Fig 1.
Map of Northern California Current.
Extent of end-to-end ecosystem model in the Northern California Current marine ecosystem. Shaded gray bins indicate the 15 ecosystem model subregions. The model domain is broken down into three bathymetric bins (inner shelf: 1–100 m; mid shelf: 101–200 m; and outer shelf: 201–1280 m) and five latitudinal bins (northern California: 40.8–42°N; southern Oregon: 42–44.4°N; northern Oregon: 44.4–46°N; Columbia River zone: 46–46.7°N; and Washington coast: 46.7–48.34°N). State outline data comes from US Department of Commerce, Census Bureau, Cartographic Boundary Files.
Fig 2.
The EcoTran ecosystem model allows direct linking of physical oceanographic forcings to the food web model, which drives primary production and the transport of plankton, detritus, and nutrients across model domain boundaries. The currency of a time-dynamic EcoTran model (see Figs 9 and 10 for examples) is nitrogen input to the system as nitrate and ammonium at the base of the food web via upwelling and detritus remineralization by bacteria. The ecosystem model is driven by nutrient flux that is important for bottom-up food web processes using the coastal upwelling transport index (CUTI) [42]. The CUTI time series (in daily time steps) drives advection (red arrows) of nutrients across the shelf. Primary production is supported by nutrient uptake and, in turn, supports grazing and predation by higher trophic level groups and catch by fishing fleets. Adapted from Ruzicka et al., 2016 [33].
Table 1.
NCC EcoTran parameters.
Table 2.
Functional group definitions.
Fig 3.
Subregional heatmaps of functional group spatial distributions.
Each functional group was broken out into the 15 subregions via survey data, fisheries landings data, or species distribution models (see Methods), or distributional assumptions when there was a lack of available information. The color of each subregional cell is a gradient denoting the proportion of biomass (for each functional group) that is within each subregional cell (with red being the highest and pale yellow being the lowest proportions, respectively). The proportion of biomass in each subregion sums to 1 across all subregions. See “SubRegions/” in supplemental data and code to reproduce this plot, and for plots of all functional groups). Adult Chinook, common murre, and sooty shearwater distributions are based off of the juvenile salmon and ocean ecosystem survey (JSOES); hake distributions are from the hake acoustic trawl survey; herring, jack mackerel, and sardine distributions are from the coastal pelagic species (CPS) acoustic trawl survey; and Southern resident killer whale (SRKW) distributions are based off of movement data from satellite-tagged Southern resident killer whales [130]. State outline data comes from US Department of Commerce, Census Bureau, Cartographic Boundary Files.
Fig 4.
Biomass density and trophic level of model functional groups.
A) Ecosystem-wide biomass density input values (y-axis) for the aggregated ecosystem model (no subregion-specific values) are based on survey data / stock assessments (black points), were borrowed from a previous ecosystem model (Ruzicka et al. 2012; blue points), or are estimated by the model during mass-balancing (green points; see methods). The trophic level of each functional group is estimated by the model and is based on the diet matrix. Numbers indicate functional groups, identified in Table 1. B) Same as A, but on a log (base 10) axis scale. Red points indicate detritus groups, which are not used in estimation of the regression line (Link 2010). Purple point is aggregated seabird groups 71–76 to display how the choice to aggregate groups affects how far away from the line the point falls. Equation for slope is estimated only with black points and is defined as log(y) = 3.5–1.085 x.
Fig 5.
Diet matrix of ecosystem model by trophic level.
The diet matrix is visualized here as a weighted, directed graph. Numbered nodes are functional groups (see Table 1 for functional group numbers), whereas arrows indicate directed edges from predator group towards prey group. The shade of blue indicates strength of interaction (higher diet preference results in darker blue network edges) up to a value of 0.10, at which point network edges get thicker with higher values. This aesthetic choice was made to not overly clutter the diagram and to make visualization of strong interactions more apparent. The y-axis values are the estimated trophic level of each functional group, and the x-axis is value-less and only used to help visualize multiple groups.
Fig 6.
EcoTran trophic network of ecosystem model.
The EcoTran trophic network is visualized here as a weighted, directed graph. Numbered nodes are functional groups (see Table 1 for numbers), while arrows indicate directed edges (energy flows from producer groups towards consumer groups). The shade of green indicates strength of interaction (higher diet preference and prey biomass results in darker green network edges) up to a value of 0.10, at which point network edges get thicker with higher values, as in Fig 5. This graph includes detritus groups (86–90), which dominate the network.
Fig 7.
EcoTran trophic network of non-detritus groups by trophic level.
The EcoTran trophic network is visualized here as a weighted, directed graph with detritus groups (86–90) removed (see Fig 6). Numbered nodes are functional groups (see Table 1 for numbers); arrows indicate directed edges (energy flows from producer groups towards consumer groups). The shade of green indicates strength of interaction (higher diet preference and prey biomass results in darker green network edges) up to a value of 0.025, at which point network edges get thicker with higher values.
Fig 8.
Validation; biomass-specific production and consumption values by trophic level.
(A) The y-axis is the log-scaled production to biomass ratio (biomass-specific production) and the x-axis numbers indicate the different functional groups (see Table 1) sorted by increasing trophic level (far left is low trophic levels, e.g., phytoplankton, and far right is high trophic levels, e.g., marine mammals). (B) The y-axis is the log-scaled consumption to biomass ratio (biomass-specific consumption) and the x-axis is the same as in panel A. The y axes are on the log scale, so negative values indicate P/B and C/B values (for A and B, respectively) less than 1 on the original scale.
Fig 9.
Assessing stability: 150-year model simulation driven by average upwelling time series.
Each living consumer functional group is pictured here as a smoothed individual line. The proportion of abundance relative to starting values are plotted on the y-axis and the daily simulation timestep is plotted on the x-axis (axis in yearly units). The simulation here is driven by the average (day of year average for 1988–2021, see Methods) Coastal Upwelling Transport Index (CUTI) to reduce interannual variation for assessment of equilibrium and model stability. No functional groups are changing by more than 5% in the final 20 years of simulation (see S6 Table). That is, all groups have reached equilibrium (i.e., are within this 5% threshold). No functional groups go extinct over the 150-year simulation.
Fig 10.
Primary production time series as both ecosystem model output and vertically generalized production model.
Satellite imagery-derived estimates of daily primary productivity via a vertically generalized production model (VGPM, blue lines) [50, 51] are plotted against ecosystem model-derived estimates of primary production driven by an upwelling time series (black lines are aggregated large and small phytoplankton functional groups). The average values of the final eight years of the VGPM (2014–2021) are used to inform the starting conditions (values) for the ecosystem model. After this point, the ecosystem model is driven entirely by nutrient inputs to the system as determined by the coastal upwelling transport index (CUTI) [42], and any resemblance to the VGPM time series is an indication that the ecosystem model is capturing the appropriate dynamics in primary productivity.
Fig 11.
Invertebrate functional group time series as both ecosystem model output and independent estimates.
Independently derived estimates (blue points; blue lines = locally estimated scatterplot smoothing lines) of relative biomass via a Juvenile Salmon and Ocean Ecosystem Survey (JSOES; large jellies), fisheries landings (small cephalopod aggregate = market squid), and a pre-season abundance model (Dungeness crabs [83]) are plotted against ecosystem model-derived estimates of matching functional groups (black lines). The ecosystem model is driven entirely by nutrient inputs to the system as determined by the coastal upwelling transport index (CUTI) [42] and trophic relationships, and any resemblance of the two time series is an indication that the ecosystem model is matching independently-observed dynamics.
Fig 12.
Coastal pelagic fish functional group time series as both ecosystem model output and independent estimates.
Independently derived estimates (blue points; blue lines = locally estimated scatterplot smoothing lines) of relative biomass of sardine, anchovy, jack mackerel, and Pacific chub mackerel via stock assessments [56, 84, 86, 87] are plotted against ecosystem model-derived estimates of matching functional groups (black lines).
Fig 13.
Seabird and mammal functional group time series as both ecosystem model output and independent estimates.
Independently derived estimates (blue lines and points) of relative biomass via a Juvenile Salmon and Ocean Ecosystem Survey (JSOES; common murre and sooty shearwaters), a humpback whale mark-recapture study (baleen whales [88]), and counts of the well-monitored Southern resident killer whale population [https://www.whaleresearch.com/orca-population] are plotted against ecosystem model-derived estimates of matching functional groups (black lines).
Fig 14.
Simplified Chinook salmon centric EcoTran trophic network.
The EcoTran trophic network is visualized here as a weighted, directed graph with detritus groups (86–90) and those without direct energy flow to or from Chinook salmon groups removed. That is, this is a simplified version of Fig 7, which allows for a focused perspective on Chinook salmon. Numbered nodes are functional groups (see Table 1 for numbers); arrows indicate directed edges (energy flows from producer groups towards consumer groups). The color intensity and line thickness indicates strength of interaction. Higher diet preference and prey biomass results in darker network edges up to a value of 0.025, at which point network edges get thicker with higher values (direct salmon connections are in red, all other connections in grey).