Anthropogenic disturbance and modification of natural habitats are major drivers of the systemic decline of biodiversity worldwide (Ojima et al., 1994; Waycott et al., 2009). Physical disturbance from human activities (e.g. clearing and building) can generate sharp environmental boundaries, but can also introduce secondary processes that disturb surrounding habitats. These so-called ‘secondary disturbances’ include changes in light, water, sound and nutrient regimes, disrupting animal movements and behaviour, or assisting the spread of invasive pests (Walker et al., 1987; Laurance and Yensen, 1991; Murcia, 1995). Secondary disturbances can be mechanistically linked or contingent on primary disturbances, and have lower relative intensity though they may be more persistent. The difficulty of detecting secondary disturbances means they are sometimes poorly understood and often overlooked by environmental impact assessments (Walker et al., 1987; Erftemeijer and Robin Lewis, 2006). Whilst several of the key mechanisms causing biodiversity loss are well investigated, the roles of and interactions between primary and secondary disturbances are still unknown for many species and stressors (Maxwell et al., 2017).

A prominent example of an environmental impact resulting from both primary and secondary disturbances in the marine environment is the impact of boat moorings on seagrasses. Moorings physically disturb the seabed within a ‘scour zone’ and potentially introduce a range of secondary disturbances in the surrounding area. For example, clearings from scour can disturb surrounding seagrasses by increasing sediment movement (Frost et al., 1999), burying plants, driving fragmentation, and impacting resource translocation (Tuya et al., 2013). Open space can increase exposure to storm damage and give fast colonising species (e.g. species of the genera Zostera and Halophila) a competitive advantage, which may outcompete slower-growing species (Kirkman, 1985). Shading by moored vessels is another potential secondary disturbance that extends beyond the scour zone (Evans et al., 2017). Each of these secondary disturbances that extend into the surrounding meadow may accumulate with increasing mooring density.

Elsewhere, efforts to conserve seagrasses have included mitigation of direct impacts (e.g. no anchoring areas), reducing pollution from neighbouring watersheds (e.g. Bryars and Rowling, 2009), or active restoration (e.g. Paling et al., 2007). The limited success of restoration and ongoing declines of seagrasses globally suggest that losses are challenging to reverse, emphasising the need to understand and mitigate the processes causing loss (Cullen-Unsworth and Unsworth, 2016). One option to reduce impacts from boating is through environmentally friendly mooring designs, which reduce physical disturbance and potentially some secondary disturbances. Trials of new moorings that avoid physical disturbance have shown some success (Demers et al., 2013), posing a potential strategy to mitigate impacts on seagrasses (Glasby and West, 2018).

Coastal monitoring programs are our chief method for detecting ecological changes, but most lack sufficient spatial and temporal resolution to attribute and manage change before losses occur (Field et al., 2007; Schultz et al., 2015). Without sampling at higher spatial and temporal resolution, we may fail to detect or mitigate local processes that contribute to declines in difficult-to-restore habitats, or miss the early-warning signs of ecosystem collapse (Duarte, 2002; Dakos et al., 2012; Kéfi et al., 2014; Schultz et al., 2015). This is especially true for secondary disturbances, which are typically more subtle and spatially distributed than primary disturbances. Most marine and coastal ecosystems lack sufficient spatial and temporal data to detect small-scale losses before they escalate (Pressey, 2004) or impacts of secondary disturbances.

Methods from landscape ecology use detailed observations over large areas to untangle the local and regional processes that drive ecological patterns, but practical limitations have typically precluded this style of study from marine subtidal environments (Pittman, 2017). However, recent technological developments have facilitated the use of these methods in marine systems, including sonar- and accelerometer-based positioning systems, remote autonomous vehicles (Hill et al., 2014), and high-resolution imaging converted to ecological data via computer-vision and machine-learning software (e.g. Beijbom et al., 2015; Griffin et al., 2017). Spatial analytical approaches allow for investigating relationships between impact exposure and species distributions, with taxonomic and spatial resolution limited only by the sampling methodology (Elith and Leathwick, 2009). When matching spatial data is available to describe the intensity of the stressor, distribution modelling can assess the relationship and predict species’ response across space (Elith and Leathwick, 2009). Spatially explicit ‘seascape’ analyses are still relatively rare in sub-tidal marine systems but might improve our understanding of sensitive habitats and processes that cause their decline (Brown and Hamilton, 2018). Similar approaches can be applied to risk analysis, impact assessment, and multiple-stressor studies, and can improve our ability to simulate and test the effectiveness of regulation and management in complex ecosystems (Dumbauld and McCoy, 2015; Martin et al., 2018).

In this study, we use a landscape approach to assess the impacts of coastal infrastructure on the threatened seagrass Posidonia australis, which forms a dominant and valuable habitat in southern Australia. Declines since the 1950s (Glasby and West, 2015) and ongoing risks from coastal development (Clark and Johnston, 2016) have led to P. australis being listed as a regionally endangered ecological community in some areas of New South Wales under the NSW Fisheries Management Act 1994 in 2010 and the Australian Environment Protection and Biodiversity Conservation Act 1999 in 2015. Despite the legislative protection, P. australis meadows in some estuaries have declined by up to 40% in recent years (Evans et al., 2018). Moorings have removed an estimated ~130,000 m² of the species (Glasby and West, 2018). Individual scars may take centuries to recover (Meehan and West, 2000) due to slow growth and the fact that P. australis relies on intact rhizome material to stabilise sediments for new shoots (Meehan and West, 2000). Investigations into moorings effects on P. australis have concentrated on the rate and area of physical removal (Demers et al., 2013; Evans et al., 2018; Glasby and West, 2018). These losses, however, could be difficult to predict if secondary disturbances that accumulate with mooring density result in non-linear relationships between distance (exposure) to moorings and seagrass survival (Unsworth et al., 2017).

We aimed to investigate whether environmentally friendly moorings (EFM) reduced the primary and secondary disturbance from boat moorings for the threatened seagrass P. australis and two non-threatened and co-occurring species of seagrass (Zostera muelleri and Halophila ovalis). We surveyed a seagrass meadow at Manly Cove, Sydney Harbour, at a high spatial resolution (<1 m) to investigate the response of P. australis, Z. muelleri and H. ovalis to the proximity and density of the two mooring designs. We separately tested the effects of mooring design (i) close to the mooring in the area subject to physical disturbance by scour (0-5m from moorings), and (ii) further from the mooring (5–25m away) where there is no scour, but secondary disturbance may be operating. We also demonstrate how a spatially explicit approach can aid decision-making by simulating the implications of three potential management actions at the site. Specifically, we used simulations to predict how the cover of P. australis may change in response to (i) replacing any single traditional mooring with an EFM, (ii) replacing all EFMs with traditional moorings, and (iii) replacing all traditional moorings with EFMs.

Our results suggest that in addition to boat moorings physically disturbing seagrasses, they also reduce seagrass cover in the surrounding meadow, which may reflect secondary disturbance processes. Furthermore, we found indications that reduced physical disturbance was associated with a concomitant improvement in the cover of the threatened seagrass Posidonia australis in the surrounding meadow. In areas subjected to direct disturbance, moorings designed to avoid contact with the seabed (EFMs) were associated with significantly higher seagrass cover than traditional ‘swing’ moorings. Potential impacts of secondary disturbance from moorings (i.e. impacts outside scour zones) were only detected for the threatened species Posidonia australis, a slow-growing seagrass with low recovery potential, while two faster-growing seagrasses, Zostera muelleri and Halophila ovalis showed no significant relationships with moorings outside of the scour zone. Mooring density was associated with lower P. australis cover, but this effect was reduced when most moorings were EFMs. We suggest that this amelioration of cumulative effects due to EFMs might reflect a reduction in the secondary disturbances that are contingent on physical scour (e.g. lower rates of sediment transport and edge effects) rather than by moored boats (e.g. shading, pollution).

Simulations found that the benefits of replacing traditional moorings with EFMs were greater in the non-scour than scour zones, increasing seagrass cover at the meadow scale. These results support the use of mooring designs that avoid physical disturbance to seagrasses, which in turn are likely to reduce secondary disturbances to the surrounding meadow. By modelling the density of seagrass, we were able to add details to estimates of P. australis at the site beyond what would be obtainable from satellite or aerial imagery. The sampling methodology was an example of extending traditional landscape ecology ideas (e.g. Walker and Walker, 1991) into underwater seascapes, and similar approaches may be more frequently or extensively conducted using hyperspectral remote sensing or automated vehicles in the future. This study demonstrates both the applicability of seascape ecology for impact identification and simulation; and emphasises the importance of identifying the key scales at which impacts might accumulate.

Bare circular clearings of vegetation are a characteristic signal of the impact of boat moorings in seagrass meadows, but this was the first effort to distinguish between direct physical disturbance from mooring gear and potential secondary disturbances due to habitat clearing (e.g. increased sediment transport, fragmentation and edge effects) or the moored vessels (e.g. shading). Studies of boat mooring impacts on seagrasses have previously quantified the extent of complete clearings by measuring the area of sandy vs vegetated seabed in aerial images (Glasby and West, 2018), diver-surveys (Demers et al., 2013), and/or camera-drops (Unsworth et al., 2017). These studies established relationships between the size of mooring scars and seagrass species, depth, mooring age, and vessel size. They determined that impacts from individual moorings scale up to cause substantial impacts at a regional scale. Glasby and West (2018) demonstrated that mooring scars in P. australis grew linearly after installation (over a 4.8-year period) despite no perceived change in the mooring equipment. We suggest that the expanding decline of seagrasses around moorings is partly due to secondary disturbance contingent upon the physical disturbance, and that these processes have significant implications for the persistence of seagrasses where traditional moorings are installed, especially in meadows of threatened P australis.

Species traits may explain the relative sensitivity of P. australis (c.f. Z. muelleri or H. ovalis) to the secondary disturbance from physical fragmentation by moorings and, in turn, the stronger response of P. australis when physical disturbance is reduced. Mooring scars in P. australis meadows are larger than those found in meadows of smaller, faster-growing species and expand faster over time (Glasby and West, 2018). Coloniser or early successional genera like Zostera, Heterozostera and Halophila are often interspersed within fragmented P. australis meadows and can be indicators of environmental stress (Irving et al., 2013). The study of human impacts in marine and terrestrial systems has benefitted from studies that identify how community dynamics can modify stressor responses (Johnston and Keough, 2003). Accordingly, more investigation is required to determine whether the patterns we observed are representative of monospecific meadows, especially for Z. muelleri and H. ovalis.

Apart from physical impacts, increasing mooring/boat density should result in less light reaching the seabed. Seagrasses are susceptible to major reductions in light - but the effects of chronic low-level shading are less well understood (Bulmer et al., 2016) and are likely to be both site and species-specific (Statton et al., 2018). Z. muelleri has been shown to have phenotypic plasticity in response to subtle changes in light conditions with negligible penalty (Manassa et al., 2017), whereas H. ovalis appears more sensitive to shading (Longstaff et al., 1999), and P. australis responses can vary based on genotypic diversity (Evans et al., 2017). These effects invite new considerations for both planning and conserving seagrasses and emphasise the need to factor in the cumulative effects of multiple primary and secondary stressors.

The implications of this study for the ongoing management of P. australis are clear - the physical disturbance of P. australis should be avoided, and non-contact designs such as EFMs should be used where seagrasses meadows exist to avoid creating clearings that are likely to expand. The density of moorings has a significant negative effect on P. australis and should not be overlooked as a source of long-term chronic decline. Our findings are likely to be conservative given that the study site was heavily impacted by a dense installation of traditional moorings and a legacy of disturbance associated with urbanisation. A hypothetical ‘pristine’ site is likely to benefit more from EFM designs than our study site by maintaining meadow contiguity, assuming moorings are necessary and traditional designs are the alternative option. The study site was chosen specifically as one of two remaining P. australis meadows in Sydney Harbour, from which we hoped to gain new insights into this species’ management (Johnston et al., 2015).

Variations in the effectiveness of non-contact mooring designs in reducing impacts on seagrasses found elsewhere indicate that further investigation and ongoing maintenance are needed (Hastings et al., 1995; Demers et al., 2013). Despite EFM density being positively associated with P. australis, the maximum density of EFMs we could assess was much less than that of traditional moorings and as such, it should be assumed that high densities of EFMs could also introduce smaller, but still detrimental effects for seagrasses. Further investigation is needed to allow us to generalise these patterns for other seagrasses, meadows, ‘non-contact’ mooring designs and impact sources, and to better understand the non-linear relationship between seagrass and mooring density. In scour areas, the lack of remnant seagrass is likely to make natural recovery slow, and active restoration, such as using beach-cast fragments, offers hope for recolonizing scour areas Ferretto et al., 2021).

There are many situations where environmental decision-making can be improved using data and simulation, in this case by assessing the effect of complex spatial arrangements of interacting stressors. Given the forecasted synergistic effects of climate change and urbanisation (Hughes et al., 2003; Mayer-Pinto et al., 2015), there is an urgent need to improve our understanding of factors driving coastal habitat loss and to implement effective mitigation. Most species continue to diminish after being listed as threatened, and whilst conservation efforts can reduce the loss of individual species, extinction has accelerated into the 21st century (Barnosky et al., 2011). An important aspect to combat this is the monitoring of fine-scale effects to avoid landscape-scale impacts (e.g. Romero et al., 2015), but many species/stressor combinations are yet to be investigated (Dafforn et al., 2016). Other sources of secondary impacts in marine systems (e.g. light or noise pollution) may have similar cumulative effects on species or communities (Bolton et al., 2017). In urbanised marine environments, multiple overlapping sources of pollutants or sediment input could have a similar spatially driven effect, along with other areas where human-made structures modify the marine environment (e.g. pylons, seawalls, and aquaculture installations) (Dafforn et al., 2015). When applied to the stabilisation and recovery of threatened species, simulations might also consider the use of rehabilitation and restoration techniques, especially where (as in the case of P. australis) the species has limited ability to recover naturally.