Ecosystem engineers and functional dominance: when species loss matters for ecosystem function

Understanding when and why species loss affects community function in natural communities remains an open question in ecology. Recent cross‐system analyses have shown that functional dominance is associated with whether species loss matters: in communities dominated by few high‐performing species, lost species contribute little to function. But the question is what generates functional dominance in the first place. Here we examine whether an ecosystem engineer can generate it. Along a natural density gradient of the intertidal mussel Mytilaster minimus (202–2317 ind. m −2 , 54 samples, 93 associated taxa) in the Mediterranean Sea, we applied the extended Price equation to decompose between‐site variation in community abundance into five additive components. Context dependence (CDE) – the inflation of shared species – is the largest component of the change in community function along the density gradient, accounting for ~46% of total absolute variation across Price components. Species loss is non‐random: rare, low‐functioning species are preferentially lost at low engineer density. Functional dominance increases with engineer density and the importance of species loss is lower where dominance is higher, replicating the cross‐system dominance pattern within a single engineer‐structured system. Each Price component correlates with the quantity and quality of organic matter trapped by the mussel matrix: the CDE tracks trapped organic matter (ρ = −0.45), non‐random species identity tracks total biopolymeric carbon (ρ = +0.44) and species loss tracks microalgal biomass (partial r = −0.52 controlling for density). The ecosystem engineer thus generates a resource‐mediated pathway: density drives trapping, trapping feeds the dominant species, dominance buffers function against species loss. This mechanism is consistent with species richness per se being a poor predictor of function in engineer‐structured systems and points to engineer density as a highly informative conservation parameter.