Testing the deep reef refugia hypothesis in temperate mesophotic ecosystems using sponges as model organisms

Temperate mesophotic ecosystems (TMEs) occur in cool waters across the globe and occupy the depth band between the shallow euphotic zone and the aphotic deep sea. Situated at the lower limits of light availability for photosynthesis, these ecosystems are typically characterised by reduced environmental variability and a lower frequency and intensity of physical disturbances compared to shallower environments. TMEs are often dominated by slow-growing, long-lived, sessile invertebrates such as sponges, bryozoans and cnidarians, which contribute significantly to benthic complexity and biodiversity. Due to their relative isolation from surface-driven perturbations, TMEs have been hypothesised to function as climate refugia for species inhabiting adjacent shallow-water ecosystems. This concept, commonly referred to as the Deep Reef Refugia Hypothesis (DRRH), proposes that deeper habitats may be protected from environmental change and disturbances, such as storm impacts and marine heatwaves (MHWs), and serve as sources of larvae to support the persistence of shallow populations. For the DRRH to be possible several assumptions must hold true: environmental conditions must be less stressful or more stable at depth; there must be species that occur across both shallow and mesophotic zones (depth-generalists); and there must be effective biological connectivity, particularly through larval exchange, between these zones. While this framework has received increasing attention in tropical systems, its applicability in temperate regions remains poorly understood. As the impacts of climate change on marine systems, including the frequency and intensity of MHWs, continue to increase, it is important to understand whether communities inhabiting TMEs are likely to benefit from any refuge effect. The main aim of my thesis was to critically evaluate support for the DRRH in temperate mesophotic ecosystems, using sponges as model organisms due to their ecological dominance, physiological sensitivity, and the well-established microbial symbioses that underpin their ecological functions. To achieve this aim, I integrated long-term observational datasets, depth-structured community surveys, and laboratory-based experimental manipulations. Collectively, this approach provided insights into the extent to which TMEs can function as climate refugia, the factors constraining this potential, and the biological mechanisms underpinning sponge responses to thermal stress. In my first data chapter, I analysed a seven-year dataset of benthic community monitoring from four coastal and offshore sites in the Fiordland Marine Area (FMA) and the Poor Knights Marine Reserve (PKMR) in New Zealand (with field monitoring prior to 2022 conducted by a previous student, and from 2022 onwards by myself). Each site hosted benthic communities spanning a vertical depth gradient from shallow to mesophotic zones, with the rocky reef wall ending between 70 and 110 m depending on the site. Using SCUBA and remotely operated vehicle (ROV) surveys, I characterised patterns in benthic community composition and diversity through time and space. Temporal analyses revealed considerable site- and depth-specific variation in community dynamics. At two sites, mesophotic communities exhibited relatively low temporal variability compared to shallower assemblages, broadly supporting the prediction that deeper communities are more stable. However, at the remaining two sites, mesophotic communities displayed comparable or greater interannual fluctuations than shallow assemblages, challenging assumptions about the buffering capacity of depth. Across all sites, compositional change was generally stochastic, with limited evidence for directional succession or cyclic turnover. These findings suggest that environmental stability at depth does not necessarily confer ecological stability, and that local-scale factors such as hydrodynamics, localised disturbances, and biological interactions may exert a stronger influence on community trajectories than depth alone. In my second data chapter I explored patterns of vertical distribution and species turnover within sponge assemblages. To achieve this, I conducted a depth-gradient survey of sponge assemblages at three sites within the PKMR, known for high sponge diversity. Abundance data were collected for sponge taxa from 5 to 65 metres depth and analysed using multivariate ordination and diversity metrics. The resulting patterns revealed high beta diversity across the depth gradient, with clear vertical structuring of assemblages. The vast majority of sponge operational taxonomic units (OTUs) were confined to either the shallow or mesophotic zones, and only a small subset (18 of 64 OTUs) were consistently observed at multiple depths. Moreover, even among these depth-generalists, abundance varied significantly between depths, suggesting restricted distributions. These results indicate that while some sponge species occur across depth zones, the broader assemblage exhibits limited vertical overlap, constraining the potential for TMEs to function as refugia for most sponge species. In my third data chapter I assessed the physiological resilience of a depth-generalist sponge to MHW conditions. I selected Rowella lancifera, a calcareous sponge identified as a common depth-generalist in the FMA, for a controlled laboratory experiment simulating MHW conditions. Sponges were collected from shallow (10 m) and upper-mesophotic (30 m) depths and exposed to elevated temperatures mimicking and exceeding recent MHW temperatures from the FMA. I measured stress responses including respiration rate, buoyant weight change, and mortality, as well as differences in microbial community composition between control and MHW-treated sponges. In situ temperature monitoring revealed significant differences in temperature and variability between source depths, with upper-mesophotic sponges consistently experiencing lower mean and maximum temperatures throughout the preceding year. Despite this, sponges from both depths exhibited similar stress responses under simulated MHW conditions, with marked increases in respiration, reductions in buoyant weight, and increased mortality compared to controls. I also found evidence for shifts in microbial community composition, potentially due to reductions in thermally sensitive taxa. These findings suggest limited intraspecific variation in thermal tolerance between shallow and upper-mesophotic sponges despite differing thermal regimes. Notably, the experimental thermal stress scenario was only 1 °C above previous MHW maxima recorded at the study site, underscoring how close sponges in both depth zones may already be to their thermal tolerance thresholds. My fourth and final data chapter assessed the role of microbial symbionts in thermal stress resilience. To achieve this, I conducted an experiment investigating how disruptions to the sponge microbiome influence host responses to thermal stress. Using broad-spectrum antibiotics, I induced microbial dysbiosis in R. lancifera specimens and exposed both healthy and microbiome-disrupted individuals to sub-lethal MHW conditions. Microbial community changes were quantified via 16S rRNA gene amplicon sequencing, while physiological stress was assessed via respiration rate and qualitative health indicators. While dysbiosis alone did not induce visible stress, its interaction with thermal stress revealed synergistic effects. Dysbiotic sponges exhibited significantly higher respiration rates under MHW exposure compared to non-dysbiotic counterparts. Microbial sequencing data showed reductions in key nitrifying and aerobic taxa and increases in anaerobic microbial groups, suggesting a functional shift that may impair nutrient cycling and host resilience. These results demonstrate that the sponge microbiome plays a key role in facilitating resilience to environmental stress and that microbial instability can exacerbate the impacts of thermal anomalies. Taken together, the findings presented in my thesis contribute a nuanced and integrative perspective on the DRRH as it applies to sponge-dominated TMEs. While certain sites and species display traits supporting depth-mediated buffering and the potential for repopulation of shallow habitats, the overall evidence suggests this potential is highly variable and constrained by ecological structure and physiological limits. Importantly, limited vertical overlap in sponge species, coupled with the susceptibility of depth-generalists to heat stress, indicates that TMEs may not reliably serve as refugia under intensifying climate change. Furthermore, the strong influence of microbial integrity on sponge resilience highlights the need to consider host–microbiome interactions in ecological forecasting and conservation planning. In conclusion, my work underscores the importance of adopting a depth-integrated and mechanistic approach to understanding climate vulnerability in marine ecosystems. It highlights that the potential for deeper habitats to act as refugia is not universal and must be evaluated in a context-specific manner, accounting for community composition, physiological thresholds, and microbial dynamics. Conservation efforts have historically focussed on shallow coastal reefs, often overlooking deeper mesophotic ecosystems. My findings highlight the need for management strategies that incorporate depth-specific environmental monitoring and protection. Given the vulnerability and ecological importance of mesophotic communities, expanding marine protected areas to include deeper zones could enhance conservation outcomes, particularly where these habitats support sensitive species.