Pacific herring (Clupea pallasii) is an ecologically and commercially valuable forage fish in the North Pacific Ocean. However, knowledge gaps exist around the abiotic and biotic drivers behind its variable population dynamics–as well as on the ability of the species to show spatially structured trends that stabilize population portfolios in the face of environmental change. Here we examined how historical hydroclimatic variability in the San Francisco Estuary (California) has driven age-0 Pacific herring population dynamics over 35 years. First, we used wavelet analyses to examine spatio-temporal variation and synchrony in the environment, focusing on two key variables: salinity and temperature. Next, we fitted Multivariate Autoregressive State-Space models to environmental and abundance time series to test for spatial structure and to parse out abiotic (salinity and temperature) from biotic influences (spawning and density dependence). Finally, we examined the stabilizing effects of spatially asynchronous population fluctuations (i.e., portfolio effects) across the estuary. Our results showed that temperature, but not salinity, fluctuated synchronously across regions on seasonal and decadal timescales. The top-ranked model showed strong evidence of regional population structure and regional variation in population responses to the environment. As expected, age-0 herring were generally associated with cooler, saltier conditions in spring. Density dependence was strong in all regions, suggesting that local factors influencing rearing conditions limited juvenile population growth across the estuary. Notably, age-0 abundance fluctuations were on average 29% more stable across the estuary than in individual regions, demonstrating that portfolio effects arising from population asynchrony have been stabilizing recruitment across the estuary over the past four decades. We contend that ecosystem-based fishery management strategies to restore eelgrass and tidal-marsh-rearing habitats could increase the carrying capacity of the estuary, further stabilizing the herring population and reducing the risk of fishery closures.

These data were derived from the following resources available in the public domain: (1) Department of California Fish and Wildlife (CDFW) San Francisco Bay Study (URL: https://filelib.wildlife.ca.gov/Public/BayStudy/) and (2) Department of California Fish and Wildlife (CDFW) Pacific Herring Team (public data available upon request - submit request here: https://wildlife.ca.gov/Regions/Marine/Contact/Ask-Marine). These derived data were processed according to the following steps. For the age-0 catch-per-unit-effort (CPUE) data extracted from the San Francisco Bay Study midwater trawl from 1981–2015, stations where sampling yielded catch >0 in 50% of the time steps were retained, to ensure convergence of the MARSS models. This resulted in 30 stations (4 regions). For the spawning stock biomass (SSB) index, no filtering occurred because it represents a single estimate for the entire estuary. All values were log-transformed and z-score transformed prior to modeling. Variables used in analysis and provided in attached data files. Region codes: C, Central Bay; S, South Bay; SP, San Pablo Bay; SU, Suisun Bay. Variable Survey name Gear type Spatial extent Period Pacific herring Mean CPUE: Age 0 CDFW SF Bay Study Midwater trawl S, C, SP, SU April to June Spawning stock biomass (SSB) index: Age 2+ CDFW Herring Team Egg deposition and spawner surveys Estuary (S, C, SP combined) October to April Environmental Mean salinity (PSU) CDFW SF Bay Study Water quality sonde: mean of surface and bottom profiles S, C, SP, SU October to June Mean temperature (°C) CDFW SF Bay Study Water quality sonde: mean of surface and bottom profiles S, C, SP, SU October to June

Please see the README document ("README_Dataset-PacificHerringSFBay_v0.1.0.txt") and the accompanying article: Pak, N., D. Colombano, T. Greiner, J. Hobbs, S. Carlson, and A. Ruhi. 2023. Disentangling abiotic and biotic controls of age-0 Pacific herring population stability across the San Francisco Estuary. Ecosphere. Accepted. R Markdown script is provided to reproduce model results based on the derived datasets: Multivariate autoregressive state space (MARSS) model script specifies model structures, fits, diagnostics, AICc comparisons, and coefficients.Funding provided by: California Sea Grant, University of California, San DiegoCrossref Funder Registry ID: http://dx.doi.org/10.13039/100005522Award Number: 62407