Coral cover loss, reef structural complexity declines, transitions in coral community composition, and declines in reef-associated species abundance, have been fundamental aspects of the on-going degradation of coral reefs globally (Hoey et al., 2016). These changes have had, and are continuing to have, profound impacts on many of the ecosystem goods and services that reefs provide to society (Pendleton et al., 2016). Of equally pressing concern however are the impacts that these ecological declines are now having on the processes of carbonate production and bioerosion on reefs (Perry et al., 2014a,b). These changes are especially important because they control the ability of reefs to both maintain, and add to, their framework structures, and which in turn influences their capacity to track rising sea levels (Perry et al., 2013b, 2015b; Saunders et al., 2016). Understanding the magnitude of reef carbonate budget changes, and how they have changed within and between reef-building regions has thus been a focus of considerable recent research interest. This has encompassed assessments of both reef framework carbonate production (Perry et al., 2014a; Pratchett et al., 2015) and bioerosion rates (Perry et al., 2014a; Weinstein et al., 2014), and considerations of how rates of both may respond to future climate change drivers (e.g., Fang et al., 2013; Barkley et al., 2015; DeCarlo et al., 2015; Januchowski-Hartley et al., 2017; Schönberg et al., 2017). In addition, there has been considerable effort aimed at improving the understanding of how net reef carbonate budget states vary between reefs and how they may respond to ecological change, these being based on either census (e.g., Stearn et al., 1977; Scoffin et al., 1980; Hubbard et al., 1990; Eakin, 1996; Mallela and Perry, 2007; Perry et al., 2012, 2013b) or hydrochemical (e.g., Smith and Kinsey, 1976; Gattuso et al., 1996; Andersson and Gledhill, 2013; Shaw et al., 2016) in-situ measurements. In some cases a combination of both methods have been applied (e.g., Courtney et al., 2016) and up-scaling, based on remotely sensed imagery, applied to derive reef-wide production rates (e.g., Andréfouët and Payri, 2001; Moses et al., 2009; Hamylton et al., 2013b). In a few cases this budget work has also been extended to considerations of future reef growth potential as a function of contemporary budget states (Perry et al., 2015b; Perry and Morgan, 2017), and to the modeling of carbonate budgets under future change scenarios (Alvarez-Filip et al., 2013; Kennedy et al., 2013). Finally, there has been some recent work aimed at using census-based approaches to quantify rates of reef-scale carbonate sediment generation, using habitat specific census-data to explore the linkages between reefs and adjacent reef islands (Perry et al., 2015a; Morgan and Kench, 2016a), and remotely sensed imagery to up-scale production rates for specific sediment producing taxa (e.g., Doo et al., 2017).

Whilst this body of research provides an improved understanding of both site specific and intra-regional scale carbonate budget states, and of the key biological agents of inorganic carbonate production and erosion on reefs, there has been a tendency in much of this work to focus only on assessments of the most highly productive reef and coral-dominated zones (see Yamano et al., 2000; Harney and Fletcher, 2003; Hart and Kench, 2007 for useful exceptions). There are, of course, many valid reasons for doing this, most especially in settings where reefs form semi-continuous linear, or circum-reef platform, structures, and thus play an especially critical role as wave protecting structures (Ferrario et al., 2014), and as key sites of sediment supply (Perry et al., 2015a). However, in many settings reef development can be far more discontinuous. Indeed, in the case of individual reef platforms or coastal shelf environments, the coral-dominated reefal habitats may actually comprise a relatively small proportion of the overall contemporary platform surface. Instead, complex mosaics of hardgrounds, sand- or rubble-dominated, and seagrass habitats can occupy a greater proportional area of the platform or shelf surface and, as earlier studies on reef carbonate production rates suggested, these lower productivity zones can result in significantly reduced reef-scale carbonate budgets (Smith and Kinsey, 1976; Buddemeier and Smith, 1988; Kinsey and Hopley, 1991). In order to improve our understanding of net inorganic carbonate cycling rates on reefs, to derive more accurate measures of platform scale carbonate budgets, and to better understand how carbonate budgets may influence the development and stability of proximal reef islands or beaches, requires as many of the following as possible to be factored for: (1) rates of both reef framework production and loss based on habitat specific ground-truthed data; (2) measures of sediment production and dissolution rates (see for example recent work by Cyronak et al., 2013; Eyre et al., 2014; Andersson, 2015 which demonstrates the budgetary relevance of the latter); and (3) to account for how both may vary between habitats and, where relevant, how rates may vary with energy (Hamylton et al., 2013a) or depth (Hustan, 1985). Quantifying both the framework and sediment carbonate components of reef budgets is especially desirable in a changing global environmental context, since it is realistic to assume that the processes driving the framework and sediment generation parts of budgets will respond in non-uniform ways to climate change and direct anthropogenic factors.

Here we explore these ideas in the context of an atoll rim reef platform in the central Maldives (Vavvaru, Lhaviyani Atoll), and demonstrate how habitat-specific census-based approaches, combined with remotely-sensed and ground-truthed habitat mapping, can be used to refine estimates of reef scale carbonate productivity. Specifically, we combine recently adapted census-based methodologies to provide quantitative estimates of: (1) net reef framework budgets, based on habitat specific measures of framework carbonate production and bioerosion; and (2) net reef sediment budgets, based on quantification of the sediment derived from both direct sources and from the bioerosional reworking of reef framework, and factored to account for off-reef sediment flushing and dissolution. We calculate both parts of this whole reef carbonate budget approach within each of eight delineated reef-lagoon habitats around Vavvaru, and factor for habitat extent in subsequent assessments of the total net reef carbonate budget (G, where G = kg CaCO₃ m⁻² year⁻¹). We then use the resultant data to explore spatial (between-habitat) variations in both framework and sediment production and loss rates, and consider the implications for platform-scale budgets and for the reef-associated landforms that these platforms support.

The carbonate budget of a coral reef strongly influences the development and maintenance of its structural complexity, which is a key control on ecosystem service provisioning, and its long term net growth potential (Kinsey and Hopley, 1991; Perry et al., 2008). In this context, the early pioneering work undertaken on reef carbonate budgets reported that very high rates (in the range 10–12 G) of gross carbonate production may define those shallow water reefs having high branching coral cover (Smith and Kinsey, 1976; Kinsey, 1981; Kinsey and Hopley, 1991), and that this may equate to potential vertical growth rates of around 7 mm year⁻¹ (Kinsey and Hopley, 1991). Lower, but still significant rates (around 4 G) were reported to define reef flat habitats, whilst rates of around 0.5 G were reported for sand and rubble dominated reef habitats (Smith and Kinsey, 1976; Kinsey and Hopley, 1991).

Most of these early calculations were underpinned by novel hydrochemical based methodologies and these significantly advanced our capacities to start quantifying the carbonate budgets of different reef types. Indeed, the above, and other reported rates (e.g., Chave et al., 1972), were used as the basis for making some of the first spatial scale assessments of reef-wide carbonate production, both for a range of idealized reef system types (e.g., for the different reef types that define the Great Barrier Reef; Kinsey and Hopley, 1991), and for a range of reef geomorphic types (i.e., atoll, barrier, fringing; Chave et al., 1972). Subsequent reviews of reef carbonate production rates helped further define the optimal rates associated with different coral biogeographic provinces, and the different production rates that define different depth zones within each of these provinces (Vecsei, 2001). This collective body of data has thus provided a framework to support local scale reef production rate estimates (Smith and Kinsey, 1976), to contribute to global scale assessments (Milliman and Droxler, 1995), and more recently to support revised reef- and regional-scale estimates of reef carbonate production based on improved spatial mapping techniques (e.g., Hamylton et al., 2013b; Leon and Woodroffe, 2013).

Such metrics have growing relevance, however, because of the profound changes that are now impacting coral reefs globally, caused both by direct anthropogenic disturbances and by climate change drivers (especially temperature-induced coral bleaching events), and both of which may radically alter reef carbonate budget states (Eakin, 1996; Edinger et al., 2000; Perry and Morgan, 2017). As a result, there has been a growing interest in using reef budgets as a reef health indicator (e.g., Perry et al., 2008; Mace et al., 2014). This has merit because of the underpinning role that carbonate budgets play in controlling many physical functional aspects of reefs, including their growth potential, their role as generators of coastal sediment, and their role in supporting many economically important reef-associated species. However, to more usefully utilize budget datasets in support of these issues, and to be able to more successfully monitor how contemporary budget states vary both between reefs and between different reef habitats, requires a number of advances to be made. We need to be able to more fully quantify whole reef-scale carbonate budgets that account for the different drivers of both framework and sediment production. We need to be able to better predict how these rates may change as different drivers of carbonate production and erosion change over time (e.g., Kennedy et al., 2013), and we need to better understand how reef-scale rates may vary as a function of the types and extent of different reef habitats.

Here, we have outlined an approach to mapping reef-scale carbonate budgets that allows us to factor for estimates of both framework and sediment production and erosion, to identify the different drivers of both, and to assess these processes within different habitat types whose spatial extent has been constrained by a combination of satellite image analysis and ground truthing. Based on the application of this approach at Vavvaru in the Maldives we calculate that the overall platform budget averages 1.15 G. At this site, this results from very small areas of high productivity reef, but much larger areas that are defined either by net negative budgets, or by very low net positive budgets. Highest rates of framework carbonate production occur, not surprisingly, in the two reef habitats, and in the two other habitats where corals are relatively abundant (the shelf-edge hardground habitat and the Porites bommie zone; Tables 2–4), whilst bioerosion-derived sediment production in these same zones is also responsible for a very high proportion of the sediment generation we calculate. Low rates of framework carbonate production and bioerosion, but also of new sediment generation, define the sand-dominated lagoon habitats around Vavvaru (Tables 2–4). Although, not quantified it must thus be assumed that these sand-dominated lagoon zones represent “sinks” for a high proportion of the sediment (new or reworked) that is produced in the more productive reef habitats (e.g., Purdy and Gischler, 2005; Schlager and Purkis, 2013). However, in terms of constructing habitat specific platform scale budgets, rates of framework and sediment export and loss also need to be taken into account. Such rates are not simplistic to measure in situ and so, as outlined in the methods, we have utilized published data on rates of framework rubble export and sediment export from other sites in the Maldives, and also factored a generic literature-derived sediment dissolution rate in our calculations, and then applied these to appropriate habitats around Vavvaru (see Tables 2–3).

The net effect of these calculations i.e., accounting for net framework and sediment generation, less sediment dissolution and sand/rubble export as appropriate, is to exacerbate patterns of inter-habitat heterogeneity. Resultant net total carbonate budgets range from 5.99 and 7.05 G within the two reef habitats, to −0.07 to −0.29 G within the sand-dominated nearshore and lagoonal habitats (Figure 5C). Rates thus differ by more than one order of magnitude between habitats. The implications for the wider platform system are, however, much more profound when account is made of the spatial extent of the different habitats. Most noticeable is that the two habitats with the highest framework, sediment and overall net budgets (reef habitats 1 and 2; Figures 5A–C) and which thus drive most of the carbonate production on the platform actually occupy the smallest proportion of the reef platform (1.84 and 6.52%, respectively; Tables 2–4). Conversely, the habitats that occupy the largest spatial areas, the western nearshore habitat (Zone 7; 30.4% of platform area) and the large lagoonal area on the northern side of the platform (Zone 8; 23.3% of platform area) have amongst the lowest net budget rates (both being net negative; −0.07 and −0.29 G, respectively; Figure 5C, Tables 2–4). If these individual habitat rates for framework and sediment production are then multiplied by the total habitat area and summed they generate estimates of the respective total rates of framework and sediment production and loss for the entire platform. On this basis we calculate that 220,880 kg of framework CaCO₃ year⁻¹ is produced across the entire platform, against a framework breakdown and export figure of 123,749 kg CaCO₃ year⁻¹. This equates to a net platform scale reef framework budget of only 0.12 G (Table 4). Sediment production is calculated to total 889,818 kg CaCO₃ year⁻¹, and sediment dissolution and export 67,918 kg CaCO₃ year⁻¹. This equates to a net platform scale budget of 1.04 G, and thus a combined whole platform net carbonate budget of only 1.15 G (Table 4).

As evident from these figures, and an implication of the classic early carbonate budget studies outlined above, is that a focus on only the high productivity, high coral cover reef zones on reefs will thus generate much higher budget estimates than are representative of most reef systems as a whole. Indeed, our data suggests that there can be close to an order of magnitude difference between the budgets of the platform margin reef habitats and the much lower budgets that define the overall reef system, a distinction that is again in line with the ideas presented in many of the early studies which compared reef, reef flat and lagoon habitat rates (see Kinsey, 1981; Kinsey and Hopley, 1991). However, upscaling habitat specific rates in the past often proved problematic due to field mapping constraints, but advances in satellite image resolution (such as utilized here) are making such assessments increasingly accurate (e.g., Andréfouët and Payri, 2001; Moses et al., 2009; Hamylton et al., 2013b). Such data can thus be combined with appropriate field-derived budget estimates to develop not only much improved estimates of reef-scale carbonate production, but also key insights into the most quantitatively significant species that drive the resultant patterns. This, in turn, has the potential to feed into modeling efforts aimed at understanding the implications of different species transitions, caused by environmental change, on reef carbonate production and erosion rates (see Perry et al., 2011 for conceptual models of such changes). Such approaches have recently been applied to project reef framework budget transitions for a hypothetical Caribbean reef under different climate change and management intervention scenarios (Kennedy et al., 2013), but increasingly there is the potential to undertake similar work for the wider carbonate producing system.

At Vavvaru the high coral carbonate production rates measured within the reefal zones along the platform margin were driven by high branching Acropora spp. cover, whilst parrotfish were identified as the major framework bioeroders and thus as major generators of carbonate sand. Parrotfish are well known to produce very large quantities of sediment as a by-product of their substrate grazing (Morgan and Kench, 2016b), and indeed have been identified as the major sand generators in atoll interior faro settings elsewhere in the Maldives (Perry et al., 2015a). Maintenance of both thriving shallow water coral communities, even when occupying a relatively small area of the wider reef system, and of the associated parrotfish populations, are thus critical to maintaining the framework and sediment budgets of this system in net positive states.

A key issue that arises from the above observations are the impacts of the overall platform-scale carbonate budget on the stability, or lack thereof, of platform top reef islands. The formation and maintenance of these islands depends on there being a net positive supply of sediment from the generation sites to the nodal points of island formation (Gourlay, 1988; Mandlier and Kench, 2012). Given that the overall sediment budget at Vavvaru was calculated to be low (only 1.03 G), and thus that our best estimates suggest that only a little more sediment is produced overall within the platform than is lost, it would seem likely that the platform island sediment system at this site probably already exists close to a budget related stability threshold. In this context it is especially significant to note that Vavvaru Island has been highly mobile on its platform surface over the last ~45 years, the island shoreline migrating 129 m between 1969 and 2008 (a mean rate of 3.3 m year⁻¹). Whilst other factors, such as island size (e.g., Perry et al., 2013a) may contribute to this high mobility state, it is reasonable to hypothesize that the low rates of sediment supply as suggested by the current sediment budget study of the platform may be limiting island expansion and reducing the potential for the island position to stabilize.

Such issues become increasingly significant in the context of on-going reef degradation which can drive both progressive (Edinger et al., 2000) and rapid declines (Eakin, 1996) in reef carbonate budgets. Of immediate and pressing concern in a Maldivian context are the impacts of the major bleaching of mid-2016 caused by the very strong El Niño conditions. At sites in the southern Maldives recent work has reported on the very major declines in coral carbonate production rates and the widespread transition to net negative budget states that occurred at those sites (Perry and Morgan, 2017). Assuming similar transitions have occurred at Vavvaru, and reports from mid-2016 do indicate that bleaching of a similar magnitude was widespread across the Maldives (IUCN, 2016), it is entirely reasonable to hypothesize that carbonate production rates at Vavvaru will also have declined markedly since our work was undertaken in 2015. This would be especially significant because, as outlined above, these narrow coral-dominated zones make a disproportionally important contribution to maintaining the budget of the wider platform in an albeit low, but still net positive, budget state. Indeed, if similar bleaching-driven transitions have occurred at Vavvaru it is realistic to assume that the budget of the overall reef platform may now have become net negative.

Assuming the reef carbonate budget at Vavvaru has followed similar trajectories to sites in the southern Maldives (Perry and Morgan, 2017), there may thus be a number of implications for the island both positive and negative, and that will influence future island stability over different timescales. An initial major impact, would be for a likely overall transition in the framework budget to a net negative state. This may have only a minor impact on the stability of the island itself in the short-term, but if low coral cover states persist then the growth potential of the reef areas will be diminished, limiting any wave protecting functions that these discrete areas of reef provide. At bleaching impacted sites in the southern Maldives, however, an additional consequence of coral die-off was an initial marked increase in parrotfish bioerosion (Perry and Morgan, 2017). If this has been repeated at Vavvaru the short-term budget impacts may actually be a spike in the generation of parrotfish derived sediment, which as the major sand producer on these reefs may result in a short-term additional pulse of sediment to the island. However, long-term persistence of low coral cover states combined with progressive bioerosion pressure would act to progressively reduce the structural complexity of the reefs, and which provides essential shelter and food availability (e.g., van Rooij et al., 1996). Thus, any short-term enhancement of the sediment budget may be offset by a progressive reduction in parrotfish biomass. Thus, one can hypothesize that future benthic community transitions that impact upon different aspects of carbonate production and erosion, may have rather profound implications for the stability of reef-associated landforms at this, and other similar, sites. Quantifying and mapping the rates and sources of carbonate production and erosion at reef-wide scales thus provides us with tools to consider these landform stability issues in the context of overall budget states and of habitat diversity and extent.

Early studies of reef carbonate budgets provided not only novel estimates of reef production rates, but also made the critical point that large areas of most reef complexes i.e., encompassing reef flats and lagoon areas, are likely to have much lower carbonate production rates. It was thus suggested that overall reef-wide carbonate budgets are likely to be much lower, perhaps averaging only round 1.5 G. In other words the average reef system might have an approximately one order of magnitude lower production rate overall compared to that measured in the most productive (shallow, coral-dominated) zones. Our findings support this idea but extend this knowledge by developing an approach that allows us to better quantify which carbonate producing and eroding taxa drive these patterns across reef complexes, and how rates of both reef framework and carbonate sediment production and loss vary between habitats. This is an approach that has value for supporting efforts to up-scale estimates of inorganic carbonate production and cycling to reef-wide and, indeed, regional scales, and as outlined above is an issue with relevance beyond simply calculating reef-wide carbonate budget metrics. These processes of carbonate production, cycling, and loss are likely to have implications for the development and stability of proximal reef-associated landforms, including beaches and, in this example, reef islands. The approach we outline thus has potential to allow us to better explore the complex nature of geo-ecological linkages in reef systems and along reef-fronted coastlines, and to consider the budgetary implications on the form and development of reef-associated landforms. Such approaches are of course only as good as the empirical data that drive the production and erosion estimates, and whilst these are improving for many key aspects of framework and sediment production, there is a pressing need for additional data on coral calcification rates from many regions and for improved approaches to calculating sediment generation rates (see Perry et al., 2016 for such approaches in the content of the carbonate producing algae Halimeda). However, especially poorly constrained at present are data on rates of framework breakdown and export from reefs, and of sediment losses, both physical and geochemical. As recent studies have shown (Cyronak et al., 2013; Eyre et al., 2014; Morgan and Kench, 2014; Andersson, 2015) both areas are likely to be of significant budgetary relevance and urgently deserve additional attention. These data constraints aside, the current study highlights the marked differences that can exist in terms of total framework and sediment production between habitats, and the major dampening effect on the net carbonate budget at the platform scale when rates are scaled to account for habitat type and size. The net impact is that whilst individual reef systems may exhibit discrete areas of high coral cover and high associated carbonate production, platform scale budgets can be significantly lower, to the extent that any loss of coral cover and associated reductions in grazer-driven bioerosion in the main reef zones would be sufficient to tip the overall platform budget into a net negative state, with consequent implications for platform top island stability.

CP conceived the research. CP, KM, and RY contributed to field data collection and analysis, and to the writing of the paper.