Several studies have described the effects of sediments on corals, but we will not review them all here (see Johannes, 1975; Rogers, 1990; Fabricius, 2005; Erftemeijer et al., 2012 and references cited therein). Once sediments enter the marine environment, they can settle directly on corals and other reef organisms or remain in suspension, reducing the light required for photosynthesis, growth and calcification (Cortés and Risk, 1985; Richmond, 1993). Sediment characteristics including size, color, and mineralogy dictate settling/suspension dynamics and light attenuation capacities (Storlazzi et al., 2015). The coarse fraction (~>2 mm) settles on the bottom where it can smother and abrade corals during high-energy events (Jones et al., 2016). Finer sediments (~<0.0625 mm) tend to remain in suspension longer and can easily become resuspended (Ogston et al., 2004; Storlazzi et al., 2009) reducing light penetration over long distances (~10s km) offshore or down-current from points of delivery depending on river flow and ocean currents (Restrepo et al., 2016; Tarya et al., 2018; Torres-Pérez et al., 2021). Fine sediment particles settling on corals also can enhance bacterial activity on coral tissue (Risk and Edinger, 2011) and can expose corals to dissolved substances, trace metals, and other contaminants such as pesticides that may adhere to clay minerals (Glynn et al., 1989; Pait et al., 2012; Whitall et al., 2014; Jones et al., 2020).

Studies have been conducted in the laboratory and in the field. Interpretation of the results of laboratory manipulations must consider the types and particle sizes of sediments applied, the presence of contaminants, and the unnatural conditions (Hubbard and Pocock, 1972). Experiments in the field are difficult because of conditions which cannot be controlled as in a laboratory setting, such as water motion that can carry away applied sediments. Generalizations are difficult because studies have been done with terrestrial and marine sediments, with different coral species, with different particle sizes, with sediments alone or mixed with nutrients, and under both natural field conditions and in laboratories.

Corals have mechanisms to cope with sediments settling on polyps (e.g., cilia, mucus shedding, etc.) up to a point (e.g., Stafford-Smith and Ormond, 1992; Stafford-Smith, 1993). However, these all require expenditure of energy that could otherwise go into growth or other functions. Some studies describe experimental manipulations with terrestrial and marine sediments in the field or in the laboratory with a focus on short term exposure (days) to identify acute effects (Weber et al., 2006). In an early laboratory study of 26 coral species found in Florida, Hubbard and Pocock (1972) examined the sediment rejection capabilities after application of particles ranging from 62 to 2,000 μm. All of the species were able to reject silt-sized particles and most could reject all particle sizes. Only two species were described as poor or incompetent in rejecting sediments.

Building on these and similar studies, Erftemeijer et al. (2012) presented a synthesis of field and laboratory data on the sensitivity (as determined by minor sublethal to major lethal effects) of 46 coral species to turbidity and of 71 species to sedimentation. They assigned a score (the basis of which was not provided) to each species ranging from very tolerant to very sensitive, significantly including not only the exposure in terms of suspended matter and sediment input (mg cm^–2 day^–1) but also the duration of exposure, and found a significant relationship between sensitivity and growth form, for example with branching species typically less affected than massive ones. Given the range of responses and the inherent complexity, this useful but somewhat subjective ranking could be improved through re-examination, updating and substantiation with empirical data.

Many studies provide useful data on tolerance to sediments but for only a few species (e.g., Rogers, 1983; Rice and Hunter, 1992). Morphology clearly is a factor. Branching corals are less likely to retain sediments resulting in smothering and partial mortality than more rounded, massive colonies (Rogers, 1983; Erftemeijer et al., 2012.) The focus of studies of the effects of suspended sediments or deposition should not just be on short-term responses of adult corals, but also on coral growth rate and recruitment (see Jokiel et al., 2014; Perez et al., 2014; Jones et al., 2020) and changes in the ability to photosynthesize (see Phillip and Fabricius, 2003).

Sublethal responses of a limited number of Pacific and Atlantic coral species to sediments have been noted in several studies of coral photobiology, using PAM Fluorometry (Piniak and Storlazzi, 2008; Flores et al., 2012; Browne et al., 2014; Rushmore et al., 2021) and analysis of gene expression (Downs et al., 2012) suggesting that even when mortality does not occur, sediments can reduce overall energy available for growth and reproduction.

Sofonia and Unsworth (2010) proposed that accounting for duration of exposure to varying levels of Photosynthetically Available Radiation (PAR) provides a more meaningful metric to quantify the effects of turbidity than single threshold values. Such an approach could allow monitoring of light received by corals to define adverse levels of turbidity above which corals would be stressed, triggering specific management actions (Sofonia and Unsworth, 2010). Sofonia and Unsworth (2010) do not discuss how they quantified stress to corals or acknowledge how challenging this would be given that sublethal effects can affect reproduction and growth well before corals die partially or entirely.

Coral species differ in their ability to deal with sediment accumulation and with reduction of light transmissivity associated with turbidity. Species-specific tolerances, however, are not as well-studied as one might think given statements in the published literature. In general, presence or absence of particular coral species growing on a reef may not reveal that much about the environmental conditions or the stressors to which the corals are exposed and does not serve as a precise indicator of sediment conditions reflecting a gradient of sediment influence. We have to be cautious about thinking of certain species as “indicators.” For example, there is evidence that Montastraea cavernosa is more resistant to sediments than some other species (Lasker, 1980), but that does not mean that its presence on a reef indicates that the reef has been subjected to sediment stress.

Although there is some evidence that coral species have different tolerances/sensitivities to suspended and accumulated sediments, currently no rigorous hierarchy or well-defined gradient of coral species responses to sediment accumulation or light attenuation exists. Susceptibility (or resistance) to sediment stress could vary for each species and could differ depending on reef location (i.e., proximity to shore, depth, wave action, etc.) and particular environmental conditions (Weil, 2021). However, large shifts in coral community composition, for example as revealed by cores, can indicate major changes in environmental factors such as a degradation in water quality (Duprey et al., 2016; Cybulski et al., 2020; Cramer et al., 2021).

Stony corals are the architects of coral reefs, and as Bruno and Valdivia (2016) have concisely stated, “Coral loss is thus a direct measure of habitat degradation,” and “Macroalgal cover is an indirect measure of reef degradation.” In general, deterioration of coral reefs from excessive sedimentation or other causes, is reflected in declining amounts of coral and increasing amounts of algae (e.g., Rogers and Miller, 2006; Schutte et al., 2010; Jackson et al., 2014; Arias-González et al., 2017), although these changes are not always tightly linked. Not all shifts in relative abundance between benthic components involve corals and algae. In one study in Panama, the coral Agaricia tenuifolia replaced Porites over time apparently as a result of a loss of water quality with increased input of nutrients and sediments associated with agriculture and a rise in population density (Aronson et al., 2014).

Large quantities of sediments (deposited or suspended in the water column) can lead to adverse reef-scale effects. These may include bleaching, reduced recruitment, collapse of physical structure with associated loss of ecological services, and loss of fish habitat with associated declines in abundance and species diversity (Erftemeijer et al., 2012). Several studies show a correlation between presumed (as opposed to measured) sediment exposure (along a distance-based gradient) and lower coral cover (coral abundance), lower recruitment, and more partial mortality (e.g., Acevedo et al., 1989; Ginsburg et al., 2001; Nugues and Roberts, 2003; Dikou and van Woesik, 2006a).

Establishing how much sediment is excessive, leading to adverse effects on structure or function, from either a terrestrial or marine perspective is difficult (Rodgers et al., 2012; Tuttle and Donahue, 2020). Watersheds and coral reefs are not commonly studied as connected systems simultaneously, particularly in the Caribbean. No rigorous, long-term comprehensive, before, during and after empirical study of the effects of dredging or increased/decreased terrestrial sediment discharge on the structure and function of a Caribbean coral reef has been published. Similarly, there is no rigorous analysis of changes in marine water quality or sedimentation rates and concomitant improvement in coral reef structure resulting from watershed management actions (Kroon et al., 2014).

With increasing sediment (or other) stress, coral species are unlikely to be replaced by other coral species. More often, the relative abundance of corals of different species changes as colonies become fragmented and portions of them die. Corals are modular, colonial organisms, remnants of which can persist after other portions die. In reefs off Okinawa, Japan, partial mortality was a better indicator of the deleterious effects of sediment-laden runoff than percent live cover, colony density, generic richness, or colony size (Dikou and van Woesik, 2006b).

High sediment exposure, from runoff of terrestrial sediments or dredging projects, can have direct and indirect effects on coral communities and nearby habitats. Direct effects include smothering and mortality of corals, while indirect effects include impairment of coral recruitment and associated alteration of colony size frequency distribution (Cortés and Risk, 1985). In a study of coral reefs along an inshore to offshore gradient in the USVI, Smith et al. (2008) noted more old coral mortality (i.e., partial and not recent) and bleaching nearshore suggesting a link with higher sediment exposure from land runoff. In a study of the effects of flooding in a small bay in Kauai, Hawaii, Storlazzi et al. (2009) documented survival of adult corals but burial of sites which would have been suitable for coral settlement. Small variations in hydrodynamic processes could drive considerable changes in rates of sediment deposition and resuspension (These studies found no correlation between sedimentation and coral diseases.) In a review of the effects of dredging near or on coral reefs, Erftemeijer et al. (2012, Table 1) noted a range from minimal effects to, in one case (in Bahrain), a major loss of 22 ha of coral reefs with an associated degradation of 8 ha nearby because of increased turbidity and sedimentation. Effects were both direct and indirect, with loss of coral and declines in fish abundance and diversity, and changes in water circulation. In some cases, seagrass beds and mangroves were affected.

An early focus on the direct effects of sediments on corals based on both laboratory and field studies has been followed by more recent studies of indirect effects on corals as well as effects on other reef-associated organisms. Fewer studies have focused on the interactions between sedimentation and other stressors. Sediments that accumulate on algae (filamentous turf or macroalgae) sometimes promote and sometimes inhibit algal growth with subsequent effects on coral growth and survival (Birrell et al., 2005; Muthukrishnan and Fong, 2014). Sediment particles also inhibit recruitment/settlement and survival of both algae and corals with long-term consequences for reef persistence or recovery (e.g., Bahr et al., 2015). Often terrestrial sediments contain organic matter or other pollutants, and the deleterious effects of sediments may be attributable more to these materials than to the sediment particles themselves (see van Dam et al., 2011). Macroalgae can cause detrimental effects beyond smothering and abrasion of corals and inhibition of coral recruitment. These include decreased coral growth (Burkepile and Hay, 2008), lowered fecundity (Foster et al., 2008), and greater mortality (Nugues and Bak, 2006; Smith et al., 2008).

Sediments can interact with other stressors such as overfishing to hinder coral growth, and it is difficult to quantify the interactive effects of local stressors (Bruno and Valdivia, 2016). Any stressor affecting the susceptibility of a coral (immune response, energy availability, pathogen virulence, etc.) could be a potential driver for a disease to develop. Some evidence indicates possible links between increased sedimentation and diseases of scleractinian corals, specifically black band disease (Littler and Littler, 1996; Bruckner et al., 1997). Sediments may be a reservoir for disease pathogens, and benthic algae may trap sediments harboring disease pathogens (Nugues et al., 2004; Voss and Richardson, 2006a; Haapkylä et al., 2011).

Some studies show increased disease prevalence with increase in nutrients specifically (Vega Thurber et al., 2014; Wear and Thurber, 2015). However, disease outbreaks occur in the absence of any obvious or rigorously documented stressor (with the exception of warming seawater temperatures). Currently extensive and rapid coral mortality in Florida and portions of the Caribbean is being driven more by stony coral tissue loss disease (Alvarez-Filip et al., 2019; Muller et al., 2020; Brandt et al., 2021) than any other stressor, but no link to any environmental factor has been discovered and verified.

The most urgent problem now affecting coral reefs in Florida and some portions of the Caribbean, including the USVI and PR, is the advance of this stony coral tissue loss disease (SCTLD), first reported on Florida reefs in 2014 (Precht et al., 2016). Rates of coral mortality exceed any documented before, and some highly susceptible species are being eliminated from some locations. Although other less virulent diseases have been associated with sediment exposure or increased nutrient concentrations (Bruno et al., 2003; Kaczmarsky et al., 2005; Voss and Richardson, 2006b), and stony coral tissue loss disease has not been linked conclusively so far to any anthropogenic activity, sediment exposure, sewage, chemical pollutants, or any other environmental factor, and research continues. Recently viruses have been documented in endosymbionts in corals with SCTLD, but no definitive reason for a viral disease to appear in the endosymbionts has been determined (Work et al., 2021).

It is logical to think that increased sediment exposure as well as any other factor that stresses corals would increase their vulnerability to diseases, but the correlation between sedimentation and diseases is not well-documented (Sutherland et al., 2004). No studies rigorously show sediment stress from dredging or increased sediment discharge from watersheds is linked to disease outbreaks. One study of a reef in Barrow Island, Australia (Pollock et al., 2014) concluded that dredging had caused an increase in coral disease prevalence, but the study began a month after dredging ceased, and a second study in the same area saw increased coral mortality from dredging effects but no link between dredging and coral diseases (see detailed discussion in Stoddart et al., 2019). In a study of the effects of a major dredging project in Florida, a particularly virulent disease was associated with more coral mortality than sediment stress (Gintert et al., 2019).

Terrestrial erosion refers to the wearing away of earthen materials by runoff, ice, wind, or gravity. Human effects on upland erosion vary with the type of activity and depend on landscape and climate characteristics (Ludwig and Probst, 1998; Syvitski et al., 2014). Erosion processes typically responsible for sediment discharge from watersheds to coral reefs include those induced by overland flow, streambank erosion, and landsliding (Figure 1). Not all eroded sediment reaches nearshore waters because it can be retained for variable periods by sediment sinks such as floodplains or wetlands (Walling, 1999). Sediment discharge represents the total mass of sediment delivered by a watershed per unit time (e.g., Megagrams per year [Mg = 10⁶ g]). The term sediment yield refers to the area-normalized discharge rate (e.g., Mg km^–2 year^–1; Milliman and Farnsworth, 2013). The term “runoff” refers simply to water flowing on the land surface and not to sediment discharge.

Runoff (i.e., freshwater) and sediment discharge from watersheds to coastal waters may be: (a) ephemeral (i.e., sporadically delivering runoff to coastal waters ∼10 times per year during wet conditions and only for a few hours or days at a time. Examples of these are the watersheds draining the USVI and Oahu-Hawaii; Wolanski et al., 2009; Ramos-Scharrón and LaFevor, 2016); (b) perennial (i.e., with constant streamflow year-round like many rivers draining Puerto Rico and Colombia; Larsen and Webb, 2009; Restrepo et al., 2016); or (c) a combination of both as is the case for Australian watersheds flowing toward the Great Barrier Reef (Joo et al., 2012). This is a relevant coral reef stress factor as the temporality of sediment delivery in part controls the nature (i.e., sporadic or chronic) and duration of exposure, factors known to determine sediment impacts on corals (Erftemeijer et al., 2012). However, neither sediment discharge rate nor its duration exactly equates to exposure levels as those also commonly depend on the distance from the outlet to the corals, sediment particle size distribution and mineralogy, and oceanic factors controlling direction, mode and rate of sediment transport and deposition.

The global peak of terrestrial sediment delivery to the oceans occurred during the first half of the 20th century (Hooke, 2000). This period predates the post-WWII proliferation of dams, which now act as sediment sinks that have reduced worldwide sediment discharge (Syvitski and Kettner, 2011). However, it has been known for a very long time (e.g., Marsh, 1867) that most coral reefs are far from the large watershed outlets included in these planetary-scale assessments (McLaughlin et al., 2003). Therefore, these global delivery estimates are not necessarily representative of sediment discharges to coral-bearing ocean waters (Browning and Sawyer, 2021). For example, there are numerous studies showing an acceleration of terrestrial sediment accumulation since the mid-20th century in marine environments of the northeastern Caribbean (e.g., Jessen et al., 2008; Ryan et al., 2008; Brooks et al., 2015). Although historical sediment delivery rates from coastal watersheds into nearshore systems have been documented at some locations (e.g., Bartley et al., 2018), limited empirical data exist for most coral reefs.

Watershed sediment discharges into coastal ecosystems can range significantly and corresponding impacts to corals are relatively unknown. Consequently, there is no specific watershed sediment discharge threshold above which it becomes deleterious to coral reefs. For example, sediment delivery from the 130,000 km² Burdekin watershed in Australia averages 5.6 million Mg year^–1 with disputable effects on the Great Barrier Reef (Neil et al., 2002; De’ath et al., 2012). In contrast, much smaller watersheds in moist to wet tropical areas of PR and St. Lucia have far lower sediment loads (ca. 5,000–450,000 Mg year^–1) that are also believed to degrade corals (Larsen and Webb, 2009; Bégin et al., 2014). Similarly, sediment delivered to fringing reefs in the USVI is considered a threat to coral reefs despite delivery rates of only 20–275 Mg year^–1 (Ramos-Scharrón and MacDonald, 2007a). It is important to note that watershed discharge or yield rates are typically reported in units of Mg year^–1 or Mg km^–2 year^–1, and while these are useful for communicating information among earth scientists, the meaning of these values in terms of coral stress is unknown. It is not known how these values translate into biologically-meaningful reductions in light transmissivity or sediment settling rates on corals occurring at higher temporal resolution (i.e., hours, days; Weber et al., 2006).

Sediment discharge does not equal the sum of all erosion occurring within a watershed:

where, “Erosion” is in units of mass per unit time (e.g., Mg year^–1) and equals the sum of all sediment produced by overland flow, landsliding, streambank erosion, and any other relevant erosive process. Sediment discharge ranges from ∼10 to ∼100% of that eroded within the watershed (Walling, 1983) [although values >100% have been reported (de Vente et al., 2007; Church, 2017)]. That is because, in broad terms, sediment discharge equals the difference between sediment released by erosion and that retained either briefly (i.e., hours to a few years) or over long-time frames (i.e., decades to thousands of years) by sediment sinks (e.g., bottom of hillslopes, floodplains, estuaries, salt ponds, etc.) (Meybeck and Vörösmarty, 2005). This can be expressed by the following equation:

where all terms are in units of mass per unit time (e.g., Mg year^–1), “Sediment discharge” represents the rate of sediment output, “Erosion” is as defined for Equation 1, and “Δ Storage” is the change in sediment accumulation in sinks over the same time period represented by sediment discharge and erosion. Δ Storage values may be both positive or negative. A positive value indicates a gain in the amount of sediment stored [e.g., a river going over its banks and adding sediment to its floodplain (i.e., retention)], while a negative value means a loss of sediment storage [e.g., a river widening or changing course during a high flow event facilitated by cutting into (i.e., releasing) previously stored floodplain sediments]. Therefore, sediment delivery can increase by: (a) accelerated erosion, (b) a decrease in sediment storage (i.e., releases of sediments stored in sinks), (c) reductions in the trapping efficiency of sinks, or (d) increases in sediment transport capacity of rivers due to urbanization and channelization, changes in river geometry induced by extreme flood events, or increased flow rates associated with climate change (Wohl, 2015). Conversely, decreases in sediment output can occur by reduced erosion and/or gains in sediment storage.

The role of sediment storage is crucial in defining how any increases in erosion manifest themselves as sediment discharge. In cases where a large portion of the eroded sediments become stored for thousands of years, erosion can occur without any noticeable increases in sediment discharge (Fryirs et al., 2007). In cases where eroded sediment is stored for decades, these sediments can become an important source of contemporary sediment discharge (Wohl, 2015). This is the presumed case in some portions of PR where ‘legacy sediments’ originally eroded from hillslopes during the period of extensive cultivation on the island were originally stored near river margins and are now believed to contribute to a significant portion of watershed-scale sediment discharges (Clark and Wilcock, 2000; Larsen and Santiago Román, 2001).

The degree to which terrestrial sediments discharged from watersheds can potentially influence coral reefs depends not only on watershed processes controlling erosion, deposition and discharge but also on the level of connectivity between watersheds and coral reefs (Bainbridge et al., 2018). Throughout the USVI and PR, studies have shown that watersheds can exert an influence on coral environments through the following approaches: (a) spatial distribution of coral species ranked by perceived level of sediment tolerance in relation to distance from watershed outlets (e.g., Acevedo et al., 1989); (b) a proven temporal synchrony between measured seasonal watershed runoff and isotopic signals of dissolved inorganic carbon in coral cores (Moyer and Grottoli, 2011); (c) decreases in net skeletal growth rates determined from coral cores associated with seasonal increases in watershed runoff (Miller and Cruise, 1995); (d) mineralogical fingerprinting of sediment sources (Takesue et al., 2021); and (e) sediment plume mapping using remotely sensed techniques (Hernández et al., 2020). However, none of these studies have demonstrated temporal trends in coral reef conditions in response to changes in sediment discharge corresponding to varying land use, extreme events, or the implementation of erosion mitigation strategies.

Our review of the literature reveals that the science related to sediment dynamics within the marine environment in the northeastern Caribbean lacks the monitoring and modeling capacities implemented in Micronesia and the Great Barrier Reef (e.g., Wolanski et al., 2004). Additionally, the successes in curtailing stresses related to land-based sources of pollution on corals in Micronesia (Richmond et al., 2007) are not easily replicated throughout the northeastern Caribbean. In the islands of Palau, Guam, and Pohnei reductions in pollutant discharges to waters sustaining coral reefs were effective due to the singularity of the stressor and its human cause (i.e., terrigenous sediment discharge associated with accelerated overland flow erosion caused by land clearing) and a unique grassroots governance structure that allowed for widespread implementation of recommendations (i.e., reducing deforestation). In contrast, watersheds in the USVI and PR deliver a wide variety of pollutants to coastal waters including nutrients, fecal pollution, agrochemicals, and heavy metals, among many others (e.g., Bonkosky et al., 2009; Larsen and Webb, 2009; Oliver et al., 2011, 2018; Norat-Ramírez et al., 2019). Additionally, sediments are produced by a variety of sources and are related to numerous activities including road building, agricultural production, and urbanization. Finally, land and coral reef governance in the USVI and PR is complex as management responsibilities fall under a variety of jurisdictions and this makes it difficult to reach a widespread consensus on the actions needed to reduce sediment and other pollutant discharges into coastal waters.

Quantification or estimation of sediment discharge is not an exact science. That is because the processes controlling sediment availability and transport are very complex and variable in time (White, 2005) and this leads to inexact estimates [±50% accuracy in annual sediment yields according to Milliman and Farnsworth (2013)]. Therefore, relying on empirical data to establish statistically significant changes in sediment discharge as a response to land use or climatic forcing is difficult. To add to this challenge, most streams are ungauged, particularly in developing countries (e.g., the Insular Caribbean; Lumbroso et al., 2011) where the majority of the world’s corals are located (Veron and Phinney, 2006). Therefore, most sediment discharge assessments rely on indirect methods. Most sediment-delivery assessments either have been qualitatively inferred through evaluation of changes in land use or estimated using uncalibrated and unvalidated models (e.g., Burke et al., 2011) and this is particularly true for the Caribbean. Due to the lack of data, it is not possible to rigorously evaluate the performance of the most sophisticated models versus other more simplistic approaches. An in-depth discussion on the pros and cons of the variety of available approaches is beyond the scope of this article. Rather, here we provide a brief critical commentary on how land cover assessments and models have been used for the purpose of estimating watershed sediment discharge to coral-bearing coastal waters.

The USVI and the island of Culebra-PR are drained by ephemeral streams within micro-watersheds ranging in size from a few hectares to about 10 km² (Ramos-Scharrón and LaFevor, 2016). Even though land use at both locations is legally restricted by the intended enforcement of the Coastal Zone Management Act and more than thirty protected areas, terrestrial sediment is still considered an important coral-reef stressor (Rogers et al., 2008, Commonwealth of PR and NOAA, 2010). Progression in how watershed sediment discharge to coral reefs has been addressed in the USVI and Culebra follows the same general timeline for most of the world. The earliest evidence of sediment-related concerns with respect to coral reefs in the USVI dates to the late 1970s, when water-turbidity monitoring was initiated and relatively high values were detected at some locations (US-EPA, 1983). Anecdotal evidence from the 1980s includes the documentation of noticeable sediment entering coral-bearing waters adjacent to recently developed watersheds (Beets et al., 1986; Rogers and Teytaud, 1988) and the tendency for coral-colonized substrates to be located away from stream outlets and large watersheds (Hubbard, 1986; Hubbard et al., 1987). Anecdotal evidence in Culebra was first documented in the early 1990s (Collazo et al., 1992; Hernández-Delgado, 1992) and formally evaluated through modeling (Ramos-Scharrón et al., 2012a) and, much later, monitoring (Otaño-Cruz et al., 2017; Gómez-Andújar and Hernández-Delgado, 2020).

Sediment-budget studies combining various forms of empirical data, field-based and remotely-sensed reconnaissance of erosion processes by trained personnel, and watershed modeling led to the identification of the unpaved road network as the main source of sediment on many of the islands’ watersheds (Figures 2a–c; Anderson and MacDonald, 1998). Unpaved roads stand out as sediment contributors due to the combination of: (a) erosion rates that are up to four orders of magnitude greater than natural rates (Ramos-Scharrón and MacDonald, 2005; McLaughlin, 2019); (b) the relatively minor contribution from other sediment sources due to a low degree of development and lack of widespread agricultural activity; and (c) the absence of widespread landsliding (MacDonald et al., 1997). Runoff and sediment delivery in these ephemeral-stream settings naturally occur only a handful of times per year (Larson et al., 2015), but unpaved roads can potentially induce a tenfold increase in the magnitude and frequency of sediment delivery to coastal waters (Ramos-Scharrón and LaFevor, 2018). Suspended sediment concentrations of runoff discharged from roaded watersheds are in the 100s to 1,000 mg L^–1 range while those from undisturbed watersheds are only in the 10s to less than 100 mg L^–1 (Ramos-Scharrón, 2004). Studies from a variety of terrestrial and marine perspectives agree that sediment loads into coral reefs are highly dependent on the density of unpaved roads (Nemeth and Nowlis, 2001; Brooks et al., 2007; Ramos-Scharrón and MacDonald, 2007a; Gray et al., 2008; Otaño-Cruz et al., 2017) and that sediment-related impacts on adjacent coral reefs are strongly dependent on the distance between them and updrift watershed outlets (Smith et al., 2008; Otaño-Cruz et al., 2019).

The small size and steepness of watersheds in the USVI and Culebra, in combination with the low erosion rates from undisturbed areas (2–8 Mg km^–2 year^–1) and the potential for land disturbance to increase erosion by several orders of magnitude set the stage for drastic changes in sediment discharges even with relatively limited development (Ramos-Scharrón and MacDonald, 2007b). The degree to which increases in hillslope erosion may lead to greater sediment discharge is referred to as “sensitivity” (Fryirs et al., 2007). This is a function of connectivity (see Box 1) and is dependent on the proximity of sediment sources to the stream network, stream gradient, and the effectiveness of sediment-storage elements (e.g., salt ponds), among other factors. Where coastal wetlands that are capable of retaining eroded sediment are absent, the proportion of sediment mobilized within USVI and Culebra watersheds that is delivered to coastal waters is expected to be quite high (∼75%; Ramos-Scharrón and MacDonald, 2007a; Ramos-Scharrón, 2021).

The 69-ha, short (∼0.8 km from ridgeline to coastline) and steep (average slope of 25°) Hawksnest Bay watershed in St. John lies mostly within Virgin Islands National Park boundaries and is, therefore, largely undisturbed. However, development has occurred and has led to varying degrees of soil exposure. The main source of sediment up until the 1970s was a limited network of unpaved roads that occupied ca. 3.5% of the watershed. Application of the locally and empirically derived STJ-EROS model (Ramos-Scharrón and MacDonald, 2007a) suggests that the unpaved road network in the 1970s likely led to sediment yields ∼90 times above background rates (Figure 3). Paving of a significant stretch of a road in the late 1970s reduced the proportion of watershed with exposed soils by half and that may have reduced sediment discharge into the bay by about a third. Land clearing associated with the construction of a medical clinic that began in October of 1980 (Hubbard et al., 1987) increased the proportion of exposed soil to 4.6% and sediment discharge to ∼130 times above background, which is likely the highest the watershed has ever experienced. Construction of this clinic coincided with the occurrence of one of the island’s highest recorded daily rainfall totals (18 April 1983; total: 381–483 mm and max intensities 132–152 mm h^–1 at Caneel Bay and Trunk Bay, respectively: Curtis, 1985; Boulon. personal communication, 13 March 2021). During this brief event, about 215 Mg of sediment are estimated to have reached the bay in a single day (∼85% from the construction site). This pulse of sediment may have been a cause of short-lived decreases in the growth rate of the dominant coral species Orbicella annularis (formerly Montastraea annularis) documented in Hawksnest Bay (Hubbard et al., 1987). The sediment pulse may also have been responsible for the mortality of Acropora palmata colonies in eastern Hawksnest closest to the watershed outlet (Beets et al., 1986). However, even in this situation where a significant pulse of sediment was evidently delivered directly to coral reefs, there is a lack of direct evidence of sediments being responsible for coral losses.

Currently, the estimated sediment yield is ∼3,000% above undisturbed conditions even though only portions of the developed areas remain unvegetated and the limited new-road segments have been mostly paved. Our estimates indicate that sediment delivery from this watershed can increase 25-fold over background for every 1% increase in land area converted to bare soil. This highlights the sensitivity of these small and steep watersheds to even minor levels of disturbance. It also points at the need to limit erosion by quickly introducing vegetation, paving existing roads and enhancing sediment-storage capacity with detention ponds in order to protect nearshore coral reefs from sediments resulting from new or already existing development.

Anecdotal evidence and scientific studies relying on modeling and both field-based and remotely-sensed evidence within the USVI and Culebra (e.g., MacDonald et al., 1997) have helped to develop watershed-management strategies that target decreasing erosion from unpaved roads and reducing connectivity (Commonwealth of PR and NOAA, 2010; The Territory of USVI and NOAA, 2010; Reed, 2012; HWG and PDC, 2017). Strategies have included cutting erosion rates along unpaved roads by a third through road-drainage improvements (Ramos-Scharrón, 2012), retaining sediment in detention ponds (Ramos-Scharrón et al., 2012b; Figure 2d), and redirecting runoff to hillsides and ephemeral streambeds to reduce the frequency of sediment delivery to coastal waters, among others (Reed, 2012; Ramos-Scharrón and LaFevor, 2018).