Predicting larval alewife transport in Lake Michigan using hydrodynamic and Lagrangian particle dispersion models

Abstract

Several species of fish in large lakes and marine environments have a pelagic larval stage, and are subject to variable transport that can ultimately regulate survival and recruitment success. Alewife, Alosa pseudoharengus, are subject to transport by complex coastal currents during their pelagic larval stage (~30 d). We assessed backward-trajectory simulations, consisting of a Lagrangian particle dispersion model linked to the Finite Volume Community Ocean Model, to estimate likely hatch locations of aged larval alewife collected from locations on both the eastern and western sides of Lake Michigan during July 2015. We used four deployments of three satellite-tracked drifter buoys in coastal waters to assess model skill in estimating the origin of a drifter from its final location. We found that the trajectories of drifters varied greatly, depending on wind events and associated coastal transport processes, including upwelling/downwelling and coastal jet currents. In 2 of 12 cases, the backward trajectory simulations failed to predict the drifter origin, associated with transport of 170 km in a narrow coastal jet current. In the remaining 10 cases, the known drifter origin was within 3.5 km of the spatial patch of predicted possible origins for a scenario of horizontal diffusivity (188 m 2 s À1 ) consistent with the offshore model grid resolution. Modeled backward trajectories estimated that alewife originated from the same side of the lake where they were collected, within ~100 km of the collection site. Our paper demonstrates the utility of hydrodynamic models to estimate a region of origin for aged larval fish.

A Larval “Recruitment Kernel” to Predict Hatching Locations and Quantify Recruitment Patterns

“…Particle‐boundary interaction : How particles behave at a solid boundary is rather complex, even in forward tracking models (Chaput et al., 2023), let alone backward tracking models. Some simulations moved particles back to previous positions when they intersected domain boundaries (Chaput et al., 2023; Rowe et al., 2022); however, this does not necessarily reproduce the reverse trajectories. In our simulations, particles encountering a solid boundary were removed from the domain.…”

“…Measurements of daily otolith growth increments were used to estimate larval age. This approach has been done for alewife Alosa pseudoharengus (Eppehimer et al., 2019; Höök et al., 2007; Rowe et al., 2022), bluefin tuna Thunnus thynnus (Hernández et al., 2022), sand lance Ammodytes dubius (Suca et al., 2022), and Lake Whitefish Coregonus clupeaformis (Dunlop et al., 2023). From the Lake Erie larval samples, 240 and 361 Lake Whitefish larvae were selected for otolith analyses in 2018 and 2019, respectively (Table S2 in Supporting Information ).…”

“…We assume that the larvae act as passive particles during their PLD and, thus, were only subjected to hydrodynamic forcing (advection and diffusion). We ignore their diel vertical migration and active swimming behavior (Di Stefano et al., 2023; Rowe et al., 2022; Suca et al., 2022). The newly hatched Lake Whitefish larvae have poorly developed fins and little swimming ability (Mitz et al., 2019) until fin development near 17 mm length (Hoagman, 1974; Taylor & Freeberg, 1984), when they can swim at ∼2 cm/s (Hart, 1931).…”

“…By tracking the trajectories of successful recruits, larval mortality and transport to unsuitable nursery grounds are implicitly accounted for. The locations of larval particles, at the end of simulation, are then identified as potential spawning/hatching locations (Bauer et al., 2014; Christensen et al., 2007; Compaire et al., 2021; Di Stefano et al., 2023; Gargano et al., 2022; Hernández et al., 2022; Rowe et al., 2022; Suca et al., 2022; Thygesen, 2011; Torrado et al., 2021). However, in this approach there is still unreal recruitment, because the estimated particle origins may not all be suitable hatching sites (e.g., model uncertainty may cause larvae to be backtracked to soft substrates).…”

Modeling Blue Crab (Callinectes sapidus) Larval Transport and Recruitment Dynamics in a Shallow Lagoon‐Inlet‐Coastal Ocean System

“…, where x → and v → ( x → ,t) are the particle position and its vector velocity at time t, is solved using an explicit fourth order Runge-Kutta multi-step method (Chen et al, 2013). The FVCOM-based particletracking model has been successfully applied in fisheries and coastal environmental sciences (Liu et al, 2023;Nguyen et al, 2024;Premathilake & Khangaonkar, 2019;Rowe et al, 2022;Tian et al, 2009;Xia et al, 2011).…”