Global disparity in the resilience of coral reefs

Transcription

1 Review Global disparity in the resilience of coral reefs George Roff and Peter J. Mumby Marine Spatial Ecology Lab and Australian Research Council Centre of Excellence for Coral Reef Studies, School of Biological Sciences, University of Queensland, St Lucia, QLD 472, Australia The great sensitivity of coral reefs to climate change has raised concern over their resilience. An emerging body of resilience theory stems largely from research carried out in a single biogeographic region; the Caribbean. Such geographic bias raises the question of transferability of concepts among regions. In this article, we identify factors that might predispose the Caribbean to its low resilience, including faster rates of macroalgal growth, higher rates of algal recruitment, basin-wide iron-enrichment of algal growth from aeolian dust, a lack of acroporid corals, lower herbivore biomass and missing groups of herbivores. Although mechanisms of resilience are likely to be ubiquitous, our analysis suggests that Indo-Pacific reefs would have to be heavily degraded to exhibit bistability or undergo coral macroalgal phase shifts. Coral reef resilience and geographic bias The resilience of ecosystems has been studied for almost four decades [1]. In a resilience context, one of the most intensively studied systems has been coral reefs, in part because many have exhibited profound changes in community state that might be stable [2]. Many observers have reported a reduction in coral and a general increase in the quantity of seaweed (macroalgae), referred to as a coral algal phase shift [3]. Such phase shifts have been reported globally, but an analysis of citations reveals that there have been more than twice as many studies carried out in the Caribbean than in the Indo-Pacific. Indeed, the emerging body of theory of reef resilience has drawn heavily from the study of Caribbean systems [4 7]. This bias towards the Caribbean probably reflects the great motivation and tractability of studying resilience in this region; reef health has deteriorated strikingly on many Caribbean reefs [8] and their relatively low biodiversity compared with Indo- Pacific reefs [9] simplifies the challenge of modeling their dynamics. However, as with any bias, one must ask whether the principles developed from one region can be applied meaningfully elsewhere. Are processes of resilience qualitatively different among regions or merely in need of some minor changes in parameter values? Answering this question is important as researchers continue to develop management approaches to bolster the resilience of coral reefs to a plethora of disturbances [1]. We begin by revisiting the differences between the two major biogeographic provinces of coral reefs: the Caribbean and Indo-Pacific. We consider their present ecological state Corresponding author: Roff, G. (g.roff@uq.edu.au) and differences in their observed recovery. We then discuss hypotheses that might account for large-scale differences in system resilience. Some of these hypotheses have been raised previously, and others are proposed for the first time. Fundamental differences in the recovery and state of Caribbean versus Indo-Pacific reefs It has long been appreciated that the wider Caribbean and Indo-Pacific regions differ in many ways. Early workshops and review papers throughout the 197s and 198s sought to identify key biogeographic differences between these regions, including trophic structure [11], faunal composition [12], grazing pressure [13], life-history traits [11], habitat connectivity [14], coral reproductive traits [15] and species richness [16]. More recently, it has become clear that these major biogeographic provinces also differ in their ecological resilience [17]. In analyzing long-term studies of coral cover from the mid- 196s to the mid-199s, Connell [17] showed clear evidence of trajectories of decline in Caribbean reefs, with no observations of recovery following disturbance. Indo-Pacific trajectories differed significantly from those of the Caribbean region (Appendix S1 in the supplementary material online), in that 46% of reefs showed evidence of recovery following disturbance (Figure 1). Over 83% of Caribbean reefs exhibited either no recovery (47%) or no decline (36%) compared with 39% of Indo-Pacific reefs exhibiting either no recovery (24%) or no decline (15%). Applying Connell s criteria to an analysis of 38 Caribbean reefs published since his paper, we find only a single case [18] documenting recovery and reversal of macroalgal phase shifts (Table S1 in the supplementary material online). Meanwhile, despite an increase in the frequency of coral-bleaching events in the Indo-Pacific and a reduction in average health, many reefs continue to show trajectories of recovery (e.g., [19]). Differences in the trajectories of reefs from the Caribbean and from Indo-Pacific are borne out in their overall benthic state [3,7 9]. We combined survey data from Bruno et al. [2] with a large Caribbean database for reefs from between 1995 and 25 [21]. Our analysis showed significant differences in coral cover and macroalgal cover between the Indo-Pacific and Caribbean regions (Figure 2, Appendix S2 in the supplementary material online), consistent with previous studies that show a greater proportion of macroalgal dominance and a higher severity of phase shifts in the Caribbean than in the Indo-Pacific [7,17,2]. Given the evidence presented above, it seems /$ see front matter ß 212 Elsevier Ltd. All rights reserved. Trends in Ecology and Evolution, July 212, Vol. 27, No. 7

2 1 (a) 1 Indo-Pacific 8 8 Proportion of studies Recovery No decline No recovery Inconclusive Coral cover (%) Figure 1. Differential recovery of Indo-pacific and Caribbean coral assemblages following disturbance. Recovery was observed on 46% of Indo-Pacific reefs (open bars), whereas % of Caribbean reefs (black bars) exhibited recovery following disturbance. Categories are defined following [17]: no decline indicates either declines of <33% (relative to the initial cover) or increases in cover; recovery indicates a decline of at least 33% with a subsequent increase of at least 5% of the amount lost in the decline; no recovery, indicates a decline of at least 33% in coral cover with less than 5% of recovery from the initial decline; and inconclusive indicates where further observation was needed to determine recovery or decline. Redrawn from [17]. fair to characterize the Caribbean as having slow coral recovery and a greater likelihood of high macroalgal cover. Why then does this regional difference occur? We propose six hypotheses, none of which are mutually exclusive. We group hypotheses into three categories; those concerning biodiversity, bottom-up forcing and top-down forcing. (b) 1 8 Coral cover (%) Caribbean Biodiversity-based hypotheses Hypothesis 1: loss of Caribbean Acropora led to the emergence of alternate reef trajectories Studies of reef ecosystem resilience have found considerable sensitivity to the growth rate of corals [22]. We hypothesize that the loss of fast-growing corals in the Caribbean might partly explain the low resilience of coral in this region. The growth of corals falls broadly into two categories: fast-growing branching corals from the families Acroporidae (Caribbean and Indo-Pacific) and Pocilloporidae (Indo-Pacific only), capable of extending >1 mm year 1, versus slower corals, which tend to have massive or plating morphologies and extend at less than 5 mm year 1 [23]. Although the averaged growth rates of all taxa are comparable between Indo-Pacific and Caribbean reefs [23], their species composition and diversity differ dramatically. In the Indo-Pacific, upper reef slopes are dominated by highly diverse (>3 species [9]) assemblages of Acropora spp. whose linear growth can reach 165 mm year 1 [23] and exhibit rapid recovery after disturbance [24]. Historically, the upper reef slopes of the Caribbean were dominated by only two species of branching Acropora [9]. By the late 198s, disease had vastly reduced the abundance of Caribbean Acropora [25], and the remaining (non-acroporid) corals grew an order of magnitude more slowly. Although the loss of Acropora in the Caribbean had a direct negative impact on coral cover and reef complexity [9], we extend this hypothesis to consider the impact of Acropora loss on average coral growth. Rapid individual growth of corals not only increases the population recovery rate, but also provides a competitive advantage over algae. For a given level of primary productivity, the abundance of algae will be determined by the intensity of grazing on habitable substrate [26]. Faster-growing corals should lead to a rapid intensification of grazing, which helps constrain algae. We allowed A. cervicornis to return to a model Caribbean coral reef and found that its recovery completely removed hysteresis and alternate reef trajectories (Box 1). Therefore, we hypothesize that the decline in Caribbean Acropora, for which there was almost no functional 6 Macroalgal cover (%) 8 Figure 2. Density plots of macroalgal and coral cover for (a) Indo-Pacific (n = 145) and (b) Greater Caribbean (n = 7648) reefs. Indo-Pacific reefs are typically characterized by high coral cover and low macroalgal cover, whereas Caribbean reefs have typically lower coral cover and higher macroalgal cover. Survey data sourced from [2] for Indo-Pacific surveys (combined Indo-Pacific and Great Barrier Reef) and from [2] combined with AGGRA survey data [21] for Caribbean reefs. The density of reefs (z-axis) was plotted using a two-dimensional kernel density estimation in R statistical software (kde2d function, MASS package). 45

3 redundancy, shifted the ecosystem into a position where it became vulnerable to bistability. Hypothesis 2: higher diversity and greater functional redundancy of herbivores in the Indo-Pacific It is well established that the Caribbean has a vastly depauperate fauna compared with the Indo-Pacific, including herbivores [9]. Elevated herbivore diversity can potentially increase reef resilience by fostering greater niche diversification and creating functional redundancy. Burkepile and Hay [27] demonstrated that two complementary herbivore species were more effective at mitigating algal blooms than was a single species. To compare the richness of herbivores between the Caribbean and Indo-Pacific, we used FishBase ( Our results indicate a striking difference in the species and generic richness (Figure S1 in the supplementary material online). Of the 88 species and six genera of Acanthuridae that inhabit in the Indo-Pacific and Caribbean regions, the Caribbean is severely underrepresented, with four species represented by a single genus (Acanthurus). Thus, the Caribbean lacks acanthurids of the genus Naso, some of which consume the largest of reef macrophytes, Sargassum spp. [28]. Of the 98 species and nine genera of parrotfish, the Caribbean is also underrepresented, with 15 species comprising four genera. Siganids do not occur in the Caribbean yet represent a potentially significant herbivorous group on Indo-Pacific reefs [29,3], with 28 recorded species. The level of functional redundancy among fish fauna is somewhat unclear given recent evidence that some Indo-Pacific herbivores can have specialized diets (e.g., [31]). However, considering the large size of some Indo-Pacific genera (e.g., Scaridae, 43 spp., Figure S1 in the supplementary material online), there remains considerable scope for meaningful functional redundancy. The importance of functional redundancy among herbivores appears clearer at the scale of major taxa. The loss of a single, dominant urchin species (Diadema antillarum) from the Caribbean after a disease outbreak in [32] appears to have transformed the resilience of Caribbean reefs. Ecological models that include grazing by this echinoid predict that reefs would only exhibit a single equilibrial state and always tend towards recovery after disturbance [7] (Box 1). In other words, it is feasible that Caribbean reefs only became susceptible to alternate trajectories once either the Acropora were lost or the urchins became functionally extinct. The absence of a diseasebased extirpation of urchins in the Indo-Pacific also provides great functional redundancy at a taxon scale. For example, in the Western Indian Ocean, the effects of heavy fishing, which depletes herbivorous fish, is often compensated for by increases in density of the urchin, Echinometra mathaei, as it is released from predatory control by the redlined triggerfish [33]. Although such mechanisms prevent macroalgal blooms, switches to urchin dominance can be damaging to the reef, resulting in extensive erosion [9]. The Box 1. Phase shifts, hysteresis and coral growth rates We used an existing spatial simulation model of a Caribbean reef [7] to investigate the impact of a return of the branching coral, Acropora cervicornis. In the absence of this coral and the urchin Diadema, the ecosystem exhibits two alternate states whenever grazing levels lie between.5 (a heavily overfished system) and circa.28 (a lightly fished system, albeit without Diadema). The trajectory of the coral assemblage differs according to the state of the reef relative to an unstable equilibrium (broken black line, Figure I). Self-reinforcing negative feedbacks will tend to drive reefs below the unstable equilibrium towards a coral-depauperate stable equilibrium (bottom unbroken black line, Figure I). This downward trajectory occurs because there is insufficient coral recruitment to maintain the coral population, let alone help it recover. If the reef sits above the unstable equilibrium, then the system will begin to show coral recovery and move towards a coral-dominated equilibrial state (upward-pointing black arrows leading to upper unbroken black line). In practice, the reef might never reach either of the stable equilibria so this framework is best thought of as representing alternate trajectories that corals can follow between individual disturbances. The location of the unstable equilibrium reflects the underlying rates of key ecosystem processes (e.g., algal growth rates, coral growth rates and recruitment rates). The diagonal shape of the unstable equilibrium reflects the increase in grazing intensity that occurs as coral cover increases for a fixed abundance of herbivores. When A. cervicornis is re-introduced to the ecosystem (trajectories and stable equilibrium in red, Figure I), the system only exhibits a single equilibrial state; high coral cover. This means that coral recovery will begin after any disturbance, even when successive disturbance events have ratcheted the coral cover down to a low level. In other words, if acute disturbance is removed for a long enough duration, the reef would eventually follow a trajectory back to high coral cover, regardless of the amount of grazing. Similarly, addition of a modest density of Diadema (e.g., 1 m 2 ) increases grazing rate to such an extent that coral recruitment is sufficient to drive population recovery even when the cover of corals is low. Again, return of the urchins removes the existence of alternate stable equilibrial states. We modeled the growth of A. cervicornis by estimating the radial linear extension of the patch rather than that of individual branches. We estimated a radial extension of 8 cm year 1 based on published rates of increased cover and empirical data on average patch size and density. Reducing the growth rate to as low as 3 cm year 1 had the same qualitative effect. Coral cover (%) Herbivorous fish only Fish + urchins Proportion of reef grazed Figure I. Potential increase in the resilience of Caribbean forereefs through reintroduction of the fast-growing coral Acropora cervicornis and return of the urchin Diadema antillarum. 46

4 legitimacy of other potential biodiversity-based hypotheses remains unclear at this stage. For example, although the diversity of herbivores is greater in the Indo-Pacific (Figure S1 in the supplementary material online), it is not clear whether macroalgae have evolved a more potent suite of chemical defences [34] to combat this threat. Bottom-up hypotheses: macroalgae bloom faster in the Caribbean Even if grazing communities were equivalent between the Caribbean and Indo-Pacific, algal blooms would be more likely in areas where algal productivity is higher [35]. To examine differences in the growth rate of macroalgal assemblages, we combined data sets from herbivore-exclusion experiments conducted in shallow reef environments (<12 m depth) from the Caribbean [36 42] and Indo-Pacific [43 5]. Our results indicate a clear difference in the rate and extent of macroalgal growth between regions (Figure 3). Following removal of grazing pressure, increases in macroalgal growth were almost instantaneous in all Caribbean studies (within 1 4 weeks) but took significantly longer in the Indo-Pacific (Appendix S3 in the supplementary material online) at upwards of 1 weeks (Figure 3 inset). At 2 weeks, Caribbean studies reached a maximum of approximately 9% macroalgal cover, whereas Indo-Pacific studies report a maximum of approximately 25% (Figure 3). Furthermore, at heavily fished sites in the Indo-Pacific where total herbivore biomass has been reduced to levels similar to the Caribbean (3 17 g m 2 ), macroalgal cover remains comparatively low (<22%, Figure 4). By contrast, levels of macroalgal cover in the Caribbean under similar total herbivore biomass are considerably higher (up to 78%, Figure 4). Thus, herbivores seem less able to contain algal blooms in the Caribbean than in the Indo-Pacific. Although seasonality can influence macroalgal growth [4], the distribution of dates among studies allows seasonality to be rejected as a confounding factor in this analysis (Table S2 in the supplementary material online). This leaves four alternative explanations of the apparent fourfold higher growth rate of fleshy macroalgae in the Caribbean (Figure 3). First, given that inherent growth rates vary among algal species, could the Caribbean studies each represent a faster-growing subset of species by chance? Although one cannot discount the importance of interspecific differences in growth, it seems an unlikely explanation given the broad geographic coverage within each region, the variety of habitats studied and the variety of algal assemblages involved. Second, few, if any, caging studies manage to exclude all possible herbivores. It is possible that either a higher diversity or density of micrograzers occurs in the (a) 1 8 Key: Caribbean Florida Keys a Florida Keys b Florida Keys c Percent Cover 6 4 Belize d Belize e St Croix f Jamaica g 2 (b) 1 8 Percent Cover 6 4 Key: Indo-Pacific GBR h GBR i GBR j GBR k Guam l m Fiji Hawaii n Hawaii o Time (weeks) Time (weeks) Figure 3. Comparisons of macroalgal growth rates in studies using cages to exclude herbivorous fish assemblages from (a) the Caribbean (a g; [36 42]) and (b) Indo-Pacific (h m; [43 5]) indicates significant macroalgal growth rates in the Caribbean in the absence of grazing. Analysis of longer term studies (4 13 weeks) indicates that macroalgal cover can reach approximately 55% on Indo-Pacific reefs in the absence of grazing [43,49], which falls short of the levels of macroalgal dominance observed in comparable Caribbean studies (9 95%, [39,4]). The graphs on the right are enlarged versions of the detail from those on the left. 47

5 Macroalgal cover (%) y =.21x r 2 =.52*** y = 4.65x r 2 =.8*** Total herbivore biomass (g m 2 ) Figure 4. Relationships between herbivore biomass and macroalgal cover for Indo-Pacific [77,84] reefs and Caribbean [85] reefs. At heavily fished Indo-Pacific locations (open circles), where total herbivore biomass has been reduced to Caribbean levels, macroalgal cover remains low in comparison with Caribbean reefs (black circles). See Figure S3 in the supplementary material online for the full data set. Indo-Pacific and that caging studies are less effective in this region. Unfortunately, there are insufficient data to evaluate this hypothesis. However, it should be borne in mind that the difference in algal blooms appears to be robust across a profound diversity gradient within the Indo-Pacific, stretching from species-rich Australia to species-poor Hawaii, and this casts some doubt on the existence of a systematic difference in micrograzer impact between major biogeographic provinces. This leaves two, mutually compatible hypotheses that might explain higher rates of algal blooms in the Caribbean: faster algal blooms could occur because of higher rates of algal recruitment and/or higher rates of algal growth. Hypothesis 3: higher rates of algal recruitment in the Caribbean Little is known about the spatial scales of recruitment in reef macroalgae, but available data reveal high levels of local recruitment, often within meters of parent plants [51,52]. In the Indo-Pacific, the extent of some macroalgal taxa (e.g., Sargassum) are limited by dispersal and recruitment outside areas of established adult populations [53]. Given the higher adult stock of macroalgae on current Caribbean reefs (Figure 2), it seems reasonable that recruitment could currently be greater in this region. However, it is questionable whether this mechanism could account for the rapid increase in macroalgal cover in Caribbean cage studies that were carried out before the loss of Diadema [39] or the rapid reef-wide increase in macroalgae that occurred within months of the urchin dieoff [42,54]. In short, higher algal recruitment might occur in the Caribbean, but this explanation seems unable to account for the speed of algal blooms at sites where herbivory was intense. This leaves the fourth hypothesis: that macroalgal growth rates tend to be higher in the Caribbean. No direct comparisons of algal growth rates have been made between regions, but if true, it implies that bottom-up forcing might be greater in the Caribbean. This raises the question: why? Hypothesis 4: higher nutrient levels in the Caribbean Most studies of bottom-up control in algae have focused on the concentrations of nitrogen and phosphorus-based nutrients, such as nitrate, ammonium and phosphate [55]. Water-column nutrient concentrations represent the net sum of internal cycling, algal assimilation and external inputs, relative to macroalgal growth demands [56], and provide a measure of nutrient excess on coral reefs [57]. It has long been argued that that Caribbean reef decline is associated with destructive land-use practices and widespread deforestation in a semi-enclosed basin [58] compared with remote Indo-Pacific reefs with lower population densities. A recent meta-analysis of Caribbean reefs suggests that coastal development and land cultivation associated with human density are statistically associated with coral mortality and macroalgal abundance [59]. Despite a clear anthropogenic role in the degradation of Caribbean reefs, the role of nutrients in driving macroalgal growth is unclear [6]. For example, large-scale patterns of nutrient flux via episodic upwelling events far surpass local sources of nutrient inputs [61], and coral algal phase shifts in the Caribbean have occurred on reefs remote from any major nutrient input [62]. We compared Caribbean and Indo-Pacific concentrations of nitrate and phosphate from at least 5 published data sets and found no systematic differences (mean nitrate values.43 mg l 1 vs.41 mg l 1 and mean phosphate values.24 mg l 1 vs.24 mg l 1 from the Caribbean and Indo-Pacific respectively; Figure S2 and Table S3 in the supplementary material online). Some of the highest recorded nitrate (>2 mg l 1 ) and phosphate (>.4 mg l 1 ) levels are not from the Caribbean, but from reefs located across the central and eastern Pacific and the Arabian Gulf, where macroalgal cover is low [63]. Of course, nutrients can be assimilated rapidly from the water column by phytoplankton [64] and possibly benthic algae. Therefore, it remains possible that rates of particulate matter and nutrient supply from terrestrial sources might be higher in the Caribbean (e.g., Mississippi, Amazon and Orinoco deltas, [65]) and that a higher proportion of nutrient is utilized by algae, leaving a comparable level of excess nutrient in the water column to that found across the Indo-Pacific. We anticipate the improved remote sensing of coastal water quality (e.g., [66]) will shed further light on spatiotemporal trends in both biogeographic regions. Hypothesis 5: reduced nutrient limitation in the Caribbean If nutrient concentrations cannot solely account for putative differences in algal growth rate between ocean provinces, productivity might be limited by the availability of essential trace elements [67]. Indeed, even if terrestrial nutrient inputs are relatively large in the Caribbean, algae would have to utilize a greater proportion of available nutrients for there to be similar levels of excess nutrient 48

6 Box 2. Aeolian dust and Caribbean phase shifts Annually, hundreds of millions of tons of dust is transported from the African Sahel and the Sahara deserts across the Atlantic Ocean [87]. These events are composed largely of mineral dust, along with microand macronutrients, microorganisms and trace metals that are deposited across the oligotrophic waters of the Caribbean [88] and the Gulf of Mexico [89] (Figure I). Atmospheric contributions from dust act as the primary source of iron (Fe) in surface waters of many of the worlds oceans [69]. Iron is considered a limiting nutrient in marine ecosystems, and is essential for a range of physiological processes, including photosynthesis, respiration and nitrogen fixation [9,91]. Much of the atmospheric iron transport to the oceans is present as soluble Fe 2+ [69], which has profound effects on primary productivity in pelagic oligotrophic environments. Although numerous mesoscale iron-enrichment experiments (notably IRONEX I and II, [7]) have demonstrated the capacity of iron to stimulate pelagic phytoplankton blooms, few studies have investigated the effects of iron addition on the growth rate of benthic macroalgae. The limited information available from an experimental study in the Pacific indicates that the growth of coral reef associated algae is iron limited [68,92]. On a global scale, the annual transport of dust to the Caribbean and Atlantic region is significant (ca t year 1, [93]) compared with tropical regions in the Indo-Pacific (Figure I). The highest influx of dust to the Caribbean region coincides with the summer months [86], when algal productivity on coral reefs is greatest [94]. During the early to mid 2th century, dust transport to the Caribbean region remained relatively low [71]. Following widespread drought in North Africa during the late 196s, a steady increase has been observed in dust transport to the region, with in situ measurements revealing a tenfold increase dust concentrations recorded during the early 198s compared with pre-drought conditions [95]. Although this timing coincides with the documented decline in coral cover and increase in macroalgae during the 198s [8], analysis of iron in speleotherm records from the Bahamas indicate that the Caribbean region has been influenced by influxes of African dust for 4 years [96]. Although we do not discount the possibility that a recent increase in dust could have exacerbated the recent bloom of macroalgae, we consider the millennial history of iron enrichment to be more important. It is conceivable that benthic algae in the Caribbean have been poised to take advantage of enabling conditions for thousands of years but that the required enabling conditions, be they nutrient additions or reduced grazing, are a relatively recent phenomenon. (a) (b) 3-3 Figure I. Owing to the larger source regions for dust north of the equator, approximately eight times more atmospheric iron is deposited in the northern hemisphere than in the southern hemisphere [69]. (a) Spatial distribution of atmospheric iron fluxes in the summer months (<.1 to >8 mg m 2 month 1, [86]), (b) distribution of Caribbean (blue) and Indo-Pacific (red) reefs. 49

7 in the water column (see Hypothesis 4). In either case, an argument can be made that algal growth could be less limited in the Caribbean. Algal growth is limited by various metals but iron has long been known to limit photosynthesis and by extension algal growth on coral reefs [68]. Unlike nitrogen and phosphorous, the supply of bioavailable iron (Fe 2+ ) is likely to vary at the scale of oceanic provinces. For millennia, aeolian dust has blown westwards from the deserts of North Africa to the Atlantic and Caribbean (Box 2). Aeolian dust is an important source of Fe 2+ [69], and has been strongly implicated in explaining blooms of phytoplankton in the Atlantic [7]. We extend this hypothesis to consider the effects of aeolian Fe 2+ on the productivity of benthic algae in the Caribbean. We hypothesize that millennia of aeolian Fe 2+ input, which has increased in recent decades [71], could have reduced iron limitation in Caribbean algae and therefore facilitated their ability to bloom when other enabling conditions are met. This does not imply that the Caribbean has always had more macroalgae; rather, algal blooms can occur more easily if grazing levels are reduced and/or nitrogen and phosphorous increased. Testing this hypothesis will require large-scale coordination to identify potential reservoirs of iron in the water column or sediment. However, this hypothesis is consistent with experimental and observational studies in the Pacific. For example, the outer Great Barrier Reef is iron limited [68] and the experimental addition of nitrogen and phosphorous in the absence of additional iron had very limited effect on algal communities [72] (Box 3). Similarly, recent studies have found evidence for iron limitation in some remote Pacific islands [73,74]. Top-down hypotheses: herbivores are more abundant in the Indo-Pacific Hypothesis 6: higher absolute grazing in the Indo-Pacific We compared data sets of herbivore biomass (a measure directly related to grazing intensity, [75]) from a diversity of Caribbean reefs (Jamaica, Barbados, Belize, Grand Cayman and Cuba, [76]) and across two contrasting Indo-Pacific systems: the highly diverse reefs of the Philippines [77] and those of Hawaii, which have some of the lowest diversities of the Indo-Pacific [78]). Total herbivorous fish biomass in the Indo-Pacific averaged 29. g m 2 ( 4.1 SE) and was threefold greater than that of the Caribbean (9.25 g m 2.8, Figure S3 in the supplementary material online). Parrotfish biomass in the Indo-Pacific was nearly twice that of the Caribbean ( g m 2 vs g m 2 ). Even more strikingly, the biomass of surgeonfish was more than fourfold higher in the Indo-Pacific than in the Caribbean ( g m 2 vs g m 2 ) and the maximum biomass was 12-fold higher than in the Caribbean. Not only are the maximum herbivore levels generally higher in the Indo-Pacific, but the functional form of relationships with macroalgae also differs (Figure S3 in the supplementary material online), implying that the relative importance of groups might differ among regions (e.g., such that acanthurids might be more important in the Indo-Pacific; Figure S3 in the supplementary material online). Box 3. Implications of geographic variability in resilience for reef management Although the existence of alternate equilibrial states seems credible for the Caribbean, we anticipate that Indo-Pacific systems would need to be pushed considerably harder to exhibit bistability or even experience a coral macroalgal phase shift. However, although our analysis suggests that Indo-Pacific reefs are more resilient, this does not warrant complacency in their management. Understanding of resilience in the Indo-Pacific is still in its infancy, yet it is clear that the prevailing reef management paradigms remain unchanged: efforts should be made to minimize pollution and the loss of grazers. However, the implementation of these principles may need to be adapted from one region to another. For example, our analysis of functional relationships between herbivore biomass and macroalgal cover suggests that surgeonfish have a more important role in the Indo-Pacific (Figure S3 in the supplementary material online). Thus, management of herbivory in this region might need to consider the many diminutive, yet highly abundant surgeonfish that often dominate the system. That is not to say that parrotfish are unimportant in the Indo-Pacific. Indeed, the full functional role of larger, less abundant herbivores is not yet clear. It is possible, for example, that parrotfish also have an important role controlling other coral competitors, such as sponges [97] and tunicates [98]. Some qualitative differences in system function may need to be considered when adapting management priorities from one region to another. For example, if the Indo-Pacific is more strongly iron limited than the Caribbean, then greater importance might be placed on reducing sources of iron contamination (e.g., [74]). In a similar vein, the diagnosis of a degraded reef may need to be adapted among regions (and probably also within regions). If high levels of fleshy macroalgae are relatively uncommon in the Indo- Pacific, then it would be irresponsible to base warning thresholds on the elevated levels observed in the Caribbean. Thus, management targets should focus on the key drivers of resilience at a local scale, even though many of the general principles, such as the need to control algal blooms, are universal. Although current levels of herbivore biomass differ dramatically between the Caribbean and Indo-Pacific, it is unclear whether this has always been the case. Artisanal fisheries have resulted in significant depletions of herbivores in both the Caribbean [79] and Indo-Pacific [8]. However, fishing alone seems to be an inadequate explanation for region-wide differences in herbivore biomass. Even though large-bodied parrotfish were depleted decades ago at several relatively lightly fished sites, such as Belize [81], some locations, such as Bonaire, have had very little history of herbivore exploitation [79], yet are represented in Figure 4. Moreover, surgeonfish biomass is much lower in the Caribbean, yet there has been no known fishing mortality on this genus for several of the locations in Figure 4. However, it is possible that declines in habitat quality might have affected the density of some herbivores in the Caribbean [82]. Unfortunately, few archival data reveal long-term trends in the biomass of herbivores in either the Caribbean or Indo-Pacific. Concluding remarks Although much of the theory of reef resilience was founded on Caribbean systems, we see that these coral habitats differ markedly from those of the Indo-Pacific in terms of their present state, scope for recovery, processes of herbivory and the rate at which algal blooms become established. It follows that Indo-Pacific reefs are likely to have greater resilience than those of the Caribbean. In fact, Potts [83] 41

8 remarked upon this over 25 years ago by describing the Indo-Pacific as a globally robust and resilient mosaic of regional and local, oceanic and shelf, systems, linked by long-distance dispersal, but buffered by substantial spatial barriers, which contrasted with the Caribbean, which behaves locally, regionally and globally as a single system with few barriers to dispersal and little buffering from biological or physical disturbance. We propose six hypotheses that might account for these global differences in potential resilience. Several of these hypotheses are likely to act in concert and their individual importance will vary both geographically and over time. Two of the conclusions drawn from our analysis of published literature are difficult to reconcile at present. Our analysis suggests that algal blooms occur more readily in the Caribbean. If this reflects higher algal productivity (from a release of iron limitation and/or eutrophication) then it is difficult to reconcile lower levels of herbivore biomass in the same region. Possible explanations include greater niche partitioning of herbivory in the Indo-Pacific, that herbivore biomass in the Caribbean lies below potential carrying capacity, and that habitat rather than food is limiting Caribbean herbivore biomass. Differences in the rate of algal blooms within herbivore exclusion experiments and empirical relationships between herbivore biomass and macroalgal cover suggest that top-down explanations are insufficient to account for interoceanic differences in reef dynamics. The emerging picture of the Caribbean is one that might be fundamentally predisposed to algal blooms and with limited scope for top-down control, in terms of low herbivore biomass and diversity. Disease has had a pervasive role in shaping the present state of the Caribbean, causing functional loss of the only major urchin and both acroporid species. Taken individually or in concert, these losses might have fundamentally changed the resilience of the Caribbean from a system that usually follows a trajectory of coral recovery (at least until the next disturbance event), to one that might exhibit alternate community trajectories and stable algal-dominated states. Why has the Caribbean suffered so much misfortune? Perhaps a combination of being dealt a weak evolutionary set of cards, with limited functional redundancy, together with high levels of oceanographic connectivity that facilitates the spread of epizootics and pollutants. Nonetheless, our analysis has generated several testable hypotheses to advance understanding of global disparity in reef resilience, and has implications for the management of coral reefs (Box 3). Acknowledgments This study was funded by a Laureate Fellowship from the Australian Research Council to PJM. Additional funding was provided by a Pew Fellowship in Marine Conservation and the EU FORCE project. We thank various colleagues for discussions on these issues over the years including Bob Steneck, Mark Hay, Giles Wiggs, Kenny Broad, Rob van Woesik, and Carl Safina. We also thank Alan Friedlander and John Bruno for making the data behind their figures available. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at j.tree References 1 Holling, C.S. (1973) Resilience and stability of ecological systems. Annu. Rev. Ecol. Syst. 4, Nystrom, M. et al. (2) Coral reef disturbance and resilience in a human-dominated environment. Trends. Ecol. Evol. 15, Done, T.J. (1992) Phase shifts in coral reef communities and their ecological significance. Hydrobiologia 247, Elmhirst, T. et al. (29) Connectivity, regime shifts and the resilience of coral reefs. Coral Reefs 28, Anthony, K.R.N. et al. (211) Ocean acidification and warming will lower coral reef resilience. Global Change Biol. 17, Baskett, M.L. et al. (21) Conservation management approaches to protecting the capacity for corals to respond to climate change: a theoretical comparison. Global Change Biol. 16, Mumby, P.J. et al. (27) Thresholds and the resilience of Caribbean coral reefs. Nature 45, Gardner, T.A. et al. (23) Long-term region-wide declines in Caribbean corals. Science 31, Bellwood, D.R. et al. (24) Confronting the coral reef crisis. Nature 429, Maynard, J.A. et al. (21) Building resilience into practical conservation: identifying local management responses to global climate change in the southern Great Barrier Reef. Coral Reefs 29, Birkeland, C. (1988) Geographic comparisons of coral reef community processes. In Proceedings of the 6th International Coral Reef Symposium (Choat, J.H. et al., eds), pp , 6th International Coral Reef Symposium Executive Committee 12 Sammarco, P.W. (1985) The Great Barrier Reef vs. the Caribbean: comparisons of grazers, coral recruitment patterns and reef recovery. In Proceedings Of The Fifth International Coral Reef Congress (Gabrie, C. and Salvat, B., eds), pp , Antenne Museum-EPHE 13 Birkeland, C. (1971) Grazing pressure in benthic communities on the Caribbean and Pacific coasts of Panama. Bull. Ecol. Soc. Am. 52, 5 14 Ogden, J.C. and Gladfelter, E.R. (1982) Coral Reefs, Seagrass Beds and Mangroves: Their Interaction in the Coastal Sons of the Caribbean, UNESCO Reports in Marine Science 15 Richmond, R.H. and Hunter, C.L. (199) Reproduction and recruitment of corals: comparisons among the Caribbean, the Tropical Pacific, and the Red Sea. Mar. Ecol. Prog. Ser. 6, Vermeij, G.J. (1978) Biogeography and Adaption. Patterns of Marine Life, Harvard University Press 17 Connell, J.H. (1997) Disturbance and recovery of coral assemblages. Coral Reefs 16, S11 S Idjadi, J.A. et al. (26) Rapid phase-shift reversal on a Jamaican coral reef. Coral Reefs 25, Osborne, K. et al. (211) Disturbance and the dynamics of coral cover on the Great Barrier Reef ( ). PLoS ONE 6, e Bruno, J.F. et al. (29) Assessing evidence of phase shifts from coral to macroalgal dominance on coral reefs. Ecology 9, Kramer, P.A. (23) Synthesis of coral reef health indicators for the western Atlantic: results from the AGGRA Program. Atoll Res. Bull. 496, Hoegh-Guldberg, O. et al. (27) Coral reefs under rapid climate change and ocean acidification. Science 318, Dullo, W.C. (25) Coral growth and reef growth: a brief review. Facies 51, Diaz-Pulido, G. et al. (29) Doom and boom on a resilient reef: climate change, algal overgrowth and coral recovery. PLoS ONE 4, e Bythell, J. and Sheppard, C. (1993) Mass mortality of Caribbean shallow corals. Mar. Pollut. Bull. 26, Williams, I.D. et al. (21) Limits to grazing by herbivorous fishes and the impact of low coral cover on macroalgal abundance on a coral reef in Belize. Mar. Ecol. Prog. Ser. 222, Burkepile, D.E. and Hay, M.E. (28) Herbivore species richness and feeding complementarity affect community structure and function of a coral reef. Proc. Natl. Acad. Sci. U.S.A. 42, Hoey, A.S. and Bellwood, D.R. (21) Cross-shelf variation in browsing intensity on the Great Barrier Reef. Coral Reefs 29, Fox, R.J. and Bellwood, D.R. (28) Remote video bioassays reveal the potential feeding impact of the rabbitfish Siganus canaliculatus (f: Siganidae) on an inner-shelf reef of the Great Barrier Reef. Coral Reefs 27,

9 3 Cheal, A.J. et al. (21) Coral macroalgal phase shifts or reef resilience: links with diversity and functional roles of herbivorous fishes on the Great Barrier Reef. Coral Reefs 29, Lefèvre, C.D. and Bellwood, D.R. (211) Temporal variation in coral reef ecosystem processes: herbivory of macroalgae by fishes. Mar. Ecol. Prog. Ser. 422, Lessios, H.A. (1988) Mass mortality of Diadema antillarum in the Caribbean: what have we learned? Annu. Rev. Ecol. Syst. 19, McClanahan, T.R. (1998) Predation and the distribution and abundance of tropical sea urchin populations. J. Exp. Mar. Biol. Ecol. 221, Hay, M.E. (1997) The ecology and evolution of seaweed herbivore interactions on coral reefs. Coral Reefs 16, S67 S76 35 Mumby, P.J. and Steneck, R.S. (28) Coral reef management and conservation in light of rapidly evolving ecological paradigms. Trends Ecol. Evol. 23, Sotka, E.E. and Hay, M.E. (29) Effects of herbivores, nutrient enrichment, and their interactions on macroalgal proliferation and coral growth. Coral Reefs 28, Miller, M.W. et al. (1999) Effects of nutrients versus herbivores on reef algae: a new method for manipulating nutrients on coral reefs. Limnol. Oceanogr. 44, Miller, M.W. and Hay, M.E. (1998) Effects of fish predation and seaweed competition on the survival and growth of corals. Oecologia 113, Lewis, S.M. (1986) The role of herbivorous fishes in the organization of a Caribbean reef community. Ecol. Monogr. 56, Ferrari, R. et al. (212) Interaction of herbivory and seasonality on the dynamics of Caribbean macroalgae. Coral Reefs /s Carpenter, R.C. (1986) Partitioning herbivory and its effects on coral reef algal communities. Ecol. Monogr. 56, Morrison, D. (1988) Comparing fish and urchin grazing in shallow and deeper coral reef algal communities. Ecology 69, Hughes, T.P. et al. (27) Phase shifts, herbivory, and the resilience of coral reefs to climate change. Curr. Biol. 17, Ceccarelli, D.M. et al. (211) Interactions between herbivorous fish guilds and their influence on algal succession on a coastal coral reef. J. Exp. Mar. Biol. Ecol. 399, Hatcher, B.G. and Larkum, A.W.D. (1983) An experimental analysis of factors controlling the standing crop of epilithic algal community on a coral reef. J. Exp. Mar. Biol. Ecol. 69, Scott, F.J. and Russ, G.R. (1987) Effects of grazing on species composition of the epilithic algal community on coral reefs of the central Great Barrier Reef. Mar. Ecol. Prog. Ser. 39, Thacker, R.W. et al. (21) Effects of herbivore exclusion and nutrient enrichment on coral reef macroalgae and cyanobacteria. Coral Reefs 19, Rasher, D.B. et al. (212) Effects of herbivory, nutrients, and reef protection on algal proliferation and coral growth on a tropical reef. Oecologia 169, Hixon, M.A. and Brostoff, W.N. (1996) Succession and herbivory: effects of differential fish grazing on Hawaiian coral-reef algae. Ecol. Monogr. 66, Smith, J.E. et al. (21) The effects of top-down versus bottom-up control on benthic coral reef community structure. Oecologia 163, de Ruyter van Steveninck, E.D. and Breeman, A.M. (1987) Deep water vegetations of Lobophora variegata (Phaeophyceae) in the coral reef of Curaçao: population dynamics in relation to mass mortality of the sea urchin Diadema antillarum. Mar. Ecol. Prog. Ser. 36, Stiger, V. and Payri, C.E. (1999) Spatial and temporal patterns of settlement of the brown macroalgae Turbinaria ornata and Sargassum mangarevense in a coral reef on Tahiti. Mar. Ecol. Prog. Ser. 191, McCook, L.J. (1997) Effects of herbivory on zonation of Sargassum spp. within fringing reefs of the central Great Barrier Reef. Mar. Biol. 129, Steneck, R.S. (1994) Is herbivore loss more damaging to reefs than hurricanes? Case studies from two Caribbean reef systems ( ). In Global Aspects of Coral Reefs: Health, Hazards, and History (Ginsburg, R.N., ed.), pp. C32 C37, University of Miami 55 Burkepile, D.E. and Hay, M.E. (26) Herbivore vs. nutrient control of marine primary producers: context-dependent effects. Ecology 87, Littler, M.M. and Littler, D.S. (27) Assessment of coral reefs using herbivory/nutrient assays and indicator groups of benthic primary producers: a critical synthesis, proposed protocols, and critique of management strategies. Aquat. Conserv. Mar. Freshw. Ecosyst. 17, Lapointe, B.E. (1997) Nutrient thresholds for bottom-up control of macroalgal blooms on coral reefs in Jamaica and southeast Florida. Limnol. Oceanogr. 42, Hallock, P. et al. (1993) Coral reef decline. Natl. Geogr. Res. Exploration 9, Mora, C. (28) A clear human footprint in the coral reefs of the Caribbean. Proc. R. Soc. Lond. B: Biol. Sci. 275, McClanahan, T.R. et al. (24) Phosphorus and nitrogen enrichment do not enhance brown frondose macroalgae. Mar. Pollut. Bull. 48, Leichter, J.J. et al. (23) Episodic nutrient transport to Florida coral reefs. Limnol. Oceanogr. 48, Miller, M.W. and Gerstner, C.L. (22) Reefs of an uninhabited Caribbean island: fishes, benthic habitat, and opportunities to discern reef fishery impact. Biol. Conserv. 16, Kleypas, J.A. et al. (1999) Environmental limits to coral reef development: where do we draw the line? Am. Zool. 39, Furnas, M. et al. (25) In the other 9%: phytoplankton responses to enhanced nutrient availability in the Great Barrier Reef Lagoon. Mar. Pollut. Bull. 51, Jacobs, M.B. and Ewing, M. (1969) Mineral source and transport in waters of the Gulf of Mexico and Caribbean Sea. Science 163, Lee, Z. et al. (27) Determination of primary spectral bands for remote sensing of aquatic environments. Sensors 7, Morel, F.M.M. et al. (1991) Limitations of productivity by trace metals in the sea. Limnol. Oceanogr. 36, Entsch, B. et al. (1983) Indications from photosynthetic components that iron is a limiting nutrient in primary producers on coral reefs. Mar. Biol. 73, Duce, R.A. and Tindale, N.W. (1991) Atmospheric transport of iron and its deposition in the ocean. Limnol. Oceanogr. 36, Boyd, P.W. et al. (27) Mesoscale iron enrichment experiments : synthesis and future directions. Science 315, Prospero, J.M. and Lamb, P.J. (23) African droughts and dust transport to the Caribbean: climate change implications. Science 32, Koop, K. et al. (21) ENCORE: The effect of nutrient enrichment on coral reefs. Synthesis of results and conclusions. Mar. Pollut. Bull. 42, Schroeder, R.E. et al. (28) Long-term effects of a ship-grounding on coral reef fish assemblages at Rose Atoll, American Samoa. Bull. Mar. Sci. 82, Wegley Kelly, L. et al. (211) Black reefs: iron-induced phase shifts on coral reefs. ISME J. 6, Mumby, P.J. (26) The impact of exploiting grazers (scaridae) on the dynamics of Caribbean coral reefs. Ecol. Appl. 16, Williams, I.D. and Polunin, N.V.C. (2) Large-scale associations between macroalgal cover and grazer biomass on mid-depth reefs in the Caribbean. Coral Reefs 19, Stockwell, B. et al. (29) Trophic and benthic responses to no-take marine reserve protection in the Philippines. Mar. Ecol. Prog. Ser. 389, Friedlander, A.M. et al. (27) Defining reef fish habitat utilization patterns in Hawaii: comparisons between marine protected areas and areas open to fishing. Mar. Ecol. Prog. Ser. 351, Hawkins, J.P. and Roberts, C.M. (24) Effects of artisanal fishing on Caribbean coral reefs. Conserv. Biol. 18, Russ, G.R. et al. (25) Inferring versus measuring rates of recovery in no-take marine reserves. Mar. Ecol. Prog. Ser. 292, Mumby, P.J. et al. (24) Mangroves enhance the biomass of coral reef fish communities in the Caribbean. Nature 427, Paddack, M.J. et al. (29) Recent region-wide declines in Caribbean reef fish abundance. Curr. Biol. 19,