A multi-evidence causal investigation of water quality influences on Atlantic menhaden ( Brevoortia tyrannus ) habitat use and abundance in Narragansett Bay, USA

Abstract Atlantic menhaden (Brevoortia tyrannus) are migratory forage fish that act as key trophic linkages in the Atlantic shelf—connecting the energy from plankton to higher-level consumers. As a highly mobile fish, menhaden must balance the energetic costs of evading predation, foraging, and maintaining homeostasis when choosing suitable habitat. Prior knowledge indicates that poor water quality conditions, including hypoxia, force fish to adjust their movement behaviour in real time. Menhaden frequent the Narragansett Bay (NB), Rhode Island (USA) estuary, which has an established history of eutrophication, hypoxic events, and fish kills. In this work, we present a novel multi-evidence causal investigation that synthesizes and combines mechanistic expectations from the literature with formal statistics to build causal knowledge about the effect of water quality on menhaden movement ecology. Specifically, we leverage two statistical modelling frameworks: generalized additive models (GAM) and dynamic structural equation models (DSEM)—whose design was guided by our causal knowledge analysis—to evaluate in situ water quality influences on menhaden habitat use (presence) and abundance in discrete regions of NB from 2008 to 2021. GAMs allowed us to evaluate region-specific associations between water quality [chlorophyll a, dissolved oxygen (DO), salinity, pH] and menhaden habitat use/abundance by allowing for nonlinearity in the responses. By contrast, the DSEM framework assumes responses are linear but explicitly addresses time-series structure to capture temporal dependences. We found a positive effect of bottom DO and surface chlorophyll a on menhaden habitat use (presence), with evidence of reduced habitat use at DO levels above the 2.9 mg/L established hypoxia level, suggesting sub-hypoxic conditions may still result in habitat squeeze. Predictive power was limited for abundance models, suggesting that menhaden movement ecology is an inherently variable phenomenon influenced by complex population dynamics and distal oceanographic drivers. More broadly, this work establishes a multi-evidence causal framework that can be applied to further mechanistic understanding of the relationship between water quality and forage fish behaviour in estuarine ecosystems.