Complexity-functioning relationships differ across different environmental conditions.

Habitat complexity is widely considered an important determinant of biodiversity, and enhancing complexity can play a key role in restoring degraded habitats. However, the effects of habitat complexity on ecosystem functioning - as opposed to biodiversity and community structure - are relatively poorly understood for artificial habitats, which dominate many coastlines. With Greening of Grey Infrastructure (GGI) approaches, or eco-engineering, increasingly being applied around the globe, it is important to understand the effects that modifying habitat complexity has on both biodiversity and ecological functioning in these highly modified habitats. We assessed how manipulating physical (primary substrate) and/or biogenic habitat (bivalves) complexity on intertidal artificial substrata affected filtration rates, net and gross primary productivity (NPP and GPP, respectively) and community respiration (CR) - as well as abundance of filter feeders and macro-algae and habitat use by cryptobenthic fish across six locations in three continents. We manipulated both physical and biogenic complexity using 1) flat or ridged (2.5 cm or 5 cm) settlement tiles that were either 2) unseeded or seeded with oysters or mussels. Across all locations, increasing physical and biogenic complexity (5 cm seeded tiles) had a significant effect on most ecological functioning variables, increasing overall filtration rates and community respiration of the assemblages on tiles but decreasing productivity (both GPP and NPP) across all locations. There were no overall effects of increasing either type of habitat complexity on cryptobenthic fish MaxN, total time in frame or macro-algal cover. Within each location, there were marked differences in the effects of habitat complexity. In Hobart, we found higher filtration, filter feeder biomass and community respiration on 5 cm tiles compared to flat tiles. However, at this location, both macro-algae cover and GPP decreased with increasing physical complexity. Similarly in Dublin, filtration, filter feeder biomass and community respiration were higher on 5 cm tiles compared to less complex tiles. In Sydney, filtration and filter feeder biomass were higher on seeded than unseeded tiles, and fish MaxN was higher on 5 cm tiles compared to flat tiles. On unseeded tiles in Sydney, filter feeder biomass also increased with increasing physical complexity. Our findings suggest that GGI solutions via increased habitat complexity are likely to have trade-offs among potentially desired functions, such as productivity and filtration rates, and variable effects on cryptobenthic fish communities. Importantly, our results show that the effects of GGI practices can vary markedly according to the environmental context and therefore should not be blindly and uniformly applied across the globe.