The 2014-2016 Northeast Pacific Marine Heatwave (MHW) induced the warmest 3-year period on record in the California Current Ecosystem. We tested whether larval fish assemblage structure, phenology and diversity dynamics were comparable to past warming events from 1951-2013. First, we hypothesized, based on past observations of biological effect of warming, that mesopelagic species with southern distributions relative to southern California and Pacific sardine Sardinops sagax (a coastal pelagic species) would increase during the MHW while northern mesopelagics and northern anchovy Engraulis mordax (coastal pelagic) abundances would decline. Similar to past warming, southern mesopelagics increased and northern mesopelagics decreased. Unexpectedly, however, a common southern mesopelagic, Mexican lampfish Triphoturus mexicanus, was approximately three times more abundant than the previous annual high. Further, whereas sardine abundance did not increase, larval anchovy abundance rose to near-record highs in summer 2016. Second, we hypothesized that fishes would spawn earlier during the MHW. Fishes did not spawn in an earlier season within a year, but five of six southern mesopelagic taxa spawned earlier than typical within winter and spring. Third, we predicted that species richness would increase moderately due to an influx of southern and exodus of northern species. Richness, however, was very high in all seasons and the highest ever during the summer as multiple species with primarily southern distributions were recorded spawning for the first time in southern California. The richness of northern species was also unexpectedly high during the MHW. Northern species likely persisted in the study area because in addition to the warm water, pockets of cold water were consistently present. If, as predicted, conditions similar to the MHW become more common as oceans warm, this unique and largely unexpected combination of fishes may reflect future biological conditions.

CalCOFI has been collecting depth-integrated plankton samples quarterly from the same 66 stations since 1951 (Fig. 1) using obliquely towed nets (505- µm mesh, 333 µm mesh on the cod end). Net contents were preserved at sea in a sodium borate-buffered 5% formalin solution (Kramer, Kalin, Stevens, Thrailkill, & Zweifel, 1972; P. Smith, 1974; P. E. Smith & Richardson, 1977). All larval fishes were then sorted, identified and archived at the Ichthyoplankton Ecology Lab at the Southwest Fisheries Science Center in La Jolla, CA. Raw larval abundances were divided by the percent of the sample that was sorted (samples with very thick zooplankton were split using a Folsum Plankton Splitter) and multiplied by a standard haul factor that accounted for the depth of a tow and the filtered volume of water (Kramer et al., 1972). We ultimately expressed abundances as the number of larvae under a circle of water with an area of 10 m². Although CalCOFI now collects samples in winter, spring, summer and fall, historical collections were often sparser in fall than other seasons; we thus focused on winter, spring and summer. We excluded winter 2014 because the research vessel broke down during that cruise and most of the stations were not sampled.

Our capacity to identify larvae to species has improved since 1951, and we can currently identify almost all species based on morphology. The most notable exceptions are rockfishes (genus Sebastes) where we can morphologically distinguish only Aurora rockfihs S. aurora, Splitnose rockfish S. diploproa, Chilipepper rockfish S. goodei, Shortbelly rockfish S. jordani, Cowcod rockfish S. levis, Mexican rockfish S. macdonaldi and Bocaccio rockfish S. paucispinis; the remainder were grouped as rockfishes Sebastes spp. Genetic barcoding of rockfish larvae collected in 2005 identified 30 species within the Sebastes spp. group and found that shortbelly and squarespot S. hopkinsi were numerically dominant (Thompson, Hyde, Watson, Chen, & Guo, 2016). We are working back chronologically to identify archived specimens to current standards, and samples from 1962-present are at present resolved to current standards. Prior to 1962, we grouped several more taxa to genus or family: bristlemouths Gonostomatidae (primarily Showy Bristlemouth Cyclothone signata with fewer Benttooth Bristlemouth C. acclinidens and Diplophos spp.), Lightfishes Vinciguerria spp. (almost all Panama Lightfish Vinciguerria lucetia), Lampfish in the genus Nannobrachium (mostly Broadfin Lampfish N. ritteri with fewer Pinpoint Lampfish N. regalis), and Sanddabs Citharichthys spp. (mostly Pacific Sanddab C. sordidus and Speckled Sanddab C. stigmaeus at approximately equal abundances). In the analyses (see below), we used the full data set with fishes at coarser taxonomic resolution (1951-2016) to evaluate major changes in assemblage composition and phenology and the shorter set (1962-2016) with all taxa identified to current standards to examine diversity dynamics.