Surf smelt accelerate usage of endogenous energy reserves under climate change

experiment to avoid repeated measurements. Photos were later analyzed using the software ImageJ [30] to determine the egg and total yolk area. These measurements were used to calcu- late the ratio of yolk area to total egg area. This ratio was used to account for initial differences in egg size. Surf smelt embryo heart rate measurements were taken as another measure of energy demand. Due to heart rate being affected by fish developmental stage [31], heart rate measurements were taken within treatments at a specific, identifiable developmental stage: when the eye spots of the developing embryos darken [26]. When embryos were removed from glass bowls to be photo- graphed for yolk measurements, if the eye spots had darkened since the previous day, the embryos were also video recorded for 10 seconds. Heart beats were later counted in each video. Energy consumption of surf smelt larvae was assessed through a 3-day experiment, during which 2 larvae were haphazardly selected each day from each bowl within each treatment basin. Larvae were anesthetized using tricaine following the methods of Massee et al. [32]. This method limits larval movement and allows still images of the larvae to be taken. Leica software and a stereomicroscope were used to photograph the anesthetized larvae. After being photo- graphed, larvae were removed from the experiment to avoid repeated measurements. Photo- graphs were later analyzed using ImageJ [30] to determine yolk area and oil globule area. Statistical analysis

Larval yolk sac, oil globule size, and embryo heart rates were analyzed using linear mixed effect models with day, pCO2, temperature, and their interactions as fixed factors, and experimental basin as a random factor. The ratio of yolk size to embryo size was analyzed using a generalized linear mixed effect model (family = binomial, link = logistic) with day, pCO2, temperature, and their interactions as fixed factors, and experimental basin as a random factor. For each response variable measured, models were built with and without the interaction terms based on a combination of p-value significance (α = 0.05) and AIC comparisons to choose the most appropriate model. Interaction terms that were not significant and did not improve the model fit were left out. Random slope or intercept terms were applied to experimental basin depend- ing on model fit determined by AIC comparisons. Data from elevated and ambient pCO2 treatments are configured categorically in figures for clarity, but pCO2 was treated as a contin- uous variable in statistical models. All statistics were completed using R [33]. Ethics approval All experimental procedures were carried out in accordance with the recommendations and approval of the Animal Care and Use Committee at Western Washington University (protocol approval 19–001), and embryos were collected in collaboration with the Washington State Department of Fish and Wildlife. Results Treatment conditions Direct injection of pure CO2 into the elevated pCO2 mixing tanks followed by equilibration of

both ambient and elevated pCO2 treatments to three temperatures resulted in two distinct pCO2/pH conditions, modified somewhat by temperature (Table 1 embryo and larval yolk and oil globule experiments) (S1 Table heartbeat experiment). The average pH variability was simi- lar between ambient and high pCO2 treatments at comparable temperatures (± 0.03 pH units) (Table 1). The CO2 amendment in the elevated pCO2 treatment resulted in an average increase in 42 μmol kg/SW relative to the ambient treatment (Table 1). Because pCO2 is derived from PLOS ONE Surf smelt response to climate co-stressors PLOS ONE | https://doi.org/10.1371/journal.pone.0270491 June 27, 2022 5 / 14