Introduction

The effects of human activity are evident and widespread in both terrestrial and marine systems (Sanderson et al. 2002; Worm et al. 2006; Airoldi and Beck 2007; Halpern et al. 2008). A recent review of the state of nature concluded that “changes currently underway were mostly negative, anthropogenic in origin, ominously large and accelerating” (Balmford and Bond 2005). This assessment could certainly apply to the status of coral reefs worldwide (e.g. Gardner et al. 2003; Pandolfi et al. 2003; Bruno and Selig 2007).

The full extent of decline is hard to assess because of a lack of historical baselines for most coral reefs. Among possible sources of historical information, the analysis of fossil reef communities is laborious to apply on a large scale, and historical aerial photographs provide very limited information about past coral communities. With the exception of shallow reef flats (Wachenfeld 1997; Cornish and DiDonato 2004), reef habitats are only accessible by diving, consequently quantitative studies of reefs span just the last 50 years, which is late in the history of anthropogenic impacts on many reefs (Jackson et al. 2001). Even in the short time since scuba diving became popular in the 1960s, the condition of coral reefs has declined in many parts of the world (Wilkinson 1993, 2004).

Most large-scale studies use percentage cover of living hard corals as an index of reef condition, since hard corals are the principal framework builders of coral reefs and constitute much of the habitat complexity of the surfaces of reefs. Declines in live coral cover reflect disturbances, while increases are a necessary (but not comprehensive) attribute of reef recovery (Connell et al. 1997). The 1980s and 1990s in particular saw declines in living coral and shifts in composition of reef communities in many parts of the world (Wilkinson 1993; Gardner et al. 2003; Bruno and Selig 2007). The causes of these changes have included pollution (Edinger et al. 1998), increased sedimentation from land clearing and coastal development (Yamazato 1987; Hodgson and Walton Smith 1994) overfishing, particularly of herbivorous species (Knowlton 1992; Hughes 1994a; Aronson and Precht 2006), extensive coral bleaching (Wilkinson 2004), widespread coral mortality from disease (Aronson and Precht 2001), and catastrophic loss of dominant herbivores due to diseases (Lessios 1988).

Australia’s Great Barrier Reef (GBR) is unusual among coral reef systems of the world in that the sparse human coastal population belongs to a developed economy and does not subsist heavily on reef resources. It is a large system of ~2,900 reefs scattered over a section of the continental shelf of tropical Queensland that is almost 2,000 km long and 30–200 km wide (Fig. 1). Most coral reefs are platform reefs on the continental shelf (2–25 km in longest dimension) or fringing reefs around continental islands. The GBR is a paradigm of resource management; the entire region has been a multi-use marine park since the 1980s (Ruckelshaus et al. 2008). In spite of these advantages, three studies based on syntheses of published results have reported a general decline in condition of GBR reefs since pre-colonial times (Pandolfi et al. 2003; but see Ridd 2007) and particularly a notable drop in living coral cover in the last few decades (Bellwood et al. 2004; Bruno and Selig 2007).

Fig. 1
figure 1

Map showing sectors of the Great Barrier Reef, Australia

Full size image

The Australian Institute of Marine Science (AIMS) has monitored coral cover on large areas of numerous reefs spread across much of the GBR since 1985, providing an opportunity to examine the distribution and timing of changes, especially declines, across this large system. In contrast to the studies mentioned above, AIMS surveys have used the same standard method throughout. Here, the results of that monitoring programme are used to assess changes in live coral cover on reefs of the GBR over 19 years, 1986–2004. We ask: How has average live coral cover on GBR reefs changed in that time? Have changes been general or localised? What causes of change can be identified? Does coral cover generally recover after disturbances? Finally, we consider the evidence for longer-term declines in living coral on the GBR.

Materials and methods

The AIMS monitoring programme has surveyed cover of living coral (henceforth coral cover) around the entire perimeters of numerous GBR reefs spread over 13° of latitude (11–24°S, Fig. 1) on multiple occasions since mid-1985 using a standard method: manta tows. Surveys were made over the austral summer, so “reporting years” (henceforth years) include surveys made between 1 July of the previous year and 30 June of the “reporting year”. Initially, >200 widely dispersed reefs were surveyed each year but this number was reduced as the programme developed. Reefs that were surveyed only once during the 19 years were excluded from the analyses, leaving 72–189 (median = 99) reefs surveyed in each year. GBR reef communities change along the major biophysical gradients across the continental shelf between the Coral Sea and the Queensland coast (Done 1982) and to a lesser extent from north to south. For this reason, the GBR province was divided into 11 approximately equal bands of latitude (sectors, Fig. 1) and into three cross-shelf zones: inshore, mid-shelf and outer shelf, within each sector. Inshore reefs are set in turbid waters relatively close to the coast (generally <20 km from land); mid-shelf reefs are mostly located within the main array of GBR reefs (and so are protected from direct oceanic influences by reefs to seaward) and lie outside the relatively deep “shipping channel” that runs the length of the GBR. Outer reefs have few shallow reefs to seaward and so are exposed to oceanic influences from the Coral Sea. No inshore reefs were monitored in the Swains and Pompey sectors and the only reefs in the Capricorn-Bunker sector are outer shelf reefs, meaning that there were potentially 29 “subregions” (combinations of sector × shelf position). Between 3 and 37 reefs in each subregion were surveyed more than once in the 19 years (with the exception of the inshore subregion of the Cape Upstart sector where only a single reef was surveyed more than once). Temporal changes in coral cover are generally consistent among reefs within a subregion (Ninio et al. 2000).

When carried out by trained observers, manta tows have been shown to give valid and reproducible estimates of living coral cover (Miller and Muller 1999) as well as a variety of other data including numbers of crown-of-thorns starfish (Acanthaster planci). The manta tow surveys involved a snorkeler with a “manta board” (hydrofoil) being towed slowly behind a small boat around the entire perimeter of each survey reef close to the reef crest so that the observer surveyed a 10-m-wide swathe of the shallow reef slope (Bass and Miller 1996). The boat stopped every 2 min to allow the observer to record the mean coral cover into one of 10 categories (Bass and Miller 1996), giving one cover estimate per “tow” (~200 m of reef edge). The number of tows required to cover the perimeter of a reef varies with currents and sea conditions but average number of tows per survey reef ranged from 3.8 to 203.5 (median = 39.5 tows or ~7.9 km). Observers were full-time monitoring staff whose estimates of coral cover were cross-calibrated annually to minimise differences among observers and between years. The boat handlers followed towpaths marked on an aerial photograph of each reef so that approximately the same parts of each reef were surveyed at each visit. As a visual technique, manta towing requires a minimum of 6 m visibility underwater; inshore reefs are poorly represented in the data set for this reason. The mean coral cover at each reef in each survey was calculated by taking the mid-point of the range of the coral cover category recorded for each 2-min tow (i.e. if the mean cover observed in one tow fell in the 10–30% category, the value for that tow would be 20%) and then averaging overall tows for that reef (Sokal and Rohlf 1995).

Our initial test concerned whether average coral cover across the entire GBR changed over the 19 years of surveys. This survey is important because net decreases in living coral cover have been taken to indicate reef health decline. Linear mixed effects models (Pinheiro and Bates 2000) were used to estimate the linear trend in average coral cover across all GBR reefs, which corresponds to the net change in average coral cover over the 19-year period. The linear mixed effects model took the form:

$$ {\text{Coral cover}} = {\text{Year}} $$

As the data were based on repeated measures from the perimeters of the same reefs over time, we accounted for reef-specific temporal autocorrelation using an additional random component of “Year”, treating each reef as a subject in the repeated measures analysis. An autoregressive (order 1) covariance structure was assumed to account for temporal autocorrelation. This prevented overestimation of linear trends due to repeated measurement of the same reefs.

Because the major biophysical gradients across the continental shelf between the Coral Sea and the Queensland coast are reflected in coral reef community structure, we also tested whether linear changes in coral cover differed between reefs in different shelf positions averaged across all sectors. We used the model:

$$ {\text{Coral cover}}\; = \;{\text{Shelf}} \; + \;{\text{Year}}\; + \; {\text{Shelf}}\; \times \;{\text{Year}} $$

with autocorrelation treated as above. This model tested the generality of net coral cover changes along the onshore–offshore environmental gradient.

We then tested whether average coral cover changed at different rates in different subregions over the study period, e.g. whether the linear trends, reflecting net increases or decreases in coral cover over time, differed among subregions. This was tested with the model:

$$ {\text{Coral cover }} = {\text{Subregion}} + {\text{Year}} + {\text{Subregion }} \times {\text{ Year}} $$

A significant interaction would indicate that linear changes varied among the subregions. Coral cover estimates for reefs were again modelled as potentially autocorrelated over time as above. The average linear trend in coral cover (with 95% confidence intervals) was estimated for the reefs in each subregion from this model. Confidence intervals for each subregion were not based on repeated tests; they were regression coefficients within a common analysis, and so 95% confidence intervals provide conservative “tests” for differences from a slope of zero, which would represent no net change in coral cover in the survey period.

While linear trends provide a summary of net change in the subregions over the entire study period, the annual changes in coral cover on survey reefs in each subregion involved both declines due to disturbances and increases through recruitment and growth. Non-linear average trends in coral cover over the 19 years were estimated for each subregion using natural splines (d.f. = 4) as additional fixed effects in the mixed models (Chambers and Hastie 1993). Natural splines are non-parametric local smoothers, which assume that no single underlying function generates non-linear trends, and as such are useful for summarising non-linear temporal patterns without imposing a deterministic underlying process. These trends were then presented graphically. Additional serial correlation over time within reefs was modelled as a continuous first-order autoregressive process.

Identifying possible ecological causes of temporal change in mean live coral cover depends on correlating changes in cover with spatial and temporal distributions of potential causal factors. While correlation does not imply causation, causation does imply some correlation between variables (Shipley 2000). We used three sources if information to attribute declines in coral cover on reefs within subregions of the GBR to one or more of three classes of disturbance: coral bleaching, tropical storms and outbreaks of crown-of-thorns starfish. Widespread coral bleaching occurred twice on the GBR in the study period, in 1998 and in 2002. In each case, the extent of bleaching was assessed over a large area mainly by aerial surveys (Berkelmans and Oliver 1999; Berkelmans et al. 2004). Based on these studies, bleaching intensity in 1998 and 2002 has been estimated and mapped for all reefs of the GBR using generalised additive models as part of the e-Atlas project (e-Atlas 2009a, b), including a bleaching risk map. The potential contribution of tropical cyclones and storms was assessed from cyclone tracks and wind records from the Bureau of Meteorology (http://www.bom.gov.au/cyclone/history/index.shtml). Information on densities of A. planci was collected by AIMS as part of this monitoring project. Relevant history of A. planci outbreaks has been summarised by Miller (2002), and annual estimates of A. planci density on the survey reefs are shown in Electronic Supplemental Material, ESM Appendix 1. The different regional dynamics of A. planci outbreaks on the GBR are also available as an animation in the e-Atlas (e-Atlas 2009c).

Results

Based on all the reefs monitored in each year, the average cover of living coral on GBR reefs declined from 28.1% in 1986 to 21.7% in 2004 (Fig. 2). The average rate of loss based on the linear model was 0.21% cover per year (95% CI: −0.10 to −0.33). However, such net decline was by no means general across the ~345,000 km2 of the GBR. Changes in coral cover varied across the GBR region (Fig. 3); the coral cover on inshore reefs declined by 0.32% per year (95% CI: −0.0007 to −0.64), coral cover on mid-shelf reefs declined by 0.30% per year (95% CI: −0.16 to −0.45), and there was no net change in coral cover on outer shelf reefs (rate of change = 0.002% per year, 95% CI: 0.2 to −0.20).

Fig. 2
figure 2

Mean per cent cover of living hard coral on perimeters of reefs of the GBR over 19 years, 1986–2004. Filled circles are annual means of coral cover based on all reefs that were surveyed that reporting year; boxes indicate ±1 quartile around the median coral cover (indicated by the horizontal line) for reefs surveyed in each year; whiskers indicate the range of coral cover values. The dashed line represents the linear trend in coral cover over time estimated from a linear mixed effects model (reefs included as random effects, with temporal autocorrelation). The average rate of decline since 1986 was 0.21% yr−1 (95% CI: −0.36 to −0.16)

Full size image
Fig. 3
figure 3

Mean per cent cover of living hard coral on perimeters of reefs of the GBR grouped by position on the continental shelf over 19 years, 1986–2004. Filled circles are annual means of coral cover based on all reefs surveyed that reporting year, boxes represent ±1 quartile, the horizontal line across the boxes indicates the median value, and whiskers indicate the range of values. The dashed lines represent the linear trends in coral cover over time estimated from a linear mixed effects model (reefs included as random effects, with temporal autocorrelation). The estimated annual rates of change over the 19 years were inshore reefs−0.32% (95% CI: −0.00007 to −0.64), mid-shelf reefs −0.30% (95% CI: −0.16 to −0.45) and outer shelf reefs 0.002% (95% CI: 0.20 to −0.20)

Full size image

Grouping survey reefs by subregion showed that much of the overall decline in average coral cover on the GBR in the survey period was due to large declines in two inshore and four mid-shelf subregions (Fig. 4). There was an increase in one outer shelf subregion and no substantial change in the remaining 21 subregions where more than one reef was monitored (Table 1). The greatest linear rates of coral loss occurred on reefs in inshore subregions of the Townsville (1.32% year−1) and Innisfail (1.04% year−1) sectors. Reefs in these subregions had relatively high coral cover at the start of the surveys, and these subregions were also those worst affected by bleaching in 1998 (see e-Atlas 2009b). Note that the relationship between extent of bleaching and subsequent coral mortality is not well documented on the GBR and is not always linear (Baird and Marshall 2002), though the two variables are likely to be positively correlated at the extreme values.

Fig. 4
figure 4

Per cent cover of live coral on perimeters of reefs in subregions of the GBR over 19 years, 1986–2004. Subregions are arranged from north to south (top to bottom) and from inshore to outer shelf (left to right). Fine lines represent profiles of coral cover on individual reef over time. Thick lines represent the smoothed (non-linear) trend in coral cover over time, estimated using natural splines. The seven subregions where average coral cover showed significant linear trends over time (Table 1) are enclosed in boxes

Full size image
Table 1 Change in mean per cent cover on perimeters of reefs grouped by GBR subregion
Full size table

Average coral cover also declined on reefs in four mid-shelf subregions: in the Cooktown-Lizard Is, Cairns, Cape Upstart, and particularly in the Swains sector (0.97% year−1, Table 1). Many reefs in these subregions suffered outbreaks of A. planci (ESM Appendix 1) as well as damage from cyclones (Table 1). The second wave of A. planci outbreaks (Reichelt et al. 1990) was first detected on reefs north of Cairns in 1979 (Miller 2002) and then progressed southwards. The AIMS monitoring programme was established in response to the many outbreaks on reefs near Townsville in the mid-1980s (ESM Appendix 1). The mid-shelf reefs in the Innisfail, Townsville, and Cape Upstart sectors all had low or declining coral cover in early surveys (with some North–South lag, Fig. 4) following large numbers of recorded outbreaks (ESM Appendix 1). Coral cover then generally increased on these mid-shelf reefs until the start of the third recorded wave of A. planci outbreaks was seen in the Cooktown-Lizard Is region in 1994 (Miller 2002). There were numerous outbreaks on mid-shelf reefs in the Cooktown-Lizard Is sector in the late 1990s. The wave of outbreaks then progressed southwards to affect many mid-shelf reefs in the Townsville region after 2000 (ESM Appendix 1). Coral cover in these subregions declined again (Fig. 4) as starfish numbers increased. The late declines in coral cover on mid-shelf reefs in the Cooktown-Lizard Is and Cairns sectors led to a declining linear trend over the survey period (Table 1). Note that many mid-shelf reefs in the Townsville sector had high densities of A. planci both at the start of the survey period and in the early 2000s (ESM Appendix 1), with a period of recovery in the 1990s (Fig. 4). This bimodal pattern of starfish outbreaks certainly affected coral cover in the subregion, but there was no significant linear decline over the survey period.

The outbreaks of A. planci on the southern Swain Reefs were not synchronised with the waves of outbreaks in the central GBR (Sweatman 2008). The substantial losses of coral recorded on the mid-shelf reefs of the Swains sector were most probably caused by persistent outbreaks of A. planci (ESM Appendix 1) that affected reefs with very high initial coral cover (Fig. 4).

While linear trends represent the average change over the 19 years, changes in coral cover on reefs and in subregions were generally not linear. Mean coral cover in subregions fluctuated and there were clear periods of recovery as well as declines (Fig. 4), though these were not synchronised among subregions. For instance, coral cover on the outer reefs of the Capricorn-Bunker sector recovered strongly after being damaged by storms in 1988 (Halford et al. 2004; Emslie et al. 2008) and fluctuations in coral cover on mid-shelf reefs in the central sectors of the GBR in response to A. planci outbreaks are discussed above. Reefs in most subregions showed periods of recovery; regeneration (here defined as a sequence of increases in average coral cover in a subregion, based on the non-linear trend in successive years, totalling more than 5% of the substrate) occurred in 24 of the 29 subregions during the 19 years (Table 1, Fig. 4). Five subregions did not show such regeneration. Coral cover on reefs in the mid-shelf subregions in the Cairns and Swains sectors declined over the survey period without any substantial periods of recovery. Coral cover on reefs in three other subregions, inshore reefs near Cairns and mid-shelf reefs in the Princess Charlotte Bay and Whitsunday sectors, did not show sustained increases, but did not decline significantly either over the 19 years (Table 1).

Discussion

Changes in coral cover on the GBR 1986–2004

Long-term coral monitoring using a standard survey method showed that the average cover on GBR reefs declined from 28.1 to 21.7% over 19 years to 2004. However, trends in coral cover varied locally, cover on reefs in the great majority of subregions fluctuated but showed no substantial net change. Declines in coral cover coincided with disturbances, notably occurrence of outbreaks of A.planci in mid-shelf subregions (Table 1), with periods of recovery in the intervals without disturbances. Coral cover declined dramatically in a few subregions from a variety of possible causes (Table 1). The greatest net declines were on inshore reefs in the Townsville and Innisfail sectors where the loss of coral coincided with thermal bleaching in 1998. Did the inshore reefs show signs of recovery in the period of surveys? There was no evidence of recovery in the monitoring surveys, but manta towing is not an effective method for detecting small colonies that are characteristic of the early stages of recovery, particularly in turbid inshore sites. There is other evidence of recovery; 6 years after bleaching, survey sites on inshore reefs in the Innisfail sector had the highest densities of juvenile corals (≤10 cm diameter) of any inshore reefs of the GBR (Sweatman et al. 2007), though the densities of juvenile corals were much lower on inshore reefs in the Townsville sector.

The other major declines in coral cover occurred on mid-shelf reefs and were correlated with waves of outbreaks of A. planci. The outbreaks of A. planci on the southern Swain Reefs were not synchronised with the waves of outbreaks in the central GBR (Sweatman 2008). The substantial losses of coral recorded on the mid-shelf reefs of the Swains sector were probably mostly caused by persistent outbreaks of A. planci affecting reefs with very high initial coral cover, but also reflected a change in sampling effort. In the early years of the programme, up to 32 reefs spread across the Swains sector were surveyed annually, but only seven reefs in the south of the Swains sector were surveyed regularly 1993–2004. Five of these seven reefs had large and persistent outbreaks of A. planci for most of the survey period, a high incidence of outbreaks that was not representative of reefs across the sector (Sweatman et al. 2008). The critical question is whether the coral on these remote reefs will recover once the starfish numbers subside.

Long-term changes in coral cover on the GBR

While the AIMS long-term data show a decline in average coral cover on GBR reefs from 28.1 to 21.7% between 1986 and 2004, two studies (Bellwood et al. 2004; Bruno and Selig 2007) based on unweighted meta-analyses have suggested that average coral cover on GBR reefs was considerably higher in the 1960s and 1970s than in the 1980s. Bellwood et al. (2004) presented a plot of mean coral cover on the GBR indicating that cover halved from ~40% in the early 1960s to ~20% in 2000, almost three times the rate of decline in the AIMS long-term monitoring data. These authors gave no information on data sources or analyses other than that they used published studies. Coral cover data from 1986 onwards were predominantly from the AIMS long-term monitoring programme (T.P. Hughes, pers. comm.). In the second study, Bruno and Selig (2007) used information from 2,667 sites to assess change across the Indo-Pacific, including the GBR as one subregion. Based on published studies including AIMS monitoring data, they found that mean coral cover on the GBR declined by ~25% (in relative terms) from the period 1968–1983 to 1984–1996 and then was relatively stable until the end of their study period in 2004. We argue that this difference is substantially due to a change in the scale of surveys and in survey methods. The most compelling evidence for this is that, in the data sets of both studies, the annual estimates of the mean for coral cover on the GBR drop abruptly in 1986, the first year of large-scale monitoring on the GBR, and then vary rather little around the new level in subsequent years (see Fig. 1a of Bellwood et al. 2004 and Fig. S1 (supplementary material) of Bruno and Selig 2007).

Very few estimates of cover are available from anywhere in the GBR province in most years prior to 1986, so meta-analyses must combine sparse estimates from small areas of a variety of habitats on just a few reefs in a few areas of the GBR, e.g. those listed by Connell (1997), that often changed from year to year. The AIMS monitoring shows that coral cover in most of the 29 GBR subregions has fluctuated asynchronously over the 19 years of surveys (Fig. 4), so estimates of annual mean values based on selected sites on a few reefs in a limited area of the GBR at any time may give a biased estimate of the system-wide mean. Another concern is that the objectives of many early reef studies predisposed them to select areas of high coral cover. An analysis of the topics of highly cited coral reef publications from all regions since 1970 found a clear shift from patterns of diversity and habitat use in the 1970s to a heavy emphasis on ecological disturbance after 2000 (Mumby and Steneck 2008). This change in research emphasis is likely to be reflected in the choice of study sites, with early studies selecting sites for their diverse communities and high coral cover where the study organisms are abundant and large samples can be found in a small area (Hughes 1994b; Connell 1997). This may be less true for the GBR where the impact of A. planci has caused public concern since the 1970s.

Most cover estimates for GBR reefs in any year after 1986 came from AIMS monitoring surveys of the entire reef perimeters of a large number of widely dispersed reefs and so were based on a swathe 10 m wide and several kilometres long that spanned reef zones with both high and low coral cover. The apparent abrupt drop in average coral cover on the GBR in 1986 is most probably due to the inclusion of AIMS monitoring data with cover estimates from small selected patches of reef from small-scale studies. Bruno and Selig (2007) identified these as potential biases in their analyses. In the particular case of the GBR, we argue that they are the most plausible explanation for much of the apparent drastic decline in coral cover in the 1980s. Such estimates from other regions of the world, where the scale of sampling did not change, would not be affected.

How much is the GBR degraded?

Whatever levels of coral cover existed prior to 1986, the homogeneous data series from the long-term monitoring programme shows firstly that the average cover on the perimeters of reefs across the GBR declined from 28.1 to 21.7% between 1986 and 2004 and secondly that this has been mainly due to large declines in some subregions rather than a consistent, system-wide decline. A reef system that is stable in the long term will still show cycles of disturbance and recovery at a subregional scale, and the growth rates of stony corals set a period of at least 5–10 year for coral cover to recover substantially (Connell 1997). In the kind of broad analysis presented here, reef resilience is manifested as substantial increases in coral cover following disturbance. In the great majority of subregions of the GBR, reefs showed both declines and substantial periods of increasing living coral cover over the 19 years of surveys, evidence that many reefs retained their regenerative capacity. This raises the question of what rates of recovery should be expected and what rates would indicate that community resilience is compromised? In particular, would the declines and recovery rates in the subregions where coral declined most allow long-term sustainability or will recovery be incomplete by the next major disturbance? There will be no simple answer. Recovery depends on many factors, but will certainly vary with the nature and extent of impact and with the growth rates and reproductive strategies of the corals that make up the diverse communities in each subregion, as well as the return time of disturbances.

The GBR has clearly changed due to human activities and many of the same processes that preceded and led to the decline of reefs in the Caribbean and other regions are evident on the GBR (Bellwood et al. 2004), including increased sedimentation and runoff, decline in abundance of megafauna, overharvesting and coral disease. But there are important differences between the GBR and the Caribbean in particular. Inshore reefs are frequently exposed to runoff (Devlin et al. 2001), but these reefs constitute less than 5% of the reef area of the GBR. While there is evidence that sediment discharge into the GBR lagoon has increased since European settlement (McCulloch et al. 2003), this represents a minor addition to the pre-existing sediment accumulation that is subject to resuspension (Larcombe and Woolfe 1999), so sedimentation and turbidity on inshore reefs (and certainly on GBR reefs in general) may not have increased substantially. Numbers of dugong, turtles and sharks have declined (Great Barrier Reef Marine Park Authority 2009), and there certainly have been examples of overharvesting: the collapse of bêche de mer and pearl shell fisheries in the late 19th century prompted early calls for scientific research on the GBR (Bowen and Bowen 2002). There is an active fishery for predatory finfish species, but unlike their Caribbean counterparts, fishers on the GBR use lines to target high-value species, rather than taking whatever they catch (Newman et al. 2006) using unselective gear such as traps (Sammarco 1985). There is currently minimal catch of herbivorous species by amateur or professional fishers, aside from a very small fishery for trochus shell. Coral diseases occur widely on the GBR (Willis et al. 2004; Bruno et al. 2007) but resulting coral mortality has been insignificant compared with the effects of white band disease in the Caribbean (Aronson and Precht 2001). The cause of the greatest changes in coral cover on the GBR in the period of surveys, outbreaks of A. planci, may have become more frequent due to eutrophication (Brodie et al. 2005; Fabricius et al. 2010) or to reduced predation (Sweatman 2008), though mechanisms remain speculative in each case. In summary, while most of the stressors that are responsible for the decline of reefs worldwide are present on the GBR, damage to date has been localised rather than system-wide.

AIMS’ long-term data show that, in the recent past, major disturbances affected GBR reefs in a few subregions at a time in an unsynchronised manner. This may favour rapid recovery because the limited area affected by each disturbance means that coral communities can potentially be replenished by larvae from undisturbed reefs nearby. We argue that the GBR is currently less degraded from its natural, resilient state than some published reports have asserted. However, the spectre of rapidly accelerating climate change and the pervasive effects of ocean acidification (e.g. Hoegh-Guldberg et al. 2007) give a pessimistic outlook because the frequency, intensity and spatial extent of impacts are all likely to increase.

Should a thermal anomaly such as that which affected many Indian Ocean reefs in 1998 (Wilkinson 2002; Smith et al. 2008) occur on the GBR, it could kill the majority of corals across very large areas at one time and cause system-wide decline. The reduced reproductive output from such decimated coral populations (Szmant and Gassman 1990; Baird and Marshall 2002) would likely depress subsequent coral recruitment, slow recovery and predispose a general shift in reef communities to alternate states that are not dominated by corals (McManus and Polsenberg 2004). System-wide decline in GBR coral cover could also result from multiple disturbances, e.g. starfish predation, coral bleaching, disease outbreaks and tropical storms (Hoegh-Guldberg 1999; Webster et al. 2005) that are interspersed by incomplete recoveries, from increased frequency of substantial disturbances, from slower recovery of reef communities because of reduced recruitment or growth rates, or both. Under more optimistic scenarios, the critical distinction between long-term stability (episodic declines followed by complete recovery) and a slow incremental degradation will only be resolved reliably by careful analyses of long-term data collected systematically over appropriate geographic scales.