Freshwater mussels (Bivalvia: Unionidae) are often described as ecosystem engineers, with the capacity to alter abiotic habitats; this can facilitate certain macroinvertebrate taxa, increasing biodiversity and potentially altering community composition. Mussel species are often implicitly considered to be broad ecological equivalents, with similar impacts on abiotic environments and benthic communities. This assumption of redundancy is important in conservation approaches that aim to preserve ecosystem functioning, rather than species identity. We tested this assumption of equivalence in two species of subtropical freshwater mussels, Lamellidens marginalis and Parreysia caerulea, studying the associations between these mussel species and macroinvertebrate communities. We measured mussel density and abundance of macroinvertebrates, identified to family level, at 50 sites within Dhanmondi Lake, a heavily polluted sub-tropical urban lake in Dhaka, Bangladesh. We investigated the associations of mussel density on different macroinvertebrate families, which we further classified into functional feeding groups (predators, scrapers, collectors and shredders). We found that densities of the two mussel species were negatively correlated in the 1-3-m nearshore zone; average density of Lamellidens was 46.4 & PLUSMN; SE 3.6 individuals/0.25 m(2), and average density of Parreysia was 6.4 & PLUSMN; SE 1.4 individuals/0.25 m(2). In total, 2319 Lamellidens and 320 Parreysia specimens were recorded. Sites containing higher densities of Lamellidens had greater similarity in macroinvertebrate community composition, while there was no effect for Parreysia. Particular macroinvertebrate families and functional feeding groups were also associated with mussels; the majority of significant associations were with Lamellidens but not Parreysia. The clearest associations were with predators and scrapers, which increased in abundance with Lamellidens density. This provides support for a possible role for mussels in structuring freshwater invertebrate communities and emphasises the importance of species identity in these effects, with one possible keystone species (Lamellidens) largely accounting for mussel-invertebrate associations. While microhabitat variables (shading and sediment type) were poor predictors of abundance in any invertebrate taxa (including mussels), manipulative studies are needed to disentangle the direct effects of mussels from the role of overlapping habitat preference between mussels and other invertebrates. Our results highlight the importance of species identity in patterns of community composition, particularly for ecosystem engineer species. We also reinforce the need to test the assumption of ecological redundancy within communities.
