henhaobio
Sign in

Nature CommunicationsSystems BiologyPeer-reviewed journal article

Marine heatwaves disrupt ecosystem structure and function via altered food webs and energy flux

Dylan G. E. GomesiD, James Joseph RuzickaiD, Lisa G. CrozieriD, David D. HuffiD, Richard D. BrodeuriD, Joshua David StewartiD

PDFDOIPreprintCodeData
In-depth full-text digest available · Jump to digest ↓

Abstract

The prevalence and intensity of marine heatwaves is increasing globally, disrupting local environmental conditions. The individual and population-level impacts of prolonged heatwaves on marine species have recently been demonstrated, yet whole-ecosystem consequences remain unexplored. We leveraged time series abundance data of 361 taxa, grouped into 86 functional groups, from six long term surveys, diet information from a new diet database, and previous modeling efforts, to build two food web networks using an extension of the popular Ecopath ecosystem modeling framework, Ecotran. We compare ecosystem models parameterized before and after the onset of recent marine heatwaves to evaluate the cascading effects on ecosystem structure and function in the Northeast Pacific Ocean. While the ecosystem-level contribution (prey) and demand (predators) of most functional groups changed following the heatwaves, gelatinous taxa experienced the largest transformations, underscored by the arrival of northward-expanding pyrosomes. We show altered trophic relationships and energy flux have potentially profound consequences for ecosystem structure and function, and raise concerns for populations of threatened and harvested species.

In-depth summary

Claude Opus 5.5 · xhigh effort · 8 Nov 2026 · from the full text of version 2 (1 Feb 2024) · CC0 1.0

Pre- versus post-heatwave Ecotran food-web models of the Northern California Current show that newly abundant pyrosomes caused the largest energy-flux disruption, diverting lower-trophic energy mostly into detritus.

Prolonged marine heatwaves are known to affect individual species and populations, but no study had estimated their whole-ecosystem effects within an affected system, so short- and long-term outcomes remained uncertain. This work traces energy flow from single functional groups to the full Northern California Current food web, which captures direct and indirect cascading heatwave effects in a productive upwelling ecosystem that supports economically important fisheries.

Key results

  1. ShownThe pelagic tunicate Pyrosoma atlanticum was nearly absent before the heatwaves. Its mass arrival was the single largest perturbation to ecosystem energy flux: it captured low- and mid-trophic energy that previously supported other groups.

    Difference network across 86 functional groups (361 taxa), comparing pre-MHW (1999-2012) and MHW (2014-2022) models. More than 98% of pyrosome biomass is routed to detritus pools. For comparability, the pre-MHW model was seeded with only a trace 0.00001 mt/km2 of pyrosomes. · Results and Discussion; Model adjustment details; Network analyses and metrics; Figure legends

  2. ShownLower trophic levels shifted more than upper ones. Krill, sardine, pteropods, pelagic amphipods, invertebrate larvae and small mesh-feeding jellies ate less phytoplankton, and larger jellies declined. Overall consumption of gelatinous prey fell even though gelatinous biomass rose.

    Footprint (consumption) and reach (contribution) differed significantly between models for many functional groups, across 1000 Monte Carlo parameterizations per model. Tests were two-tailed t-tests with Bonferroni correction; exact p-values are given only in the Source Data file. · Results and Discussion; Energetic consumption and contributions; Statistical analysis; Figure legends

  3. SuggestedWhole-network properties held steady despite the lower-trophic restructuring. Mean trophic level did not change, and connectance and link density were slightly higher in the heatwave model. The authors read this as a food web with relatively robust efficiency that is at least as robust to disturbance as before.

    Table 1 compares arithmetic and biomass-weighted mean trophic level, connectance (realized links out of 7396 possible in an 86 x 86 network) and link density using two-tailed t-tests, with 100 bootstrapped networks for connectance and density. Links carried over from the pre-MHW model without new diet support were removed. The text gives no metric values. · Results and Discussion; Network analyses and metrics; Statistical analysis; Tables

  4. SuggestedThe heatwave-era food web points to fishery consequences. Chinook salmon and cod appear to have declined. Pacific jack mackerel surged and now presses harder on lower trophic levels, yet U.S. jack mackerel landings are unchanged, hinting at an under-used resource.

    PacFIN data show Chinook salmon commercial harvest in the Northern California Current fell by nearly a factor of three (a 67% decline) since heatwave onset. Jack mackerel abundance and its impact on lower trophic levels increased after the MHW, while PacFIN jack mackerel landings showed no change. · Results and Discussion

  5. SuggestedSimple temperature scaling suggests 2 °C of warming would speed biomass turnover and raise metabolic costs, which lowers trophic transfer efficiency. Low-efficiency groups such as jack and Pacific mackerels would then need to eat more to persist.

    A hypothetical 2 °C rise gives a P/B scaling factor of 1.037 (3.7% faster turnover), using the Maureaud et al. (2017) relation with von Bertalanffy K held constant. A Q10 of two gives a metabolic scaling factor of 1.149 (14.9% increase). These are hypothetical calculations; the compared static models do not include temperature as a driver. · Results and Discussion; Estimates of biomass flow and trophic transfer…

Methods, data and reproducibility

Two static, steady-state, mass-balanced food-web models were built in Ecotran (an Ecopath extension), one per period (pre-MHW 1999-2012, MHW 2014-2022). They were compared through difference networks, network metrics and footprint/reach energy-flux indices, with uncertainty propagated by 1000 Monte Carlo draws of the production matrix.

  • Long-term survey time series from six surveys covering 361 taxa, aggregated into 86 functional groups for the Northern California Current · multiple marine taxa (Northeast Pacific) · n = 361 taxa; 86 functional groups · 10.5281/zenodo.8121889
  • Diet (trophic interaction) database used to build diet matrices and set diet uncertainty (CV 0.1, 0.5, 0.8 for the three example groups) · Chinook salmon, Pacific herring, Pyrosoma atlanticum, others · n = 39083 juvenile Chinook, 2911 herring, seven pyrosome colonies
  • Euphausiid biomass densities from the binational U.S.-Canada acoustic trawl survey that combines Pacific hake assessment with integrated ecosystem sampling · euphausiids (krill) · n = 2007, 2009, 2011, 2012 (pre-MHW); 2015, 2017, 2019 (MHW)
  • Newport Hydrographic Line (NHL) survey data used to set invertebrate egg biomass · marine invertebrates · n = NHL 2000-2012 (pre-MHW); 2014-2020 (MHW)
  • Met Office Hadley Centre HadSST monthly North Pacific sea surface temperature anomalies, 1950-2022
  • PacFIN commercial fishery landings used for fleet data and for salmon and jack mackerel harvest trends
  • Prior Northern California Current food-web models (Ruzicka et al. 2012 pre-MHW; Gomes et al. 2022 MHW), adapted and harmonized for this comparison · multiple marine taxa · n = 80 living groups, 3 nutrient, 5 detritus pools, 2 fisheries
Methods & tools
Ecotran (EcoTran end-to-end ecosystem model platform), Ecopath mass-balance food web framework, Ecopath pedigree-based uncertainty (CV) assignment, Monte Carlo production-matrix sampling (1000 models), Footprint and reach energy-flux metrics, Two-tailed t-tests with Bonferroni correction, Bootstrap node resampling (100 networks), Q10 metabolic scaling and Maureaud et al. (2017) P/B-temperature relation, MATLAB R2021a, R v 4.2.2, ggplot2, qgraph
Software & environment
MATLAB R2021a and R v 4.2.2; R packages ggplot2 and qgraph (package versions not stated).
Reproducibility gaps
Some model inputs are described as unpublished data, with details deferred to the prior models (Ruzicka et al. 2012; Gomes et al. 2022). Exact p-values for the footprint/reach tests are only in the Source Data file.
Limitations (6)
  • The analysis compares two static, steady-state model snapshots. It leaves out abiotic drivers such as temperature and mechanistic links to physiological rates, which the authors defer to future dynamic modeling. · stated by the authors
  • It remains untested whether shifts in the abundance of species such as Chinook salmon, cod and jack mackerel are directly caused by the marine heatwaves; the authors say further work is needed. · stated by the authors
  • The trophic role of pyrosomes is highly uncertain. Predators' energetic gain from eating them is unclear, and their diet rests on only seven colonies (diet CV set to 0.8). · stated by the authors
  • How the ecosystem will respond to further heatwaves, or to a temporary return to pre-heatwave conditions, is unresolved. · stated by the authors
  • Significance for footprint/reach comes from t-tests on 1000 simulated Monte Carlo draws per model. With simulated replicates, p-values depend on the number of draws chosen rather than on independent observations, so the asterisks may overstate how meaningful the changes are. · noted in this summary
  • Making the models comparable required aggregating groups, merging fleets into a single commercial fleet, rebalancing pre-MHW diets and seeding a trace of pyrosome biomass. The text does not report how sensitive the footprint/reach results are to these choices. · noted in this summary

Read this if you study marine heatwave impacts, Ecopath/Ecotran food-web modeling, gelatinous zooplankton such as pyrosomes, or ecosystem-based fisheries management in the California Current.

Citations
95
FWCI
16.73
Field percentile
top 1%

Reproducibility ledger

Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.

Sign in to file a reproducibility report.

Related papers

Similarity Map
SelectedSame fieldCross-field
THISGarzke '17Gomes '24Arafeh‐Dalm '25Vad '22Gold '24Barbosa '22
73% semantic matchEcology
Jessica Garzke, Stephanie J. Connor, Ulrich Sommer+1 more·bioRxiv·Dec 14, 2017