Diving deep to escape the heat: the importance of mesophotic reefs as thermal refuges for New Zealand coastal fishes

The oceans are warming, and marine heatwaves are increasing in frequency, extremity, and duration due to anthropogenic climate change. As most fish are ectotherms, temperatures above their optimum have a host of physiological and demographic impacts, which result in a net negative effect on population productivity. However, temperatures decline with depth, which means mesophotic ecosystems, found in the ‘twilight zone’ between approximately 30 and 150 m depth, have the potential to act as thermal refuges. A growing number of studies show fish species not only shift latitudinally poleward but can also deepen their distributions to track their preferred thermal niche, and there is evidence of reef fishes moving from euphotic to adjacent mesophotic reefs during periods of elevated temperatures.

In Chapter 2 of this thesis, I analyse temperature differences and the extent to which marine heatwaves are buffered at mesophotic depths at Tawhiti Rahi (the Poor Knights Islands) and Te Mimi o Tū Te Rakiwhānoa (the Fiordland marine area) in Aotearoa, New Zealand. During summer, mesophotic temperatures in both regions are habitually cooler. Marine heatwaves relative to 10 m depth are buffered from 30 m and below at the Poor Knights, while in Fiordland they are buffered from 70 m and below; Fiordland’s hydrological conditions mean the upper limit of the mesophotic is likely much shallower than it typically occurs. Greater depths therefore offer a thermal refuge with respect to temperature. However, while pelagic fishes have flexibility in deepening their distributions, reef fishes are dependent on benthic habitats for structural complexity and food. In temperate ecosystems, this means mesophotic rocky reefs may be of particular value. Many coastal reef fishes feed on lower trophic level organisms such as macroinvertebrates and so in Chapter 3, I analysed variation in the biomass and community composition of mobile benthic macroinvertebrates (“fish food”) between euphotic (10 m) and mesophotic (50 m) reefs. Overall biomass was similar between depths both at the Poor Knights and in Fiordland, indicating there is likely to be sufficient macroinvertebrate prey for reef fishes on mesophotic reefs. Community composition varied by depth at the Poor Knights, however, with taxonomic diversity higher on mesophotic reefs. In Fiordland, community composition and taxonomic diversity varied by site but not by depth. When considering how some of the most common macroinvertebrate prey taxa contributed to community composition, I found that at the Poor Knights, the biomasses of amphipods, copepods, isopods, mysids, polychaetes and tanaids did not vary by depth, and ostracod biomass was higher on mesophotic reefs. In Fiordland, amphipod, copepod, isopod and polychaete biomasses did not vary with depth, but mysid and ostracod biomasses were sometimes lower in the mesophotic, and tanaid biomass was consistently lower. In Chapter 4, I built a temperature-dependent multispecies size-spectrum ecosystem model for the Poor Knights and used it to simulate long-term warming and short-term marine heatwave scenarios in the presence and absence of a mesophotic thermal refuge. In long-term warming scenarios, the ability to move to mesophotic reefs mitigated most fisheries species’ biomass reductions by an average of 17% and productivity reductions by 15%. Kingfish, as well as some non-fisheries functional guilds of herbivores, invertivores, piscivores and planktivores, had biomass and productivity increases with warming, likely due to predator and/or competitive release. In heatwave scenarios, almost all heatwave strengths resulted in biomass and productivity reductions for almost all modelled species, and reductions were reversed, negated or lessened when fish were allowed to move to a mesophotic thermal refuge. The thermal refuge effect reversed fisheries’ species average biomass losses during the moderate heatwave (a 28% average difference in biomass compared to when they were forced to remain on the euphotic reef), negated biomass losses during the strong heatwave (24% difference), and mitigated biomass losses by 21% and 20% during the severe and extreme heatwaves. Under moderate heatwave conditions, productivity was similar to what it was during baseline euphotic temperatures, but under strong, severe and extreme heatwave conditions, there were progressively larger reductions. When allowed to escape to a mesophotic thermal refuge, fisheries species’ average productivity losses were almost negated (5% difference) under strong heatwave conditions, negated (17% difference) under severe conditions, and mitigated by 19% under extreme conditions.

Overall, the findings of this thesis suggest that in ecosystems where the upper bound of the mesophotic zone is around the ‘typical’ 30 m, mesophotic reefs are buffered from marine heatwaves, have similar or greater availability of benthic macroinvertebrate prey for coastal fishes, and, if used as a thermal refuge, help to mitigate against – and occasionally reverse or negate – reductions in biomass and productivity for important New Zealand fisheries species. This thermal refuge effect has significant conservation and management implications, warranting the targeted protection of mesophotic reefs where they occur.