A physics-forced spatial Ecopath with Ecosim model reproduces spring bloom timing in the Strait of Georgia
Abstract
Marine ecosystem models rarely simulate lower trophic level (LTL) production from physical conditions; production is usually prescribed or supplied by a separate biogeochemical model. We tested whether Ecopath with Ecosim (EwE) can reproduce physically controlled LTL dynamics without a dedicated biogeochemical model. The Strait of Georgia Ecosystem Model of Lower Trophic Levels (SOGEM-LTL) represents 22 pelagic functional groups in the Strait of Georgia, Canada, with simplified nutrient dynamics, and is forced offline by a physical-oceanographic hindcast and an atmospheric reanalysis. We ran it from 1980 to 2018 at a three-day timestep and evaluated it against observations of spring bloom timing, phytoplankton, dissolved inorganic nitrogen and zooplankton. Interannual variation in bloom timing emerged from the physical forcing: the spatial (Ecospace) configuration matched satellite-derived bloom timing (r = 0.77, mean absolute error 10.5 days) and its early, average or late category in 13 of 14 years, more closely than a non-spatial (Ecosim) configuration of the same food web (10 of 14). The model also reproduced the seasonal cycles of nitrogen drawdown and recovery and of crustacean zooplankton biomass. Simulated variability was generally lower than observed, and skill was higher for seasonal and phenological indicators than for point-to-cell biomass comparisons. SOGEM-LTL sits between one-dimensional bloom models and three-dimensional biogeochemical models, adding trophic resolution at low computational cost and provides a basis for end-to-end models that link ocean physics to fish through their prey.
Related
- Year-round spatial segregation and fisheries overlap of white-chinned petrels from multiple populations at South Georgia
- Fibrosis burden and spatial pattern affect left atrial appendage stasis in multi-beat multi-physics simulations
- Truecell reproduces R Seurat's single-cell analysis outputs natively in the Python ecosystem
- PhySeqNet: physics-informed fluorescent foci detection for BARseq spatial transcriptomics
- Arousal-driven critical roaming reproduces human functional connectivity dynamics
BibTeX
@misc{oldford2026physicsforced,
title = {{A physics-forced spatial Ecopath with Ecosim model reproduces spring bloom timing in the Strait of Georgia}},
author = {Greig Oldford and Villy Christensen and Brian P.V. Hunt and Carl Walters},
howpublished = {EcoEvoRxiv},
year = {2026},
month = nov,
doi = {10.32942/x2kh5z},
url = {https://ecoevorxiv.org/repository/view/15955/},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - A physics-forced spatial Ecopath with Ecosim model reproduces spring bloom timing in the Strait of Georgia
AU - Greig Oldford
AU - Villy Christensen
AU - Brian P.V. Hunt
AU - Carl Walters
PY - 2026
DA - 2026/11/06
PB - EcoEvoRxiv
DO - 10.32942/x2kh5z
UR - https://ecoevorxiv.org/repository/view/15955/
AB - Marine ecosystem models rarely simulate lower trophic level (LTL) production from physical conditions; production is usually prescribed or supplied by a separate biogeochemical model. We tested whether Ecopath with Ecosim (EwE) can reproduce physically controlled LTL dynamics without a dedicated biogeochemical model. The Strait of Georgia Ecosystem Model of Lower Trophic Levels (SOGEM-LTL) represents 22 pelagic functional groups in the Strait of Georgia, Canada, with simplified nutrient dynamics, and is forced offline by a physical-oceanographic hindcast and an atmospheric reanalysis. We ran it from 1980 to 2018 at a three-day timestep and evaluated it against observations of spring bloom timing, phytoplankton, dissolved inorganic nitrogen and zooplankton. Interannual variation in bloom timing emerged from the physical forcing: the spatial (Ecospace) configuration matched satellite-derived bloom timing (r = 0.77, mean absolute error 10.5 days) and its early, average or late category in 13 of 14 years, more closely than a non-spatial (Ecosim) configuration of the same food web (10 of 14). The model also reproduced the seasonal cycles of nitrogen drawdown and recovery and of crustacean zooplankton biomass. Simulated variability was generally lower than observed, and skill was higher for seasonal and phenological indicators than for point-to-cell biomass comparisons. SOGEM-LTL sits between one-dimensional bloom models and three-dimensional biogeochemical models, adding trophic resolution at low computational cost and provides a basis for end-to-end models that link ocean physics to fish through their prey.
ER -