition, I tested whether the number of temperature anomalies was
related to an outbreak of a coral disease, white syndrome, on the Great Barrier Reef in 2002.
The weekly sea surface temperature anomaly (WSSTA) metric, which defined an anomaly as
being at least 1°C more than the typical 21-year average temperature of a specific week, best
explained the relationship between anomaly frequency and white syndrome cases. The
relationship between the number of these temperature anomalies and white syndrome cases
6
was highly significant. In addition, there was a significant, nearly exponential relationship
between total coral cover and the number of disease cases. Together, coral cover and
WSSTAs explained nearly 75% of the variance in disease cases and both high coral cover
and WSSTAs were necessary for triggering disease outbreaks (Selig et al. 2006). These
results suggest that disease may become a key driver of coral mortality as a result of
increased ocean temperatures associated with the climate change.
Chapter 3: Evidence of increased resilience to corals from marine protected areas
Marine protected areas (MPAs) are known to be effective for managing fisheries
(Alcala et al. 2005; McClanahan et al. 2007). MPAs are thought to be able to benefit corals
indirectly either through restoring historic food web structure or by mitigating nutrient and
sediment run-off if the MPA includes a terrestrial component (Russ et al. 2004; Wolanski et
al. 2004; Mumby et al. 2006). I tested the hypothesis that MPAs can have a positive effect
on coral cover by developing spatial databases of MPAs and more than 9000 surveys of live
coral cover from 1969-2006 from peer-reviewed literature, grey literature, and publicly
available datasets like Reef Check (www.reefcheck.org). I then analyzed whether the
relationship between protection and the rate of coral cover change varied with the age of the
MPA at the time of the survey. Coral cover in MPAs did not change over time whereas coral
cover on unprotected reefs declined. The benefit from MPAs generally increased with the
number of years of protection. In the Caribbean, this trend continued until a leveling-off at
approximately 30 years of protection. On the other hand, there were two distinct trend
dynamics in the Indo-Pacific. Initially the change in coral cover was positive, but then was
reset to a trend that was not significantly different from zero after 20-30 years of protection.
7
These two trends in coral cover reflect trajectories before and after the 1998 El Niño event,
which suggest that the effects of major thermal disturbances may not be mitigated by
protection. Although my work demonstrated a resilience effect, it also highlighted the
limitations of MPAs to restore coral cover to historic levels.
Chapter 4: Temperature-driven coral decline: do marine protected areas have a role?
Results from Chapter 3 indicated that the resilience effect of MPAs may not explicitly
moderate the effect of temperature stress on coral cover. I tested whether MPAs were
successfully reducing the impact of thermal stress on coral cover and quantified how
temperature anomalies were generally affecting coral cover using the temperature database
constructed in Chapter 1 and the MPA and live coral cover databases used in Chapter 3.
MPAs did not modify the effect of temperature anomalies on coral cover over time. In
addition, the magnitude of the benefit of MPAs generally was not enough to compensate for
the negative effect of temperature anomalies on coral cover. These results suggest that
MPAs will not be able to counteract the impacts of increased thermal stress as a result of
climate change.
Overall Synthesis
The results of my dissertation work suggest that temperature anomalies are playing a
major role in the decline of coral reefs and that protection in MPAs alone will not be
sufficient to mitigate coral decline. Coral reef conservation has been limited by a lack of
understanding of the stressors responsible for coral reef decline. My work suggests that
temperature-driven coral bleaching and disease are major drivers of coral loss. In addition, I
8
demonstrated that MPAs can confer some ecosystem resilience through fisheries
management and land management practices at regional scales. However, MPAs do not
reduce coral decline at rates that can offset losses from thermal stress and other major natural
and human-caused disturbances. MPAs are obviously a key tool in the management of
fisheries and coral reef health. Nonetheless, my work underscores the need for additional
and more direct conservation measures aimed at reducing the anthropogenic activities driving
climate change.
9
References
Alcala A.C., Russ G.R., Maypa A.P. & Calumpong H.P. (2005). A long-term, spatially
replicated experimental test of the effect of marine reserves on local fish yields.
Canadian Journal of Fisheries and Aquatic Sciences, 62, 98-108.
Alker A.P., Smith G.W. & Kim K. (2001). Characterization of Aspergillus sydowii (Thom et
Church), a fungal pathogen of Caribbean sea fan corals. Hydrobiologia, 460, 105111.
Arceo H.O., Quibilan M.C., Alino P.M., Lim G. & Licuanan W.Y. (2001). Coral bleaching
in Philippine reefs: coincident evidence with mesoscale thermal anomalies. Bulletin
of Marine Science, 69, 579-593.
Aronson R.B. & Precht W.F. (2001). White-band disease and the changing face of Caribbean
coral reefs. Hydrobiologia, 460, 25-38.
Aronson R.B. & Precht W.F. (2006). Conservation, precaution, and Caribbean reefs. Coral
Reefs, 25, 441-450.
Babcock R.C., Kelly S., Shears N.T., Walker J.W. & Willis T.J. (1999). Changes in
community structure in temperate marine reserves. Marine Ecology-Progress Series,
189, 125-134.
Barry J.P., Baxter C.H., Sagarin R.D. & Gilman S.E. (1995). Climate-Related, Long-Term
Faunal Changes in a California Rocky Intertidal Community. Science, 267, 672-675.
Beaugrand G., Brander K.M., Lindley J.A., Souissi S. & Reid P.C. (2003). Plankton effect on
cod recruitment in the North Sea. Nature, 426, 661-664.
Behrenfeld M.J., O'Malley R.T., Siegel D.A., McClain C.R., Sarmiento J.L., Feldman G.C.,
Milligan A.J., Falkowski P.G., Letelier R.M. & Boss E.S. (2006). Climate-driven
trends in contemporary ocean productivity. Nature, 444, 752-755.
Bellwood D.R., Hughes T.P., Folke C. & Nystrom M. (2004). Confronting the coral reef
crisis. Nature, 429, 827-833.
Birrell C.L., McCook L.J. & Willis B.L. (2005). Effects of algal turfs and sediment on coral
settlement. Marine Pollution Bulletin, 51, 408-414.
Blanchon P. & Shaw J. (1995). Reef drowning during the last deglaciation - evidence for
catastrophic sea-level rise and ice-sheet collapse. Geology, 23, 4-8.
Boyett H.V., Bourne D.G. & Willis B.L. (2007). Elevated temperature and light enhance
progression and spread of black band disease on staghorn corals of the Great Barrier
Reef. Marine Biology, 151, 1711-1720.
10
Brown B.E. (1997). Coral bleaching: causes and consequences. Coral Reefs, 16, S129-S138.
Bruno J.F. & Bertness M.D. (2001). Habitat modification and facilitation in benthic marine
communities. In: Marine Community Ecology (eds. Bertness MD, Hay ME & Gaines
SD). Sinauer Sunderland, MA, pp. 201-218.
Bruno J.F. & Selig E.R. (2007). Regional decline of coral cover in the Indo-Pacific: timing,
extent, and subregional comparisons. PLoS One, 2, e711.
Bruno J.F., Selig E.R., Casey K.S., Page C.A., Willis B.L., Harvell C.D., Sweatman H. &
Melendy A.M. (2007). Thermal stress and coral cover as drivers of coral disease
outbreaks. Plos Biology, 5, 1220-1227.
Bruno J.F., Siddon C.E., Witman J.D., Colin P.L. & Toscano M.A. (2001). El Nino related
coral bleaching in Palau, Western Caroline Islands. Coral Reefs, 20, 127-136.
Coles S.L., Jokiel P.L. & Lewis C.R. (1976). Thermal tolerance in tropical versus subtropical
Pacific reef corals. Pacific Science, 30, 159-166.
Diaz-Pulido G. & McCook L.J. (2002). The fate of bleached corals: patterns and dynamics of
algal recruitment. Marine Ecology-Progress Series, 232, 115-128.
Edmunds P.J. (2002). Long-term dynamics of coral reefs in St. John, US Virgin Islands.
Coral Reefs, 21, 357-367.
Evans R.D. & Russ G.R. (2004). Larger biomass of targeted reeffish in no-take marine
reserves on the Great Barrier Reef, Australia. Aquatic Conservation-Marine and
Freshwater Ecosystems, 14, 505-519.
Gardner T.A., Cote I.M., Gill J.A., Grant A. & Watkinson A.R. (2003). Long-term regionwide declines in Caribbean corals. Science, 301, 958-960.
Glynn P.W. (1993). Coral reef bleaching - ecological perspectives. Coral Reefs, 12, 1-17.
Glynn P.W., Cortes J., Guzman H.M. & Richmond R.H. (1988). El Niño (1982-1983)
associated coral mortality and relationship to sea surface temperature deviations in
the tropical eastern Pacific. Proceedings of the Sixth International Coral Reef
Symposium, 2, 693-698.
Glynn P.W., Mate J.L., Baker A.C. & Calderon M.O. (2001). Coral bleaching and mortality
in Panama and Ecuador during the 1997-1998 El Niño-Southern Oscillation event:
spatial/temporal patterns and comparisons with the 1982-1983 event. Bulletin of
Marine Science, 69, 79-109.
11
Graham N.A.J., Wilson S.K., Jennings S., Polunin N.V.C., Bijoux J.P. & Robinson J. (2006).
Dynamic fragility of oceanic coral reef ecosystems. Proceedings of the National
Academy of Sciences of the United States of America, 103, 8425-8429.
Grimsditch G.D. & Salm R.V. (2005). Coral reef resilience and resistance to bleaching. The
World Conservation Union, Gland.
Halpern B.S. (2003). The impact of marine reserves: do reserves work and does reserve size
matter? Ecological Applications, 13, S117-S137.
Harley C.D.G., Hughes A.R., Hultgren K.M., Miner B.G., Sorte C.J.B., Thornber C.S.,
Rodriguez L.F., Tomanek L. & Williams S.L. (2006). The impacts of climate change
in coastal marine systems. Ecology Letters, 9, 228-241.
Harvell C.D., Kim K., Burkholder J.M., Colwell R.R., Epstein P.R., Grimes D.J., Hofmann
E.E., Lipp E.K., Osterhaus A.D.M.E., Overstreet R.M., Porter J.W., Smith G.W. &
Vasta G.R. (1999). Emerging marine diseases--climate links and anthropogenic
factors. Science, 285, 1505-1510.
Harvell C.D., Mitchell C.E., Ward J.R., Altizer S., Dobson A.P., Ostfeld R.S. & Samuel
M.D. (2002). Climate warming and disease risks for terrestrial and marine biota.
Science, 296, 2158-2162.
Hoegh-Guldberg O. (1999). Climate change, coral bleaching and the future of the world's
coral reefs. Marine & Freshwater Research, 50, 839-866.
Hoegh-Guldberg O., Mumby P.J., Hooten A.J., Steneck R.S., Greenfield P., Gomez E.,
Harvell C.D., Sale P.F., Edwards A.J., Caldeira K., Knowlton N., Eakin C.M.,
Iglesias-Prieto R., Muthiga N., Bradbury R.H., Dubi A. & Hatziolos M.E. (2007).
Coral reefs under rapid climate change and ocean acidification. Science, 318, 17371742.
Hubbard D.K., Zankl H., Van Heerden I. & Gill I.P. (2005). Holocene reef development
along the northeastern St. Croix shelf, Buck Island, US Virgin Islands. Journal of
Sedimentary Research, 75, 97-113.
Hughes T.P., Baird A.H., Bellwood D.R., Card M., Connolly S.R., Folke C., Grosberg R.,
Hoegh-Guldberg O., Jackson J.B.C., Kleypas J., Lough J.M., Marshall P., Nystrom
M., Palumbi S.R., Pandolfi J.M., Rosen B. & Roughgarden J. (2003). Climate change,
human impacts, and the resilience of coral reefs. Science, 301, 929-933.
Hughes T.P., Rodrigues M.J., Bellwood D.R., Ceccarelli D., Hoegh-Guldberg O., McCook
L., Moltschaniwskyj N., Pratchett M.S., Steneck R.S. & Willis B. (2007). Phase
shifts, herbivory, and the resilience of coral reefs to climate change. Current Biology,
17, 360-365.
12
Jones R.J., Bowyer J., Hoegh-Guldberg O. & Blackall L.L. (2004). Dynamics of a
temperature-related coral disease outbreak. Marine Ecology Progress Series, 281, 6377.
Kuta K.G. & Richardson L.L. (2002). Ecological aspects of black band disease of corals:
relationships between disease incidence and environmental factors. Coral Reefs, 21,
393-398.
McClanahan T.R., Graham N.A.J., Calnan J.M. & MacNeil M.A. (2007). Toward pristine
biomass: reef fish recovery in coral reef marine protected areas in Kenya. Ecological
Applications, 17, 1055-1067.
McClanahan T.R. & Mangi S. (2000). Spillover of exploitable fishes from a marine park and
its effect on the adjacent fishery. Ecological Applications, 10, 1792-1805.
McWilliams J.P., Cote I.M., Gill J.A., Sutherland W.J. & Watkinson A.R. (2005).
Accelerating impacts of temperature-induced coral bleaching in the Caribbean.
Ecology, 86, 2055-2060.
Menzel A. & Fabian P. (1999). Growing season extended in Europe. Nature, 397, 659-659.
Mumby P.J., Dahlgren C.P., Harborne A.R., Kappel C.V., Micheli F., Brumbaugh D.R.,
Holmes K.E., Mendes J.M., Broad K., Sanchirico J.N., Buch K., Box S., Stoffle R.W.
& Gill A.B. (2006). Fishing, trophic cascades, and the process of grazing on coral
reefs. Science, 311, 98-101.
Mumby P.J., Harborne A.R., Williams J., Kappel C.V., Brumbaugh D.R., Micheli F., Holmes
K.E., Dahlgren C.P., Paris C.B. & Blackwell P.G. (2007a). Trophic cascade facilitates
coral recruitment in a marine reserve. Proceedings of the National Academy of
Sciences of the United States of America, 104, 8362-8367.
Mumby P.J., Hastings A. & Edwards H.J. (2007b). Thresholds and the resilience of
Caribbean coral reefs. Nature, 450, 98-101.
Murawski S.A., Brown R., Lai H.L., Rago P.J. & Hendrickson L. (2000). Large-scale closed
areas as a fishery-management tool in temperate marine systems: the Georges Bank
experience. Bulletin of Marine Science, 66, 775-798.
Parmesan C. & Yohe G. (2003). A globally coherent fingerprint of climate change impacts
across natural systems. Nature, 421, 37-42.
Patterson K.L., Porter J.W., Ritchie K.E., Polson S.W., Mueller E., Peters E.C., Santavy D.L.
& Smiths G.W. (2002). The etiology of white pox, a lethal disease of the Caribbean
elkhorn coral, Acropora palmata. Proceedings of the National Academy of Sciences
of the United States of America, 99, 8725-8730.
13
Precht W.F. & Aronson R.B. (2004). Climate flickers and range shifts of reef corals.
Frontiers in Ecology and the Environment, 2, 307-314.
Reaka-Kudla M.L. (1997). The global biodiversity of coral reefs: a comparison with
rainforests. In: Biodiversity II: Understanding and protecting our natural resources
(eds. Reaka-Kudla ML, Wilson DE & Wilson EO). Joseph Henry/National Academy
Press Washington, DC, pp. 83-108.
Richardson L.L. & Kuta K.G. (2003). Ecological physiology of the black band disease
cyanobacterium Phormidium corallyticum. Fems Microbiology Ecology, 43, 287-298.
Roberts C.M., Bohnsack J.A., Gell F., Hawkins J.P. & Goodridge R. (2001). Effects of
marine reserves on adjacent fisheries. Science, 294, 1920-1923.
Roberts C.M., McClean C.J., Veron J.E.N., Hawkins J.P., Allen G.R., McAllister D.E.,
Mittermeier C.G., Schueler F.W., Spalding M., Wells F., Vynne C. & Werner T.B.
(2002). Marine biodiversity hotspots and conservation priorities for tropical reefs.
Science, 295, 1280-1284.
Rogers C.S. & Beets J. (2001). Degradation of marine ecosystems and decline of fishery
resources in marine protected areas in the US Virgin Islands. Environmental
Conservation, 28, 312-322.
Root T.L., Price J.T., Hall K.R., Schneider S.H., Rosenzweig C. & Pounds J.A. (2003).
Fingerprints of global warming on wild animals and plants. Nature, 421, 57-60.
Russ G.R., Alcala A.C., Maypa A.P., Calumpong H.P. & White A.T. (2004). Marine reserve
benefits local fisheries. Ecological Applications, 14, 597-606.
Schmitz O.J., Post E., Burns C.E. & Johnston K.M. (2003). Ecosystem responses to global
climate change: moving beyond color mapping. Bioscience, 53, 1199-1205.
Schutte V.G.W., Selig E.R. & Bruno J.F. (in prep). Resilience on Caribbean coral reefs:
spatiotemporal trends in coral and macroalgal cover.
Selig E.R., Harvell C.D., Bruno J.F., Willis B.L., Page C.A., Casey K.S. & Sweatman H.
(2006). Analyzing the relationship between ocean temperature anomalies and coral
disease outbreaks at broad spatial scales. In: Coral reefs and climate change: science
and management (eds. Phinney JT, Hoegh-Guldberg O, Kleypas J, Skirving W &
Strong A). American Geophysical Union Washington, DC, pp. 111-128.
Sheppard C. & Rioja-Nieto R. (2005). Sea surface temperature 1871-2099 in 38 cells in the
Caribbean region. Marine Environmental Research, 60, 389-396.
Sheppard C.R.C. (2003). Predicted recurrences of mass coral mortality in the Indian Ocean.
Nature, 425, 294-297.
14
Sutherland K.P., Porter J.W. & Torres C. (2004). Disease and immunity in Caribbean and
Indo-Pacific zooxanthellate corals. Marine Ecology Progress Series, 266, 273-302.
Vitousek P.M., Mooney H.A., Lubchenco J. & Melillo J.M. (1997). Human domination of
Earth's ecosystems. Science, 277, 494-499.
Voigt W. & al. e. (2003). Trophic levels are differentially sensitive to climate. Ecology, 84,
2444-2453.
Walther G.R., Post E., Convey P., Menzel A., Parmesan C., Beebee T.J.C., Fromentin J.M.,
Hoegh-Guldberg O. & Bairlein F. (2002). Ecological responses to recent climate
change. Nature, 416, 389-395.
Wapnick C.M., Precht W.F. & Aronson R.B. (2004). Millennial-scale dynamics of staghorn
coral in Discovery Bay, Jamaica. Ecology Letters, 7, 354-361.
Ward J.R. & Lafferty K.D. (2004). The elusive baseline of marine disease: are diseases in
ocean ecosystems increasing? PloS Biology, 2, 542-547.
Wilkinson C. (2000). Status of Coral Reefs of the World: 2000. Australian Institute of Marine
Science, Cape Ferguson.
Willis B.L., Page C.A. & Dinsdale E.A. (2004). Coral disease on the Great Barrier Reef. In:
Coral Health and Disease (eds. Rosenberg E & Loya Y). Springer-Verlag Berlin, pp.
69-104.
Wolanski E. & De'ath G. (2005). Predicting the impact of present and future human land-use
on the Great Barrier Reef. Estuarine Coastal and Shelf Science, 64, 504-508.
Wolanski E., Richmond R.H. & McCook L. (2004). A model of the effects of land-based,
human activities on the health of coral reefs in the Great Barrier Reef and in Fouha
Bay, Guam, Micronesia. Journal of Marine Systems, 46, 133-144.
15
CHAPTER 1
Mapping the variability in frequency, magnitude and spatial extent of coral reef
temperature anomalies between 1985 and 2005
Abstract
Increases in ocean temperature due to climate change are expected to have extensive
impacts on marine ecosystems. Coral reefs are widely considered to be particularly
vulnerable to these changes in ocean temperatures, yet we understand little about the broadscale spatiotemporal patterns and fine-scale variability in ocean temperatures underlying
many coral reef ecosystem changes. We created a novel 4-km global ocean temperature
anomaly database to measure these temperature patterns on coral reefs. Then we tested
whether key characteristics in temperature anomalies, including their size and spatial extent,
varied regionally and annually. Our analyses found substantial variability in anomaly
frequency, duration and magnitude across and within tropical regions. We also discovered
that even during major climatic events like El Niños, locations affected by these anomalies
were idiosyncratic. In addition, we found that 48% of bleaching-related, and 44% of diseaserelated, anomalies were less than 50 km2, smaller than the resolution of most models used to
forecast climate changes. These analyses describe a striking degree of fine-grained
variability in temperature anomalies. Quantifying this variability is essential to our ability to
understand the scales at which temperature is affecting coral reef ecosystem health and to
make accurate predictions about the potential impacts of climate change.
Introduction
Human-induced climate change is altering the physical processes that structure natural
communities (Chapin et al. 2000; Parmesan & Yohe 2003; Thomas et al. 2004; Harley et al.
2006). Indeed, climate change is expected to continue to have wide-ranging effects on
marine ecosystems due to warming ocean temperatures, sea level rise, ocean acidification,
and changes in ocean circulation patterns (IPCC 2007). Differences in these physical
properties could affect species ranges, species composition and diversity, and productivity
(Harley et al. 2006), which could lead to major social and economic consequences (Stern
2006). Although we are already seeing evidence of climate change effects on a range of
ecosystems, we still do not fully comprehend how the underlying physical patterns being
modified by climate change are linked with biological patterns (Walther et al. 2002).
Without understanding the interaction between physical patterns and biological responses,
we cannot accurately predict ecosystem responses to climate change or develop the necessary
management interventions.
Globally, ocean surface temperatures have increased 0.52°C ± 0.19°C from 1850 to
2004 (Rayner et al. 2006). Ocean temperature changes can have a range of impacts on
marine species from physiological stress to mortality (Harley et al. 2006). Climate models
predict global rises in ocean temperature over at least the next 100 years (Rayner et al. 2006;
IPCC 2007). Coincident with anthropogenically driven ocean warming are predicted
17
increases in the severity of ocean temperature anomalies, defined as periods of unusually
warm ocean temperatures, in all of the tropical oceans (Hoegh-Guldberg 1999; Sheppard
2003; Sheppard & Rioja-Nieto 2005). Although temperature anomalies can be a natural
consequence of climate variability, recent work has indicated that the recent thermal event in
the Caribbean in 2005 was at least partially due to anthropogenic climate changes (Donner et
al. 2007).
Scleractinian or reef-building corals are believed to be particularly vulnerable to
these temperature anomalies because they already live near their thermal limits (Jokiel &
Coles 1990; Glynn 1993; Berkelmans & Willis 1999). Warmer than usual temperatures play
a role in two processes that can lead to coral mortality: coral bleaching and disease.
Mortality from bleaching and disease is known to be a major cause of the unprecedented
decline in coral reef health in the last 30 years in both the Caribbean (Gardner et al. 2003)
and Pacific (Bruno & Selig 2007). Coral bleaching has become more frequent and more
extensive in the last twenty years (Hoegh-Guldberg 1999). Mass bleaching due to severe El
Niño events in 1982-1983 and 1997-1998 led to exceptional regional-scale coral mortality
(Glynn et al. 1988; Wilkinson 2000; Bruno et al. 2001; Glynn et al. 2001). In order to
predict the consequences of future thermal stress events and design effective management
actions, we need to quantify the temperature patterns causing these coral declines.
Long-term sea surface temperature datasets have provided valuable historical records
of temperature patterns in coral reef systems (Folland et al. 2001; Rayner et al. 2006). These
data have also been used to model possible increases in the temperature anomalies on coral
reefs and put current temperatures in a climate context (Sheppard 2003; Barton & Casey
2005; Sheppard & Rioja-Nieto 2005). However, these datasets have resolutions of 100 km or
18
more, which may be too coarse to pair with biological data and ecological processes if there
is substantial fine-grained variability (Sheppard 2003; Sheppard & Rioja-Nieto 2005).
Patterns in coral abundance and diversity can vary at scales of 1-10 km (Edmunds & Bruno
1996) and ecological processes like dispersal similarly occur at scales of 1-10s of kilometers
for many coral reef organisms (Grantham et al. 2003; Ayre & Hughes 2004). Therefore, we
need to understand the scale of variability in temperature anomalies to determine how best to
match biological and temperature data on coral reefs.
We quantified the spatial and temporal patterns of temperature anomalies associated
with bleaching and disease using a novel 4-km global ocean temperature anomaly database
from 1985-2005. First we calculated three parameters of temperature anomalies that relate to
generally accepted metrics of disturbance: frequency, magnitude, and duration (Connell
1978). These data not only enabled us to establish a baseline from which we could compare
extreme events like the 1998 El Niño, but also allowed us to test how these anomaly
characteristics varied by region. Because disturbance extent and spatial scale are also
important determinants of severity and are often used in the design of protected areas, we
measured these variables as well (Connell 1978; Pickett & Thompson 1978). By calculating
the total area affected by anomaly events and quantifying the distribution of anomaly sizes,
we evaluated the degree of clustering of anomalies and whether they were spatially pervasive
or localized over large spatial scales. Together, these analyses described the spatial structure
and variability of temperature anomalies, the results of which provide critical context for
evaluating past and future thermal stress patterns.
19
Materials and Methods
Temperature dataset
The Coral Reef Temperature Anomaly Database (CoRTAD) was developed using
data from the Pathfinder Version 5.0 collection produced by the National Oceanic and
Atmospheric Administration’s (NOAA) National Oceanographic Data Center (NODC) and
the University of Miami’s Rosenstiel School of Marine and Atmospheric Science
(http://pathfinder.nodc.noaa.gov). These data are derived from the Advanced Very High
Resolution Radiometer (AVHRR) sensor and are processed to a resolution of approximately
4.6 km at the equator. These data have the highest resolution covering the longest time
period of any satellite-based ocean temperature dataset (Fig. 1.1).
We used weekly averages of day and nighttime data with a quality flag of 4 or better,
which is a commonly accepted cutoff for “good” data (Casey & Cornillon 1999; Kilpatrick et
al. 2001). By using a day-night average, we reduced the number of missing pixels by 25%
with virtually no loss in accuracy (Table S1.1). We also added back in some data that were
initially given low quality levels. The Pathfinder algorithm eliminates any observation with a
Sea Surface Temperature (SST) more than 2°C different from a relatively coarse resolution
SST value based on the Reynolds Optimum Interpolation Sea Surface Temperature (OISST)
value, a long-term, in situ-based data set (Kilpatrick et al. 2001; Reynolds et al. 2002).
Therefore, we added observations back into the analysis if the SST was greater than the
OISST, but less than the OISST+5°C. The 5°C threshold is a reasonable selection that
allows diurnal warming events (Kawai & Wada 2007) or other spatially limited warm spots
back into the dataset without including unrealistic and erroneously warm values. Values less
than the OISST were not included because they may have been biased by cloud
20
contamination and other satellite errors, which tend to result in cooler SST estimates. These
processes resulted in a dataset with only 21.2% missing data.
To create a gap-free dataset for analysis, we used 3 x 3 pixel median spatial fill. We
then did a temporal fill using the Piecewise Cubic Hermite Interpolating Polynomial (PCHIP)
function in Matlab (The Mathworks Inc. 2006) to fill the remaining gaps. We chose this
conservative approach because it provided interpolated SSTs that are bounded by the nearest
available values in time. It also used data from only a very limited spatial domain, which is
an important consideration given the variability of coral reef environments. Using these gapfilled data, we then created site-specific climatologies for each reef grid cell to describe longterm temperature patterns over the 21-year dataset (Eqn. 1). The climatology was generated
using a harmonic analysis procedure that fits annual and semi-annual signals to the time
series of weekly SSTs at each grid cell:
SST(t)clim = Acos(2πt + B) + Ccos(4πt + D) + E
(1)
where t is time, A and B are coefficients representing the annual phase and amplitude, C and
D are the semi-annual phase and amplitude, and E is the long-term temperature mean.
Similar approaches have been used for generating climatologies because they are more robust
than simple averaging techniques, which can be more susceptible to data gaps from periods
of cloudiness (Podesta et al. 1991; Mesias et al. 2007).
Sea surface temperatures from AVHRR quantify only the temperature of the 'skin' of
the ocean, roughly the first 10 µm of the ocean surface (Donlon et al. 2007). Most field
surveys of coral cover occur between 1-15 m depth. To be useful for coupling with coral reef
21
biological data, these temperature data must be relatively accurate beyond the 'skin' of the
ocean. We used linear regression to examine how data from in situ reef temperature loggers
compared with data from the CoRTAD to determine how accurate the temperature data were
at a variety of depths and locations around the world (Table S1.1).
Temperature anomaly metrics
Several metrics could be used to link coral reef ecosystem health with temperature
including trophic structure, diversity or percent coral cover (Roberts et al. 2002; Newman et
al. 2006; Bruno & Selig 2007) . However, we focused our analysis on coral bleaching and
disease because they are key drivers of coral decline and their relationships with temperature
are better understood (Glynn 1993; Aronson & Precht 2001; Bruno et al. 2007). Analyses
were performed on two metrics (Table 1): one that is commonly known to lead to bleaching
(Glynn 1993; Liu et al. 2003; Strong et al. 2004), and one that is correlated with increased
disease severity (Selig et al. 2006; Bruno et al. 2007). Coral bleaching results when corals
lose their symbiotic zooxanthellae (Glynn 1993, 1996). Bleaching is a natural stress
response not only to warm temperatures, but also to cool temperatures (Hoegh-Guldberg &
Fine 2004) as well as light and salinity values different from the normal range (Glynn 1993).
Corals can recover from bleaching, but their ability to do so is dependent on the magnitude
and duration of the anomaly event (Glynn 1993). The temperature thresholds that result in
coral bleaching vary by location and species (Berkelmans & Willis 1999). Bleaching is often
connected to Thermal Stress Anomalies (TSA), defined as occurring when temperatures
exceed 1°C more than the climatologically warmest week of the year (Table 1, Glynn 1993).
The temperature anomaly thresholds relevant to disease have been studied in only one
22
pathogen-host system (Selig et al. 2006; Bruno et al. 2007). In that system, changes in
disease cases were correlated with Weekly Sea Surface Temperature Anomalies (WSSTAs),
temperatures that were 1°C greater than the weekly average for that location.
The best metric for predicting bleaching or disease may vary according to location,
species, and pathogen (Berkelmans 2002; Selig et al. 2006; Bruno et al. 2007). For example,
bleaching on the Great Barrier Reef was best predicted by the maximum anomaly over a 3
day period (Berkelmans et al. 2004), rather than an anomaly metric like the TSA. Although
the 7-day averaging approach in the CoRTAD may be too temporally coarse to capture all
bleaching events, it is necessary to maintain consistency and minimize gaps in the dataset
across broad spatial scales. In addition, the data will be less likely to yield false positives for
TSAs and will likely capture most WSSTA events, which have a lower temperature
threshold.
Anomaly frequency, temporal duration and magnitude
We calculated both the frequency of TSAs and WSSTAs based on the number of
anomalies in each calendar year and cumulatively over the 21-year study. Anomaly
magnitude and duration, two important predictors of coral bleaching and disease, may also
play roles in the percent mortality during a thermal event (Glynn 1993; Winter et al. 1998;
Berkelmans et al. 2004). Anomaly magnitudes were calculated over the whole time series.
We quantified anomaly duration by each anomaly event, which we defined as beginning
when ocean temperatures exceeded the threshold value for TSA or WSSTA and ending when
temperatures returned below the threshold again. Only locations that had an anomaly and
were located on a reef were included in the analysis.
23
Region delineations
We delineated regions within ocean basins to better understand how anomaly patterns varied
at regional scales (Fig. 1.2). These regions vary in total area, but they represent general
regions of similar diversity and biogeography (Roberts et al. 2002), major bathymetric
breaks, or management. These demarcations are similar to marine ecoregions recently
defined by several conservation organizations (Spalding et al. 2007). We restricted our
analysis to regions that have substantial coral reef area and were large enough not to cause
edge effects in the calculation of anomaly areas. Thus, we did not include reefs off the coasts
of Brazil, West Africa, Bermuda and parts of southeastern and western Australia.
Coral reef location data
Shallow coral reef location data were compiled from a variety of freely available datasets.
The initial GIS layer was developed from Reefs at Risk (Bryant et al. 1998) and Reefs at
Risk in Southeast Asia (Burke et al. 2002), which used data from the United Nations
Environment Program – World Conservation Monitoring Centre (UNEP-WCMC). These
data were improved with additional data from scientists and governmental agencies. We then
added GIS data from ReefBase and Reef Check to capture additional reef areas worldwide.
Data included both polygons and points, which were then gridded at a resolution of 1 km
(Fig. 1.2).
Calculation of anomaly areas
We calculated patterns in anomaly area for both WSSTAs and TSAs throughout the
tropics. For each metric, we first created 1096 weekly grids of anomaly presence and
24
absence for all the weeks in the database. We only calculated areas for anomalies that
contained at least one grid cell that overlapped a known coral reef location. We used the
bwlabel function in the Image Processing Toolbox in Matlab 7.3 (The Mathworks Inc. 2006)
to identify each anomaly and determine whether it was connected to a neighboring thermal
event in any of the eight adjacent grid cells.
Results
CoRTAD Validation
Although the AVHRR sensor theoretically measures only the 'skin' of the ocean, the
Pathfinder processing attempts to create a 'bulk' sea surface temperature more representative
of the temperature at approximately 1 m depth. In fact, we found that the temperature data in
the CoRTAD were quite accurate over a wide range of coral locations and depths down to at
least 10 m. The CoRTAD typically had r2 > 0.90 when compared with in situ temperature
loggers for most of the studied coral reef areas (Table S1.1). Accuracy will clearly depend
on local oceanographic conditions, particularly the degree of stratification and depth.
Because the AVHRR instruments are affected by atmospheric absorption of surface IR
radiance, temperature measurements can be less accurate as a result of cloud cover (water
vapor), aerosols, vapor, CO2, CH4, and NO2 (Kilpatrick et al. 2001). Validations based on
only a few satellite and in situ match up points (<50) may yield artificially lower correlations
(Table S1.1).
25
Anomaly frequency
Anomaly frequencies varied by year and by region (Figs. 1.3-4) for both WSSTA
(disease) and TSA (bleaching) metrics. Spatial patterns for the two metrics were very
similar. Because the temperature threshold for attaining a TSA is higher, the number of
TSAs was almost always lower than the number of WSSTAs. Over the 21-year period
analyzed, TSA frequency varied from a cumulative average of 13 events in the Florida Keys
to an average of 87 events in the Central Pacific (Figs. 1.3A-B). At the reef scale, the
average TSA frequency across 21 years was 37 anomalies and the maximum was 444
anomalies. In a typical year, there were an average 1-4 TSAs and 4-11 WSSTAs (Figs.
1.5A-B) on a given reef. However, frequencies changed drastically for some regions during
major events, such as El Niño.
We also examined patterns in anomaly frequencies by year to provide insight into
whether more recent years had more TSAs than earlier years in the database (Figs. 1.6-7). In
the Caribbean, the Mesoamerican Barrier Reef, Gulf of Mexico, and Eastern Caribbean
regions all had the highest number of anomalies in 1998 (Figs. 1.2, 1.6). There were also a
significant number of anomalies in 2005, particularly in the Antilles region (Figs. 1.2, 1.6).
In the Pacific, the signature for the 1998 El Niño event was very clear, particularly in the
Central Pacific, but the Indo-Pacific still had areas that did not have their maximum
frequency in that year. For example, much of the territorial waters of Indonesia between
Kalimantan and West Papua experienced greater anomaly frequencies during the 1987-1988
El Niño event (Fig. 1.7). Nonetheless, the severity of the 1998 El Niño in the Indo-Pacific
and the 2005 event in the Caribbean was clear both in the number of bleaching-related
26
anomalies and in the ratio of anomalies in those years compared with other years (Figs. 1.89).
Anomaly durations
The duration of temperature anomalies has long been known to affect bleaching events
(Glynn et al. 1988), although its effect on coral disease severity remains poorly understood.
Mean TSA durations varied considerably by region (Fig. 1.5C). The Central America,
Persian Gulf, and Pacific Central America regions had the longest mean anomaly durations,
averaging approximately 1.8 weeks (Figs. 1.5C, S1.1, and S1.2). During the 1998 El Niño
event, TSA anomaly durations were markedly higher in Pacific Central America, South
China Sea, and the Persian Gulf, where durations averaged 6, 3.1 and 4.4 weeks, respectively
(Fig. S1.3; Persian Gulf not shown). Although anomaly durations were higher for other
regions during El Niño years, they were within the standard deviation of the long-term
averages (Graham et al. 2006). Anomaly durations for WSSTAs were longer, particularly in
Pacific Central America and Taiwan and Japan, where they averaged more than 2 weeks
(Fig. 1.5D). Like TSAs, WSSTA durations were greater during El Niño events in nearly
every region except those in the Caribbean (Fig. S1.4).
Anomaly magnitudes
Absolute anomaly magnitudes between WSSTAs and TSAs were not directly comparable
because of the different temperature thresholds. Anomaly magnitudes were highest for the
Red Sea and Persian Gulf regions. Typical mean TSA magnitudes over the whole time series
varied between 1.4°C in the Western Pacific to 1.8°C in the Persian Gulf (Fig. 1.5E).
27
WSSTAs displayed a similar overall pattern to TSAs. Over the whole time series mean
magnitudes of WSSTAs ranged from 1.3°C in the Western Pacific to 1.8°C in the Persian
Gulf (Fig. 1.5F).
Anomaly areas
Anomaly areas displayed relatively similar patterns for WSSTAs and TSAs (Fig. 1.10) with
most anomalies having small areas and a large number of occurrences. Spatially very large
anomalies (<500 km2) occurred, but represented less than 10% of the data. There was a
marked difference between TSAs and WSSTAs in the total number of events, although this
probably results from the higher temperature threshold required to be counted as a TSA.
From 1985-2005, there were 398,931 TSAs and 1,000,525 WSSTAs. However, both metrics
had relatively the same percentage of anomalies in different area categories. For example,
33% of TSAs and 29% of WSSTAs were between 1-25 km2 in size, the smallest size
category, which corresponds roughly to one grid cell. More than 90% of TSAs and WSSTAs
were 1-425 km2 and 1-500 km2, respectively.
Discussion
The frequency, magnitude, and extent of temperature anomalies can be used to
quantify disturbance severity and its role in the regulation of ecological communities
(Connell 1978; Menge & Sutherland 1987). We found substantial variability in the
frequency, duration, and magnitude of the temperature anomalies that can lead to bleaching
and disease both within and across regions and across years. Anomalies were highly
28
pervasive even in non-extreme years, but nearly 50% of the anomalies that can cause
bleaching were less than 50 km2. Overall, these results suggest that although temperature
anomalies are spatially extensive, their fine-scale spatial structure will result in highly
localized effects on coral disease and bleaching.
Patterns in anomaly frequency, magnitude, and temporal duration affect the
occurrence and severity of bleaching and disease events, although several other physical
(e.g., light) and biological (e.g., coral cover) parameters may be necessary or synergistic
factors (Glynn 1993; Bruno et al. 2007). Frequencies for both bleaching- and disease-related
anomalies were highly regionally-specific and strongly correlated with large-scale climatic
patterns like El Niño events (Figs. 1.3-4). Temperature anomalies are known to exhibit high
spatial heterogeneity on reefs according to depth, site, and season (Leichter et al. 2006). This
heterogeneity was common across reefs globally, although regional patterns and differences
between bleaching and disease-related anomalies also emerged (Fig. 1.5A-F). The highest
mean TSA frequencies in the Caribbean were found in the Central America region whereas
the Florida Keys experienced the highest WSSTA frequencies. Differences in the
characteristics of TSAs and WSSTAs could be indicative of greater regional vulnerability to
bleaching or disease. Although the presence of an anomaly connotes conditions that may be
favorable to the development of bleaching or disease, they are not necessarily predictive of
actual changes in coral condition. Nonetheless, these regional patterns describe baseline
characteristics to which future changes in anomalies or coral condition can be compared.
Our results also provided spatial context for how severe the 1998 El Niño and the
2005 warm event in the Caribbean were compared to major thermal events during 19852005. The severe 1998 El Niño event in the Indo-Pacific (Fig. 1.8) was clearly
29
unprecedented in the last 20 years, as was the 2005 warm event in the eastern Caribbean (Fig.
1.9). Yet even during these extreme events, substantial local and regional variability existed
(Figs. 1.6-7). Although 33% of reefs in the Caribbean had their 21-year peak in number of
anomalies during 2005, 18% of reefs had the highest number of anomalies during 1998 (Fig.
1.6). Similarly, during the 1998 El Niño, 21% of reefs in the Indo-Pacific experienced the
highest recorded anomaly frequencies. Yet, across much of eastern Indonesia and the Great
Barrier Reef, frequencies were higher during the 1988 and 2002 El Niño events (Fig. 1.7).
Throughout the Indo-Pacific, 10% of reefs had their highest anomaly frequencies during
2002. Indeed, although the Great Barrier Reef experienced extensive bleaching in 1998, the
most severe and extensive bleaching on record occurred in 2002 (Berkelmans et al. 2004).
These results suggest that anomalies are spatially ephemeral in nature and may have different
patterns even when associated with large climatic patterns like El Niños, which have wellestablished spatial signatures.
We also found that the total spatial extent of anomalies was extensive, but there was
substantial fine-scale spatial structure. From 1985 to 2005, all coral reef locations within our
study regions experienced at least one anomaly. Even in individual years, a considerable
percentage of reefs had at least one anomaly. During 1985, a non-extreme year, the Great
Barrier Reef had at least one bleaching-related anomaly in over 67% of the total area in the
region. In comparison, during the 1998 El Nino event, 84% of the Great Barrier Reef was
affected. In spite of the large area anomalies can affect, globally the anomalies themselves
were typically small in size. Over 21 years, 48% of bleaching anomalies and 44% of disease
anomalies were less than 50 km2, smaller than the spatial resolution of many climate models
and temperature anomaly datasets (Fig. 1.10). Therefore, even though a substantial portion
30
of total reef area may be affected during a thermal event, bleaching-related coral losses may
be spatially patchy.
Coral bleaching and disease can have dramatic effects on the abundance and
composition of corals and coral reef ecosystem health (Aronson et al. 2002; Graham et al.
2006; Ledlie et al. 2007). During major thermal events like the 1998 El Niño, more than
50% of live coral cover in affected sites can be lost (Sheppard 1999; Aronson et al. 2000;
Bruno et al. 2001). Although the dynamics of temperature anomalies and loss of coral cover
through disease outbreaks are less well understood, several coral diseases are associated with
elevated temperatures (Kuta & Richardson 2002; Patterson et al. 2002; Bruno et al. 2007).
Declines in coral cover through disease and bleaching can have drastic effects on the entire
coral reef ecosystem (Aronson & Precht 2001; Jones et al. 2004; Graham et al. 2006).
Studies of individual reefs that have undergone coral mortality from the 1998 El Niño event
have found subsequent parallel declines in fish abundance and diversity (Jones et al. 2004)
and erosion of the reef framework itself (Graham et al. 2006) even on protected and isolated
reefs. Decreased shoreline protection from storms and declines in fisheries from loss of the
reef structure potentially jeopardize US$129-$375 billion/year in value from coral reef
ecosystem services (Costanza et al. 1997).
The dramatic decline in coral reefs have made monitoring and predicting
vulnerability to coral reef bleaching and disease key priorities for scientists, conservationists,
and policymakers. Our results characterize the frequency, magnitude, and extent of
bleaching and disease-related temperature anomalies and their natural variability. These data
can be used for a variety of applications to better understand how climate change is affecting
the temperature anomalies causing coral reef decline. Time series analyses of specific areas
31
may help elucidate whether the frequency, magnitude, and duration of anomalies are
increasing over time. In addition, these data can be used to generate spatial maps of anomaly
durations and magnitudes to get a picture of what the typical patterns would be for specific
locations. Managers can then determine how unusual future events are for their areas. If
paired with in situ observations of coral cover, bleaching, and disease, these data could also
be used to develop local and regional portraits of vulnerability to temperature stress.
Understanding the scale of variability in temperature anomalies is a critical step
towards characterizing ocean temperature patterns and their relationship to ecosystem
response. Even minor changes in the local dynamics of physical factors like temperature can
have dramatic effects on ecosystems. Models in the Caribbean predict that rises of only
0.1°C in regional ocean temperatures could trigger 35% and 42% increases in the geographic
extent and intensity of coral bleaching, respectively (McWilliams et al. 2005). However,
documenting these patterns alone clearly will not be enough to predict ecosystem response to
climate change. Local conditions could be affected by management interventions, the
synergistic effects of other stressors, and species composition (Harley et al. 2006). Our
results highlight the importance of incorporating finer scale data into climate models so that
we can more accurately assess vulnerability to temperature changes and design relevant
management interventions.
32
Figure 1.1 Comparison of the 4 km Pathfinder with other satellite datasets. An
example of the benefits of increasing resolution on the Great Barrier Reef from; (A)
50 km HotSpot data to (B) 9 km Pathfinder data to (C) 4 km Pathfinder data. In the 4
km data, there is less missing data, allowing for greater coverage of coastal areas
where many reefs occur. In addition, the 4 km data displays more spatial structure
and precision in the temperature values. Data are from January climatological
averages, monthly for the 50 km and from the first week of January for the 9 km and
4 km data.
33
Figure 1.2 Region delineations for the CoRTAD analysis. (GBR = Great Barrier
Reef, MBR = Mesoamerican Barrier Reef, and SCS = South China Sea). Purple
dots are locations of coral reefs.
34
Figure 1.3 Cumulative number of TSAs (1985-2005) in (A) Indo-Pacific and (B)
Caribbean.
A
B
35
Figure 1.4 Cumulative number of WSSTAs (1985-2005) in (A) Indo-Pacific and
(B) Caribbean.
A
B
36
Figure 1.5 Mean frequency, duration, and magnitude of TSAs and WSSTAs.
Temporal average of mean TSA (A) and WSSTA (B) frequency and 95% confidence
intervals (bars) from 1985-2005 for the studied regions (see Figure 2). Mean
duration for TSA (C) and WSSTA (D) as well as mean magnitude for TSA (E) and
WSSTA (F) are also shown. In each panel, the data are ordered by ocean basin:
Caribbean Sea (top), Indian Ocean (middle), and Pacific (bottom) and sorted
regionally from highest to lowest mean value.
A
B
C
D
37
F
E
38
Figure 1.6 Year of the maximum number of TSAs in the Caribbean. For each
grid cell, the year with the greatest number of TSAs was recorded. The signature of
the 2005 warm event is clearly visible in the Antilles region.
39
Figure 1.7 Year of maximum number of TSAs in the Pacific. For each grid cell,
the year with the greatest number of TSAs was recorded. Although most locations
had their maximum TSAs during the 1998 El Niño event (yellow), several other El
Niño years including 1988 (light blue) and 2002 are also visible.
40
Figure 1.8 Severity of the 1998 El Niño event. (A) The number of TSAs in 1998
and (B) the ratio of the number of TSAs in 1998 relative to the number of TSAs from
1985-2005. The 1998 El Niño event was unusual for equatorial Pacific and much of
the Indian Ocean.
A
B
41
Figure 1.9 Severity of the 2005 warm event. (A) The number of TSAs in 2005 and
(B) the ratio of the number of TSAs in 2005 relative to the number of TSAs from
1985-2005. The warm event in the Antilles region produced anomaly frequencies
unprecedented in the last 21 years.
A
B
42
Figure 1.10 Frequency distribution of anomaly areas for TSAs and WSSTAs
for 1985-2005. TSAs are shown with the solid black line and WSSTAs with the
dashed grey line. Most climate studies occur at a spatial resolution of 50 km2
(dotted grey line). Raw anomaly areas were categorized into 25 km2 bins.
43
Table 1.1 Metrics of thermal stress.
Metric
Question
Thermal stress
Is the temperature ≥1°C
anomalies (TSA) warmer than
typical summertime
highs?
Weekly sea
surface
temperature
anomalies
(WSSTA)
Is the temperature ≥1°C
warmer than usual for
that week of the year?
44
Application
Bleaching
References
(Glynn 1993;
Podesta & Glynn
2001; Liu et al.
2003)
Disease
(Selig et al. 2006;
Bruno et al. 2007)
Acknowledgements
For temperature data: Erich Bartels of MOTE Marine Laboratory and Peter Edmunds of
California State University-Northridge; T. Kristiansen and M. O'Connor for comments on the
manuscript.
45
References
Aronson R.B. & Precht W.F. (2001). White-band disease and the changing face of Caribbean
coral reefs. Hydrobiologia, 460, 25-38.
Aronson R.B., Precht W.F., Macintyre I.G. & Murdoch T.J.T. (2000). Ecosystems - coral
bleach-out in Belize. Nature, 405, 36-36.
Aronson R.B., Precht W.F., Toscano M.A. & Koltes K.H. (2002). The 1998 bleaching event
and its aftermath on a coral reef in Belize. Marine Biology, 141, 435-447.
Ayre D.J. & Hughes T.P. (2004). Climate change, genotypic diversity and gene flow in reefbuilding corals. Ecology Letters, 7, 273-278.
Barton A.D. & Casey K.S. (2005). Climatological context for large-scale coral bleaching.
Coral Reefs, 24, 536-554.
Berkelmans R. (2002). Time-integrated thermal bleaching thresholds of reefs and their
variation on the Great Barrier Reef. Marine Ecology Progress Series, 229, 73-82.
Berkelmans R., De'ath G., Kininmonth S. & Skirving W.J. (2004). A comparison of the 1998
and 2002 coral bleaching events on the Great Barrier Reef: spatial correlation,
patterns, and predictions. Coral Reefs, 23, 74-83.
Berkelmans R. & Willis B.L. (1999). Seasonal and local spatial patterns in the upper thermal
limits of corals on the inshore Central Great Barrier Reef. Coral Reefs, 18, 219-228.
Bruno J.F. & Selig E.R. (2007). Regional decline of coral cover in the Indo-Pacific: timing,
extent, and subregional comparisons. PLoS One, 2, e711.
Bruno J.F., Selig E.R., Casey K.S., Page C.A., Willis B.L., Harvell C.D., Sweatman H. &
Melendy A.M. (2007). Thermal stress and coral cover as drivers of coral disease
outbreaks. Plos Biology, 5, 1220-1227.
Bruno J.F., Siddon C.E., Witman J.D., Colin P.L. & Toscano M.A. (2001). El Nino related
coral bleaching in Palau, Western Caroline Islands. Coral Reefs, 20, 127-136.
Bryant D., Burke L., McManus J. & Spalding M. (1998). Reefs at Risk: A Map-Based
Indicator of Potential Threats to the World's Coral Reefs. World Resources Institute,
Washington, DC.
Burke L., Selig E.R. & Spalding M. (2002). Reefs at Risk in Southeast Asia. World
Resources Institute, Washington, DC.
Casey K.S. & Cornillon P. (1999). A comparison of satellite and in situ-based sea surface
temperature climatologies. Journal of Climate, 12, 1848-1863.
46
Chapin F.S., Zavaleta E.S., Eviner V.T., Naylor R.L., Vitousek P.M., Reynolds H.L., Hooper
D.U., Lavorel S., Sala O.E., Hobbie S.E., Mack M.C. & Diaz S. (2000).
Consequences of changing biodiversity. Nature, 405, 234-242.
Connell J.H. (1978). Diversity in tropical rain forests and coral reefs. Science, 199, 13021310.
Costanza R., dArge R., deGroot R., Farber S., Grasso M., Hannon B., Limburg K., Naeem S.,
Oneill R.V., Paruelo J., Raskin R.G., Sutton P. & vandenBelt M. (1997). The value of
the world's ecosystem services and natural capital. Nature, 387, 253-260.
Donlon C., Robinson I., Casey K.S., Vazquez-Cuervo J., Armstrong E., Arino O.,
Gentemann C., May D., LeBorgne P., Piolle J., Barton I., Beggs H., Poulter D.J.S.,
Merchant C.J., Bingham A., Heinz S., Harris A., Wick G., Emery B., Minnett P.,
Evans R., Llewellyn-Jones D., Mutlow C., Reynolds R.W., Kawamura H. & Rayner
N. (2007). The global ocean data assimilation experiment high-resolution sea surface
temperature pilot project. Bulletin of the American Meteorological Society, 88, 11971213.
Donner S.D., Knutson T.R. & Oppenheimer M. (2007). Model-based assessment of the role
of human-induced climate change in the 2005 Caribbean coral bleaching event.
Proceedings of the National Academy of Sciences of the United States of America,
104, 5483-5488.
Edmunds P.J. & Bruno J.F. (1996). The importance of sampling scale in ecology: kilometerwide variation in coral reef communities. Marine Ecology-Progress Series, 143, 165171.
Folland C.K., Rayner N.A., Brown S.J., Smith T.M., Shen S.S.P., Parker D.E., Macadam I.,
Jones P.D., Jones R.N., Nicholls N. & Sexton D.M.H. (2001). Global temperature
change and its uncertainties since 1861. Geophysical Research Letters, 28, 26212624.
Gardner T.A., Cote I.M., Gill J.A., Grant A. & Watkinson A.R. (2003). Long-term regionwide declines in Caribbean corals. Science, 301, 958-960.
Glynn P.W. (1993). Coral reef bleaching - ecological perspectives. Coral Reefs, 12, 1-17.
Glynn P.W. (1996). Coral reef bleaching: facts, hypotheses and implications. Global Change
Biology, 2, 495-509.
Glynn P.W., Cortes J., Guzman H.M. & Richmond R.H. (1988). El Niño (1982-1983)
associated coral mortality and relationship to sea surface temperature deviations in
the tropical eastern Pacific. Proceedings of the Sixth International Coral Reef
Symposium, 2, 693-698.
47
Glynn P.W., Mate J.L., Baker A.C. & Calderon M.O. (2001). Coral bleaching and mortality
in Panama and Ecuador during the 1997-1998 El Niño-Southern Oscillation event:
spatial/temporal patterns and comparisons with the 1982-1983 event. Bulletin of
Marine Science, 69, 79-109.
Graham N.A.J., Wilson S.K., Jennings S., Polunin N.V.C., Bijoux J.P. & Robinson J. (2006).
Dynamic fragility of oceanic coral reef ecosystems. Proceedings of the National
Academy of Sciences of the United States of America, 103, 8425-8429.
Grantham B.A., Eckert G.L. & Shanks A.L. (2003). Dispersal potential of marine
invertebrates in diverse habitats. Ecological Applications, 13, S108-S116.
Harley C.D.G., Hughes A.R., Hultgren K.M., Miner B.G., Sorte C.J.B., Thornber C.S.,
Rodriguez L.F., Tomanek L. & Williams S.L. (2006). The impacts of climate change
in coastal marine systems. Ecology Letters, 9, 228-241.
Hoegh-Guldberg O. (1999). Climate change, coral bleaching and the future of the world's
coral reefs. Marine & Freshwater Research, 50, 839-866.
Hoegh-Guldberg O. & Fine M. (2004). Low temperatures cause coral bleaching. Coral Reefs,
23, 444-444.
IPCC (2007). Climate Change 2007: the Physical Science Basis. Contribution of Working
Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate
Change. Cambridge University Press, Cambridge.
Jokiel P.L. & Coles S.L. (1990). Response of Hawaiian and other Indo-Pacific reef corals to
elevated temperature. Coral Reefs, 8, 155-162.
Jones G.P., McCormick M.I., Srinivasan M. & Eagle J.V. (2004). Coral decline threatens fish
biodiversity in marine reserves. Proceedings of the National Academy of Sciences of
the United States of America, 101, 8251-8253.
Kawai Y. & Wada A. (2007). Diurnal sea surface temperature variation and its impact on the
atmosphere and ocean: A review. Journal of Oceanography, 63, 721-744.
Kilpatrick K.A., Podesta G.P. & Evans R. (2001). Overview of the NOAA/NASA advanced
very high resolution radiometer Pathfinder algorithm for sea surface temperature and
associated matchup database. Journal of Geophysical Research-Oceans, 106, 91799197.
Kuta K.G. & Richardson L.L. (2002). Ecological aspects of black band disease of corals:
relationships between disease incidence and environmental factors. Coral Reefs, 21,
393-398.
48
Ledlie M.H., Graham N.A.J., Bythell J.C., Wilson S.K., Jennings S., Polunin N.V.C. &
Hardcastle J. (2007). Phase shifts and the role of herbivory in the resilience of coral
reefs. Coral Reefs, 26, 641-653.
Leichter J.J., Helmuth B. & Fischer A.M. (2006). Variation beneath the surface: quantifying
complex thermal environments on coral reefs in the Caribbean, Bahamas and Florida.
Journal of Marine Research, 64, 563-588.
Liu G., Skirving W. & Strong A.E. (2003). Remote sensing of sea surface temperatures
during 2002 Barrier Reef coral bleaching. EOS, 84, 137-144.
McWilliams J.P., Cote I.M., Gill J.A., Sutherland W.J. & Watkinson A.R. (2005).
Accelerating impacts of temperature-induced coral bleaching in the Caribbean.
Ecology, 86, 2055-2060.
Menge B.A. & Sutherland J.P. (1987). Community Regulation - Variation in Disturbance,
Competition, and Predation in Relation to Environmental-Stress and Recruitment.
American Naturalist, 130, 730-757.
Mesias J.M., Bisagni J.J. & Brunner A. (2007). A high-resolution satellite-derived sea
surface temperature climatology for the western North Atlantic Ocean. Continental
Shelf Research, 27, 191-207.
Newman M.J.H., Paredes G.A., Sala E. & Jackson J.B.C. (2006). Structure of Caribbean
coral reef communities across a large gradient of fish biomass. Ecology Letters, 9,
1216-1227.
Parmesan C. & Yohe G. (2003). A globally coherent fingerprint of climate change impacts
across natural systems. Nature, 421, 37-42.
Patterson K.L., Porter J.W., Ritchie K.E., Polson S.W., Mueller E., Peters E.C., Santavy D.L.
& Smiths G.W. (2002). The etiology of white pox, a lethal disease of the Caribbean
elkhorn coral, Acropora palmata. Proceedings of the National Academy of Sciences
of the United States of America, 99, 8725-8730.
Pickett S.T.A. & Thompson J.N. (1978). Patch dynamics and the design of nature reserves.
Biological Conservation, 13, 27-37.
Podesta G.P., Brown O.B. & Evans R.H. (1991). The annual cycle of satellite-derived seasurface temperature in the southwestern Atlantic Ocean. Journal of Climate, 4, 457467.
Podesta G.P. & Glynn P.W. (2001). The 1997-98 El Niño event in Panama and Galapagos:
An update of thermal stress indices relative to coral bleaching. Bulletin of Marine
Science, 69, 43-59.
49
Rayner N.A., Brohan P., Parker D.E., Folland C.K., Kennedy J.J., Vanicek M., Ansell T.J. &
Tett S.F.B. (2006). Improved analyses of changes and uncertainties in sea surface
temperature measured in situ sice the mid-nineteenth century: The HadSST2 dataset.
Journal of Climate, 19, 446-469.
Reynolds R.W., Rayner N.A., Smith T.M., Stokes D.C. & Wang W.Q. (2002). An improved
in situ and satellite SST analysis for climate. Journal of Climate, 15, 1609-1625.
Roberts C.M., McClean C.J., Veron J.E.N., Hawkins J.P., Allen G.R., McAllister D.E.,
Mittermeier C.G., Schueler F.W., Spalding M., Wells F., Vynne C. & Werner T.B.
(2002). Marine biodiversity hotspots and conservation priorities for tropical reefs.
Science, 295, 1280-1284.
Selig E.R., Harvell C.D., Bruno J.F., Willis B.L., Page C.A., Casey K.S. & Sweatman H.
(2006). Analyzing the relationship between ocean temperature anomalies and coral
disease outbreaks at broad spatial scales. In: Coral reefs and climate change: science
and management (eds. Phinney JT, Hoegh-Guldberg O, Kleypas J, Skirving W &
Strong A). American Geophysical Union Washington, DC, pp. 111-128.
Sheppard C. & Rioja-Nieto R. (2005). Sea surface temperature 1871-2099 in 38 cells in the
Caribbean region. Marine Environmental Research, 60, 389-396.
Sheppard C.R.C. (1999). Coral decline and weather patterns over 20 years in the Chagos
Archipelago, central Indian Ocean. Ambio, 28, 472-478.
Sheppard C.R.C. (2003). Predicted recurrences of mass coral mortality in the Indian Ocean.
Nature, 425, 294-297.
Spalding M.D., Fox H.E., Allen G.R., Davidson N., Ferdana Z.A., Finlayson M., Halpern
B.S., Jorge M.A., Lombana A., Lourie S.A., Martin K.D., McManus E., Molnar J.,
Recchia C.A. & Robertson J. (2007). Marine ecoregions of the world: a
bioregionalization of coastal and shelf areas. Bioscience, 57, 573-583.
Stern N.H. (2006). The Economics of Climate Change: the Stern Review. Cambridge,
Cambridge, UK.
Strong A.E., Liu G., Meyer J., Hendee J.C. & Sasko D. (2004). Coral Reef Watch 2002.
Bulletin of Marine Science, 75, 259-268.
The Mathworks Inc. (2006). Matlab. In. The Mathworks Inc. Natick, MA.
Thomas C.D., Cameron A., Green R.E., Bakkenes M., Beaumont L.J., Collingham Y.C.,
Erasmus B.F.N., de Siqueira M.F., Grainger A., Hannah L., Hughes L., Huntley B.,
van Jaarsveld A.S., Midgley G.F., Miles L., Ortega-Huerta M.A., Peterson A.T.,
Phillips O.L. & Williams S.E. (2004). Extinction risk from climate change. Nature,
427, 145-148.
50
Walther G.R., Post E., Convey P., Menzel A., Parmesan C., Beebee T.J.C., Fromentin J.M.,
Hoegh-Guldberg O. & Bairlein F. (2002). Ecological responses to recent climate
change. Nature, 416, 389-395.
Wilkinson C. (2000). Status of Coral Reefs of the World: 2000. Australian Institute of Marine
Science, Cape Ferguson.
Winter A., Appeldoorn R.S., Bruckner A., Williams E.H. & Goenaga C. (1998). Sea surface
temperatures and coral reef bleaching off La Parguera, Puerto Rico (northeastern
Caribbean Sea). Coral Reefs, 17, 377-382.
51
CHAPTER 2
Analyzing the relationship between ocean temperature anomalies and
coral disease outbreaks at broad spatial scales
Abstract
Ocean warming due to climate change could increase the frequency and severity of infectious
coral disease outbreaks by increasing pathogen virulence or host susceptibility. However,
little is known about how temperature anomalies may affect disease severity over broad
spatial scales. We hypothesized that the frequency of warm temperature anomalies increased
the frequency of white syndrome, a common scleractinian disease in the Indo-Pacific. We
created a novel 4 km satellite temperature anomaly dataset using data from NOAA’s
Pathfinder program and developed four different temperature anomaly metrics, which we
correlated with white syndrome frequency at 47 reefs spread across 1500 km of the Great
Barrier Reef. This cross-sectional epidemiological analysis used data from disease field
surveys conducted by the Australian Institute of Marine Science six to twelve months after
the summer of 2002, a year of extensive coral bleaching. We found a highly significant
positive relationship between the frequency of warm temperature anomalies and the
frequency of white syndrome. There was also a highly significant, nearly exponential
relationship between total coral cover and the number of disease cases. Further, coral cover
modified the effect of temperature on disease frequency. Both high coral cover ( > 50%) and
anomalously warm water appear to be necessary for white syndrome outbreaks to occur and
these two risk factors explained nearly 75% of the variance in disease cases. These results
suggest that rising ocean temperatures could exacerbate the effects of infectious diseases on
coral reef ecosystems.
Introduction
Over the last four decades, coral cover has declined dramatically on reefs worldwide
(Gardner et al. 2003; Bellwood et al. 2004). Several factors are thought to be responsible for
this decline including overfishing (Jackson 1997; Pandolfi et al. 2003), terrestrial run-off
(Fabricius 2005), climate change (Hoegh-Guldberg 1999; Hughes et al. 2003), and infectious
disease (Aronson & Precht 2001). There is growing recognition that we need to focus on
possible synergisms among these and other stressors (Hughes & Connell 1999; Lenihan et al.
1999). In the last several years, the relationship between climate conditions and disease has
received more attention as researchers have connected factors such as temperature and
precipitation with increases in human and wildlife diseases (Pascual et al. 2000; Patz 2002;
Kutz et al. 2005; Pounds et al. 2006). Yet, few studies have focused on the effects of climate
change on diseases in the ocean, particularly at broad spatial scales.
Disease has already had significant effects on coral reef ecosystems. Several diseases
have altered the landscape of Caribbean reefs, causing the near extirpation of the keystone
herbivore Diadema antillarum (Lessios 1988) and dramatic losses of Acropora cervicornis
and Acropora palmata (Aronson & Precht 2001; Aronson et al. 2002). These disease
outbreaks mediated a shift from coral- to algal-dominated communities in the Caribbean
53
(Aronson & Precht 2001). The scale and severity of coral loss on many Caribbean reefs is
unprecedented in the paleontological record (Aronson & Precht 2001; Wapnick et al. 2004)
of many reefs and indicative of the emergence of a novel stressor (Aronson & Precht 2001).
Recent studies quantifying both disease reports (Ward & Lafferty 2004) and the number of
described coral diseases (Sutherland et al. 2004) suggest that the frequency of marine
diseases is increasing and many of these diseases are the result of previously unknown
pathogens. Reports from the Pacific indicate that diseases of reef-building corals may be far
more widespread than previously believed (Sutherland et al. 2004; Willis et al. 2004; Aeby
2005; Raymundo et al. 2005). Although coral disease is likely underreported in the Pacific
due to a lack of disease research, increases in coral disease cases have been detected since the
late 1990s (Willis et al. 2004). These lines of evidence provide strong support for the
hypothesis that there has been a real increase in the number of coral disease reports over the
last three decades (Harvell et al. 1999; Ward & Lafferty 2004).
The causes underlying recent increases in coral disease outbreaks are complex and
poorly understood, in part because of a paucity of knowledge about the identity and sources
of most coral pathogens (Sutherland et al. 2004). Pathogens have a variety of purported
vectors and reservoirs including algae (Nugues et al. 2004), invertebrates (Rosenberg &
Falkovitz 2004; Williams & Miller 2005), sediment transported from the Sahel region of
northern Africa (Shinn et al. 2000), and sewage effluent (Patterson et al. 2002). In addition,
some diseases are hypothesized to be associated with changes in corals’ microbial
communities, rather than the result of an external infectious pathogen.
Abiotic factors like nutrients and temperature can exacerbate disease severity, but the
scale of their effects may vary. For example, nutrients increased the severity of two coral
54
diseases in experimental manipulations (Bruno et al. 2003). However, inputs of terrestrial
pollution, including nutrients, do not always result in elevated disease levels on affected reefs
(Weil 2004; Willis et al. 2004). Although localized inputs of nutrients may play some role in
disease outbreaks, the regional scale of most outbreaks (Aronson & Precht 2001; Kim &
Harvell 2004) indicates that a climatic variable like temperature may be a critical driver of
disease dynamics. For example, disease prevalence, or the proportion of the total population
that is diseased, may be related to the frequency and magnitude of warm temperature
anomalies. Extensive work clearly links these anomalies to coral bleaching events (Glynn et
al. 1988; Glynn & D'croz 1990; Hoegh-Guldberg 1999; Bruno et al. 2001; Fitt et al. 2001;
Liu et al. 2003; Strong et al. 2004). Since bleaching is a sign of physiological stress in corals
(Glynn 1993; Brown 1997), it is expected that bleached or thermally-stressed corals would be
more susceptible to opportunistic and residential pathogens (Hayes et al. 2001; Rosenberg &
Ben-Haim 2002).
Forecast sea surface temperature (SST) models predict that the frequency and severity
of warm temperature anomalies will increase with climate change (Hoegh-Guldberg 1999;
Sheppard 2003; Sheppard & Rioja-Nieto 2005). Ocean temperature has already increased,
on average, 0.4-0.8ºC (Folland et al. 2001) from 1861 to 2000, with some regional variation
(Casey & Cornillon 2001). These anomalies and the general warming of the ocean could
have several effects on disease. One possible outcome of global warming is that the summer
“disease season” may become more severe, as summer temperature maxima increase, and
longer, as these elevated thermal regimes start earlier and persist later in the season.
Seasonal variability in coral disease abundance has been found in multiple field studies in
different regions. In the Caribbean, several coral diseases, including black band (Edmunds
55
1991; Kuta & Richardson 2002), white pox (Patterson et al. 2002), and dark spots disease
(Gil-Agudelo & Garzon-Ferreira 2001), are more prevalent or spread across colonies more
rapidly during summertime than during cooler seasons. On the Great Barrier Reef (GBR),
white syndrome frequency was greater in summer than winter on surveyed reefs (Willis et al.
2004). Similarly, in the summer of 2002 on the GBR, Jones et al. (2004) documented a
localized outbreak of atrementous necrosis. In addition, shorter or warmer winters may
release some infectious diseases from the low temperature control that provides hosts with a
seasonal escape from disease (Harvell et al. 2002). Climate warming has also been predicted
to alter the geographic distributions of infectious diseases by shifting host and pathogen
latitudinal ranges pole-ward (Marcogliese 2001; Harvell et al. 2002).
Testing the hypothesis that the frequency or intensity of temperature anomalies can
influence coral disease dynamics requires high quality data on temperature and disease
frequency. Here we discuss how analyses of these broad scale questions are now possible
using a newly developed high-resolution satellite temperature anomaly dataset. We then
present a case study where we examine the effects of the warm temperature anomalies on
white syndrome, an emergent sign of disease on GBR corals. We also discuss future
research directions for testing the relationship between temperature and disease.
Using remote sensing data to explore the climate warming disease outbreak hypothesis
Most documented coral disease outbreaks have been at the scale of ocean basins
(Lessios 1988; Aronson & Precht 2001; Willis et al. 2004). Previously, correlating these
outbreaks with ocean temperature was complicated by a scale mismatch. Temperature has
56
typically been measured with relatively high accuracy at very fine scales or with less
accuracy at coarse scales. In situ temperature loggers can be highly effective at capturing
small-scale variability (1-100 m) (Leichter & Miller 1999; Castillo & Helmuth 2005;
Leichter et al. 2005), but are limited in their spatial extent. On the other hand, the 50 km
HotSpot mapping (Fig. 2.1) (Strong et al. 2004), although effective for predicting bleaching
at broad scales (Bruno et al. 2001; Berkelmans et al. 2004; Strong et al. 2004), is too coarse
to accurately represent the temperature of waters surrounding many reefs and may not
sufficiently capture local variability (Toscano et al. 2000). Therefore, the development of
new, higher resolution remote sensing products is required to detect correlations between
temperature anomalies and disease outbreaks.
We developed a novel satellite temperature anomaly product to investigate the
relationship between sea surface temperature and disease at broad spatial scales. Our dataset
used the 4 km Advanced Very High Resolution Radiometer (AVHRR) Pathfinder Version
5.0 SST dataset produced by NOAA’s National Oceanographic Data Center and the
University of Miami’s Rosenstiel School of Marine and Atmospheric Science
(http://pathfinder.nodc.noaa.gov). These data now provide the longest sea surface
temperature record at the highest resolution of any global satellite dataset. The Pathfinder
Version 4.2 dataset had a 9 km resolution (Kilpatrick et al. 2001), a marked improvement
from the 50 km data. However, inaccuracies in the land mask, which defines the extent of
the data, resulted in coverage of only 60% of reef areas (Fig. 2.1). Using a more refined
climatology and more accurate coefficients for generating temperatures, the 4 km Pathfinder
Version 5.0 dataset reprocessed the full AVHRR record from 1985 through 2005 to remove
inaccuracies. With these improvements, temperature records are now available for more than
57
98% of reefs worldwide. Appropriate use of satellite temperature data requires validation
with in situ loggers (Reynolds et al. 2002). In theory, an infrared observing satellite like
AVHRR only measures an integrated temperature over approximately the top 10
micrometers of the ocean surface. Therefore, we validated the assumption that the 4 km
Pathfinder Version 5.0 SST values reflect temperatures on shallow reefs, where the majority
of reef-building corals are found, by comparing satellite temperature estimates to in situ
temperature logger data collected by the Australian Institute of Marine Science
(www.reeffutures.org) on nine shallow reefs on the GBR (5 m to 9 m depth). We fit a linear
multilevel model and found a highly significant relationship (p < 0.0001) with a common
value for the slope of all the reefs of 0.96 (SE = 0.017). We found the difference between
satellite-derived and benthic temperatures was generally less than the 0.2ºC error of most in
situ loggers. We also ran linear regression analyses for each reef to investigate the generality
of this relationship at all nine reefs (Table 2.1) and found that the satellite temperature
measurements were a very good predictor of benthic water temperature.
One key advantage of satellite data is their ability to provide a long-term temporal
record of temperature values, which enables users to create accurate climatologies, or
average long-term patterns (Fig. 2.2). These climatologies are then used as the basis for
calculating deviations from typical weekly, monthly, seasonal, or annual temperatures. For
disease analyses, relevant data include the frequency of deviations or temperature anomalies,
the duration of anomaly events, occurrence of wintertime anomalies, and the geographic
extent of the anomaly.
58
Case study on the Great Barrier Reef
The Great Barrier Reef (GBR) is the largest barrier reef system in the world and one of the
most highly managed, with more than a third of reef area in marine reserves, or no-take areas
(Fernandes et al. 2005). Recent surveys suggest that disease is more prevalent than
previously believed on the GBR (Willis et al. 2004). Black band, white syndrome, brown
band, and skeletal eroding band have all been reported on the GBR (Willis et al. 2004). In
spite of the well-documented relationship with increased temperatures and black band
disease in the Caribbean (Edmunds 1991; Kuta & Richardson 2002), it did not vary in
frequency over the course of our study (Willis et al. 2004). Instead we focused our analysis
on white syndrome, which increased dramatically in 2002-2003 on some reefs. We tested
specific hypotheses about how temperature might affect disease frequency by using an
information-theoretic approach to evaluate different thermal stress metrics.
Disease surveys
The disease data used for this analysis were collected by the Australian Institute of
Marine Science’s Long-term Monitoring Program (AIMS LTMP) during their 2002-2003
surveys. Surveys were performed at 47 reefs in a stratified design. Five permanent 50 m x 2
m belt transects were surveyed at three sites on each reef at approximately 6-9 m depth (Full
methods in Sweatman et al. 2003). For each transect, the number of colonies infected with
white syndrome was quantified. Reefs were grouped by latitudinal sectors, which together
cover more than 1500 km and a variety of different oceanographic regimes (Fig. 2.3). White
syndrome is characterized by a band of recently-exposed white skeleton, sometimes preceded
by a band of bleached tissue at the tissue-skeleton interface. The white band moves across
59
the colony as the front of tissue mortality progresses, potentially resulting in mortality of the
whole colony (Fig. 2.4). White syndrome has been recorded on at least 17 species from 4
families including Acroporidae, Pocilloporidae and Faviidae, which together constitute a
significant percentage of overall coral cover on the GBR (Willis et al. 2004). White
syndrome is similar to Caribbean white diseases such as white band I, white band II, white
plague I, and white plague II in its disease signs. Because the pathogen(s) that cause white
disease in the Pacific are not known, we cannot state with certainty that the syndrome does
not represent more than one distinct disease (Willis et al. 2004) or that the underlying
etiology is infectious.
Temperature data
Consistent with previous analyses of satellite data in coral studies, we used nighttime
weekly temperature averages (Liu et al. 2003; Strong et al. 2004). Nighttime daily averaged
data had too many gaps and would have required extensive interpolation. Although they may
not capture short duration events, weekly data provide substantially more continuity in the
record and still represent a time scale short enough to capture most thermal stress events that
negatively impact corals (Glynn 1993; Podesta & Glynn 2001). To measure anomalies, we
first calculated weekly climatologies, or the mean values at each calendar week from 19852004 for each 4 x 4 km pixel (Fig. 2.2). Missing data in the climatologies were interpolated
using a Piecewise Cubic Hermite Interpolating Polynomial (PCHIP) function in MATLAB
(The Mathworks Inc. 2005). The climatologies were then smoothed using a five-week
running mean to minimize unusual fluctuations from periods of limited data availability.
60
Gaps in weekly temperature observations were also interpolated using the PCHIP function
without modifying the original data.
Thermal stress metrics
To test the hypothesis that temperature affects disease frequency, we first designed
temperature metrics relevant to disease. Infectious diseases are interactions between hosts
and pathogens. Temperature can increase host susceptibility, but it can also increase
pathogen growth rate, transmission rate, and over-wintering survival (Harvell et al. 2002).
Because of the potential complexity of this relationship and the paucity of data about most
pathogens, no specific algorithm exists for disease prediction. However, data on how corals
respond to stress and epidemiological theory provide a general guide to temperature
thresholds that may be applicable to disease dynamics. We developed a series of temperature
metrics for our case study based on metrics known to have a physiological effect on coral
health. Three of our four thermal stress metrics are based on an anomaly threshold of 1°C,
because this is widely assumed to estimate the point at which a warm temperature anomaly
induces a measurable physiological stress in a coral host, (Glynn 1993; Winter et al. 1998;
Hoegh-Guldberg 1999; Berkelmans 2002)and in general, increases of ≥1°C above normal
summertime temperatures are thought to induce bleaching (Glynn 1993, 1996; Winter et al.
1998; Hoegh-Guldberg 1999; Berkelmans 2002). All of the metrics measured deviations
from the location-specific climatologies we created. Because disease surveys were
conducted at different times of year (November-March), we standardized all of our metrics to
include the number of anomalies during the 52 weeks prior to each disease survey. We then
61
used the latitude and longitude of each reef to match it with its corresponding pixel in the
satellite data.
We developed each temperature metric to investigate a specific hypothesis related to
the relationship between thermal stress and disease. The first three metrics are locationspecific and assume that corals are acclimated to the thermal regime at their location. Work
by Berkelmans and Willis (1999) suggests that corals exhibit some degree of local
acclimation or adaptation. The fourth metric assumes that temperature is affecting disease
rates at regional scales. If thermal anomalies are acting on the pathogen itself, by increasing
growth or reproductive rates, they would likely be acting at a regional scale due to high
pathogen mobility (McCallum et al. 2003).
Metric descriptions
1) WSSTA = Weekly Sea Surface Temperature Anomalies = deviations of 1°C or greater from
the mean climatology during a particular week at a given location from 1985-2003. This
metric is designed to be both location and season-specific by determining whether the
temperature is unusual for that location at a particular week of the year.
2) TSA= Thermal Stress Anomalies = deviations of 1°C or greater from the mean maximum
climatological weekly temperature from 1985-2003. The mean maximum climatological
week is the warmest week of the 52 weekly climatologies. This metric is also site-specific
but designed to detect deviations from typical summertime highs and is similar to metrics
used by the Coral Reef Watch program (Liu et al. 2003; Strong et al. 2004).
62
3) LTSA = Local Temperature Stress Anomalies = deviations from the upper 2.5% of all
weekly measurements taken at that location from 1985-2003. This metric is site-specific. It
is designed to detect whether temperatures are unusual based on the distribution and
extremes of all measured temperatures, regardless of calendar week.
4) RTSA = Regional Thermal Stress Anomalies = deviations from the upper 2.5% of all
weekly measurements taken at all reefs. For this metric, local temperatures are compared to
regional average values.
Metric selection and statistical analysis
We evaluated the different thermal stress metrics (Table 2.2) using Akaike
Information Criterion (AIC) (Akaike 1973). Identifying which thermal stress metric best
explains the relationship between temperature and disease can facilitate the development of
an appropriate model for a specific coral-pathogen system. Different diseases may be
affected by different temperature characteristics. For example, TSAs may be highly
correlated with an increase of one disease while another disease may be more correlated with
WSSTAs. AIC is an estimate of expected relative Kullback-Leibler information, an
information-theoretic measure of the distance between models where smaller AIC values
should be preferred. In practice, only the relative differences in AIC between models are
meaningful. When these differences are normalized as Akaike weights, (wi), they can also
be given a probabilistic interpretation. Each wi is the weight of evidence favoring model i as
the best model among the models under consideration (Anderson et al. 2000; Anderson &
Burnham 2002).
63
Coral cover is clearly related to the number of white syndrome cases on a reef (Fig.
2.5A) and was also included in the analysis. We used multiple, nonlinear regression analysis
(StataCorp LP 2006) to analyze the relationship between the two independent variables,
thermal stress and coral cover, and the dependent variable, the total number of white
syndrome cases at each reef (i.e., the number/1500m2). We used a Poisson model and a
quadratic function for the thermal stress metrics and a linear function for coral cover.
Results
The austral summer in 2002 triggered the most extensive bleaching event ever
documented on the GBR (Berkelmans et al. 2004). Surveys conducted after this summer
found a major increase in disease frequency with some sampled reef areas having more than
300 cases of disease (Willis et al. 2004). AICC, a small-sample (second order) bias-adjusted
variant of AIC (Burnham & Anderson 2002), indicated that the number of WSSTAs provided
the best model fit (Table 2.2) and was used in the main analysis as the thermal stress metric.
The whole Poisson regression model (r2 = 0.73) and both main effects (WSSTAs and coral
cover) were all highly statistically significant at the reef scale (all P < 0.0001, n = 47). There
was also a highly significant interaction between coral cover and WSSTAs (P < 0.0001). At
low and intermediate cover (0-40%), the number of white syndrome cases was greatest when
annual WSSTAs were 4-6 and declined slightly with increasing WSSTA (Fig. 2.6).
However, when cover was high (> 41%), white syndrome frequency was more than 3 times
greater when the frequency of WSSTAs was > 7 than when it was 4-6 (Fig. 2.6).
64
Discussion
Temperature anomalies are known to be the underlying cause of mass coral bleaching
(Podesta & Glynn 2001; Liu et al. 2003; Berkelmans et al. 2004; Strong et al. 2004), but
their relationship with infectious disease dynamics is not well understood. Our results
suggest that warm temperature anomalies can significantly affect the frequency of white
syndrome on the GBR, especially where coral cover is high. Total coral cover was clearly
related to the number of white syndrome cases on a reef (Fig. 2.5A) (Willis et al. 2004).
With few exceptions, white syndrome frequency was relatively low (< 30 cases/1500 m2, the
area surveyed at each reef) when coral cover was < 50% (Fig. 2.5A) and was very high, 192
cases ± 46 (mean ± 1 SE) at the eight reefs where coral cover was > 50%. The nearly
exponential relationship between coral cover and disease cases (Fig. 2.5A) suggests there is a
threshold coral cover of approximately 50% that is generally required to for an outbreak of
white syndrome to occur. This could be due to higher host densities on high coral cover
reefs. Host density is widely known to influence disease dynamics (Anderson & May 1986).
White syndrome has been documented in the major coral families on the GBR, including the
abundant staghorn and tabular species of Acropora (Willis et al. 2004). Therefore, total coral
cover is likely to be directly related to white syndrome host cover. However, the relationship
between host density and prevalence is not always clear for infectious coral diseases such as
sea fan aspergillosis (Kim & Harvell 2004) possibly because secondary transmission is rare
among host colonies (Edmunds 1991). Additionally, several other aspects of total coral
cover could also influence the dynamics of white syndrome and other infectious coral
diseases. For example, coral cover could be positively related to animal vectors (i.e. coral
predators and mutualists), which could increase rates secondary transmission, although Willis
65
et al. (2004) found that the density of the corallivorous snail Drupella spp. was unrelated to
white syndrome frequency on the GBR.
Temperature measured as WSSTA frequency also had a strong effect on the number
of disease cases (Fig. 2.5B). The selection of the WSSTA metric as the best model using an
AIC approach suggests that the effect of thermal stress on white syndrome is both seasonand location-specific. The WSSTA metric is the only metric we tested that incorporates
temperature anomalies throughout the year. These findings are consistent with coral
physiology studies, which have found that at the beginning of cooler months, zooxanthellae
densities increase and coral tissue is built up (Brown et al. 1999). With warmer or longer
than usual summers or warmer winters these accumulations may not occur, increasing corals’
vulnerability to future stress (Fitt et al. 2001). Higher WSSTA frequencies may lead to
chronic stress, which could increase host susceptibility and disease prevalence. However,
WSSTAs could also have influenced disease frequency by increasing pathogen virulence.
The relationship between WSSTA and white syndrome cases (Fig. 2.5B) is also suggestive of
a weak threshold response when the number of annual WSSTAs exceeds approximately
seven. There were several reefs with > 7 WSSTAs and a low number of cases (< 10),
however, all were low coral cover reefs (mean cover 15.1 ± 3.0, n = 7). In fact, the only reefs
with > 200 cases had > 50% cover and a WSSTA frequency > 7 (Fig. 2.5B), suggesting that
both conditions are necessary for white syndrome outbreaks to occur. Furthermore, these
two risk factors explained nearly 75% of the variance in disease cases.
White syndrome frequency varied substantially among the six sectors, possibly due in
part to regional temperature variation. Average WSSTA frequency within sectors was
positively related to average disease frequency, largely because Capricorn Bunkers, the
66
sector with by far the greatest number of cases also had the greatest number of WSSTAs
(Fig. 2.7) and the highest average coral cover (Willis et al. 2004). Although other studies
have found that higher nutrient levels are associated with increased disease severity (Kuta &
Richardson 2002; Bruno et al. 2003), Willis et al. (2004) found that outer shelf reefs had
higher levels of white syndrome than inner shelf reefs. Assuming that distance from shore is
a proxy for nutrient availability, these findings suggest that nutrient and sediment input may
not be primary contributors to increases in white syndrome (Willis et al. 2004). In fact,
several inner shelf reefs relatively close to shore in the Cairns sector had high WSSTA
frequencies but very few disease cases in 2002, possibly because host density was also low or
because other abiotic or biotic conditions inhibited white syndrome (Willis et al. 2004).
The importance of different abiotic and biotic factors driving white syndrome and
other diseases is likely to vary with scale and the host-pathogen system (Bruno et al. 2003).
Ocean currents may facilitate spread or isolation of different pathogens (McCallum et al.
2003), but no empirical studies have yet quantified potential dispersal patterns. Biotic factors
like host age or size structure are also likely to have significant effects on disease prevalence
(Anderson & May 1986; Dube et al. 2002; Lafferty & Gerber 2002). Older or larger hosts
may be more vulnerable to disease (Dube et al. 2002; Borger & Steiner 2005), which could
be particularly devastating for coral populations where older or larger individuals are likely
to have higher reproductive output (Hall & Hughes 1996; Sakai 1998; Dube et al. 2002).
Uncertainty about the identity and source of most pathogens represents a major
challenge in understanding the factors that determine pathogen survival and development.
Of the currently described coral diseases, only 5 of 18 (Sutherland et al. 2004) have been
identified through fulfillment of Koch’s postulates, which require a putative pathogen to be
67
1) isolated from a diseased individual, 2) grown in pure culture, and 3) transferred to a
healthy organism where it induces the disease state (Koch 1882). An additional postulate,
that the pathogen be reisolated from the infected organism, was not formulated by Koch, but
is also typically recommended (Fredricks & Relman 1996; Richardson 1998; Sutherland et
al. 2004). Fulfilling Koch’s postulates for coral pathogens has been challenging, in part due
to the complex nature of the host-pathogen relationship, the possibility of multiple disease
agents, and the difficulty of natural inoculation (Richardson 1998; Sutherland et al. 2004).
Identifying pathogens through Koch’s postulates is not essential for epidemiological study
(Fredricks & Relman 1996), but without isolating pathogens, it has been difficult to
determine the mechanisms behind disease dynamics. For example, with a known, cultured
pathogen, manipulative experiments on both the isolated pathogen and the host-pathogen
system could yield insights into whether temperature is affecting disease by increasing
expression of pathogen virulence factors, increasing pathogen growth or reproductive rates,
or increasing host susceptibility (Harvell et al. 1999; Harvell et al. 2002).
Although the mechanisms are not known, this study found a strong correlative
relationship between white syndrome and warm temperature anomalies. This relationship
could have several implications for coral communities on the GBR. White syndrome affects
key reef-building species on the GBR including the competitively dominant tabular acroporid
corals that constitute a substantial percentage of total coral cover (Baird & Hughes 2000;
Connell et al. 2004). Reductions in the abundance of these corals could cause shifts in
species assemblages or abundances (Baird & Hughes 2000). In the Caribbean, loss of
acroporid corals due to white band disease has led to a shift in dominance to Agaricia sp. on
some reefs (Aronson et al. 2002) and precipitated a shift to macroalgal dominance on others.
68
Predicted increases in the frequency and severity of thermal stress anomalies with global
climate change (Hoegh-Guldberg 1999) could exacerbate these kinds of disease effects. Our
results and the balance of the published evidence from field studies comparing coral disease
prevalence among seasons (Edmunds 1991), years (Willis et al. 2004) and sites (Kuta &
Richardson 2002) strongly suggest that water temperature plays a substantial role in coral
disease dynamics. Because reef building corals are irreplaceable as marine foundation
species (Bruno & Bertness 2001), the synergism between temperature and disease could have
cascading effects throughout reef ecosystems (Bruno et al. 2003).
Future research directions
The development of 21 years of consistently processed satellite sea surface
temperature and anomaly data for the GBR region represents a meaningful advancement in
understanding the effects of temperature on several parameters of coral health including
disease. Further validation of the satellite estimates will provide a better understanding of its
limitations and enable more productive and effective use of the dataset. For example, in
areas with persistent cloud cover, low data availability may decrease accuracy (Kilpatrick et
al. 2001). The relationship of the satellite-measured surface temperatures to temperatures at
different depths will depend on bottom topography and the presence of oceanographic
features like internal tidal bores (Leichter & Miller 1999), which can alter the temperature
regime experienced by corals. Assessing patterns of frequency, extent, and intensity of
anomalies over the full time period of the dataset could also identify areas that are more or
less vulnerable to thermal stress. The presence or absence of these correlations could help
69
determine whether marine protected areas can be used to protect areas of greater resilience or
resistance to thermal stress (West & Salm 2003; Obura 2005; Wooldridge et al. 2005).
To facilitate regional scale investigations of coral disease risk factors and dynamics,
refinement of remote sensing tools must be complemented with rigorous disease monitoring
protocols. Much of the current monitoring is idiosyncratic, often in response to a disease
outbreak with little baseline or long-term monitoring data (but see Kim & Harvell 2004;
Willis et al. 2004; Santavy et al. 2005). Long-term longitudinal and cross-sectional
epidemiological studies are an essential component of elucidating density-dependence in
disease dynamics, susceptible age classes, possible pathogens or modes of transmission, and
potential effects on reproductive output and population dynamics. Monitoring programs
intended for use in conjunction with satellite temperature data should sample from different 4
km grid cells as defined by the Pathfinder dataset so that there is adequate replication for
analysis. In addition, surveys should be conducted within a relatively close time frame so
that the data do not covary with other temporal patterns. Finally, manipulative laboratory
experiments are also an essential complement to these correlative studies to identify
mechanisms driving correlations between temperature and rates of infection and spread.
Conclusions
Until very recently it was impossible to correlate satellite-derived temperature
anomalies with in situ disease surveys. Temporally-consistent satellite temperature data at 4
km resolution enabled us to observe a positive relationship between temperature anomalies
and outbreaks of white syndrome at broad spatial scales. A continuing challenge in assessing
70
the importance of temperature anomalies as factors in disease dynamics will be to measure
the role of other variables such as proximity to pathogen sources, local water quality, and
physical oceanographic patterns. In some cases, these variables could be more important risk
factors than temperature. A priority for coral reef scientists is the implementation of disease
surveys on a global scale to adequately test the hypotheses that temperature anomalies and
other factors drive coral disease. It is likely that there will be different thermal stress
thresholds in different biogeographic regions for each disease. Work is underway to describe
disease syndromes worldwide and develop a repeatable method for surveying coral disease
(Willis et al. 2004) that can be implemented globally. In combination with satellite
temperature data and other data sources, we can then begin to understand disease dynamics at
regional scales.
71
Figure 2.1 Developing the 4 km temperature dataset. An example of the benefits
of increasing resolution on the Yucatan peninsula and Caribbean Sea from; (A) 50
km HotSpot data to (B) 9 km Pathfinder data to (C) 4 km Pathfinder data. In the 4
km data, there is less missing data, allowing for greater coverage of coastal areas
where many reefs occur. In addition, the 4 km data displays more spatial structure
and precision in the temperature values. Data are from January climatological
averages, monthly for the 50 km data and from the first week of January for the 9 km
and 4 km data.
72
Figure 2.2 Mean sea surface temperature (ºC) from 1985-2003. Averages from
this time period were used to create weekly climatologies for each surveyed reef.
73
Figure 2.3 Number of white syndrome cases in 2002-2003. These data were
collected by the Australian Institute of Marine Science’s Long-term Monitoring
Program. Each of the six latitudinal areas or sectors are labeled with their names
and abbreviations. Cooktown / Lizard Island (CL) has nine surveyed reefs, Cairns
(CA) has ten, Townsville (TO) has eight, Whitsundays (WH) has nine reefs, Swains
(SW) has seven, and Capricorn Bunkers (CB) has four. For each reef, three
different sites were surveyed. Disease cases were highest in the Cooktown/Lizard
Island and Capricorn Bunkers sectors.
74
Figure 2.4 Example of white syndrome spreading across Acropora cytherea.
White areas are coral skeleton that have been recently exposed following die-off
behind the disease front.
75
Figure 2.5 Relationships between the number of white syndrome cases and
(A) total percent coral cover (B) and the number of WSSTAs. Solid points in B
represent reefs with > 50% coral cover. Each point represents the values from a
single reef (n=47).
White syndrome cases
400
B
A
300
200
100
0
0
20
40
60
80 0
Percent total coral cover
5
10
WSSTAs
76
15
Figure 2.6 The effect of total coral cover and the number of WSSTAs on the
number of white syndrome cases. Values are mean ±1 SE. Values above error
400
WSSTAs
5
Low (0-3)
300
Medium (4-6)
200
High (>7)
4
6
4
4
0
9
5
3
Medium (21-40)
7
High (>41)
100
Low (0-20)
White syndrome cases
bars are the number of surveyed reefs in each of the nine categories.
Percent total coral cover
77
Figure 2.7 Relationship between mean thermal stress (WSSTAs) and the mean
number of white syndrome cases at the sector scale. See Figure 2.3 for sector
delineations and abbreviations. Values are mean ±1 SE.
White syndrome cases
350
300
CB
250
200
150
100
CL CA
50
SW
TO WH
0
0
3
6
9
12
WSSTAs
78
Table 2.1 Relationship between weekly averaged satellite and in situ
temperatures at nine reefs on the GBR reefs at 5-9 m depth. Field data were
collected by the Australian Institute of Marine Science in cooperation with CRC Reef
Research Centre and the Great Barrier Reef Marine Park Authority.
Reef
Latitude
Longitude
Period
n
P
R2
Agincourt
-16.0384
145.8688
1996-2004
173
p<.0001
0.91
Chicken
-18.6521
147.7217
1996-2004
244
p<.0001
0.94
Davies
-18.8060
147.6686
1996-2004
200
p<.0001
0.96
Dip
-18.3999
147.4519
1997-2004
197
p<.0001
0.92
East Cay
-21.4698
152.5665
1995-2004
248
p<.0001
0.96
John Brewer
-18.6188
147.0815
1996-2004
239
p<.0001
0.94
Lizard Island
-14.6915
145.4692
1996-2004
134
p<.0001
0.90
Myrmidon
-18.2572
147.3813
1995-2004
205
p<.0001
0.93
Turner Cay
-21.7031
152.5601
1997-2004
204
p<.0001
0.96
79
Table 2.2. Akaike Information Criterion for thermal stress metrics. For each
thermal metric we calculated: the AIC, a measure of expected relative Kullback
Leibler information; the AICC, AIC corrected for small sample size; the difference, Δ i
between each model and the best model in the set; and the Akaike weight, wi . The
sum of all the Akaike weights is equal to 1.0 and each weight represents the
approximate likelihood that a specific model is the best model of those compared, in
a Kullback-Leibler information sense. For the models that involve a transformed
response, likelihoods were adjusted with the Jacobian of the transformation to make
the AICC values comparable.
AICc
Δi
wi
Metric
AIC
WSSTA
1340
1342
0
1
TSA
1464
1465
123
0
LTSA
1479
1481
138
0
RTSA
1476
1477
135
0
80
Acknowledgements
We thank A. Alker, A. Barton, K. France, S. Lee, A. Melendy Bruno, S. Neale, M.
O’Connor, N. O’Connor, L. Stearns, all past and present members of the Australian Institute
of Marine Science’s Long-term Monitoring Program involved in collecting the disease data,
and two anonymous reviewers for their comments. This research was funded in part by the
National Science Foundation (OCE-0326705 to J.F.B), an EPA STAR Fellowship to E.R.S,
the NOAA Coral Reef Conservation Program and its NESDIS Coral Reef Watch project, and
the University of North Carolina at Chapel Hill.
81
References
Aeby G.S. (2005). Outbreak of coral disease in the Northwestern Hawaiian Islands. Coral
Reefs, 24, 481-481.
Akaike H. (1973). Information theory and an extension of the maximum likelihood principle.
In: Second International Symposium on Information Theory (eds. Petrov BN & Csaki
F). Akademiai Kaido Budapest, pp. 267-281.
Anderson D.R., Burnham K.P. & Thompson W.L. (2000). Null hypothesis testing: Problems,
prevalence, and an alternative. Journal of Wildlife Management, 64, 912-923.
Anderson D.R. & Burnham K.R. (2002). Avoiding pitfalls when using information-theoretic
methods. Journal of Wildlife Management, 66, 912-918.
Anderson R.M. & May R.M. (1986). The invasion, persistence and spread of infectious
diseases within animal and plant communities. Philosophical Transactions of the
Royal Society of London Series B-Biological Sciences, 314, 533-570.
Aronson R.B., MacIntyre I.G., Precht W.F., Murdoch T.J.T. & Wapnick C.M. (2002). The
expanding scale of species turnover events on coral reefs in Belize. Ecological
Monographs, 72, 233-249.
Aronson R.B. & Precht W.F. (2001). White-band disease and the changing face of Caribbean
coral reefs. Hydrobiologia, 460, 25-38.
Baird A.H. & Hughes T.P. (2000). Competitive dominance by tabular corals: an
experimental analysis of recruitment and survival of understorey assemblages.
Journal of Experimental Marine Biology and Ecology, 251, 117-132.
Bellwood D.R., Hughes T.P., Folke C. & Nystrom M. (2004). Confronting the coral reef
crisis. Nature, 429, 827-833.
Berkelmans R. (2002). Time-integrated thermal bleaching thresholds of reefs and their
variation on the Great Barrier Reef. Marine Ecology Progress Series, 229, 73-82.
Berkelmans R., De'ath G., Kininmonth S. & Skirving W.J. (2004). A comparison of the 1998
and 2002 coral bleaching events on the Great Barrier Reef: spatial correlation,
patterns, and predictions. Coral Reefs, 23, 74-83.
Berkelmans R. & Willis B.L. (1999). Seasonal and local spatial patterns in the upper thermal
limits of corals on the inshore central Great Barrier Reef. Coral Reefs, 18, 219-228.
Borger J.L. & Steiner S.C.C. (2005). The spatial and temporal dynamics of coral diseases in
Dominica, West Indies. Bulletin of Marine Science, 77, 137-154.
82
Brown B.E. (1997). Coral bleaching: causes and consequences. Coral Reefs, 16, S129-S138.
Brown B.E., Dunne R.P., Ambarsari I., Le Tissier M.D.A. & Satapoomin U. (1999).
Seasonal fluctuations in environmental factors and variations in symbiotic algae and
chlorophyll pigments in four Indo-Pacific coral species. Marine Ecology Progress
Series, 191, 53-69.
Bruno J.F. & Bertness M.D. (2001). Habitat modification and facilitation in benthic marine
communities. In: Marine Community Ecology (eds. Bertness MD, Hay ME & Gaines
SD). Sinauer Sunderland, MA, pp. 201-218.
Bruno J.F., Petes L.E., Harvell C.D. & Hettinger A. (2003). Nutrient enrichment can increase
the severity of coral diseases. Ecology Letters, 6, 1056-1061.
Bruno J.F., Siddon C.E., Witman J.D., Colin P.L. & Toscano M.A. (2001). El Nino related
coral bleaching in Palau, Western Caroline Islands. Coral Reefs, 20, 127-136.
Burnham K.P. & Anderson D.R. (2002). Model Selection and Multimodel Inference.
Springer-Verlag, New York.
Casey K.S. & Cornillon P. (2001). Global and regional sea surface temperature trends.
Journal of Climate, 14, 3801-3818.
Castillo K.D. & Helmuth B.S.T. (2005). Influence of thermal history on the response of
Montastraea annularis to short-term temperature exposure. Marine Biology, 148, 261270.
Connell J.H., Hughes T.E., Wallace C.C., Tanner J.E., Harms K.E. & Kerr A.M. (2004). A
long-term study of competition and diversity of corals. Ecological Monographs, 74,
179-210.
Dube D., Kim K., Alker A.P. & Harvell C.D. (2002). Size structure and geographic variation
in chemical resistance of sea fan corals Gorgonia ventalina to a fungal pathogen.
Marine Ecology Progress Series, 231, 139-150.
Edmunds P.J. (1991). Extent and effect of black band disease on a Caribbean reef. Coral
Reefs, 10, 161-165.
Fabricius K.E. (2005). Effects of terrestrial runoff on the ecology of corals and coral reefs:
review and synthesis. Marine Pollution Bulletin, 50, 125-146.
Fernandes L., Day J., Lewis A., Slegers S., Kerrigan B., Breen D., Cameron D., Jago B., Hall
J., Lowe D., Innes J., Tanzer J., Chadwick V., Thompson L., Gorman K., Simmons
M., Barnett B., Sampson K., De'ath G., Mapstone B., Marsh H., Possingham H., Ball
I., Ward T., Dobbs K., Aumend J., Slater D. & Stapleton K. (2005). Establishing
83
representative no-take areas in the Great Barrier Reef: Large-scale implementation of
theory on marine protected areas. Conservation Biology, 19, 1733-1744.
Fitt W.K., Brown B.E., Warner M.E. & Dunne R.P. (2001). Coral bleaching: interpretation of
thermal tolerance limits and thermal thresholds in tropical corals. Coral Reefs, 20, 5165.
Folland C.K., Rayner N.A., Brown S.J., Smith T.M., Shen S.S.P., Parker D.E., Macadam I.,
Jones P.D., Jones R.N., Nicholls N. & Sexton D.M.H. (2001). Global temperature
change and its uncertainties since 1861. Geophysical Research Letters, 28, 26212624.
Fredricks D.N. & Relman D.A. (1996). Sequence-based identification of microbial
pathogens: A reconsideration of Koch's postulates. Clinical Microbiology Reviews, 9,
18-33.
Gardner T.A., Cote I.M., Gill J.A., Grant A. & Watkinson A.R. (2003). Long-term regionwide declines in Caribbean corals. Science, 301, 958-960.
Gil-Agudelo D.L. & Garzon-Ferreira J. (2001). Spatial and seasonal variation of Dark Spots
Disease in coral communities of the Santa Marta area (Colombian Caribbean).
Bulletin of Marine Science, 69, 619-629.
Glynn P.W. (1993). Coral reef bleaching - ecological perspectives. Coral Reefs, 12, 1-17.
Glynn P.W. (1996). Coral reef bleaching: facts, hypotheses and implications. Global Change
Biology, 2, 495-509.
Glynn P.W., Cortes J., Guzman H.M. & Richmond R.H. (1988). El Niño (1982-1983)
associated coral mortality and relationship to sea surface temperature deviations in
the tropical eastern Pacific. Proceedings of the Sixth International Coral Reef
Symposium, 2, 693-698.
Glynn P.W. & D'croz L. (1990). Experimental-evidence for high-temperature stress as the
cause of El-Niño-coincident coral mortality. Coral Reefs, 8, 181-191.
Hall V.R. & Hughes T.P. (1996). Reproductive strategies of modular organisms:
Comparative studies of reef-building corals. Ecology, 77, 950-963.
Harvell C.D., Kim K., Burkholder J.M., Colwell R.R., Epstein P.R., Grimes D.J., Hofmann
E.E., Lipp E.K., Osterhaus A.D.M.E., Overstreet R.M., Porter J.W., Smith G.W. &
Vasta G.R. (1999). Emerging marine diseases--climate links and anthropogenic
factors. Science, 285, 1505-1510.
84
Harvell C.D., Mitchell C.E., Ward J.R., Altizer S., Dobson A.P., Ostfeld R.S. & Samuel
M.D. (2002). Climate warming and disease risks for terrestrial and marine biota.
Science, 296, 2158-2162.
Hayes M.L., Bonaventura J., Mitchell T.P., Prospero J.M., Shinn E.A., Van Dolah F. &
Barber R.T. (2001). How are climate and marine biological outbreaks functionally
linked? Hydrobiologia, 460, 213-220.
Hoegh-Guldberg O. (1999). Climate change, coral bleaching and the future of the world's
coral reefs. Marine Freshwater Research, 50, 839-866.
Hughes T.P., Baird A.H., Bellwood D.R., Card M., Connolly S.R., Folke C., Grosberg R.,
Hoegh-Guldberg O., Jackson J.B.C., Kleypas J., Lough J.M., Marshall P., Nystrom
M., Palumbi S.R., Pandolfi J.M., Rosen B. & Roughgarden J. (2003). Climate change,
human impacts, and the resilience of coral reefs. Science, 301, 929-933.
Hughes T.P. & Connell J.H. (1999). Multiple stressors on coral reefs: a long-term
perspective. Limnology and Oceanography, 44, 932-940.
Jackson J.B.C. (1997). Reefs since Columbus. Coral Reefs, 16, S23-S32.
Jones R.J., Bowyer J., Hoegh-Guldberg O. & Blackall L.L. (2004). Dynamics of a
temperature-related coral disease outbreak. Marine Ecology Progress Series, 281, 6377.
Kilpatrick K.A., Podesta G.P. & Evans R. (2001). Overview of the NOAA/NASA advanced
very high resolution radiometer Pathfinder algorithm for sea surface temperature and
associated matchup database. Journal of Geophysical Research-Oceans, 106, 91799197.
Kim K. & Harvell C.D. (2004). The rise and fall of a six-year coral-fungal epizootic.
American Naturalist, 164, S52-S63.
Koch R. (1882). Die aetiologie der tuberculose. Berl Klinische Wochenschrift, 19, 221-230.
Kuta K.G. & Richardson L.L. (2002). Ecological aspects of black band disease of corals:
relationships between disease incidence and environmental factors. Coral Reefs, 21,
393-398.
Kutz S.J., Hoberg E.P., Polley L. & Jenkins E.J. (2005). Global warming is changing the
dynamics of Arctic host-parasite systems. Proceedings of the Royal Society BBiological Sciences, 272, 2571-2576.
Lafferty K.D. & Gerber L.R. (2002). Good medicine for conservation biology: The
intersection of epidemiology and conservation theory. Conservation Biology, 16, 593604.
85
Leichter J.J., Deane G.B. & Stokes M.D. (2005). Spatial and temporal variability of internal
wave forcing on a coral reef. Journal of Physical Oceanography, 35, 1945-1962.
Leichter J.J. & Miller S.L. (1999). Predicting high-frequency upwelling: Spatial and temporal
patterns of temperature anomalies on a Florida coral reef. Continental Shelf Research,
19, 911-928.
Lenihan H.S., Micheli F., Shelton S.W. & Peterson C.H. (1999). The influence of multiple
environmental stressors on susceptibility to parasites: an experimental determination
with oysters. Limnology and Oceanography, 44, 910-924.
Lessios H.A. (1988). Mass mortality of Diadema-Antillarum in the Caribbean - what have we
learned. Annual Review of Ecology and Systematics, 19, 371-393.
Liu G., Skirving W. & Strong A.E. (2003). Remote sensing of sea surface temperatures
during 2002 Barrier Reef coral bleaching. EOS, 84, 137-144.
Marcogliese D.J. (2001). Implications of climate change for parasitism of animals in the
aquatic environment. Canadian Journal of Zoology-Revue Canadienne De Zoologie,
79, 1331-1352.
McCallum H., Harvell C.D. & Dobson A.P. (2003). Rates of spread in marine pathogens.
Ecology Letters, 6, 1062-1067.
Nugues M.M., Smith G.W., Hooidonk R.J., Seabra M.I. & Bak R.P.M. (2004). Algal contact
as a trigger for coral disease. Ecology Letters, 7, 919-923.
Obura D.O. (2005). Resilience and climate change: lessons from coral reefs and bleaching in
the western Indian Ocean. Estuarine Coastal and Shelf Science, 63, 353-372.
Pandolfi J.M., Bradbury R.H., Sala E., Hughes T.P., Bjorndal K.A., Cooke R.G., McArdle
D., McClenachan L., Newman M.J.H., Paredes G., Warner R.R. & Jackson J.B.C.
(2003). Global trajectories of the long-term decline of coral reef ecosystems. Science,
301, 955-958.
Pascual M., Rodo X., Ellner S.P., Colwell R.R. & Bouma M.J. (2000). Cholera dynamics and
El Nino-Southern Oscillation. Science, 289, 1766-1769.
Patterson K.L., Porter J.W., Ritchie K.E., Polson S.W., Mueller E., Peters E.C., Santavy D.L.
& Smiths G.W. (2002). The etiology of white pox, a lethal disease of the Caribbean
elkhorn coral, Acropora palmata. Proceedings of the National Academy of Sciences
of the United States of America, 99, 8725-8730.
Patz J.A. (2002). A human disease indicator for the effects of recent global climate change.
Proceedings of the National Academy of Sciences of the United States of America, 99,
12506-12508.
86
Podesta G.P. & Glynn P.W. (2001). The 1997-98 El Niño event in Panama and Galapagos:
An update of thermal stress indices relative to coral bleaching. Bulletin of Marine
Science, 69, 43-59.
Pounds J.A., Bustamante M.R., Coloma L.A., Consuegra J.A., Fogden M.P.L., Foster P.N.,
La Marca E., Masters K.L., Merino-Viteri A., Puschendorf R., Ron S.R., SanchezAzofeifa G.A., Still C.J. & Young B.E. (2006). Widespread amphibian extinctions
from epidemic disease driven by global warming. Nature, 439, 161-167.
Raymundo L.J., Rosell K.B., Reboton C.T. & Kaczmarsky L. (2005). Coral diseases on
Philippine reefs: genus Porites is a dominant host. Diseases of Aquatic Organisms,
64, 181-191.
Reynolds R.W., Rayner N.A., Smith T.M., Stokes D.C. & Wang W.Q. (2002). An improved
in situ and satellite SST analysis for climate. Journal of Climate, 15, 1609-1625.
Richardson L.L. (1998). Coral diseases: what is really known? Trends in Ecology &
Evolution, 13, 438-443.
Rosenberg E. & Ben-Haim Y. (2002). Microbial diseases of corals and global warming.
Environmental Microbiology, 4, 318-326.
Rosenberg E. & Falkovitz L. (2004). The Vibrio shiloi/Oculina patagonica model system of
coral bleaching. Annual Review of Microbiology, 58, 143-159.
Sakai K. (1998). Effect of colony size, polyp size, and budding mode on egg production in a
colonial coral. Biological Bulletin, 195, 319-325.
Santavy D.L., Summers J.K., Engle V.D. & Harwell L.C. (2005). The condition of coral
reefs in South Florida (2000) using coral disease and bleaching as indicators.
Environmental Monitoring and Assessment, 100, 129-152.
Sheppard C. & Rioja-Nieto R. (2005). Sea surface temperature 1871-2099 in 38 cells in the
Caribbean region. Marine Environmental Research, 60, 389-396.
Sheppard C.R.C. (2003). Predicted recurrences of mass coral mortality in the Indian Ocean.
Nature, 425, 294-297.
Shinn E.A., Smith G.W., Prospero J.M., Betzer P., Hayes M.L., Garrison V. & Barber R.T.
(2000). African dust and the demise of Caribbean coral reefs. Geophysical Research
Letters, 27, 3029-3032.
StataCorp LP (2006). Stata 9. In. StataCorp LP College Station, TX.
Strong A.E., Liu G., Meyer J., Hendee J.C. & Sasko D. (2004). Coral Reef Watch 2002.
Bulletin of Marine Science, 75, 259-268.
87
Sutherland K.P., Porter J.W. & Torres C. (2004). Disease and immunity in Caribbean and
Indo-Pacific zooxanthellate corals. Marine Ecology Progress Series, 266, 273-302.
Sweatman H., Abdo D., Burgess S., Cheal A., Coleman G., Delean S., Emslie M.J., Miller I.,
Osborne K., Page C.A. & Thompson A. (2003). Long-Term Monitoring of the Great
Barrier Reef: Status Report Number 6. Australian Institute of Marine Science,
Townsville.
The Mathworks Inc. (2005). Matlab. In. The Mathworks Inc. Natick, MA.
Toscano M.A., Liu G., C. G.I., Casey K.S., Strong A. & Meyer J.E. (2000). Improved
prediction of coral bleaching using high-resolution HotSpot anomaly mapping. In:
Ninth International Coral Reef Symposium (eds. Moosa MK, Soemodihardjo S,
Nontji A, Soegiarto A, Romimohtarto K, Sukarno & Suharsono). Ministry of the
Environment, the Indonesian Institute of Sciences, and the International Society for
Reef Studies Bali, Indonesia, pp. 1143-1148.
Wapnick C.M., Precht W.F. & Aronson R.B. (2004). Millennial-scale dynamics of staghorn
coral in Discovery Bay, Jamaica. Ecology Letters, 7, 354-361.
Ward J.R. & Lafferty K.D. (2004). The elusive baseline of marine disease: are diseases in
ocean ecosystems increasing? PloS Biology, 2, 542-547.
Weil E. (2004). Coral Reef Diseases in the Wider Caribbean. In: Coral Health and Disease
(eds. Rosenberg E & Loya Y). Springer Berlin.
West J.M. & Salm R.V. (2003). Resistance and resilience to coral bleaching: implications for
coral reef conservation and management. Conservation Biology, 17, 956-967.
Williams D.E. & Miller M.W. (2005). Coral disease outbreak: pattern, prevalence and
transmission in Acropora cervicornis. Marine Ecology-Progress Series, 301, 119-128.
Willis B.L., Page C.A. & Dinsdale E.A. (2004). Coral disease on the Great Barrier Reef. In:
Coral Health and Disease (eds. Rosenberg E & Loya Y). Springer-Verlag Berlin, pp.
69-104.
Winter A., Appeldoorn R.S., Bruckner A., Williams E.H. & Goenaga C. (1998). Sea surface
temperatures and coral reef bleaching off La Parguera, Puerto Rico (northeastern
Caribbean Sea). Coral Reefs, 17, 377-382.
Wooldridge S., Done T., Berkelmans R., Jones R. & Marshall P. (2005). Precursors for
resilience in coral communities in a warming climate: a belief network approach.
Marine Ecology-Progress Series, 295, 157-169.
88
CHAPTER 3
Evidence of increased resilience to corals from marine protected areas
Abstract
Marine protected areas (MPAs) have become a widely used and effective tool for fisheries
management. However, their effects on foundation species like corals, which have declined
at unprecedented rates, are not well understood. By restoring historic food web structure,
MPAs are hypothesized to benefit corals by increasing resilience to disturbances through the
control of macroalgal growth. We tested this hypothesis by compiling spatial databases of
MPAs and 8540 coral cover surveys from 4456 reefs around the world. Using a multilevel
model, we analyzed how MPAs affect the rate of coral cover change and whether this rate
was modified by MPA characteristics like the number of years of protection. Protection in
MPAs essentially maintained coral cover over time with trends not significantly different
from zero while coral cover on unprotected reefs continued to decline. Within MPAs 53% of
reefs experienced increases in coral cover, whereas only 33% of unprotected reefs had
increases. In addition, we found that the number of years of protection changed the
magnitude of the MPA effect on coral cover, with more years of protection generally
resulting in more positive rates of coral cover change. Although these results provide
evidence for an effect of resilience on corals in MPAs, the weakness of the MPA effect
underscores the need for additional and more direct conservation measures including those
aimed at reducing climate change.
Introduction
Continued species loss in a variety of ecosystems has heightened the need for tools to
mitigate and manage the effects of human exploitation of natural resources (Chapin et al.
2000; Brooks et al. 2002). Protected areas, which restrict access or prevent extraction, have
become one of the primary tools for species conservation. Although they have long been a
mainstay of terrestrial conservation efforts, there has been an increased emphasis on
expanding the number of protected areas in the ocean in the past few decades (Agardy 1994;
Gubbay 1995), largely due to the worldwide decline of fish stocks (Pauly et al. 1998; Worm
et al. 2005). Marine protected areas (MPAs) are broadly defined as places in the ocean
where human activities are restricted. Most MPAs specifically restrict commercial or
recreational fishing. However, the continued decline of marine ecosystems has generated
renewed interest in using MPAs more broadly for protecting species and restoring ecosystem
services (Balmford et al. 2004).
MPAs have had success as fisheries management tools (McClanahan & Mangi 2000;
Murawski et al. 2000). Most research on the effects of MPAs has focused on marine
reserves, also called no-take reserves, which prohibit all extractive activities. Research in a
variety of marine habitats has shown that the biomass, density, and size of organisms, as well
as the diversity of several taxa increased inside reserves (Halpern 2003; Williamson et al.
2004). Through protection in MPAs, exploited fish species are able to grow older and bigger
90
(Halpern 2003). Reserves have also been shown to benefit adjacent fisheries through a
spillover effect of larvae and adult fish (McClanahan & Mangi 2000; Roberts et al. 2001;
Russ et al. 2004; Murawski et al. 2005). Models indicate that this export of adults and larvae
may be able to compensate for the loss of available fishing area (Halpern et al. 2004), but this
has not always been found in field studies and may be dependent on the age of the protected
area (McClanahan & Mangi 2000).
More recently, MPAs have been proposed as a strategy for reducing coral decline.
Globally, coral reefs have undergone a period of rapid, extensive change in the last 20-30
years, which may be unprecedented in the Holocene paleontological record (Aronson &
Precht 1997; Aronson et al. 2002; Wapnick et al. 2004; but see Hubbard et al. 2005). Much
of the optimism surrounding the effects of MPAs on corals has focused on the theory that a
restoration to historic trophic structure, or better management of land-based sources of
pollution with inclusion of a terrestrial component, will increase the resilience of coral reef
ecosystems to disturbance events (Holling 1973; Grimsditch & Salm 2005; Hughes et al.
2005; Wolanski & De'ath 2005). Resilience is defined as the rate of return to a pre-disturbed
state, which is a measure of the persistence of relationships within an ecosystem (Holling
1973). Using MPAs to slow coral decline through a resilience effect is intriguing given the
limited range of management options, but it is still largely untested.
Evidence that marine protected areas may have positive effects on coral cover
through fisheries management is equivocal. Results from several studies have found that
protection in reserves increased the abundance and biomass of grazers that consume
macroalgae, which compete with corals (Mumby et al. 2006; Newman et al. 2006; Hughes et
al. 2007). Increased grazing can subsequently facilitate greater coral recruitment and
91
survivorship (Mumby et al. 2007). However, restoration of historic structure may not have
predictably positive effects on coral cover. Greater predatory fish abundance from protection
in marine reserves could actually result in decreased abundance of herbivorous fish (Hixon &
Beets 1993; Graham et al. 2003). Certain fish species may also have negative effects on
corals. Although parrotfish are key grazers on macroalgae (Carpenter 1986), they are also
corallivorous and can impede coral recovery from disturbance events (Rotjan et al. 2006). In
addition, the reduction in algal cover from greater herbivore abundance does not always
translate to an increase in coral cover (Jones et al. 2004; Graham et al. 2006; Newman et al.
2006; Ledlie et al. 2007). The effects of reserves on both fish and coral are further
complicated by studies that have found that reductions in live coral appear to have led to
subsequent reductions in fish abundance and diversity even within MPAs because many fish
and invertebrates are obligate or facultative dependents on live coral (Jones et al. 2004).
In spite of these potential benefits, no research has examined whether protection in
MPAs can mitigate coral decline through increased ecosystem resilience. Here we use
spatial databases of live coral cover surveys and MPA locations to examine whether trends in
live coral cover differ systematically inside and outside of MPAs in the Caribbean Sea and
Indian and Pacific Oceans. We then determine if attributes of MPAs – size, year of
establishment, and World Conservation Union (IUCN) category – affect within-MPA trends.
The results of these analyses will help determine if there is evidence of MPAs conferring
resilience on coral ecosystems and which characteristics of MPAs may increase a potential
positive effect.
92
Methods
Coral cover surveys
We developed a global database of 8540 surveys, which quantified the percentage of
the shallow water hard substratum covered by living hard (scleractinian) coral tissue on 4456
reefs (Fig. 3.1A) from 1969 to 2006. Surveys on 933 of 4456 reefs or 22% of the data were
repeated over time. Of the repeatedly monitored reefs, 385 had five or more observations.
Data were collected from a variety of sources including peer-reviewed literature, published
government and non-governmental organization reports, and existing databases (i.e., Reef
Check, Reefs at Risk in Southeast Asia, ReefBase). We used the ISI Web of Science and
Google Scholar to find appropriate peer-reviewed and grey literature from management
agencies and environmental organizations, targeting our search with the terms "coral" and
"cover" and "reef" and "health". We also examined all available issues of journals that were
likely to contain coral reef studies including Atoll Research Bulletin, Coral Reefs, Marine
Pollution Bulletin, and the Proceedings of the International Coral Reef Symposium. Only
surveys above 15 m depth were included to avoid depth biases (for more detail on surveys in
the Indo-Pacific regions, see Bruno & Selig 2007).
Marine protected area database
The World Database on Protected Areas (WDPA) maintains a freely available spatial
dataset on protected areas worldwide including MPAs (http://sea.unep-wcmc.org/wdbpa).
For the Caribbean, some protected areas were updated with data from The Nature
Conservancy. In addition, boundaries for the United States' national marine sanctuaries,
monuments, and parks were obtained from the National Oceanographic and Atmospheric
93
Administration, the US Department of the Interior, or the United States Geological Survey
where available. We also obtained exact boundaries from the Great Barrier Reef Marine
Park Authority for the Great Barrier Reef Marine Park for the pre-2005 zoning, which was
the management plan under which all of our data was collected except for one year. Even
MPAs which did not have polygon boundaries had data about their area. For the final
analysis, 27 of the 311 total MPAs (Fig. 3.1B) analyzed lacked exact boundary information.
In these cases, we followed the protocol of similar work (Mora et al. 2006) and delineated a
circular buffer around the point to create a boundary of the stated area of the MPA.
Many MPAs (e.g. the Florida Keys National Marine Sanctuary and the Great Barrier
Reef Marine Park) contain areas with different IUCN categories or have areas that are
spatially disjunct from one another. Therefore, we analyzed the data based on the individual
polygons and their associated zoning rather than the name of the larger MPA. The MPA
database also contained additional information about the MPAs themselves including their
size, IUCN protection category (see Appendix 2.1, Table S2.1), and the year that the MPA
was established. Many countries including Australia and the United States have not formally
assigned IUCN categories to their MPAs. In these cases, IUCN categories were assigned
based on the goals in the management plans and their corresponding level of protection on
the IUCN scale. Although the WDPA database had information on the year of establishment
for many of these categories, additional data was found by researching government and nongovernmental websites.
94
Statistical analyses
We used a multilevel model to examine the relationship between live coral cover and
protection over time. Because some surveys were repeated over time, the data could not be
considered independent. In addition, reefs were spatially clustered and this autocorrelation
had to be accounted for in the statistical model. By using a multilevel model, we were able
to include such spatial and temporal structure explicitly (McMahon & Diez 2007). This
approach focused on patterns at the scale of individual reefs while still providing population
averages for reefs at regional scales (McMahon & Diez 2007). We used the nlme library
(Pinheiro et al. 2007) in R 2.5.1 (R Development Core Team 2007) and WinBUGS 1.4.3
(Lunn et al. 2000) to analyze our models.
Percent coral cover is bounded data, which could be described by various probability
models including the beta, logistic-normal, and arcsine-normal distributions. We applied a
logit transformation to percent coral cover to obtain a logistic-normal distribution, which we
then treated as normally distributed (Lesaffre et al. 2007, see Appendix 2.2). We also
centered our time regressor in the models we examined. Centering not only reduced
correlation between the random effects and facilitated model convergence, but also made the
intercept interpretable as the coral cover in the year selected. We found that centering the
models on the year 1996 yielded the least degree of correlation between the random effects
(Singer & Willet 2003).
Model development
We began model development by constructing a generic multilevel model to deal
with the spatial and temporal correlation inherent in our data. The assignment of random
95
effects in a multilevel model accounts for unexplained spatial and temporal heterogeneity. In
our model, there were three possible levels, each of which separate observation errors from
temporal and spatial effects. At level 1, individual observations on a reef were assigned a
unique random effect. Surveys on the same reef over time were assigned a unique random
effect at level 2, accounting for temporal correlation. Level 3 contained the spatial unit
where protected and unprotected reefs were paired, accounting for spatial correlation.
The spatial units for level 3 were constructed based on MPA boundaries. All reefs
within an MPA with the same management criteria were grouped together. Then we paired
unprotected reefs with protected reefs based on their distance to the nearest MPA boundary
to create reasonably homogenous spatial grouping units. These spatial groupings were used
to assign random effects in level 3. We selected the optimal distance using maximum
likelihood estimatation to compare how a range of distances between 0 and 4000 km
influenced the MPA effect on slope (coral cover change). The loglikelihood was maximized
at a pairing distance of 200 km (Fig. S2.1). Reefs within 200 km of an MPA were paired
with the nearest MPA that contained a coral cover survey; approximately 60% of reefs were
paired (see Appendix 2.3 for more details). Any remaining unprotected or protected reefs
were left unpaired but grouped into spatial units. For the MPA-only analyses, the spatial
groupings were the MPAs.
We began with an ordinary regression model with logit-transformed coral cover
(hereafter, logit coral cover) as the response and centered year as the only predictor. We then
added temporal structure and spatial structure and used information-theoretic measures (AIC)
to conclude that we needed to account for both (Akaike 1973; Burnham & Anderson 2002).
The basic trend model fit to each reef is shown in Eqn. (1). In this model, Y is logit coral
96
cover, T is year, Tc is the centering constant, and ε is the random error. The subscript i
denotes the spatial grouping unit, j denotes the reef within that spatial unit, and k denotes the
individual survey measurement for that reef.
Level 1:
(
Yijk = β 0ij + β 1ij (Tijk − Tc ) + ε ijk , ε ijk ~ N 0, σ 2
)
(1)
The presence of the subscript j in the parameters β 0ij (intercept) and β1ij (slope) indicates that
these parameters were allowed to vary from reef to reef. The manner in which they vary is
described by the level 2 equation, Eqn. (2).
Level 2:
β 0ij = β 0i + β 2i X ij + u0ij
β1ij = β1i + β3i X ij + u1ij
⎛ ⎡0⎤ ⎡τ 2 τ 01 ⎤ ⎞
⎡u 0ij ⎤
⎟
, ⎢ ⎥ ~ N⎜ ⎢ ⎥ , ⎢ 0
2 ⎥⎟
⎜ ⎣0 ⎦ τ
τ
01
1
⎣ u1ij ⎦
⎣
⎦
⎝
⎠
(2)
Here, X is a reef-level predictor that varies among reefs but is constant for observations taken
on the same reef; u0ij and u1ij are the random effects for reef j in spatial grouping unit i and
are assumed to have a joint normal distribution as shown. Protection status (protected versus
unprotected) is an example of variable X. Random effects from different reefs were assumed
to be independent and also independent of the level 1 error terms. The subscript i on the
slope and intercept terms in Eqn. (2) indicates that reefs coming from the same spatial
grouping unit may also have characteristics in common.
Level 3:
β 0i = β 0 + β 4 Z i + v0i ⎡v0i ⎤
~
,
β1i = β1 + β5 Z i + v1i ⎢⎣ v1i ⎥⎦
97
⎛ ⎡0⎤ ⎡ ω 2 ω 01 ⎤ ⎞
⎟
N⎜ ⎢ ⎥ , ⎢ 0
2 ⎥⎟
⎜ ⎣0 ⎦ ω
ω
1 ⎦⎠
⎣ 01
⎝
(3)
In Eqn. (3), Z is a level 3 predictor and the terms v0i and v1i are the error terms. Examples of
the variable Z include MPA area, IUCN category or ocean basin. For all models, we
included random intercepts and random slopes at levels 2 and 3 (Table S2.2), which allowed
trend and coral cover to vary among spatial grouping units as well as among reefs within
those units.
MPA versus non-MPA model
We began with the basic form of the multilevel model described in Eqns. (1), (2), and
(3) and added necessary predictors and spatial structure. As a result of the pairing, the
original status of the reef, protected or not, varied with the spatial grouping unit. We treated
protection status as a level 2 variable, which could be used to model the level 1 intercept,
slope, or both, playing the role of variable X ij in Eqn. (2). Then we fit all possible models
that included MPA as a level 2 predictor. Once we selected the appropriate MPA versus
non-MPA model using AIC, we explored whether adding a variable that distinguished
between ocean basins was needed as a level 3 predictor (Fig. S2.2 compares all the models
using AIC). With a model that incorporated protection status as a level 2 predictor, we could
determine whether the trend in coral cover varied between protected and unprotected reefs
over time.
MPA-only model
The MPA-only model used reefs inside MPAs to explore how MPA characteristics—
size, IUCN status, or number of years in protection—affected coral cover and its trend over
time. For the MPA-only model, we analyzed data from within MPAs to determine which of
98
the three characteristic variables explained the trend and overall coral cover within MPAs.
The number of years of protection was a dynamic variable defined as the number of years the
reef was protected at the time of the coral cover survey.
As with the MPA versus non-MPA model, we began with the basic model described
by Eqns. (1), (2), and (3) that included random intercepts and random slopes at levels 2 and
3. We then used AIC to compare whether the addition of any of the three variables improved
the model (Akaike 1973) either by modifying the slope or intercept. Area and IUCN status
varied by the spatial grouping unit (MPA) and were therefore modeled as level 3 variables.
Because the number of years of protection at the time of the coral cover survey varied by the
individual measurement on a reef, it was modeled as a level 1 variable. We then determined
whether adding an additional variable for ocean basin was appropriate.
Model evaluation
In order to obtain more realistic estimates of parameter precision, the final AIC best
models were refit as Bayesian models in which uninformative priors were used for all
parameters. Samples from the parameter posterior distributions were obtained using Markov
chain Monte Carlo (MCMC) as implemented in WinBUGS 1.4.3 (for more details see
Appendix 2.4). We also summarized the fit of our final model using a modification of R 2 ,
the coefficient of determination, generalized to multilevel models (Gelman & Pardoe 2006).
In a multilevel model, there are multiple sources of variation to be explained. At each level,
we can write θ = μ + ε where μ is the linear predictor, ε is the random variation at that
level, and θ corresponds to the individual data points (level 1) or a regression parameter
99
(levels 2 and 3). Eqn. (4) is a generalized version of the equation above for multilevel
models:
R2 = 1−
E(Var(ε i ))
E(Var(θ i ))
(4)
In this notation, E is the expectation operator. The expectations in the ratio are calculated by
taking the means of the posterior distributions of the variance obtained from a Bayesian fit of
the model.
Results
MPA versus non-MPA model
The best model for analyzing the differences in coral cover trends over time between
protected and unprotected reefs included random effects for both the change in coral cover
(slope) and intercept at levels 2 and 3. Protection status was a significant predictor of both
the slope and intercept (P < .001). We also found that we needed an additional variable (level
3 predictor), which accounted for the ocean basin of the survey—Pacific Ocean, Caribbean
Sea or Indian Ocean. The ocean variable was only needed to modify the intercept (Fig. S2.2
lists the models we fit and compares them using AIC). The AIC best model is shown in Eqn.
(5):
logit ( pijk ) = β 0 + v0i + β 2 Protection Statusij + β 4 Indiani + β 5 Pacifici + u0ij
+ (β1 + v1i + β 3 Protection Statusij + u1ij )(Tijk − Tc ) + ε ijk
100
(5)
Protection in MPAs resulted in a positive effect on the population-average trend (Fig.
3.2). Change in coral cover was slightly positive but not significantly different from zero in
MPAs and negative outside of MPAs (Figs. 3.3, S2.3). We assessed how well individual reef
behavior compared to the overall population average using caterpillar plots and found that
the slopes generally conform to the model estimates, but the intercepts had more variability
(Fig. S2.4). The model explained 46% and 45% of the variation in intercept and slope,
respectively, at level 2 and 39% of the variation in intercept at level 3 (Table 3.1).
MPA-only model
To determine which characteristics of MPA affected coral cover within MPAs, we
examined three variables: year of protection, MPA size, and IUCN category. Based on AIC,
MPA size did not improve the model and thus was not included as a predictor. The inclusion
of IUCN category as a predictor marginally improved the model, but it had radically different
distributions in the three oceans (Tables S2.3-4) and its interpretation in the model was
problematic. For example, IUCN models fit separately by ocean did not have an effect, but
when the data were pooled across oceans the stricter IUCN categories had a weak positive
effect on coral cover. Pooling across oceans yielded a more balanced distribution of data
across the IUCN categories. Because the lack of sufficient data in some categories made the
model unreliable, we discarded IUCN as a predictor.
The number of years of protection variable significantly improved the model when
added as a level 1 predictor. The years of protection variable also had different distributional
patterns by year of MPA establishment (Fig. S2.5-6) and by ocean (Fig. S2.7). Unlike the
IUCN predictor, though, the effect of years of protection was consistent when all data were
101
included or if they were separated by ocean. To better capture any ocean differences, we
developed two models: one for the Caribbean and one for the Indian and Pacific Oceans
(hereafter Indo-Pacific). When we explored different model forms using generalized additive
mixed models, we found that a changepoint or breakpoint model (Toms & Lesperance 2003)
was needed for the Indo-Pacific (Fig. S2.8A), but the evidence for such a model in the
Caribbean was weak (Fig. S2.8B). For the Caribbean, we fit the following linear model
(Eqn. 6):
logit ( pijk ) = β 0 + β 2Yijk + v0i + u 0ij
+ (β 1 + β 3Yijk + v1i + u1ij )(Tijk − TC ) + ε ijk
(6)
In the Indo-Pacific changepoint model, the years of protection effect on trend was different
for early and late years. Our final model for the Indo-Pacific (Eqn. 7) described years of
protection having a different effect on trend depending on whether the number of years of
protection occurs before or after the changepoint:
logit ( pijk ) = β 0 + β 2Yijk + v0i + u 0ij
(
)
+ β1 + β 3Yijk + β 4 (Yijk − YCP ) + v1i + u1ij (Tijk − TC ) + ε ijk
+
(7)
Here T is the year, Y is the number of years the reef had been protected at the time of the
coral cover survey, TC is the centering constant for years, YCP is the breakpoint for years
protected, and (Y − YCP ) is the indicator function defined as follows.
+
102
(Y − YCP )+
0, if Y − YCP ≤ 0
⎧
=⎨
⎩Y − YCP , if Y − YCP > 0
We found that the Caribbean and Indo-Pacific models explained 91% and 81% of the
variation at level 1, respectively (Table 3.2, Figs. 3.5B-C). To obtain parameter estimates
and credibility intervals (hereafter CI), we fit both models in WinBUGS using equivalent
parameterizations of Eqns. (6) and (7). The number of years of protection affected the rate of
change in coral cover. In the Caribbean, the rate of change in coral cover shifted from
negative to positive at 15 years and then began to level off after 30 years (Fig. 3.4A). For the
Indo-Pacific, we determined that there were two distinct rates of coral cover change. Using
the median of the posterior distribution of the changepoint model, we determined that the
switch in the two rates occurred at 21.7 years (20.0-24.9 95% credibility interval). The
location of the changepoint was likely the result of a major cohort of reefs, which had been
protected for roughly 20 years, experiencing declines due to the 1998 El Niño event. Before
the changepoint, the rate of change in coral cover shifted from negative to positive after 5
years and continued increasing until the changepoint at approximately 21 years (Fig. 3.4B).
After the changepoint, the rate of change in coral cover became negative again, although not
significantly different from zero (Fig. 3.4B).
Discussion
Protection in MPAs is widely viewed as a potential way to mitigate the decline of
coral reefs (Grimsditch & Salm 2005; Hughes et al. 2005; Sandin et al. 2008). We analyzed
103
how protection in MPAs affected coral cover on coral reefs globally. MPAs had a weakly
positive effect on coral cover rate of change, but coral cover trend was not significantly
different from zero. In comparison, unprotected reefs continued to decline (Fig. 3.3).
Annual rates of change depended on the existing coral cover. In 2005, protected reefs in the
Pacific and Indian Oceans averaged increases of 0.08% in coral cover. Average coral cover
increases were 0.05% in the Caribbean (Fig. 3.3). By contrast, unprotected reefs averaged
declines ranging from -0.27% in the Caribbean to -0.43% and -0.41% in the Pacific and
Indian Oceans, respectively (Fig. 3.3). In total, 53% of protected reefs had increases in coral
cover whereas only 33% of unprotected reefs had increases. Our results suggest that MPAs
conferred a weak resilience on corals and that their overall effect was to maintain coral cover,
preventing the declines occurring outside of MPAs.
The effect of MPAs on coral cover was dependent on reef location and how long the
reef had been protected when it was surveyed. In the Caribbean, coral cover continued to
decline for the first 15 years after protection, perhaps due to the time required for fisheries to
rebound from exploitation, which has been affecting some areas in the region for several
hundred years (Jackson 1997; Wing & Wing 2001). The coral cover trend then became
positive and continued to be positive, leveling off at approximately 30 years (Fig. 3.4A).
This leveling off pattern is consistent with several studies of protection on reef fish (Maliao
et al. 2004; Abesamis & Russ 2005; McClanahan et al. 2007) and at least one field study on
coral cover (Halford et al. 2004), which found that populations' recovery reached a saturation
point. Because many reef-building corals are slow-growing, recovery rates and leveling off
values could vary according to the dominant suite of species on the reef. A rebound in coral
cover may also reflect a shift to faster-growing or more stress-tolerant species and a
104
subsequent change in species composition (Aronson & Precht 1997; Aronson et al. 2004).
Shifts in species composition may not only substantially alter the structural complexity of the
reef, but also alter the vulnerability of the foundation to disturbances like storms (Madin &
Connolly 2006).
In the Indo-Pacific, the effect of the number of years of protection on coral cover
displayed two distinct trends. Rates of change in coral cover continued to be negative for the
first 5 years, but became positive until approximately 20 years of protection (Fig. 3.4B). At
approximately 20 years, coral cover began to decline again and then recovered to a trend that
was not significantly different from zero. This reset pattern coincided with a major cohort of
reefs that had been protected for 20-25 years when the strong El Niño event occurred in 1998
(Fig. 3.4B). The 1998 El Niño caused high rates of mortality on reefs across the Indo-Pacific
region including within MPAs (Arceo et al. 2001; Bruno et al. 2001; Graham et al. 2007). A
substantial portion of surveyed reefs in the Indo-Pacific had been protected for 20-25 years in
1998, which was reflected in the transition to a new trend at around 20 years. The IndoPacific can experience major mortality events like the 1998 El Niño on a periodic basis (Yu
et al. 2006). The recovery and reset in coral cover trend we observed in the 30 years of data
we analyzed may be indicative of a cyclical pattern, but establishing a longer-term protection
pattern will require more data. For the Caribbean and some years in the Pacific, our MPAonly analyses estimated a more positive effect than seen in the overall MPA versus non-MPA
model (Figs. 3.3 and 3.4). Almost 60% of the surveys in our analysis were from MPAs that
were less than 15 years old. The young age of many of the studied MPAs means we may be
seeing less of a protection effect. Therefore, when the MPA effect was analyzed in aggregate
as in the MPA versus non-MPA model, we saw a weaker overall MPA effect.
105
MPA size can be an important determinant in MPA effectiveness, particularly for
mobile species like fish (Claudet et al. in press). We did not find a positive effect of MPA
size on coral cover over time. The lack of an effect of MPA size could be due to the
relatively short dispersal distances of coral larvae (Ayre & Hughes 2000). Short dispersal
distances may mean that even small MPAs may benefit from the larval supply within their
boundaries. Synthetic research on MPA effects has also found that even small MPAs can
confer positive effects on species biomass and abundance in a variety of marine habitats
(Halpern 2003). Although larger MPAs will be able to better protect highly mobile predatory
and grazing fish, there is evidence that the benefit increases directly with MPA size.
Therefore, protection in small MPAs should still yield a more complete fish assemblage than
in unprotected areas (Halpern 2003).
The interpretation of our MPA-specific variables like IUCN category was
complicated by the fact that many of the MPAs in our analysis do not have adequate
enforcement of their regulations and the categories themselves are self-reported and
inconsistent (McClanahan 1999; Burke et al. 2002). We could not exclude poorly managed
MPAs from our analysis, because we did not have data on enforcement for most of our
MPAs. However, routinely monitored MPAs like the ones that dominate this analysis may
be more likely to have had adequate enforcement because they possessed the staff and
resources to carry out repeated surveys. Our ability to detect a strong effect of MPAs may
also have been compromised by the relatively low percentage of protected coral reefs
worldwide. Only 13.4% of reefs are currently protected in non-extractive or multi-use MPAs
and only 1.4% are in no-take reserves (Mora et al. 2006). If a greater percentage of reefs
106
were protected, there could be a positive synergistic effect by having more connected
protected populations.
Several additional data issues could have also confounded the results of our analysis.
The multilevel modeling approach weights reefs with more time-series data more heavily in
the model. Because the Florida Keys National Marine Sanctuary and the Great Barrier Reef
Marine Park were the most highly surveyed MPAs in our analysis, our results could have
been biased by their overrepresentation in the model. Robustness analyses where we
dropped one or both of those subregions indicate that the qualitative outcomes of the model
would not change. In addition, the generally less negative trend in coral cover for protected
reefs could be due to a more constrained data range within MPAs with cover more likely to
increase simply because it was already so low.
Regional and global stressors like increases in ocean temperature and coral diseases
are likely to be the greatest counterbalances of positive MPA effects (Berkelmans et al. 2004;
Bruno et al. 2007). The weakness of the MPA benefit suggests that in many areas, the
effects of these events are overwhelming MPA effects. For example, the GBR experienced
major bleaching episodes in 1998 and 2002 with the 2002 episode being the most extensive
and severe mass bleaching event on record (Berkelmans et al. 2004). Because these
disturbances were relatively recent, it is possible that there was not sufficient time to
demonstrate a resilience effect. Yet it seems probable that MPAs subject to broad-scale
stressors like increasing ocean temperatures and disease will continue to experience declines
in coral cover or fail to attain pre-disturbance coral cover levels.
MPAs can play a critical role in the protection of coral reef ecosystems, particularly
fisheries (Russ & Alcala 2003). Contextualizing the success of MPAs will require analyzing
107
other metrics of resilience including species abundance and richness of other reef taxa
(Bellwood et al. 2006). Changes in coral or fish species assemblages could have far-reaching
effects on the structure and function of coral reef ecosystems (Aronson et al. 2004; Wilson et
al. 2006). Our analysis indicated that MPAs can help to maintain coral populations.
However, the weakness of the effect suggests that it may not be enough to offset coral losses
from major disease outbreaks and bleaching events, which are both predicted to increase in
frequency with climate change (Sheppard 2003; Ward & Lafferty 2004; Sheppard & RiojaNieto 2005). MPAs have long been acknowledged to be an effective yet imperfect response
to anthropogenic threats (Allison et al. 1998; Lubchenco et al. 2003). Reversing coral
decline will require both the local management of stressors through MPAs and international
policies aimed at reducing the human activities causing climate change.
108
Figure 3.1 Location of unprotected (orange) and protected (green) coral cover
surveys in the (A) Caribbean and (B) Indo-Pacific (not pictured: Central Pacific and
Eastern Indian Ocean reefs).
A
B
109
Figure 3.2 Coefficient estimates for the MPA versus non-MPA model. The 95%
credibility intervals (thin grey line) and the 50% credibility intervals (thick black line)
as well as point estimates (median) of the posterior distributions for all parameters in
the MPA versus non-MPA model using a Bayesian approach to fit the model in Eqn.
(1). There is a 95% probability that the true value lies within the 95% credibility
interval. The MPA x 10-Year Trend term should be contrasted with the 10-Year
Trend term, which is the trend for non-MPAs. The MPA x 10-Year Trend term is an
effect and gets added to the 10-Year Trend term when MPA = 1 to obtain the trend
for MPAs.
110
Figure 3.3 The change in percent coral cover from 2004 to 2005 inside and
outside of MPAs. The 95% credibility intervals (error bars) are also shown. Reefs
protected in MPAs had slightly positive changes in coral cover, although not
significantly different from zero (dashed line). Percent coral cover was obtained by
back-transforming the predicted logit from the model.
111
Figure 3.4 The effect of the number of years of protection on the 1-year
change in coral cover. The 95% credibility intervals (dark grey bands) and the 50%
credibility intervals (light grey bands) as well as the median (white line) of the
posterior distributions for the year of protection models in the (A) Caribbean and (B)
Pacific and Indian Oceans in 1996.
A
B
112
Figure 3.5 Comparisons of the average coral cover per year as predicted by
the models. in simulated data sets (light grey lines) with the observed mean coral
cover per year (thick black line) for (A) MPA versus control model, (B) MPA-only
Caribbean years of protection model, and (C) MPA-only Indo-Pacific years of
protection model.
The histograms at the bottom of the figures display the relative
sample sizes at each year for the actual data. Note that the x-axes differ for the
plots due to data availability.
A
113
B
C
114
2
Table 3.1 R for MPA versus non-MPA model. NA denotes the lack of predictor
for calculation R 2 .
Model level
Level 1
Level 2
Level 3
R 2 at each level
y (logit coral cover)
intercept
0.84
—
—
0.46
—
0.39
115
slope
—
0.45
NA
2
Table 3.2 R for MPA-only models in (A) Caribbean and (B) Indo-Pacific. R 2 can
only be calculated at level 1 for these models.
Caribbean
Indo-Pacific
y (logit coral cover)
0.91
0.80
116
Acknowledgments
T. Jobe, J. Lang, M. O'Connor, J. Weiss, P. White for helpful comments on the manuscript.
V. Schutte helped create the coral cover database. The Great Barrier Reef Marine Park for
the GIS data for the Great Barrier Reef zoning and The Nature Conservancy for MPA
boundaries for the Caribbean.
117
References
Abesamis R.A. & Russ G.R. (2005). Density-dependent spillover from a marine reserve:
long-term evidence. Ecological Applications, 15, 1798-1812.
Agardy T. (1994). Advances in marine conservation: the role of marine protected areas.
Trends in Ecology & Evolution, 9, 267-270.
Akaike H. (1973). Information theory and an extension of the maximum likelihood principle.
In: Second International Symposium on Information Theory (eds. Petrov BN & Csaki
F). Akademiai Kaido Budapest, pp. 267-281.
Allison G.W., Lubchenco J. & Carr M.H. (1998). Marine reserves are necessary but not
sufficient for marine conservation. Ecological Applications, 8, S79-S92.
Arceo H.O., Quibilan M.C., Alino P.M., Lim G. & Licuanan W.Y. (2001). Coral bleaching
in Philippine reefs: coincident evidence with mesoscale thermal anomalies. Bulletin
of Marine Science, 69, 579-593.
Aronson R.B., MacIntyre I.G., Precht W.F., Murdoch T.J.T. & Wapnick C.M. (2002). The
expanding scale of species turnover events on coral reefs in Belize. Ecological
Monographs, 72, 233-249.
Aronson R.B., MacIntyre I.G., Wapnick C.M. & O'Neill M.W. (2004). Phase shifts,
alternative states, and the unprecedented convergence of two reef systems. Ecology,
85, 1876-1891.
Aronson R.B. & Precht W.F. (1997). Stasis, biological disturbance, and community structure
of a Holocene coral reef. Paleobiology, 23, 326-346.
Ayre D.J. & Hughes T.P. (2000). Genotypic diversity and gene flow in brooding and
spawning corals along the Great Barrier Reef, Australia. Evolution, 54, 1590-1605.
Balmford A., Gravestock P., Hockley N., McClean C.J. & Roberts C.M. (2004). The
worldwide costs of marine protected areas. Proceedings of the National Academy of
Sciences of the United States of America, 101, 9694-9697.
Bellwood D.R., Hoey A.S., Ackerman J.L. & Depczynski M. (2006). Coral bleaching, reef
fish community phase shifts and the resilience of coral reefs. Global Change Biology,
12, 1587-1594.
Berkelmans R., De'ath G., Kininmonth S. & Skirving W.J. (2004). A comparison of the 1998
and 2002 coral bleaching events on the Great Barrier Reef: spatial correlation,
patterns, and predictions. Coral Reefs, 23, 74-83.
118
Brooks T.M., Mittermeier R.A., Mittermeier C.G., da Fonseca G.A.B., Rylands A.B.,
Konstant W.B., Flick P., Pilgrim J., Oldfield S., Magin G. & Hilton-Taylor C. (2002).
Habitat loss and extinction in the hotspots of biodiversity. Conservation Biology, 16,
909-923.
Bruno J.F. & Selig E.R. (2007). Regional decline of coral cover in the Indo-Pacific: timing,
extent, and subregional comparisons. PLoS One, 2, e711.
Bruno J.F., Selig E.R., Casey K.S., Page C.A., Willis B.L., Harvell C.D., Sweatman H. &
Melendy A.M. (2007). Thermal stress and coral cover as drivers of coral disease
outbreaks. PLoS Biology, 5, e124.
Bruno J.F., Siddon C.E., Witman J.D., Colin P.L. & Toscano M.A. (2001). El Nino related
coral bleaching in Palau, Western Caroline Islands. Coral Reefs, 20, 127-136.
Burke L., Selig E.R. & Spalding M. (2002). Reefs at Risk in Southeast Asia. World
Resources Institute, Washington, DC.
Burnham K.P. & Anderson D.R. (2002). Model Selection and Multimodel Inference.
Springer-Verlag, New York.
Carpenter R.C. (1986). Partitioning herbivory and its effects on coral reef algal communities.
Ecological Monographs, 56, 345-364.
Chapin F.S., Zavaleta E.S., Eviner V.T., Naylor R.L., Vitousek P.M., Reynolds H.L., Hooper
D.U., Lavorel S., Sala O.E., Hobbie S.E., Mack M.C. & Diaz S. (2000).
Consequences of changing biodiversity. Nature, 405, 234-242.
Claudet J., Osenberg C.W., Benedetti-Cecchi L., Domenici P., Garcia-Charton J.-A., PerezRuzafa A., Badalamenti F., Bayle-Sempere J., Brito A., Bulleri F., Culioli J.-M.,
Dimech M., Falcon J.M., Guala I., Milazzo M., Sanchez-Meca J., Somerfield P.J.,
Stobart B., Vandeperre F., Valle C. & Planes S. (in press). Marine reserves: size and
age do matter. Ecology Letters.
Gelman A. & Pardoe L. (2006). Bayesian measures of explained variance and pooling in
multilevel (hierarchical) models. Technometrics, 48, 241-251.
Graham N.A.J., Evans R.D. & Russ G.R. (2003). The effects of marine reserve protection on
the trophic relationships of reef fishes on the Great Barrier Reef. Environmental
Conservation, 30, 200-208.
Graham N.A.J., Wilson S.K., Jennings S., Polunin N.V.C., Bijoux J.P. & Robinson J. (2006).
Dynamic fragility of oceanic coral reef ecosystems. Proceedings of the National
Academy of Sciences of the United States of America, 103, 8425-8429.
119
Graham N.A.J., Wilson S.K., Jennings S., Polunin N.V.C., Robinson J., Bijoux J.P. & Daw
T.M. (2007). Lag effects in the impacts of mass coral bleaching on coral reef fish,
fisheries, and ecosystems. Conservation Biology, 21, 1291-1300.
Grimsditch G.D. & Salm R.V. (2005). Coral reef resilience and resistance to bleaching. The
World Conservation Union, Gland.
Gubbay S. (1995). Marine protected areas - past, present, and future. In: Marine protected
areas - principles and techniques for management (ed. Gubbay S). Chapman & Hall
London.
Halford A., Cheal A.J., Ryan D. & Williams D.M. (2004). Resilience to large-scale
disturbance in coral and fish assemblages on the Great Barrier Reef. Ecology, 85,
1892-1905.
Halpern B.S. (2003). The impact of marine reserves: do reserves work and does reserve size
matter? Ecological Applications, 13, S117-S137.
Halpern B.S., Gaines S.D. & Warner R.R. (2004). Confounding effects of the export of
production and the displacement of fishing effort from marine reserves. Ecological
Applications, 14, 1248-1256.
Hixon M.A. & Beets J.P. (1993). Predation, prey refuges, and the structure of coral-reef fish
assemblages. Ecological Monographs, 63, 77-101.
Holling C.S. (1973). Resilience and stability of ecological systems. Annual Review of
Ecology and Systematics, 4, 1-23.
Hubbard D.K., Zankl H., Van Heerden I. & Gill I.P. (2005). Holocene reef development
along the northeastern St. Croix shelf, Buck Island, US Virgin Islands. Journal of
Sedimentary Research, 75, 97-113.
Hughes T.P., Bellwood D.R., Folke C., Steneck R.S. & Wilson J. (2005). New paradigms for
supporting the resilience of marine ecosystems. Trends in Ecology & Evolution, 20,
380-386.
Hughes T.P., Rodrigues M.J., Bellwood D.R., Ceccarelli D., Hoegh-Guldberg O., McCook
L., Moltschaniwskyj N., Pratchett M.S., Steneck R.S. & Willis B. (2007). Phase
shifts, herbivory, and the resilience of coral reefs to climate change. Current Biology,
17, 360-365.
Jackson J.B.C. (1997). Reefs since Columbus. Coral Reefs, 16, S23-S32.
Jones G.P., McCormick M.I., Srinivasan M. & Eagle J.V. (2004). Coral decline threatens fish
biodiversity in marine reserves. Proceedings of the National Academy of Sciences of
the United States of America, 101, 8251-8253.
120
Ledlie M.H., Graham N.A.J., Bythell J.C., Wilson S.K., Jennings S., Polunin N.V.C. &
Hardcastle J. (2007). Phase shifts and the role of herbivory in the resilience of coral
reefs. Coral Reefs, 26, 641-653.
Lesaffre E., Rizopoulos D. & Tsonaka R. (2007). The logistic transform for bounded
outcome scores. Biostatistics, 8, 72-85.
Lubchenco J., Palumbi S.R., Gaines S.D. & Andelman S. (2003). Plugging a hole in the
ocean: the emerging science of marine reserves. Ecological Applications, 13, S3-S7.
Lunn D.J., Thomas A., Best N. & Spiegelhalter D. (2000). WinBUGS - a Bayesian modelling
framework: concepts, structure, and extensibility. Statistics and Computing, 10, 325337.
Madin J.S. & Connolly S.R. (2006). Ecological consequences of major hydrodynamic
disturbances on coral reefs. Nature, 444, 477-480.
Maliao R.J., Webb E.L. & Jensen K.R. (2004). A survey of stock of the donkey's ear abalone,
Haliotis asinina L. in the Sagay Marine Reserve, Philippines: evaluating the
effectiveness of marine protected area enforcement. Fisheries Research, 66, 343-353.
McClanahan T.R. (1999). Is there a future for coral reef parks in poor tropical countries?
Coral Reefs, 18, 321-325.
McClanahan T.R., Graham N.A.J., Calnan J.M. & MacNeil M.A. (2007). Toward pristine
biomass: reef fish recovery in coral reef marine protected areas in Kenya. Ecological
Applications, 17, 1055-1067.
McClanahan T.R. & Mangi S. (2000). Spillover of exploitable fishes from a marine park and
its effect on the adjacent fishery. Ecological Applications, 10, 1792-1805.
McMahon S.M. & Diez J.M. (2007). Scales of association: hierarchical linear models and the
measurement of ecological systems. Ecology Letters, 10, 437-452.
Mora C., Andrefouet S., Costello M.J., Kranenburg C., Rollo A., Veron J., Gaston K.J. &
Myers R.A. (2006). Coral reefs and the global network of marine protected areas.
Science, 312, 1750-1751.
Mumby P.J., Dahlgren C.P., Harborne A.R., Kappel C.V., Micheli F., Brumbaugh D.R.,
Holmes K.E., Mendes J.M., Broad K., Sanchirico J.N., Buch K., Box S., Stoffle R.W.
& Gill A.B. (2006). Fishing, trophic cascades, and the process of grazing on coral
reefs. Science, 311, 98-101.
Mumby P.J., Harborne A.R., Williams J., Kappel C.V., Brumbaugh D.R., Micheli F., Holmes
K.E., Dahlgren C.P., Paris C.B. & Blackwell P.G. (2007). Trophic cascade facilitates
121
coral recruitment in a marine reserve. Proceedings of the National Academy of
Sciences of the United States of America, 104, 8362-8367.
Murawski S.A., Brown R., Lai H.L., Rago P.J. & Hendrickson L. (2000). Large-scale closed
areas as a fishery-management tool in temperate marine systems: the Georges Bank
experience. Bulletin of Marine Science, 66, 775-798.
Murawski S.A., Wigley S.E., Fogarty M.J., Rago P.J. & Mountain D.G. (2005). Effort
distribution and catch patterns adjacent to temperate MPAs. Ices Journal of Marine
Science, 62, 1150-1167.
Newman M.J.H., Paredes G.A., Sala E. & Jackson J.B.C. (2006). Structure of Caribbean
coral reef communities across a large gradient of fish biomass. Ecology Letters, 9,
1216-1227.
Pauly D., Christensen V., Dalsgaard J., Froese R. & Torres F. (1998). Fishing down marine
food webs. Science, 279, 860-863.
Pinheiro J., Bates D., DebRoy S. & Sarkar D. (2007). nlme: Linear and Nonlinear Mixed
Effects Models. R package version 3.1-83 edn.
R Development Core Team (2007). R: A language and environment for statistical computing.
In. R Foundation for Statistical Computing Vienna, Austria.
Roberts C.M., Bohnsack J.A., Gell F., Hawkins J.P. & Goodridge R. (2001). Effects of
marine reserves on adjacent fisheries. Science, 294, 1920-1923.
Rotjan R.D., Dimond J.L., Thornhill D.J., Leichter J.J., Helmuth B., Kemp D.W. & Lewis
S.M. (2006). Chronic parrotfish grazing impedes coral recovery after bleaching.
Coral Reefs, 25, 361-368.
Russ G.R. & Alcala A.C. (2003). Marine reserves: rates and patterns of recovery and decline
of predatory fish, 1983-2000. Ecological Applications, 13, 1553-1565.
Russ G.R., Alcala A.C., Maypa A.P., Calumpong H.P. & White A.T. (2004). Marine reserve
benefits local fisheries. Ecological Applications, 14, 597-606.
Sandin S.A., Smith J.E., DeMartini E.E., Dinsdale E.A., Donner S.D., Friedlander A.M.,
Konotchick T., Malay M., Maragos J.E., Obura D., Pantos O., Paulay G., Richie M.,
Rohwer F., Schroeder R.E., Walsh S., Jackson J.B.C., Knowlton N. & Sala E. (2008).
Baselines and degradation of coral reefs in the northern Line Islands. PLoS ONE, 3,
e1548.
Sheppard C. & Rioja-Nieto R. (2005). Sea surface temperature 1871-2099 in 38 cells in the
Caribbean region. Marine Environmental Research, 60, 389-396.
122
Sheppard C.R.C. (2003). Predicted recurrences of mass coral mortality in the Indian Ocean.
Nature, 425, 294-297.
Singer J.D. & Willet J.B. (2003). Applied longitudinal data analysis: modeling change and
event occurrence. Oxford University Press, Oxford, UK.
Toms J.D. & Lesperance M.L. (2003). Piecewise regression: a tool for identifying ecological
thresholds. Ecology, 84, 2034-2041.
Wapnick C.M., Precht W.F. & Aronson R.B. (2004). Millennial-scale dynamics of staghorn
coral in Discovery Bay, Jamaica. Ecology Letters, 7, 354-361.
Ward J.R. & Lafferty K.D. (2004). The elusive baseline of marine disease: are diseases in
ocean ecosystems increasing? PloS Biology, 2, 542-547.
Williamson D.H., Russ G.R. & Ayling A.M. (2004). No-take marine reserves increase
abundance and biomass of reef fish on inshore fringing reefs of the Great Barrier
Reef. Environmental Conservation, 31, 149-159.
Wilson S.K., Graham N.A.J., Pratchett M.S., Jones G.P. & Polunin N.V.C. (2006). Multiple
disturbances and the global degradation of coral reefs: are reef fishes at risk or
resilient? Global Change Biology, 12, 2220-2234.
Wing S.R. & Wing E.S. (2001). Prehistoric fisheries in the Caribbean. Coral Reefs, 20, 1-8.
Wolanski E. & De'ath G. (2005). Predicting the impact of present and future human land-use
on the Great Barrier Reef. Estuarine Coastal and Shelf Science, 64, 504-508.
Worm B., Sandow M., Oschlies A., Lotze H.K. & Myers R.A. (2005). Global patterns of
predator diversity in the open oceans. Science, 309, 1365-1369.
Yu K.F., Zhao J.X., Shi Q., Chen T.G., Wang P.X., Collerson K.D. & Liu T.S. (2006). Useries dating of dead Porites corals in the South China sea: evidence for episodic coral
mortality over the past two centuries. Quaternary Geochronology, 1, 129-141.
123
CHAPTER 4
Temperature-driven coral decline: do marine protected areas have a role?
Abstract
Climate change is predicted to have devastating impacts on coral reef health. One
proposed conservation strategy for managing the effects of climate change on coral reef
ecosystems is the use of marine protected areas (MPAs). This strategy assumes that MPAs
can indirectly protect coral populations from ocean warming through fisheries management
and reductions in terrestrial run-off. Using temperature anomaly and coral cover databases,
we quantified the effect of temperature anomalies and MPA protection on the trend in coral
cover from 1987-2005 globally. Protection did not mitigate the negative effect of warm
temperature anomalies on coral cover. In addition, the magnitude of the MPA benefit was
generally not enough to compensate for the negative effect of temperature anomalies on coral
cover. MPAs also did not capture the natural range of temperature variability on reefs. The
current placement of MPAs overprotects areas with very few temperature anomalies and
underprotects areas with moderate numbers of anomalies. Field research suggests that
locations with moderate anomalies may be more acclimated to thermal stress and better able
to withstand bleaching. If MPAs are located more strategically to protect naturally
acclimated populations, protection may provide a synergistic benefit. However, the failure of
MPAs to lessen the negative effect of thermal stress on coral cover suggests that although
MPAs are important conservation tools, they need to be complemented with additional
policies aimed at reducing the activities responsible for climate change.
Introduction
The pervasiveness of climate change presents unique challenges for resource
management. Traditional management tools like parks and reserves have the potential to be
effective in restoring endangered populations, preventing habitat destruction (Bruner et al.
2001), and preserving ecosystem services (Balmford et al. 2002). However, climate change
is expected to alter the physical factors necessary for habitat development, complicating
management within static boundaries like parks or protected areas. Within park boundaries,
changes in temperature and precipitation may shift species ranges and abundances potentially
threatening the persistence of the ecosystem itself (Parmesan & Yohe 2003). Through
distribution shifts and fundamental changes in community structure, non-dynamic protected
areas could end up failing to even include the species or ecosystems they were designed to
conserve.
In spite of these challenges, there is still optimism that protected areas can be useful
for conservation in the face of climate change if they can increase ecosystem resilience.
Ecosystem resilience (sensu Holling 1973) is broadly defined as the rate of return to a predisturbed state, a measure of the persistence of relationships within an ecosystem. Marine
protected areas (MPAs) are thought to be able to increase resilience to climate change by
directly mitigating other stressors like overfishing (Hughes et al. 2003; Grimsditch & Salm
2005). In addition, it is theorized that conservation benefits can be maximized by protecting
125
naturally resilient areas, which have higher diversity or connectivity (Gunderson 2000;
Bengtsson et al. 2003; Hughes et al. 2003; West & Salm 2003; Bellwood et al. 2004; Adger
et al. 2005). If protected areas can enhance ecosystem resilience, they may be able to
moderate climate change effects by creating the conditions necessary for recovery from
disturbances.
To date, much of the speculation about the use of protected areas to increase
resilience to climate change effects has focused on coral reef ecosystems (Hughes et al.
2003; West & Salm 2003; Bellwood et al. 2004; Obura 2005; Wooldridge et al. 2005).
Climate change could threaten coral reef ecosystems through rising sea levels (Blanchon &
Shaw 1995) and ocean acidification (Hoegh-Guldberg et al. 2007). In the short-term,
though, corals are particularly vulnerable to climate change because reef-building corals
already live near their upper thermal limits (Jokiel & Coles 1990). When corals experience
temperatures more than 1°C beyond typical maximum summertime averages, they can lose
their symbiotic algae or zooxanthellae in a process known as coral bleaching (Glynn 1993).
Although corals can recover from bleaching in response to relatively short and weak
temperature anomalies, prolonged high temperatures like those associated with the 1998 El
Niño can cause extensive bleaching and mortality on individual reefs (Arceo et al. 2001;
Bruno et al. 2001; Graham et al. 2007). Warm temperature anomalies also triggered
subsequent mass bleaching events in 2002 on the Great Barrier Reef and 2005 in the eastern
Caribbean (Berkelmans et al. 2004; Whelan et al. 2007). Most climate change scenarios
predict an increase in the frequency and severity of these thermal stress events (HoeghGuldberg 1999; Sheppard 2003; Sheppard & Rioja-Nieto 2005).
126
Marine protected areas (MPAs) have been proposed as tools for mitigating the effects
of warm temperature anomalies by reducing or eliminating fishing. Prohibiting fisheries
exploitation in temperate and coral reef MPAs has been shown to restore natural trophic
structure (Shears & Babcock 2003; Guidetti 2006; Mumby et al. 2007a). Recent work has
found that individual MPAs can have a positive effect on coral cover through trophic
cascades, which limit algal growth and facilitate coral recruitment (Mumby et al. 2006;
Mumby et al. 2007b). Large scale analyses have also found that protection in MPAs
maintained coral cover levels over time, while unprotected reefs continued to decline (Selig
& Bruno in prep).
However, studies supporting the hypothesis that MPAs could be useful in
conservation efforts to protect corals from thermal stress are equivocal and generally
restricted to single reserves. A more complete suite of herbivorous fish could theoretically
suppress macroalgal growth following bleaching events. After bleaching events, algae can
colonize dead coral skeleton (Diaz-Pulido & McCook 2002), potentially reducing coral
settlement (Birrell et al. 2005) and suppressing coral growth (River & Edmunds 2001).
Nonetheless, the effects of a restoration to historic trophic structure are not well understood
and may not necessarily facilitate coral recovery after thermal events. An analysis of Cousin
Island Marine Reserve in the Seychelles found little recovery from major coral mortality
following the 1998 El Niño event in spite of unchanging high herbivore abundance over the
study period (Ledlie et al. 2007). In addition, Jones et al. (2004) found that protection did
not mitigate a decline in coral cover or a subsequent decline in coral-dependent fish. From
these studies, it remains unclear whether the reduction in fishing in MPAs can increase
recovery from thermal stress at regional scales.
127
Globally, the oceans have warmed 0.52°C ± 0.19°C between 1850 and 2004 with
some regional variability (Casey & Cornillon 2001; Rayner et al. 2006). If warming
continues at a similar rate, current thermal limits for most corals will be exceeded in the next
100 years without rapid acclimation or adaptation (Donner et al. 2005; Hoegh-Guldberg et
al. 2007). Even rises of 0.1°C could increase the geographic extent of bleaching in the
Caribbean by 42% (McWilliams et al. 2005). Therefore, testing whether MPAs can be
useful in mitigating temperature-associated coral loss is critical so that we can determine
what management interventions are needed before there is further loss. At least one strategy
suggests that placing MPAs in locations with specific temperature profiles may increase
overall coral reef ecosystem resilience (West & Salm 2003; Grimsditch & Salm 2005). Data
are needed for assessing whether these strategies could work and for guiding management
decisions about designing MPAs with the goal of reducing the effects of thermal stress.
We used spatial databases on temperature anomalies (Selig et al. in prep), MPAs, and
live coral cover to test whether locations with similar numbers of temperature anomalies
experienced fewer declines in coral cover inside MPAs than in unprotected areas. Failure to
mitigate the effect of temperature anomalies could be explained by at least two factors. If
MPAs are too small, they may be susceptible to widespread mortality from a single anomaly
event and not have enough healthy populations to restore degraded areas. Therefore, we
analyzed patterns in temperature anomaly areas to compare them with MPA areas. In
addition, if MPAs are not protecting natural temperature variability on reefs, they may have
populations that are less thermally tolerant to stress. We assessed whether temperature
anomaly patterns of reefs within MPAs were similar to those of all reefs. With this analysis,
we were able to determine if MPAs were disproportionately protecting reefs with low,
128
medium, or high numbers of anomalies. Overall, these analyses enabled us to evaluate
whether MPAs can buffer the effects of temperature anomalies and to identify possible
shortcoming in the design of the current set of MPAs.
Methods
Temperature anomaly database
We created a 21-year dataset of temperature anomalies covering the entire range of
tropical reef-building corals (37°N to 37°S) using the National Oceanic and Atmospheric
Administration’s (NOAA) National Oceanographic Data Center and the University of
Miami’s Rosenstiel School of Marine and Atmospheric Science Pathfinder Version 5.0
temperature data (http://pathfinder.nodc.noaa.gov). These data have the highest spatial
resolution for the longest time period of any publicly available satellite temperature data. To
avoid diurnal biases (Casey 2002) and reduce the number of missing pixels by 25%, we
averaged the day and night measurements. In addition, we used data with a quality flag of 4
or better (Casey & Cornillon 1999; Kilpatrick et al. 2001). We initially filled the remaining
data gaps (21.2% of the data) using a 3 x 3 pixel median spatial fill; remaining gaps were
filled temporally using the Piecewise Cubic Hermite Interpolating Polynomial (PCHIP)
function in Matlab (The Mathworks Inc. 2006). Climatologies were created using a
harmonic analysis procedure to fit the annual and semi-annual signals to the time-series at
each grid cell location (Selig et al. in prep).
We defined thermal stress anomalies (TSA) as deviations of 1°C or greater from the
temperature during the mean maximum climatological week from 1985-2005. The mean
129
maximum climatological week is the long-term average warmest week. Although specific
temperatures that cause coral bleaching and mortality are highly species-specific (Fisk &
Done 1985; Lang et al. 1992; Glynn 1996; Berkelmans & Willis 1999; Marshall & Baird
2000; Loya et al. 2001), this is a generally accepted threshold for conditions that may result
in bleaching (Glynn 1996; Liu et al. 2003; Berkelmans et al. 2004).
Quantifying patterns in anomaly areas by region
We analyzed patterns in anomaly area throughout the tropics and by region to
determine anomaly area statistics for geographic areas with similar diversity or management
(Fig. 4.1). These regions roughly correspond to marine ecoregions designated by nongovernmental organizations for conservation priority setting (Spalding et al. 2007). Analyses
were done by region because different oceanographic patterns may lead to variability in
anomaly areas. Knowing how anomaly areas vary by region can give managers and
policymakers more precise data for their location. To determine anomaly areas, we first
created 1096 spatial grids of anomaly presence and absence for each week of the 21-year
temperature anomaly database. Only anomalies which contained at least one pixel that
overlapped a known coral reef location were included. We used the Image Processing
Toolbox in Matlab 7.3 (bwlabel function) to identify each anomaly and determine whether it
was connected to a neighboring thermal event in any of the eight adjacent pixels (The
Mathworks Inc. 2006). Each contiguous temperature anomaly group was assigned a unique
number from which we calculated the total area of the group and the mean, maximum, and
variance in area size for each time step and region.
130
Marine protected area and coral reef databases
We built a spatial database of marine protected areas primarily using data from the
publicly available World Database on Protected Areas 2005 (WDPA Consortium 2005).
These data were then supplemented and updated with MPA data from The Great Barrier Reef
Marine Park Authority, The Nature Conservancy, NOAA, the US Geological Survey and the
US Department of Interior. Most of the MPA data had exact boundaries, but some locations
had only information about the center latitude and longitude point of the MPA location. Of
the 298 MPAs used in this analysis, approximately 25 had information on total area but not
actual boundaries. The extent of these areas had to be approximated and area calculations
represent estimates based on the best available data. For areas in which we only had point
data, we created artificial circular boundaries based on the total known area of the MPA
(Mora et al. 2006). Because we were interested only in coral reef MPAs, we selected only
parks that fell within 50 km of a known coral reef location. Coral reef location data varies in
quality for much of the world, but we compiled known locations from Reefbase
(www.reefbase.org), Reefs at Risk (Bryant et al. 1998), Reefs at Risk in Southeast Asia
(Burke et al. 2002), and Reefs at Risk in the Caribbean (Burke & Maidens 2004).
Live coral cover database
We developed a live coral cover database from several publicly available databases
(Reef Check, ReefBase, and Reefs at Risk in Southeast Asia) as well as peer-reviewed and
grey literature. To find relevant articles, we searched the ISI Web of Science and Google
Scholar with the terms "coral" and "cover" as well as "reef" and "health". We also examined
all available issues of journals that were likely to contain coral reef studies including Atoll
131
Research Bulletin, Coral Reefs, Marine Pollution Bulletin, and the Proceedings of the
International Coral Reef Symposium. In general, the surveys used some variant of the line
transect technique. We only used data from 15 m depth or above to avoid depth biases in
coral cover. Our final global database contained 8040 surveys over 4260 reefs from 19872005.
Statistical analyses
Multilevel model development
We used a multilevel model to evaluate whether protection in MPAs changed coral
cover responses to thermal stress from 1987-2005 compared to unprotected corals. By using
a multilevel model, we were able to estimate how protection and thermal stress drove
variability in coral cover over time on each reef as well as a population average for all reefs.
The multilevel modeling approach also allowed us to account for the temporal and spatial
correlation in the data (McMahon & Diez 2007). At each level we assigned different random
effects to account for spatial and temporal correlation. For example, at level 1, individual
observations on a reef were assigned a unique random effect. At level 2, the random effect
accounted for repeated surveys on the same reef over time. Level 3 accounted for spatial
correlation within the spatial grouping unit where protected and unprotected reefs were
paired. Unprotected reefs were paired with the nearest protected area that contained coral
reef surveys up to a distance of 200 km (see Selig & Bruno in prep for full methods). Then
we applied a logit transformation to the percent coral cover data and treated the logit as
normally distributed (Lesaffre et al. 2007). We also centered our time regressor at 1996 to
facilitate model convergence, provide an interpretable intercept, and to minimize correlation
132
among random effects (Singer & Willet 2003). Centering is the equivalent of examining
changes relative to a chosen reference year, in this case, 1996. We used the nlme library
(Pinheiro et al. 2007) in R 2.5.1 (R Core Development Team, 2007) and WinBUGS 1.4.3
(Lunn et al. 2000) to analyze our models.
After determining the basic model structure, we explored how to incorporate
temperature into the model. Studies suggest that there can be a temporal delay from the
beginning of a temperature anomaly to a biological response in coral cover or fisheries
(Graham et al. 2007). Therefore, we examined models with no lag and variations of 1, 2, and
3 year lags. Using Akaike Information Criterion (AIC) we found that a variable that
contained the number of weekly anomalies in the 2 calendar years preceding the year of each
coral cover survey (lag TSA) best explained variation in coral cover (Akaike 1973).
The basic trend model Eqn. (1) we fit to each reef described logit coral cover (Y ) in
terms of year (T) and its centering constant ( Tc = 1996), the number of anomalies in the 2
years preceding the coral cover survey (lag TSA) and the random error ( ε ). The subscripts i,
j, and k designate the spatial unit for pairing MPA and non-MPA reefs, the reefs within that
unit, and the individual survey measurement for that reef, respectively.
Level 1:
(
Yijk = β 0ij + β1ij (Tijk − Tc ) + β 2ij lag TSAijk + ε ijk , ε ijk ~ N 0,σ 2
)
(1)
The subscript j appearing in the parameters β 0ij (intercept), β1ij (trend), β 2ij (lag TSA)
signifies that these parameters are allowed to vary from reef to reef. Their variability is
described in Eqn. (2):
133
β 0ij = β 0i + β 3i Protection statusij + u0ij
Level 2:
⎛ ⎡0⎤ ⎡τ 2 τ 01 ⎤ ⎞
⎡u 0ij ⎤
⎟
β1ij = β1i + β 4i Protection statusij + u1ij , ⎢ ⎥ ~ N⎜⎜ ⎢ ⎥, ⎢ 0
2 ⎥⎟
0
τ
τ
⎦
⎣
01
1
⎣ u1ij ⎦
⎦
⎣
⎠
⎝
β 2ij = β 2i
(2)
Here, Protection status is a reef-level predictor that varies among reefs (protected versus
unprotected) but is constant for observations taken on the same reef. Random effects for reef
j in spatial unit i ( u0ij , u1ij , and u 2ij ) are assumed to have a joint normal distribution as shown.
Protection status is a dummy variable where 1 = protected and 0 = unprotected. Random
effects from different reefs are assumed to be independent and also independent of the level 1
error terms. The subscript i on the trend, lag TSA, and intercept parameters, as well as the
coefficients of the level 2 predictor Protection Statusij in Eqn. (2) indicates that these
parameters can vary across spatial units as described by Eqn. (3):
β 0i = β 0 + β 5 Indiani + β 6 Pacifici + v0i
β1i = β1 + v1i
Level 3:
⎛ ⎡0⎤ ⎡ ω 2 ω 01 ⎤ ⎞
⎡v ⎤
⎟
, ⎢ 0 i ⎥ ~ N⎜ ⎢ ⎥ , ⎢ 0
2 ⎥⎟
⎜ ⎣0 ⎦ ω
ω
⎣ v1i ⎦
01
1
⎣
⎦
⎝
⎠
β 2i = β 2
β 3i = β 3
(3)
β 4i = β 4
In Eqn. (3), we included ocean basin as a level 3 predictor because earlier modeling efforts
(Selig & Bruno in prep) and different temperature patterns in the different basins indicated
that coral cover may be modified by ocean basin. The Caribbean serves as the baseline level
134
in this model and corresponds to Pacific = Indian = 0. Random effects for level 3 are
designated by v0i and v1i .
To develop a final model, we determined the basic level-1 model, added random
effects where statistically necessary, and then added additional predictors (Protection status
and ocean). We tried several variations on the basic level-1 model shown in Eqn. (1)
including the interaction of time (coral cover trend) and lag TSA, but rejected them using
AIC. By fitting models to individual reefs, we observed that intercepts (logit coral cover in
1996) varied widely across reefs and spatial grouping units, whereas coral cover trend
coefficients were less variable and lag TSA coefficients had almost no variability. Random
effects account for unobserved heterogeneity in the model so we included random effects
only for the intercept and trend. Then we determined how to add Protection Status and
ocean as predictors, using AIC to identify which model was the best fit model (Akaike 1973).
To obtain more realistic estimates of parameter precision and credibility intervals, we refit
the final model as a Bayesian model with uninformative priors in WinBUGS 1.4.3 (Lunn et
al. 2000). Credibility intervals are the range within which there is a defined probability that
the true value lies. For example, within a 95% credibility interval, there is a 95% probability
that the true value lies within the given calculated range.
Monte Carlo simulations
We used Monte Carlo simulations to test whether or not temperature anomaly
patterns varied significantly between reefs in MPAs and all reefs. For this analysis, we
focused on 7 years of interest: 1985, 1988, 1995, 1998, 2000, 2002, and 2005. The years
1988, 1998, 2002, and 2005 were selected because they were El Nino years (1988, 1995,
135
1998, and 2002) or years with a documented major thermal event (2005). We included 1985
and 2000 as reference years. To assess whether the temperature anomaly patterns exhibited
by reefs in MPAs are typical of all reefs, we first treated all reefs including those within
MPAs to be a theoretical population. We then drew 10,000 random samples of size 10,555
(the number of 4 km reef pixels in MPAs that matched the temperature database pixels) from
this theoretical population of 55,626 reefs. For each sample, we calculated the variance,
interquartile range (IQR), median absolute deviation (mad), mean and median. The
frequency distribution of each summary statistic calculated from the random samples was
then used to estimate the sampling distribution of that statistic. The observed value of the
statistic for MPAs was compared to the theoretical population distribution of that statistic to
determine if the MPA value was typical. We then used Pearson's Χ2 Goodness of fit test to
get more specific information about how the general reef and MPA distributions varied.
Because of our large sample size, formal significance testing would virtually guarantee
finding a significant lack of fit. However, we were able to use the Pearson residuals to
determine which temperature anomaly categories were driving the lack of fit.
Results
Effect of protection and temperature anomalies on coral cover
The best model described coral cover as a function of time, thermal stress, protection,
and ocean. Protection status (a level-2 predictor) modified both the change in coral cover
and intercept (coral cover in 1996, the centering year), while ocean basin affected only the
intercept (level 3 predictor). Random effects were necessary for both reefs and spatial
grouping units. Using an AIC framework, we found the best model to be given by Eqn. (4):
136
logit ( pijk ) = β 0 + v0i + v2i + β 2ijTSAijk + β 3ij MPAijk + β 7 Indiani + β8 Pacifici
+ u0ij + u2ij + (β1 + v1i + β 4i MPAij + u1ij )(Tijk − Tc ) + ε ijk
(4)
We found that protection in MPAs did not mitigate the effect of temperature
anomalies on coral cover (Figs. 4.2-4.3). As with previous work, MPA status had a
significant effect on the direction and magnitude of coral cover trend (Selig & Bruno in
prep). The trend outside MPAs was negative, indicating a decline in coral cover over time.
Within MPAs, though, the coral cover trend was weakly positive, although not significantly
different from zero (Fig. 4.2). However, when we fit a model that allowed MPA to modify
the effect of temperature anomalies on coral cover, the interaction term was not significant
(p>0.1) and the model was not a better fit in an AIC context. Therefore, MPAs did not
modify the effect of thermal stress on coral cover (Fig. 4.3). In other words, MPAs may
benefit corals through the mitigation of other stressors (i.e., overfishing and terrestrial
inputs), but they are not specifically altering the effect of temperature anomalies on coral
cover.
Temperature anomalies had a significant negative effect on coral cover (P<.0001),
Although this finding is consistent with numerous studies relating temperature anomalies to
bleaching and coral mortality (Arceo et al. 2001; Bruno et al. 2001; Berkelmans et al. 2004),
our model quantified the general rate of coral cover loss as a function of the number of
temperature anomalies (Fig. 4.4). For example, across reefs, 8 anomalies of more than 1°C
above summertime averages were correlated with a 3.9% loss in coral cover.
137
Temperature anomaly patterns of MPAs versus all reefs
Protected reefs experienced less thermal stress than the general population of reefs.
Weekly temperature anomaly patterns in MPAs did not represent a random sample of those
on all reefs. For the years studied, MPAs had more reefs with low numbers of anomalies (01) than the general population of reefs. MPAs also included fewer reefs with moderate
anomaly frequencies (2-6). This result does vary among years and 1998 and 2002 were
exceptions to the overall pattern. During the 1998 El Niño, MPAs underrepresented reefs
with low numbers of anomalies, overrepresented reefs with anomaly frequencies 1-5 and
underrepresented reefs with moderate numbers of anomalies (6-12). In addition, the 1998
event was the most deviant from a random sample of temperature anomalies with nearly
every anomaly category exceeding the Pearson critical value. In contrast, low anomaly
numbers (0-2) were underrepresented in MPAs in 2002 and there was an overrepresentation
of moderate numbers of anomalies (3-8). For each year, we took the 95% percentile-based
confidence intervals from the Monte Carlo simulations and compared them to MPA anomaly
patterns. From these analyses, we determined that MPAs generally had lower mean anomaly
values (Fig. 4.5A) and lower standard deviations (Fig. 4.5B) than all reefs. Our results
suggest that locations protected within MPAs experience different temperature patterns than
all reefs and do not contain the natural range of temperature variation.
Patterns in anomaly area
The overall event size anomaly distribution illustrates that 70% of anomalies are less
than 75 km2, but that a few anomalies are over 1,000,000 km2. Anomaly event sizes varied
annually and regionally (Figs. 4.1 and 4.6). At a regional scale, variability in the size of
138
smaller anomalies was generally low (under 100 km2). Most of the variability occurred in
the tail of the distribution where there were few large anomalies (Fig. 4.6). Contrary to
expectations, the size of the region seemingly did not affect the pattern of anomaly sizes.
The smallest region in the analysis, the Florida Keys, also had some of the largest anomaly
sizes. More size variability was present in Pacific regions that are affected by El Niño events
(Fig. 4.6).
Discussion
The success of marine protected areas (MPAs) in restoring fisheries on coral reefs
(Alcala et al. 2005; McClanahan et al. 2007c) has led to increased optimism that they may
also be useful for mitigating temperature stress associated with climate change (Sandin et al.
2008). However, our global analysis found that there is no evidence to suggest that MPAs
are reducing the effects of temperature anomalies on coral cover in general. Although MPAs
helped to reduce coral decline, they did not modify the effect of thermal stress on coral cover
(Fig. 4.3).
The effects of temperature anomalies on bleaching rates and coral cover on individual
reefs have been widely documented (Glynn et al. 1988; Goreau & Hayes 1994; Arceo et al.
2001; Bruno et al. 2001; McClanahan et al. 2007b). However, the general relationship
between the number of temperature anomalies and the percentage loss in coral cover has not
been described. We found a negative relationship between the number of anomalies and
change in coral cover (Fig. 4.4A). The degree of percent coral cover loss depended on the
number of anomalies and the starting coral cover (Fig. 4.4A). For example, at 50% coral
139
cover, 4 anomalies resulted in coral cover losses of approximately 2% (Fig. 4.4B). Bleaching
as a result of temperature anomalies is highly species- and location-specific (Berkelmans &
Willis 1999) and mortality on a given reef can approach 90% during episodes of high
anomaly frequency or magnitude (Graham et al. 2006). Clearly bleaching rates on individual
reefs can be highly idiosyncratic, but our results suggest that there was a general relationship
between the number of temperature anomalies, coral cover at that time, and percent coral
cover lost during thermal events.
In addition, when the magnitude of coral loss as a result of temperature anomalies
was compared to the magnitude of the positive effects of MPAs, our results indicated that
MPAs alone may not be sufficient to mitigate coral loss due to temperature anomalies.
Under optimal conditions, MPAs generally resulted in increases in coral cover of 1-2% per
year (Selig & Bruno in prep). Therefore, the balance between coral loss from temperature
anomalies and the benefits from MPAs are unlikely to produce a net gain in coral cover over
time on most reefs.
Several factors could explain the failure of MPAs to protect against thermal stress.
One possibility is that MPAs are currently not optimally designed because of their placement
or their size. One conservation scheme prescribes protecting populations that have had a
history of little or no temperature stress and populations that have had only moderate levels
of stress (Grimsditch & Salm 2005). When we compared the thermal profiles of the current
set of MPAs to the general population of all reefs, we found that MPAs had a significantly
different number of anomalies from the distribution of anomalies on all reefs. In fact, MPAs
disproportionately protected low anomaly frequency areas. These areas of low thermal stress
likely result from local oceanographic conditions like upwelling. However, field research
140
suggests that populations that have not been subjected to temperature anomalies undergo
greater mortality when they do experience thermal stress (McClanahan et al. 2007a).
Including sites with a history of more moderate numbers of temperature anomalies
may be necessary for developing a set of MPAs capable of slowing or reversing coral decline
because these locations contain more acclimated corals. Coral acclimation and adaptation to
thermal stress through exchange of their zooxanthellae is routinely invoked as a possible
mechanism for corals to survive increases in ocean temperature (Baker 2001; Coles & Brown
2003; Baker et al. 2004; Fautin & Buddemeier 2004). Successful acclimation or adaptation
could occur through the acquisition of thermally-tolerant zooxanthellae, potentially during
warm water events that result in non-lethal bleaching (Fautin & Buddemeier 2004). For
several years during our study period, MPAs had substantially lower anomaly frequencies
and less temperature variability than all reefs (Fig. 4.5B). An analysis of field sites in East
Africa found that sites which had more temperature variability experienced less mortality
during thermal stress events (McClanahan et al. 2007a). Therefore, the lack of representation
of reefs with more temperature variability and moderate numbers of anomalies in MPAs
could partially explain why protection does not modify the effect of warm temperature
anomalies on coral cover.
In addition, our analysis of anomaly sizes suggests that MPAs may be too small to
adequately protect against thermal stress. A key component of protected area design has been
to create reserves larger than the typical natural disturbance regime so that enough species
and ecosystem function remain to recover from disturbances (Pickett & Thompson 1978).
We found that anomalies varied considerably in size based on region and year. Most
anomaly events are relatively small, but the overall distribution usually includes a small
141
number of large area, particularly during major events like the strong El Niño in 1998 when a
few weeks had anomaly sizes that exceeded 1 million km2. Because no MPA can encompass
such a large contiguous area, managers may need to design MPAs to insure against a
reasonable degree of thermal stress rather than a maximum by constructing MPAs to be
larger than an acceptable percentage of anomaly events (Fig. 4.6). For example, in the
Western Indonesia region, an MPA greater than 50 km2 would be larger than 50% of
anomaly events and at 150 km2, it would be larger than 75% of anomaly events. Currently,
about 40% of coral MPAs are 1-2 km2 (Mora et al. 2006), considerably smaller than a typical
anomaly. Therefore, most MPAs are likely to be wholly affected by a single anomaly event.
Larger MPAs could confer greater resilience from thermal anomaly events because they
would be more likely to contain unaffected populations within their boundaries. These
population might then be able to reseed those that had experienced mortality during the event
(Grimsditch & Salm 2005).
Regardless of MPA design, a resilience effect of MPAs on coral cover can only be
mediated indirectly through the two stressors that MPAs can directly affect. MPAs have
direct effects on coral reef fish populations through the restoration of historic abundance and
age structure (McClanahan & Mangi 2000; Roberts et al. 2001; Halpern 2003; Russ et al.
2004; Williamson et al. 2004). They can also have local direct effects on the reduction of
nutrients and sediments if they include a terrestrial component (Grigg 1995; Wolanski &
De'ath 2005). Through the mitigation of these stressors, MPAs can provide a mechanistic
indirect benefit to corals (Mumby et al. 2007a; Selig & Bruno in prep). Coral cover is an
important metric of resilience because live coral provides the foundation structure for the
entire reef ecosystem and many reef taxa are dependent on its framework (Bruno & Bertness
142
2001; Jones et al. 2004; Graham et al. 2006). However, other metrics of resilience may also
be indicative of reef health including coral size, coral and fish diversity, fish abundance,
macroalgal cover and connectivity to other reef communities (Bellwood et al. 2003; Birrell et
al. 2005; Grimsditch & Salm 2005; Hughes et al. 2007; McClanahan et al. 2008).
Biophysical models and field studies will be needed to determine the mechanisms and role of
other potential sources of coral reef resilience.
Researchers predict that the thermal limits of corals will need to increase at rates of
0.2–1.0°C per decade to avoid mass bleaching and mortality (Donner et al. 2005). Whether
corals can keep pace with such rates through mechanisms such as adaptive bleaching is
unclear (Sheppard 2003; Sheppard & Rioja-Nieto 2005). Our results caution against relying
on MPAs to substantially alter the course of climate change-induced coral degradation. The
benefits of MPAs to fisheries are clear and the restoration to more intact trophic structure
may explain the small benefit MPAs provide to corals. Nevertheless, MPAs do not
specifically mitigate the effect of warm temperature anomalies on coral cover. Some of the
weakness of the MPA effect could be due to the size and placement of the current group of
MPAs. However, their overall benefit is likely to be too small to offset the scale of major
thermal events. MPAs are clearly an important tool in managing coral reef health, but they
must be complemented with direct measures aimed at reducing the anthropogenic activities
responsible for climate change.
143
Figure 4.1 Region delineations for the analysis. Boundaries for regions are
based on biogeographic patterns in diversity and similar management. Purple dots
represent reef locations.
144
Figure 4.2 Coefficient estimates for the multilevel model. The 95% credibility
intervals (thin grey line) and the 50% credibility intervals (thick black line) as well as
point estimates (median) of the posterior distributions for all parameters using a
Bayesian approach to fit the model in Eqn. (4). There is a 95% probability that the
true value lies within the 95% credibility interval. The MPA x 10-Year Trend term
should be contrasted with the 10-Year Trend term, which is the trend for controls.
The MPA x 10-Year Trend term is an effect and gets added to the 10-Year Trend
term when MPA = 1 to obtain the trend for MPAs. The magnitude of the anomaly
effect varies according to the number of anomalies in the 2 years preceding the live
coral cover survey. The effect of 5 total anomalies is displayed.
145
Figure 4.3 The effect of TSA on logit coral cover for each spatial grouping unit
split by MPA and non-MPA. The bottom edge of the box denotes the 1st quartile
of the data, the top edge locates the 3rd quartile, and the horizontal line inside the
box corresponds to the median. The mean values are represented by the red stars.
Both unweighted and weighted (inverse of the standard error of the mean) T-tests
based on the observations of the coefficients for MPAs versus non-MPAs in the
figure were non-significant (p>.65). These results are consistent with the more
accurate estimates from the multilevel model (p>.15; see text).
146
Figure 4.4 Multilevel model predictions of the effect of thermal stress
anomalies (TSA) on the change in coral cover at (A) different levels of coral cover
and anomaly frequencies and (B) when coral cover is 50%. The 95% credibility
intervals (thin black line), 50% credibility intervals (thick grey line), and point
estimates (median) of the posterior distributions are shown.
A
B
147
Figure 4.5 Comparison of anomaly frequency within MPAs to anomaly
frequency of all reefs. Variability by year in (A) mean anomaly frequency (open
circles) and (B) standard deviation for the theoretical distribution of anomaly
frequency for all reefs and its 95% confidence intervals (error bars). The mean MPA
anomaly frequency and standard deviation (black star) are generally lower.
A
B
148
Figure 4.6 Anomaly sizes for different regions (Fig. 4.1). Bars represent the
anomaly size at which 50% (black), 75% (white), 90% (grey), and 95% (hashed) of
the sizes are included.
149
Acknowledgments
T. Kristiansen, M. O'Connor, P. Reynolds, and J. Weiss for helpful comments on the
manuscript. The Great Barrier Reef Marine Park for the GIS data for the Great Barrier Reef
zoning and The Nature Conservancy for MPA boundaries for the Caribbean.
150
References
Adger W.N., Hughes T.P., Folke C., Carpenter S.R. & Rockstrom J. (2005). Socialecological resilience to coastal disasters. Science, 309, 1036-1039.
Akaike H. (1973). Information theory and an extension of the maximum likelihood principle.
In: Second International Symposium on Information Theory (eds. Petrov BN & Csaki
F). Akademiai Kaido Budapest, pp. 267-281.
Alcala A.C., Russ G.R., Maypa A.P. & Calumpong H.P. (2005). A long-term, spatially
replicated experimental test of the effect of marine reserves on local fish yields.
Canadian Journal of Fisheries and Aquatic Sciences, 62, 98-108.
Arceo H.O., Quibilan M.C., Alino P.M., Lim G. & Licuanan W.Y. (2001). Coral bleaching
in Philippine reefs: coincident evidence with mesoscale thermal anomalies. Bulletin
of Marine Science, 69, 579-593.
Baker A.C. (2001). Reef corals bleach to survive change. Nature, 411, 765-766.
Baker A.C., Starger C.J., McClanahan T.R. & Glynn P.W. (2004). Corals' adaptive response
to climate change. Nature, 430, 741-741.
Balmford A., Bruner A., Cooper P., Costanza R., Farber S., Green R.E., Jenkins M., Jefferiss
P., Jessamy V., Madden J., Munro K., Myers N., Naeem S., Paavola J., Rayment M.,
Rosendo S., Roughgarden J., Trumper K. & Turner R.K. (2002). Ecology - Economic
reasons for conserving wild nature. Science, 297, 950-953.
Bellwood D.R., Hoey A.S. & Choat J.H. (2003). Limited functional redundancy in high
diversity systems: resilience and ecosystem function on coral reefs. Ecology letters, 6,
281-286.
Bellwood D.R., Hughes T.P., Folke C. & Nystrom M. (2004). Confronting the coral reef
crisis. Nature, 429, 827-833.
Bengtsson J., Angelstam P., Elmqvist T., Emanuelsson U., Folke C., Ihse M., Moberg F. &
Nystrom M. (2003). Reserves, resilience and dynamic landscapes. Ambio, 32, 389396.
Berkelmans R., De'ath G., Kininmonth S. & Skirving W.J. (2004). A comparison of the 1998
and 2002 coral bleaching events on the Great Barrier Reef: spatial correlation,
patterns, and predictions. Coral Reefs, 23, 74-83.
Berkelmans R. & Willis B.L. (1999). Seasonal and local spatial patterns in the upper thermal
limits of corals on the inshore central Great Barrier Reef. Coral Reefs, 18, 219-228.
151
Birrell C.L., McCook L.J. & Willis B.L. (2005). Effects of algal turfs and sediment on coral
settlement. Marine Pollution Bulletin, 51, 408-414.
Blanchon P. & Shaw J. (1995). Reef drowning during the last deglaciation - evidence for
catastrophic sea-level rise and ice-sheet collapse. Geology, 23, 4-8.
Bruner A.G., Gullison R.E., Rice R.E. & da Fonseca G.A.B. (2001). Effectiveness of parks in
protecting tropical biodiversity. Science, 291, 125-128.
Bruno J.F. & Bertness M.D. (2001). Habitat modification and facilitation in benthic marine
communities. In: Marine Community Ecology (eds. Bertness MD, Hay ME & Gaines
SD). Sinauer Sunderland, MA, pp. 201-218.
Bruno J.F., Siddon C.E., Witman J.D., Colin P.L. & Toscano M.A. (2001). El Nino related
coral bleaching in Palau, Western Caroline Islands. Coral Reefs, 20, 127-136.
Bryant D., Burke L., McManus J. & Spalding M. (1998). Reefs at Risk: A Map-Based
Indicator of Potential Threats to the World's Coral Reefs. World Resources Institute,
Washington, DC.
Burke L. & Maidens J. (2004). Reefs at Risk in the Caribbean. World Resources Institute,
Washington, DC.
Burke L., Selig E.R. & Spalding M. (2002). Reefs at Risk in Southeast Asia. World
Resources Institute, Washington, DC.
Casey K.S. (2002). Daytime vs nighttime AVHRR sea surface temperature data: a report
regarding Wellington et al. (2001). Bulletin of Marine Science, 70, 169-175.
Casey K.S. & Cornillon P. (1999). A comparison of satellite and in situ-based sea surface
temperature climatologies. Journal of Climate, 12, 1848-1863.
Casey K.S. & Cornillon P. (2001). Global and regional sea surface temperature trends.
Journal of Climate, 14, 3801-3818.
Coles S.L. & Brown B.E. (2003). Coral bleaching - capacity for acclimatization and
adaptation. In: Advances in Marine Biology, Vol 46, pp. 183-223.
Diaz-Pulido G. & McCook L.J. (2002). The fate of bleached corals: patterns and dynamics of
algal recruitment. Marine Ecology-Progress Series, 232, 115-128.
Donner S.D., Skirving W.J., Little C.M., Oppenheimer M. & Hoegh-Guldberg O. (2005).
Global assessment of coral bleaching and required rates of adaptation under climate
change. Global Change Biology, 11, 2251-2265.
152
Fautin D.G. & Buddemeier R.W. (2004). Adaptive bleaching: a general phenomenon.
Hydrobiologia, 530-31, 459-467.
Fisk D.A. & Done T.J. (1985). Taxonomic and bathymetric patterns of bleaching in corals,
Myrmidon Reef (Queensland). Proceedings of the Fifth International Coral Reef
Congress, Tahiti, 6, 149-154.
Glynn P.W. (1993). Coral reef bleaching - ecological perspectives. Coral Reefs, 12, 1-17.
Glynn P.W. (1996). Coral reef bleaching: facts, hypotheses and implications. Global Change
Biology, 2, 495-509.
Glynn P.W., Cortes J., Guzman H.M. & Richmond R.H. (1988). El Niño (1982-1983)
associated coral mortality and relationship to sea surface temperature deviations in
the tropical eastern Pacific. Proceedings of the Sixth International Coral Reef
Symposium, 2, 693-698.
Goreau T.J. & Hayes R.L. (1994). Coral bleaching and ocean hot-spots. Ambio, 23, 176-180.
Graham N.A.J., Wilson S.K., Jennings S., Polunin N.V.C., Bijoux J.P. & Robinson J. (2006).
Dynamic fragility of oceanic coral reef ecosystems. Proceedings of the National
Academy of Sciences of the United States of America, 103, 8425-8429.
Graham N.A.J., Wilson S.K., Jennings S., Polunin N.V.C., Robinson J., Bijoux J.P. & Daw
T.M. (2007). Lag effects in the impacts of mass coral bleaching on coral reef fish,
fisheries, and ecosystems. Conservation Biology, 21, 1291-1300.
Grigg R.W. (1995). Coral reefs in an urban embayment in Hawaii: a complex case history
controlled by natural and anthropogenic stress. Coral Reefs, 14, 253-266.
Grimsditch G.D. & Salm R.V. (2005). Coral reef resilience and resistance to bleaching. The
World Conservation Union, Gland.
Guidetti P. (2006). Marine reserves reestablish lost predatory interactions and cause
community changes in rocky reefs. Ecological Applications, 16, 963-976.
Gunderson L.H. (2000). Ecological resilience - in theory and application. Annual Review of
Ecology and Systematics, 31, 425-439.
Halpern B.S. (2003). The impact of marine reserves: do reserves work and does reserve size
matter? Ecological Applications, 13, S117-S137.
Hoegh-Guldberg O. (1999). Climate change, coral bleaching and the future of the world's
coral reefs. Marine Freshwater Research, 50, 839-866.
153
Hoegh-Guldberg O., Mumby P.J., Hooten A.J., Steneck R.S., Greenfield P., Gomez E.,
Harvell C.D., Sale P.F., Edwards A.J., Caldeira K., Knowlton N., Eakin C.M.,
Iglesias-Prieto R., Muthiga N., Bradbury R.H., Dubi A. & Hatziolos M.E. (2007).
Coral reefs under rapid climate change and ocean acidification. Science, 318, 17371742.
Holling C.S. (1973). Resilience and stability of ecological systems. Annual Review of
Ecology and Systematics, 4, 1-23.
Hughes T.P., Baird A.H., Bellwood D.R., Card M., Connolly S.R., Folke C., Grosberg R.,
Hoegh-Guldberg O., Jackson J.B.C., Kleypas J., Lough J.M., Marshall P., Nystrom
M., Palumbi S.R., Pandolfi J.M., Rosen B. & Roughgarden J. (2003). Climate change,
human impacts, and the resilience of coral reefs. Science, 301, 929-933.
Hughes T.P., Rodrigues M.J., Bellwood D.R., Ceccarelli D., Hoegh-Guldberg O., McCook
L., Moltschaniwskyj N., Pratchett M.S., Steneck R.S. & Willis B. (2007). Phase
shifts, herbivory, and the resilience of coral reefs to climate change. Current Biology,
17, 360-365.
Jokiel P.L. & Coles S.L. (1990). Response of Hawaiian and other Indo-Pacific reef corals to
elevated temperature. Coral Reefs, 8, 155-162.
Jones G.P., McCormick M.I., Srinivasan M. & Eagle J.V. (2004). Coral decline threatens fish
biodiversity in marine reserves. Proceedings of the National Academy of Sciences of
the United States of America, 101, 8251-8253.
Kilpatrick K.A., Podesta G.P. & Evans R. (2001). Overview of the NOAA/NASA advanced
very high resolution radiometer Pathfinder algorithm for sea surface temperature and
associated matchup database. Journal of Geophysical Research-Oceans, 106, 91799197.
Lang J.C., Lasker H.R., Gladfelter E.H., Hallock P., Jaap W.C., Losada F.J. & Muller R.G.
(1992). Spatial and temporal variability during periods of recovery after mass
bleaching on Western Atlantic coral reefs. American Zoologist, 32, 696-706.
Ledlie M.H., Graham N.A.J., Bythell J.C., Wilson S.K., Jennings S., Polunin N.V.C. &
Hardcastle J. (2007). Phase shifts and the role of herbivory in the resilience of coral
reefs. Coral Reefs, 26, 641-653.
Lesaffre E., Rizopoulos D. & Tsonaka R. (2007). The logistic transform for bounded
outcome scores. Biostatistics, 8, 72-85.
Liu G., Skirving W. & Strong A.E. (2003). Remote sensing of sea surface temperatures
during 2002 Barrier Reef coral bleaching. EOS, 84, 137-144.
154
Loya, Sakai, Yamazato, Nakano, Sambali & Woesik v. (2001). Coral bleaching: the winners
and the losers. Ecology Letters, 4, 122-131.
Lunn D.J., Thomas A., Best N. & Spiegelhalter D. (2000). WinBUGS - a Bayesian modelling
framework: concepts, structure, and extensibility. Statistics and Computing, 10, 325337.
Marshall P.A. & Baird A.H. (2000). Bleaching of corals on the Great Barrier Reef:
differential susceptibilities among taxa. Coral Reefs, 19, 155-163.
McClanahan T.R., Ateweberhan M., Muhando C.A., Maina J. & Mohammed M.S. (2007a).
Effects of climate and seawater temperature variation on coral bleaching and
mortality. Ecological Monographs, 77, 503-525.
McClanahan T.R., Ateweberhan M. & Omukoto J. (2008). Long-term changes in coral
colony size distributions on Kenyan reefs under different management regimes and
across the 1998 bleaching event. Marine Biology, 153, 755-768.
McClanahan T.R., Ateweberhan M., Sebastian C.R., Graham N.A.J., Wilson S.K.,
Bruggemann J.H. & Guillaume M.M.M. (2007b). Predictability of coral bleaching
from synoptic satellite and in situ temperature observations. Coral Reefs, 26, 695701.
McClanahan T.R., Graham N.A.J., Calnan J.M. & MacNeil M.A. (2007c). Toward pristine
biomass: reef fish recovery in coral reef marine protected areas in Kenya. Ecological
Applications, 17, 1055-1067.
McClanahan T.R. & Mangi S. (2000). Spillover of exploitable fishes from a marine park and
its effect on the adjacent fishery. Ecological Applications, 10, 1792-1805.
McMahon S.M. & Diez J.M. (2007). Scales of association: hierarchical linear models and the
measurement of ecological systems. Ecology Letters, 10, 437-452.
McWilliams J.P., Cote I.M., Gill J.A., Sutherland W.J. & Watkinson A.R. (2005).
Accelerating impacts of temperature-induced coral bleaching in the Caribbean.
Ecology, 86, 2055-2060.
Mora C., Andrefouet S., Costello M.J., Kranenburg C., Rollo A., Veron J., Gaston K.J. &
Myers R.A. (2006). Coral reefs and the global network of marine protected areas.
Science, 312, 1750-1751.
Mumby P.J., Dahlgren C.P., Harborne A.R., Kappel C.V., Micheli F., Brumbaugh D.R.,
Holmes K.E., Mendes J.M., Broad K., Sanchirico J.N., Buch K., Box S., Stoffle R.W.
& Gill A.B. (2006). Fishing, trophic cascades, and the process of grazing on coral
reefs. Science, 311, 98-101.
155
Mumby P.J., Harborne A.R., Williams J., Kappel C.V., Brumbaugh D.R., Micheli F., Holmes
K.E., Dahlgren C.P., Paris C.B. & Blackwell P.G. (2007a). Trophic cascade facilitates
coral recruitment in a marine reserve. Proceedings of the National Academy of
Sciences of the United States of America, 104, 8362-8367.
Mumby P.J., Hastings A. & Edwards H.J. (2007b). Thresholds and the resilience of
Caribbean coral reefs. Nature, 450, 98-101.
Obura D.O. (2005). Resilience and climate change: lessons from coral reefs and bleaching in
the western Indian Ocean. Estuarine Coastal and Shelf Science, 63, 353-372.
Parmesan C. & Yohe G. (2003). A globally coherent fingerprint of climate change impacts
across natural systems. Nature, 421, 37-42.
Pickett S.T.A. & Thompson J.N. (1978). Patch dynamics and the design of nature reserves.
Biological Conservation, 13, 27-37.
Pinheiro J., Bates D., DebRoy S. & Sarkar D. (2007). nlme: Linear and Nonlinear Mixed
Effects Models. R package version 3.1-83 edn.
Rayner N.A., Brohan P., Parker D.E., Folland C.K., Kennedy J.J., Vanicek M., Ansell T.J. &
Tett S.F.B. (2006). Improved analyses of changes and uncertainties in sea surface
temperature measured in situ sice the mid-nineteenth century: the HadSST2 dataset.
Journal of Climate, 19, 446-469.
River G.F. & Edmunds P.J. (2001). Mechanisms of interaction between macroalgae and
scleractinians on a coral reef in Jamaica. Journal of Experimental Marine Biology
and Ecology, 261, 159-172.
Roberts C.M., Bohnsack J.A., Gell F., Hawkins J.P. & Goodridge R. (2001). Effects of
marine reserves on adjacent fisheries. Science, 294, 1920-1923.
Russ G.R., Alcala A.C., Maypa A.P., Calumpong H.P. & White A.T. (2004). Marine reserve
benefits local fisheries. Ecological Applications, 14, 597-606.
Sandin S.A., Smith J.E., DeMartini E.E., Dinsdale E.A., Donner S.D., Friedlander A.M.,
Konotchick T., Malay M., Maragos J.E., Obura D., Pantos O., Paulay G., Richie M.,
Rohwer F., Schroeder R.E., Walsh S., Jackson J.B.C., Knowlton N. & Sala E. (2008).
Baselines and degradation of coral reefs in the northern Line Islands. PLoS ONE, 3,
e1548.
Selig E.R. & Bruno J.F. (in prep). Evidence of increased resilience to corals from marine
protected areas.
Selig E.R., Bruno J.F. & Casey K.S. (in prep). Spatial and temporal variation in temperature
anomalies on coral reefs: patterns in frequency, extent, and magnitude
156
Shears N.T. & Babcock R.C. (2003). Continuing trophic cascade effects after 25 years of notake marine reserve protection. Marine Ecology-Progress Series, 246, 1-16.
Sheppard C. & Rioja-Nieto R. (2005). Sea surface temperature 1871-2099 in 38 cells in the
Caribbean region. Marine Environmental Research, 60, 389-396.
Sheppard C.R.C. (2003). Predicted recurrences of mass coral mortality in the Indian Ocean.
Nature, 425, 294-297.
Singer J.D. & Willet J.B. (2003). Applied longitudinal data analysis: modeling change and
event occurrence. Oxford University Press, Oxford, UK.
Spalding M.D., Fox H.E., Allen G.R., Davidson N., Ferdana Z.A., Finlayson M., Halpern
B.S., Jorge M.A., Lombana A., Lourie S.A., Martin K.D., McManus E., Molnar J.,
Recchia C.A. & Robertson J. (2007). Marine ecoregions of the world: a
bioregionalization of coastal and shelf areas. Bioscience, 57, 573-583.
The Mathworks Inc. (2006). Matlab. In. The Mathworks Inc. Natick, MA.
WDPA Consortium (2005). World database on protected areas 2005. World Conservation
Union (IUCN), UNEP-World Conservation Monitoring Centre (UNEP-WCMC),
Cambridge.
West J.M. & Salm R.V. (2003). Resistance and resilience to coral bleaching: implications for
coral reef conservation and management. Conservation Biology, 17, 956-967.
Whelan K.R.T., Miller J., Sanchez O. & Patterson M. (2007). Impact of the 2005 coral
bleaching event on Porites porites and Colpophyllia natans at Tektite Reef, US
Virgin Islands. Coral Reefs, 26, 689-693.
Williamson D.H., Russ G.R. & Ayling A.M. (2004). No-take marine reserves increase
abundance and biomass of reef fish on inshore fringing reefs of the Great Barrier
Reef. Environmental Conservation, 31, 149-159.
Wolanski E. & De'ath G. (2005). Predicting the impact of present and future human land-use
on the Great Barrier Reef. Estuarine Coastal and Shelf Science, 64, 504-508.
Wooldridge S., Done T., Berkelmans R., Jones R. & Marshall P. (2005). Precursors for
resilience in coral communities in a warming climate: a belief network approach.
Marine Ecology-Progress Series, 295, 157-169.
157
APPENDIX 1
Supplementary material for Chapter 1
Figure S1.1 Mean duration (weeks) of TSAs, 1985-2005 in (A) Indo-Pacific and
(B) Caribbean.
A
B
159
Figure S1.2 Mean duration (weeks) of WSSTAs, 1985-2005 in (A) Indo-Pacific
and (B) Caribbean.
A
B
160
Figure S1.3 Mean duration (weeks) of TSA in 1998 in (A) Indo-Pacific and (B)
Caribbean.
A
B
161
Figure S1.4 Mean duration (weeks) of WSSTAs in 1998 in (A) Indo-Pacific and
(B) Caribbean.
A
B
162
Table S1.1 Validations of the CoRTAD data using in situ data loggers at different depths and locations. Data
presented included mean diff. (the mean difference between the logger data and the satellite data), RMS (root mean
square), N (number of observations) and r2. Logger data come from the Perry Institute for Marine Science (Bahamas), the
National Data Buoy Center (Florida Keys, Hawaii), Erich Bartels/MOTE Marine Laboratory (Florida Keys), Peter Edmunds
(U.S. Virgin Islands), and the CRC Research Centre (Great Barrier Reef).
Weekly Daytime
(Pathfinder V 5.0)
Location
163
Bahamas
Bahamas
Bahamas
Bahamas
Bahamas
Bahamas
Site
Name
Adderley
33m
Barracuda
Rocks 3m
Lee
Stocking
Island
16m
Lee
Stocking
Island
31m
Rainbow
Garden
Shark
Rock
Depth
(m)
Mean
diff.
RMS
N
Weekly Nighttime
(Pathfinder V 5.0)
Mean
diff.
r²
RMS
N
r²
Weekly CoRTAD
Mean
diff.
RMS
N
r²
33
0.52
0.85 341
0.88
0.21
0.57
334
0.92
0.4
0.72
419
0.9
3
0.21
0.82 299
0.9
-0.11
0.81
374
0.9
0.07
0.85
471
0.9
16
0.46
0.82 137
0.86
-0.11
0.63
182
0.89
0.15
0.68
244 0.87
31
0.62
1.02 125
0.79
0.08
0.8
173
0.8
0.32
0.9
229 0.77
4
0.36
0.75 312
0.9
-0.39
0.78
376
0.89
-0.01
0.73
512 0.88
3
0.31
0.79 231
0.89
-0.61
0.96
335
0.88
-0.15
0.85
485 0.85
Bahamas
164
Bahamas
Florida Keys (US)
Florida Keys (US)
Florida Keys (US)
Florida Keys (US)
Florida Keys (US)
Florida Keys (US)
Florida Keys (US)
GBR (Australia)
GBR (Australia)
GBR (Australia)
GBR (Australia)
GBR (Australia)
GBR (Australia)
GBR (Australia)
GBR (Australia)
GBR (Australia)
GBR (Australia)
GBR (Australia)
Hawaii (US)
Hawaii (US)
Virgin Islands (US)
South
Norman
South
Perry
42025
DRYF1
FWYF1
LONF1
MLRF1
SANF1
SMKF1
Agincourt
Chicken
Davies
Dip
East Cay
John
Brewer
Kelso
Lizard
Middle
Myrmidon
Turner
Cay
French
Frigate
Shoals
Kure Atoll
St. John
9m
3
0.26
0.78 148
0.89
-0.78
1.11
189
0.89
-0.23
0.99
287 0.83
16
1-1.5
1-1.5
1-1.5
1-1.5
1-1.5
1-1.5
1-1.5
6-9
6-9
6-9
6-9
6-9
0.49
0.32
0.46
0.53
0.43
0.3
0.5
0.17
0.25
0.29
0.38
0.35
0.25
0.75
0.68
0.9
0.81
1.1
0.73
0.9
0.79
0.62
0.65
0.64
0.71
0.5
115
119
394
442
351
616
571
521
241
287
241
236
315
0.92
0.93
0.94
0.94
0.91
0.93
0.93
0.92
0.9
0.92
0.94
0.9
0.95
0.09
-0.04
0.09
0.15
-0.13
0.01
0.16
-0.1
0
0.15
0.19
0.14
0.2
0.45
0.57
0.59
0.7
1.04
0.61
0.68
0.72
0.54
0.49
0.44
0.55
0.45
112
109
389
420
432
604
556
527
194
289
230
235
298
0.94
0.94
0.97
0.92
0.91
0.93
0.94
0.92
0.91
0.94
0.96
0.92
0.96
0.18
0.17
0.32
0.45
0.56
0.17
0.36
0.11
0.14
0.23
0.26
0.22
0.18
0.51
0.52
0.79
0.83
1.63
0.69
0.79
0.73
0.53
0.64
0.51
0.6
0.49
148
152
465
742
659
931
721
781
367
391
317
341
397
0.93
0.95
0.95
0.92
0.87
0.93
0.94
0.93
0.92
0.9
0.95
0.91
0.94
6-9
6-9
6-9
6-9
6-9
0.29
0.42
0.09
0.42
0.19
0.64
0.7
0.66
0.95
0.6
285
355
171
57
247
0.93
0.93
0.9
0.91
0.92
0.04
0.16
-0.12
-0.01
0.01
0.47
0.5
0.61
0.62
0.51
234
317
199
15
247
0.94
0.94
0.9
0.97
0.92
0.14
0.26
-0.02
0.83
0.09
0.57
0.55
0.6
1.5
0.51
356
470
343
438
361
0.93
0.94
0.91
0.83
0.93
6-9
0.36
0.56 278
0.96
0.28
0.49
264
0.96
0.31
0.52
355 0.96
1-1.5
1-1.5
0.18
-0.2
0.42
0.69
31
41
0.85
0.97
0.04
-0.11
0.27
0.62
29
41
0.93
0.96
0.11
0.04
0.34
0.71
35 0.88
61 0.95
9
-0.12
0.62 430
0.76
-0.44
0.8
275
0.72
-0.25
0.66
873 0.76
APPENDIX 2
Supplementary material for Chapter 3
2.1 Levels of management
There is no direct measure of management effectiveness in the MPA database, but all of the
MPAs have an assigned World Conservation Union (IUCN) category. These categories
reflect management objectives, but do not have specific criteria associated with them.
Categories I-IV are generally accepted as protected areas in the traditional sense that they are
protecting an ecosystem and permit only active restoration or population control, although
permitted recreational and commercial fishing may be allowed (The World Conservation
Union (IUCN) & World Commission on Protected Areas 2000). More controversy has
developed about the final two categories, V and VI. Their stated purposes are to protect
cultural values and sustainable use, respectively (The World Conservation Union (IUCN) &
World Commission on Protected Areas 2000). These categories have been criticized as
pushing the boundaries of what defines a protected area. Locke & Dearden (2005) propose
reclassifying V and VI as sustainable development areas.
2.2 Logistic-normal distribution
Percent coral cover is a limited range response variable, because it is constrained to
lie between the limits of 0 and 100 (Long 1997). Our dataset spans the range from near 0 to
165
100. Such a wide range of data can pose problems for ordinary linear regression particularly
when the response variable lies near one of the boundaries. These problems persist with a
log transform because the lognormal distribution is unbounded above. Standard solutions
involve transforming the response (logit or arcsine) or choosing a probability model (beta)
that constrains the response to lie within its natural limits. We chose a logit transform based
on AIC comparisons with these other approaches.
2.3 Distance pairing
In this study, we used the distance to the nearest MPA to create spatial structure units
for the non-MPA surveys, which we then used as a level 3 structural variable in the model.
This definition exercise is analogous to trying to identify the appropriate blocking unit to
which "treatments" (MPA versus non-MPA) are applied in an ANOVA context. There are
currently no studies examining what an appropriate distance is for pairing MPA and nonMPAs surveys to determine differences in their effects on coral cover. Several studies have
analyzed within MPA effects and quantified spillover into adjacent areas (McClanahan &
Mangi 2000; Roberts et al. 2001; Abesamis & Russ 2005). However, for this analysis we
were trying to pair protected and non-protected areas of relative homogeneity rather than
establish the spatial extent of MPA effect. For all non-MPA surveys, we calculated the
distance to the nearest MPA boundary that included a survey.
We then used maximum likelihood estimation to assess how changing the pairing
distance influenced the MPA effect on slope. Without theory to motivate a distance
threshold cut-off, we used a maximum likelihood approach to objectively select a distance
cut-off. We ran our model pairing data at 13 different distances ranging from 1 km to 4000
166
km. At a distance of zero, each spatial unit or block had a single treatment, MPA. When the
distance was greater than zero, non-MPA surveys were added to the spatial unit or block.
Non-MPA surveys that were not close enough to be paired with an MPA area were assigned
to their own spatial unit. We would expect that by increasing block size a treatment effect
would become easier to detect in part due to the greater power achieved by a larger sample
size. On the other hand, increasing the size of the spatial unit could increase variability
within the block by adding dissimilar areas. These features suggest that there should be some
optimal choice for distance pairing. The shape of our loglikelihood curve was concave down
(Fig. S1) with a well-defined maximum, indicating that there was indeed an optimal value.
We found that the loglikelihood was maximized at 200 km, indicating that a model that used
200 km as a distance cut-off for pairing was relatively better than the other models examined.
In addition, this cut-off allowed more than 60% of non-MPA surveys to be assigned to spatial
structural units with MPA surveys.
2.4 Bayesian fitting
All final models were refit as Bayesian models using Markov chain Monte Carlo (MCMC) to
obtain more realistic estimates of parameter precision. For the MPA versus non-MPA
model, uninformative priors were taken for all model parameters. For the MPA only model,
a slightly informative prior was used for the breakpoint to bound the Markov chains away
from zero. Three separate chains were run and mixing was assessed both graphically and
analytically. Iterations were continued until all parameters yielded R̂ values (the square root
of the variance of the mixture of all chains divided by the average within chain variance) that
167
were less than 1.1 for all parameters. The first half of each chain was discarded as a burn-in
period and the remainder was thinned to minimize serial correlations.
168
Figure S2.1 The relationship between the MPA effect on slope and the
distance of non-MPAs surveys from MPAs. The loglikelihood (solid black line) is
maximized at 200 km, where approximately 60% of the non-MPA data has been
paired in a structural unit with MPA data (dashed green line). MPA effect on slope
and confidence intervals (grey dashed line) do not vary significantly with distance.
169
Figure S2.2 AIC values for all models examined. The best model is the one with
the smallest AIC value. In this case, the best model is one in which MPA modifies
the slope and intercept and ocean modifies the intercept only. Models with AICs that
exceed 18650 are designated with arrows.
170
Figure S2.3 Coral cover slope and intercept values from the MPA versus nonMPA model. (A) The change in percent coral cover per year inside and outside of
MPAs from 1996 to 1997 and a static estimate of (B) percent coral cover in 1996
with the 95% credibility intervals (thin grey line) and the 50% credibility intervals
(thick black line) as well as point estimates (median) of the posterior distributions.
Coral cover in MPAs may be lower than non-MPAs due to biases in their placement.
By 2005, those differences disappear. Percent coral cover was obtained by backtransforming the predicted logit from the model.
B.
A.
171
Figure S2.4 Caterpillar plots for (A) the slope random effects and (B) intercept
random effects for the level-3 residuals. Each point and associated vertical line in
the plot represents the random effect for one spatial structural unit with paired
protected and unprotected reefs (n=357) and its 95% credibility interval. Lines
whose credibility intervals do not intersect the blue dashed line at zero are
considered outliers of the model. The level-3 slope random effects are well-behaved.
Only 4.2% of the data have an obvious indication of bad fit. By chance you would
expect 5% of the data to lie outside 95% confidence intervals. The model does not
describe level 3 intercepts as well, with 9.8% of the intercept random effects being
outliers. At level 2, 0.7% of the slope random effects are outliers and 10.7% of the
intercept random effects are outliers (not shown).
B
A
172
Figure S2.5 The number of reefs categorized by the year when their respective
MPAs were established in the (A) Caribbean and (B) Indo-Pacific.
A
B
173
Figure S2.6 Times series of the number of years of protection by reef in the (A)
Caribbean and (B) Indo-Pacific. Each grey line represents a single reef and shows
the number of years of protection at the first survey and how that reef continued to
be surveyed through time. A grey dashed line appears every 20 reefs. Gaps in the
survey years can be seen by white breaks between the grey lines. The Caribbean
had more reefs that had been protected for between 5-15 years and another group
which had been surveyed at between 20-35 years of protection. The Indo-Pacific
had a more even distribution across years, although there were fewer surveys in
early years of protection.
A
B
174
Figure S2.7 Distribution of years protected by ocean
175
Figure S2.8 Generalized additive models for the (A) Caribbean and (B) IndoPacific. There is no evidence of a changepoint in the Caribbean, but there is in the
Indo-Pacific.
A
B
176
Table S2.1 Percent of reef area by IUCN category. Management objectives are
defined by The World Conservation Union (IUCN) and the World Commission on
Protected Areas (2000).
IUCN
category
Management objectives
I
science or wilderness protection 4.4
II
ecosystem protection and
recreation (e.g. national park)
23.7
III
conservation of specific natural
features (e.g. natural
monument)
0.1
IV
conservation through
management intervention
6.4
V
landscape/seascape
conservation and recreation
(i.e. cultural values)
6.1
VI
sustainable use of natural
ecosystems
59.4
177
Percent of total reef
area in category
Table S2.2 List of models analyzed. The best model is the one with the smallest
AIC value. In this case, the best model is one in which we have slopes and
intercepts at levels 2 and 3.
Level 2
random
Model effects
1
4 intercepts
5 slopes
slopes and
6 intercepts
7
8
9
10 intercepts
11 intercepts
12 intercepts
13 slopes
14 slopes
15 slopes
slopes and
16 intercepts
slopes and
17 intercepts
slopes and
18 intercepts
Level 3
random
effects
K (# of
parameters) loglikelihood AIC
3
-12968.68 25943.36
4
-10155.56 20319.13
4
-12119.83 24247.66
6
4
4
-9919.098 19850.2
-11085.055 22178.11
-11977.371 23962.74
6
5
5
-11024.468 22060.94
-9516.756 19043.51
-9771.385 19552.77
7
5
5
-9293.002
18600
-10539.692 21089.38
-11679.633 23369.27
7
-10520.289 21054.58
intercepts
7
-9288.595 18591.19
slopes
slopes and
intercepts
7
-9669.387 19352.77
9
-9227.755 18473.51
intercepts
slopes
slopes and
intercepts
intercepts
slopes
slopes and
intercepts
intercepts
slopes
slopes and
intercepts
178
Table S2.3 Distribution of IUCN categories across oceans by surveys.
IUCN I-II
IUCN III-IV
IUCN V-VI
Caribbean
358
1457
55
Pacific
1021
342
895
179
Table S2.4 Distribution of IUCN categories across oceans by reefs.
IUCN I-II
IUCN III-IV
IUCN V-VI
Caribbean
235
287
52
Pacific
449
68
162
180
References
Abesamis R.A. & Russ G.R. (2005). Density-dependent spillover from a marine reserve:
Long-term evidence. Ecological Applications, 15, 1798-1812.
Locke H. & Dearden P. (2005). Rethinking protected area categories and the new paradigm.
Environmental Conservation, 32, 1-10.
Long J.S. (1997). Regression Models for Categorical and Limited Dependent Variables.
Sage Publications, Thousand Oaks, CA.
McClanahan T.R. & Mangi S. (2000). Spillover of exploitable fishes from a marine park and
its effect on the adjacent fishery. Ecological Applications, 10, 1792-1805.
Roberts C.M., Bohnsack J.A., Gell F., Hawkins J.P. & Goodridge R. (2001). Effects of
marine reserves on adjacent fisheries. Science, 294, 1920-1923.
The World Conservation Union (IUCN) & World Commission on Protected Areas (2000).
Protected areas: benefits beyond boundaries. In. International Union for Conservation
of Nature and Natural Resources (IUCN) and the World Commission on Protected
Areas Gland, Switzerland.
181