Peter A. Todd

The diversity‐habitat complexity relationship has been utilised widely in conservation and biodiversity enhancement interventions, but few studies have attempted to tease apart the components of complexity that drive this relationship. The ecological engineering of seawalls is one area where this topic has advanced, albeit not often at scales relevant to fish. We constructed habitat enhancement units (Simple, Complex and Freestyle ‘fish houses’) out of hollow concrete blocks and installed them at the base of tropical rip‐rap seawalls. Simple and Complex fish houses were cuboid, had the same surface area and volume and included 100 holes (microhabitats). The holes in Simple fish houses were all the same size, whereas 25 size variations were used in the Complex design (volume‐independent manipulation of a single complexity element). The Freestyle fish house was non‐cuboid and had more overall volume, microhabitat types and sizes. We examined the volume‐independent (Simple vs. Complex fish houses) and volume‐dependent (Freestyle fish house) effects of microhabitat complexity on fish taxonomic and functional assemblage metrics at two spatial scales and across diel cycles. We also investigated diurnal and nocturnal fish‐microhabitat size relationships. There was a modest, but not significant, volume‐independent effect of complexity on fish assemblages. The Freestyle design, however, supported significantly greater abundance, species richness, and distinct taxonomic and functional compositions. These results were dependent on spatial scales and diel cycles. Diel variation in fish activity patterns resulted in stronger size‐matching relationships between fish and microhabitats at night than during the day. Synthesis and applications . Our study shows that, to enhance fish diversity, it is important to provide three‐dimensional habitat architecture that incorporates a wide range of microhabitat sizes and types. Our findings also highlight some key considerations when assessing the performance of intervention designs, including spatial‐scale dependent effects of structural complexity, diel variation in fish‐microhabitat relationships, and choice of intervention assessment metric (i.e. taxonomic vs. functional diversity).