Restoration and non-invasive monitoring of geogenic reefs in temperate waters

Coastal marine habitats are among the most productive ecosystems on the planet that fulfil important nursery functions for marine organisms and offer numerous benefits to society such as food production, coastal protection and carbon sequestration. Yet, with an estimated 40% of the global population living within 100 km’s of the coast, these habitats also face enormous pressure from anthropogenic factors including pollution, urban development and agricultural runoff. The Baltic Sea provides a pertinent example of how a long history of intense human pressures can result in one of the most degraded coastal systems across the globe. High levels of nutrient runoff from land have turned the Baltic Sea into the largest man-made hypoxic area in the world, while systemic overfishing and habitat degradation have triggered a collapse in commercially-important Baltic cod. Geogenic (i.e. rocky) reefs constitute an important type of coastal habitat in the Baltic Sea, utilized by a wide variety of species for foraging, shelter and reproduction. However, century-long extraction of geogenic reefs to serve as construction material has degraded or completely removed large reef areas in the Baltic Sea with detrimental effects to associated flora and fauna. Whereas degraded biogenic habitats can often be recovered by excluding the pressure from the system (e.g. through protected areas), the removal of geogenic material from the seabed constitutes a permanent shift in habitat availability for reef-associated taxa that warrants active intervention for the habitat and its functions to be restored. Such restoration efforts rely heavily on empirical evidence to identify optimal restoration strategies given limited time and resource availability, which highlights the importance of robust sampling designs and consistent, long-term monitoring studies. The aim of this thesis is to advance the relatively young scientific discipline of marine habitat restoration by focusing on two independent restoration projects in the Western Baltic Sea (WBS) and by exploring the use of various non-invasive monitoring techniques to document marine communities inhabiting vulnerable habitats. Restoration efforts can potentially benefit from theories and concepts emerging from fields such as conservation biology and reserve design, yet evidence of synergies with restoration ecology are scarce. Following a general introduction and a synopsis of the main research papers in Chapters 1 and 2, respectively, Chapter 3 investigates if a long-standing debate in conservation biology, the dilemma whether single large or several small (SLOSS) habitats promote biodiversity, has important implications for restoration ecology. Accordingly, boulder reefs were restored in a ‘single large’ (SL) and ‘several small’ (SS) design and monitored using remote underwater video systems (RUVS) in a before-after control-impact (BACI) sampling design. Results revealed that both reef designs strongly promoted the abundance of gadoids (mainly cod Gadus morhua and saithe Pollachius virens), yet smallbodied mesopredators were largely associated with the SS reefs. These findings are discussed in detail regarding the potential of reef restoration to contribute to re-establishing top-down control by predatory fishes and the wider implications of the documented SLOSS effects for restoration efforts in general. A widely used extension of RUVS monitoring involves the inclusion of bait (BRUVS) that attracts marine organisms to the camera’s field of view to enhance the efficiency of sampling efforts, by reducing potential zero inflation (i.e. samples void of detectable organisms) and increasing the power of statistical inference. The BRUVS method has raised concerns regarding sampling biases, including a potential risk of obscuring species-habitat associations (SHAs) by attracting organisms away from their naturally preferred habitat. Chapter 4 investigates this concern through a comparative assessment of SHAs for seven common Baltic species as recorded by RUVS and BRUVS. We demonstrate that BRUVS are capable of capturing SHAs previously documented in the literature and even show some superiority relative to RUVS by detecting established associations that were absent when using RUVS, potentially
due to limited statistical power obtained from RUVS data. These results confirm that BRUVS constitute a suitable non-invasive method to document SHAs while producing relatively high species counts, and hence we continue to employ this sampling method in Chapters 5 and 6. Chapter 5 reports on one of the few efforts carried out to date to restore cobble (rocks with diameter between ~ 6 and 25 cm) reef habitats. While the extraction of marine boulders from Danish waters was banned in 2010, the removal of smaller cobble-sized rocks continues in dedicated areas. We show that cobble reef restoration has the potential to ameliorate local declines in biodiversity and that variations in cobble coverage on the seabed drive a large number of significant correlations among species in the study area. These results provide evidence in favor of stepping up conservation efforts of cobble habitats in the Baltic Sea, as the ongoing removal and degradation of this habitat type continue to put associated reef taxa at risk. Finally, motivated by the relatively time-demanding analysis of video data associated with RUVS monitoring, Chapter 6 explores the potential of emerging environmental DNA (eDNA) techniques to monitor marine taxa and their association with coastal habitats. eDNA sampling offers a rapid and low-cost method to sample target species or entire communities with a high degree of sensitivity, as organisms are detected from DNA traces they release into the surrounding environment that can be collected from seawater samples with little effort. We assess the potential to infer SHAs from eDNA data for four common coastal species through a direct comparison with subsequently deployed BRUVS and show that eDNA sampling is capable of detecting SHAs for sedentary species on a fine spatial scale. Given the mosaic seabed characteristics of the study area, our findings suggests that eDNA traces of these sedentary species were highly localized, yet the lack of strong correlations between eDNA signals and fish counts from BRUVS concurrently highlights the ongoing challenge of inferring relative species abundance from eDNA data. Collectively, results from this thesis work demonstrate that geogenic reef restoration can benefit marine biodiversity and promote the abundance of commercially important species such as cod and other gadoids. The BACI sampling design and Bayesian inference play a central role throughout the thesis and we show how the two components can be effectively combined to disentangle restoration effects from natural system fluctuations while offering an intuitive probabilistic interpretation of results. We argue that eDNA sampling represents a powerful supplement to (B)RUVS monitoring studies by providing high sensitivity for the detection of target species and recommend in-tandem use of these methods to more effectively guide marine conservation and restoration efforts.