Content uploaded by Jason A. Toy
Author content
All content in this area was uploaded by Jason A. Toy on Jul 25, 2022
Content may be subject to copyright.
Molecular Ecology. 2022;00:1–19.
|
1wileyonlinelibrary.com/journal/mec
1 | INTRODUCTION
Ocean acidification (OA; here defined as both increased ocean pCO2
and decreased pH) has been identified as a major threat to marine
species (Kroeker et al., 2013). Several studies have documented
changes in neurological functioning, including altered cognition,
sensory function and behaviour, in marine fish (e.g., Domenici
et al., 2012; Hamilton et al., 2013; Hamilton et al., 2017; Munday
et al., 2010; Pistevos et al., 2015), raising concerns about neurologi-
cal impacts leading to changes in the strength of species interactions
Received: 24 August 2021
|
Revised: 5 June 2022
|
Accepted: 4 July 2022
DOI : 10.1111/m ec.16 611
ORIGINAL ARTICLE
Upwelling- level acidification and pH/pCO2 variability moderate
effects of ocean acidification on brain gene expression in the
temperate surfperch, Embiotoca jacksoni
Jason A. Toy1 | Kristy J. Kroeker1 | Cheryl A. Logan2 | Yuichiro Takeshita3 |
Gary C. Longo1,4 | Giacomo Bernardi1
This is an op en access ar ticle under the terms of the Creative Commons Attribution-NonCommercial License, which permits use, distribution and reproduction
in any medium, provide d the original work is properly cited an d is not used for co mmercial purposes.
© 2022 The Authors. Molecular Ecology published by John W iley & Sons Ltd.
1Department of Ecology and Evolutionary
Biolog y, Universit y of California Santa
Cruz, Santa Cruz, C alifornia , USA
2Division of Science and Environmental
Policy, California State University
Monterey Bay, Seaside, California, USA
3Monterey Bay Aquarium Research
Instit ute, Moss La nding, California, USA
4NRC Resea rch Associ ateship Prog ram,
Northwest Fisheries Science Center,
Nationa l Marine Fisheries Ser vice,
Nationa l Oceanic and Atmospher ic
Adminis tration, Seattle, Washington, USA
Correspondence
Jason A . Toy, Depar tment of Ecology
and Evolutionary Biology, Univer sity of
Califo rnia Santa C ruz, Santa Cruz, CA ,
USA.
Email: jatoy@ucsc.edu
Funding information
David and L ucile Packard Foundation;
Packard Endowment Award via UCSC;
UCSC Co mmittee on Research; th e David
and Lucile Packard Foundation
Handling Editor: Sean Rogers
Abstract
Acidification- induced changes in neurological function have been documented in several
tropical marine fishes. Here, we investigate whether similar patterns of neurological im-
pacts are observed in a temperate Pacific fish that naturally experiences regular and often
large shifts in environmental pH/pCO2. In two laboratory experiments, we tested the
effect of acidification, as well as pH/pCO2 variability, on gene expression in the brain tis-
sue of a common temperate kelp forest/estuarine fish, Embiotoca jacksoni. Experiment 1
employed static pH treatments (target pH = 7.85/7.30), while Experiment 2 incorporated
two variable treatments that oscillated around corresponding static treatments with the
same mean (target pH = 7.85/7.70) in an eight- day cycle (amplitude ± 0.15). We found that
patterns of global gene expression differed across pH level treatments. Additionally, we
identified differential expression of specific genes and enrichment of specific gene sets
(GSEA) in comparisons of static pH treatments and in comparisons of static and variable
pH treatments of the same mean pH. Importantly, we found that pH/pCO2 variability
decreased the number of differentially expressed genes detected between high and low
pH treatments, and that interindividual variability in gene expression was greater in vari-
able treatments than static treatments. These results provide important confirmation of
neurological impacts of acidification in a temperate fish species and, critically, that natural
environmental variability may mediate the impacts of ocean acidification.
KEYWORDS
climate change, differential gene expression, environmental variability, embiotocidae, global
change, RNA- seq
2
|
TOY eT a l.
(e.g., predation). In contrast, more recent work has questioned the
generality and replicability of such impacts across studies and species
(Clark et al., 2020a). Additionally, much of the evidence of neurolog-
ical impacts comes from studies of a few tropical reef species under
static pH/pCO2 regimes (Nagelke rken & Munday, 2016). Physiological
evidence indicates these neurological impacts may be the result
of a hypercapnia- driven reversal of electrochemical gradients in
GABAergic neurons. This has been hypothesized to result from in-
ternal acid– base balance processes that lead to an accumulation of
intracellular [HCO3
−] and/or a decrease in extracellular [Cl−] (Heuer &
Grosell, 2014; Nilsson et al., 2012). This shift in ion concentrations is
thought to cause neuron depolarization upon GABA A receptor acti-
vation rather than the hyperpolarization expected under nonacidified
conditions, reversing the functional nature of these neurons from in-
hibitory to excitatory and presumably causing the observed shifts in
cognition and behaviour (Heuer & Grosell, 2 014; Nilsson et al., 2012;
Schunter et al., 2019). Given this body of evidence, altered neurologi-
cal func tion may be a major pathway through which changing seawa-
ter carbonate chemistry will impact fitness in marine fish. Continued
work elucidating the molecular mechanisms underlying these changes
is therefore critical for moving the field forward.
Many coastal ecosystems experience significant environmental
variability over a range of temporal scales, including fluctuations in
seawater pH/pCO2 (Chan et al., 2 017; Hofmann et al., 2011; Kang
et al., 2022; Kro eker et al., 2020). In upwelling regions, where deeper,
more acidic water is brought to the ocean surface, pH can vary by
half a unit over a period of weeks (Hirsh et al., 2020; Hofmann
et al., 2011). In seagrass beds, pH can vary by a whole unit over
a period of hours to weeks due to fluctuations in photosynthesis
and respiration and tidal movement (Duarte et al., 2013; Hofmann
et al., 2011). These fluctuations often reach or exceed predictions
for the mean future ocean pH under OA (Gruber et al., 2012; Hauri
et al., 2013; Takeshita et al., 2015). In previous studies, exposure to
low pH/high pCO2 seawater has affected indicators of fish neuro-
logical function anywhere from 2– 12 days after exposure has ceased
(Hamilton et al., 2013; Munday et al., 2010), but it is unclear what
duration of exposure elicits these ef fects. How temporal environ-
mental variability moderates fish responses to low pH/high pCO2
remains critically understudied, leaving many unanswered questions
about how OA may realistically affect populations in nature (but see
Jarrold et al., 2017; Jarrold & Munday, 2019; Schunter et al., 2021).
Investigations into the ef fects of pH/pCO2 variability are important
for accurate prediction of the severity of impacts acidification will
have on natural populations and ecosystems (Kroeker et al., 2020).
For example, if variability dampens the effect s documented in
studies using only static pH/pCO2 treatments (as seen in Jarrold
et al., 2017 and Jarrold & Munday, 2019, where diel pCO2 fluctua-
tions ameliorated impairments in behaviour and growth seen under
static decreases in pCO2), we may be overestimating the effects of
OA, and overlooking an important role that variability may play as
a provider of temporal refuge. Conversely, if variability exacerbates
negative impacts of acidification, acting as an additional stressor,
we may be underestimating the potential impact of acidification on
natural populations.
We expect physiological responses to OA to be reflected in the
gene expression of the affected organism (Griffiths et al., 2019;
Hamilton et al., 2017). In particular, we expect changes in brain gene
expression to be associated with shift s in neurological and cogni-
tive function (Schunter et al., 2016). Given the proposed mechanism
of OA- induced cognitive impairment described above, we expect
experimental acidification to impact expression in genes related
to the maintenance of homeostasis and neuronal signalling, such
as ion transporter and signal receptor genes, and those involved in
the GABAergic signalling pathway. Changes in expression in these
gene categories have been noted in spiny damselfish (Schunter
et al., 2018) and three- spined stickleback (Lai et al., 2016), but this
has not yet been investigated in a temperate reef species with an
evolutionary histor y of exposure to fluctuating pCO2.
Here, we present two experimental studies of the ef fects of
acidification on brain gene expression in a common temperate reef
fish, the black surfperch (Embiotoca jacksoni). Surfperches make up
a large propor tion of fish biomass on California rocky reefs (Laur &
Ebeling, 1983) and suppor t an immensely popular recreational fish-
er y. E. jacksoni is found in both upwelling reef systems and estua-
rine seagrass ecosystems, and therefore has an evolutionary history
of exposure to variable pH conditions that are often more ex treme
than those experienced by tropical reef fish (Hofmann et al., 2011).
A few studies have investigated the effec ts of acidification on tem-
perate reef fishes (e.g., Cline et al., 2020; Hamilton et al., 2017;
Kwan et al., 2017), but surfperches are unique because they exhibit
viviparity and no pelagic larval phase, with young born as devel-
oped juveniles. Additionally, E. jacksoni has limited adult dispersal
(Bernardi, 2000, 2005; Hixon, 1981). These two life- history traits
increase the likelihood of adaptation to local environmental condi-
tions in E. jacksoni, which may lead to divergent effects of acidifica-
tion in this species compared to other temperate fish (e.g., greater
physiological adaptation to OA in populations that have historically
experienced local acidification).
Experiment 1 was designed to determine the presence and ex-
tent of any impacts of acidification on E. jacksoni brain gene expres-
sion and used a static and more extreme acidified treatment (pH
~7.30). In Experiment 2, we used a less extreme static treatment
and incorporated two variable treatments with different mean pH
levels to mimic upwelling- scale pH variability. This experiment was
designed to test the potential role of temporal pH/pCO2 variability
in mediating any neurological effects of acidification. Together, the
results of these experiments provide a more comprehensive under-
standing of the impacts of acidification on marine organisms, partic-
ularly in dynamic, temperate ecosystems.
2 | MATERIALS AND METHODS
2.1 | Collections and acclimation
W e c o l l e c t e d y o u n g - o f - t h e - y e a r E. jacksoni from Elkhorn Slough
(Monterey County, CA) using a beach seine. Collected fish were
placed in coolers and driven back to UCSC- CSC , where they were
|
3
TOY eT al.
kept in outdoor flow- through containers until the start of each ex-
periment. For exact dates of collections, acclimation periods, and
experimental manipulations see Table S1.
2.2 | Experimental design
We conducted two separate experiments with similar methods in
November 2015 and September 2017 at University of California,
Santa Cruz's Coastal Science Campus (UCSC- CSC). Both experi-
ments treated E. jacksoni juveniles in outdoor flow- through seawater
systems . In Experimen t 1 (2015), we set target treatments at pH 7.85,
representing a common current upwelling condition along the coast
of Central California, and pH 7.30, representing a current extreme
estuarine event or future extreme upwelling event (Chan et al., 2017;
Hofmann et al., 2011; Lowe et al., 2019; Takeshita et al., 2015). Both
treatments in this experiment held the target pH constant (static)
over the course of the experiment. Five randomly assigned juvenile
E. jacksoni were distributed across two replicate tanks at pH 7.30 and
four were distributed across two replicate tanks at pH 7.85 (opaque
200 L plastic drums). Seawater pH treatments were replicated only
at the level of holding tanks. Replicates were then brought down to
experimental pH levels over 7 days. Tissue sampling was conducted
after 23 days of treatment (Table S1).
In Experiment 2 (2017), we incorporated upwelling- scale pH
variability into two of the treatments, and target pH levels were set
at more conservative levels. Two static pH treatments were set at
target pH levels of 7.85 and 7.70, approximating conser vative pres-
ent and future reef conditions during the upwelling season (Chan
et al., 2017; Takeshita et al., 2015). For each static treatment, there
was a corresponding variable treatment that oscillated around the
same mean pH as the static treatment with an amplitude of ±0.15
pH and a period of 8 days (Figure 1), approximating a typical upwell-
ing pattern (Hofmann et al., 2011). An additional treatment, hereaf-
ter referred to as “ambient”, had a static target pH of 8.00. However,
because our pH control system was not capable of increasing pH
above that of the incoming seawater, periodic natural decreases in
the pH of the input seawater below 8.00 caused this treatment to
exhibit an intermediate level of variabilit y between that of the static
and variable treatments (Figure S1). We used 10 header buckets
(two per treatment) to create the five different pH treatments, with
two replicate tanks per header. We randomly assigned six juvenile
E. jacksoni to replicate tanks (translucent 61 L plastic containers; 6
individuals × 4 replicate tanks × 5 pH treatments) and allowed them
to acclimate at ambient pH for 2– 3 days. We then allowed the pH of
each treatment to slowly approach its starting pH (target pH 8.00,
7.85, or 7.70) over a period of 2 days. After an additional 4– 5 days,
the variable treatments began their programmed oscillations
(Figure S1). Due to logistical restrictions, the treatments were sep-
arated into t wo groups (pH 7.85 and ambient treatments, pH 7.70
treatments) that were staggered in their timing by 1 day (Figure S1).
Using a custom- built LabView program, set points for the variable
treatments were changed throughout the experiment at intervals
of 0.003125 pH/h to create 8- day cycles. During this experiment,
fish were removed from their treatment tanks on two occasions to
conduct behavioural assays, after which they were returned to their
treatment tanks. Because we were met with logistical challenges
that precluded the proper execution of these trials, these data were
not analysed. Eight days (1 full cycle of the variable treatments) were
allowed to elapse between the last trial and tissue sampling, which
was conduc ted after 22 days (Figure S1; Table S1).
2.3 | pH control system and sampling of
seawater chemistry
Seawater pH was manipulated using a custom- built feedback con-
trol system. Two large sumps received a continuous flow of ambient
seawater. One of these sumps (“low pH”) was continuously bubbled
with CO2 gas, while the other (“ambient”) was left untreated. Lines
from both sumps fed seawater into header buckets at varying rates
to create pH treatments. The pH of each bucket was continuously
measured by Honeywell Durafet II sensors connected to Honeywell
Universal Dual Analysers (UDAs; see Kapsenberg et al., 2017).
FIGURE 1 Experiment design and data analysis pipeline for
Experiments 1 and 2.
4
|
TOY eT a l.
Seawater pH in each header was controlled through a feedback sys-
tem, where a solenoid valve determined the flow of low pH seawater
to the header to either increase or decrease pH. The mixed treat-
ment water from each header then flowed out into two replicate
holding tanks. We oxygenated and mixed seawater in each header
using air pumps/stones and/or water pumps (Experiment 2 only).
Prior to beginning each experiment, we calibrated the Durafet
sensors from the header buckets using equimolar Tris buffer
(DelValls & Dickson, 1998 ) obtained from the Dickson Laboratory
(Scripps Institution of Oceanography). In Experiment 1, discrete
water samples were taken from the replic ate tanks at seven time
points and used for characterization of carbonate chemistr y via
spectrophotometric pH analysis and open cell total alkalinity titra-
tion (Dickson et al., 20 07). In Experiment 2, samples were taken
from headers at five time points and used for post hoc calibration
of Durafet pH measurements. Using a handheld sensor (YSI), we also
measured temperature, pH, dissolved oxygen, and salinity in each
replicate tank daily and, in Experiment 2, in each header as well to
allow for calibration of YSI pH measurements to calibrated Durafet
measurements. See Tables 1 and 2 for measured and calculated sea-
water parameters.
2.4 | Experimental considerations
A heat wave struck Santa Cruz during Experiment 2. This added
stress may have contributed to the juvenile mortality observed dur-
ing this experiment (48 out of 120 fish died, unrelated to treatment;
ANOVA, F = 2.088, p = .133). Additionally, some of the Durafet
sensors (4 of 10) used to control the pH in the headers experienced
heavy fouling by microalgae toward the end of Experiment 2. This
probably led to artificially high pH measurement s for about 2 h
around midday due to photosynthesis, and thus a corresponding
over- correction by the pH control system. Because of this issue, the
pH of certain tanks was probably lower during midday than their
respective set points. To better understand the scale of this over-
correction, we conducted a test 11 days after tissue sampling, in
which all Durafets were placed in the same header with no active
pH control (Figure S2). Though the effect of fouling on recorded
pH appeared to strengthen over the 11 days since tissue sampling,
this post- hoc test revealed variability in the impact across head-
ers, and a relatively even distribution of fouling across treatments
(Figure S2). The greatest spike during this test occurred in one of
the two ambient header s with a magnitude of ~0. 5 pH units, but this
treatment was not included in most of our analyses, and thus we be-
lieve it does not affect our conclusions. Examination of experiment
Durafet readings from the ambient header (which, due to its high
set point and limited pH control, displayed the true extent of the pH
spikes) revealed that significant spikes (deviation of ~0.05 pH units
or greater) in this most affected treatment only began occurring
approximately 3 days before tissue dissection. Because our test in-
dicated that the other headers were affected to a much smaller de-
gree (Figure S2), the other treatments probably did not experience
midday spikes of >0.05 pH units for any significant duration prior
to the end of the experiment. To prevent the inclusion of spurious
pH data points in the characterization of the experimental treat-
ments, we used the continuous Durafet pH and temperature data
for the dates 1– 17 September, after which we used pH and tem-
perature data from daily YSI readings t aken from each header. This
shift in sampling frequency probably explains much of the appar-
ent increased variability of the pH treatments after 17 September
(Figure S1), as the YSI data represents only a daily snapshot of the
pH of each header. The pH sensor within the YSI is also functionally
different from those within Durafets (Martz et al., 2010). Fina lly,
outside of the daily spikes, the Durafet pH data collected 17– 24
September showed no obvious departure from the precision seen
earlier in the experiment. We therefore contend that, apart from
the midday overcorrections experienced at the end of the experi-
ment, the true precision and variability of the pH treatments was
unchanged after 17 September, and any apparent changes reflect
only a change in the pH- sensing instrument used.
2.5 | Fish care and handling
This experiment was run under the approval of UCSC IACUC pro-
ject proposals BERNG1312 and KROEK1503_A2. We performed
system checks at least daily and fed fish frozen shrimp every day
(Experiment 1) or a mix of frozen brine shrimp, Spirulina brine
shrimp, and mysis shrimp ever y other day (Experiment 2). Tanks
were cleaned and excess food removed approximately seven times
per week (Experiment 1) and at least once per week (Experiment 2).
To minimize stress, a shelter was placed in each replicate. To reduce
heat and sun exposure, shade cloth was kept over the top of the
replicate tanks whenever water monitoring, cleaning, or feeding was
not occurring.
TABLE 1 Carbonate chemistr y and environmental parameters for treatment containers in Experiment 1
Treatment pHT (spec) pCO2 (μatm) ΩTA (μmol/kg) Temp (°C) Salinity (ppt) pHT (YSI)
Target pH 7.85 7.88 ± 0.02 599 ± 36 1.50 ± 0.08 2193 ± 59 13.0 ± 0.5 33.8 ± 0.1 7.8 9 ± 0.04
Target pH 7.30 7. 35 ± 0.06 2204 ± 333 0.48 ± 0.07 2212 ± 20 12 .2 ± 0.8 33. 8 ± 0.1 7. 36 ± 0.14
Note: Aragonite saturation state (Ω) and pCO2 were calculated with the R package seacarb (Gattuso et al., 2021) using the spectrophotometric pH
and total alkalinity (TA) values from discrete bot tle samples, and salinity and temperature values from YSI readings. All values are means ± SD. Mean
pHT (spec) and TA were calculated from bottle samples taken at seven time points across the experiment. Mean pHT (YSI) was calculated from daily
readings that were calibrated using the discrete bottle samples.
|
5
TOY eT al.
2.6 | Tissue sampling
At the end of each experiment, we dissected tissue from four in-
dividuals from each treatment. Individuals were dissected one at a
time, with all dissections taking less than 10 min from the time of
fish removal from its tank. Brain and lateral muscle tissue were dis-
sected and sequenced in Experiment 1 for use in the transcriptome
assembly, but only brain tissue was sequenced in Experiment 2. In
Experiment 2, the whole brain was dissected at the approximate
time when the variable pH treatments were crossing (in the ascend-
ing direction) their target mean pH levels (Figure S1). Only brain gene
expression analysis will be further discussed here. Tissue was stored
in screw- cap tubes and flash frozen in liquid nitrogen. We stored all
tissue samples at −80°C until RNA extraction.
2.7 | RNA extraction and library preparation
Dissected whole brains were arbitrarily subsampled and homog-
enized using a Qiagen TissueLyser. A discussion of the potential
effects of subsampling are included below in the “Interindividual var-
iability in gene expression” section of the Discussion. We extracted
RNA using the Qiagen RNeasy Mini extraction kit. RNA quality and
quantity were assessed using a NanoDrop spectrophotometer and
Qubit fluorometer. RNA was stored in DEPC- treated water at −80°C.
cDNA libraries were prepared from 1 μg of total RNA using the New
England Biolabs NEBNext Ultra II RNA Library Prep kit. Prepared
libraries were sequenced on an Illumina HiSeq 4000 (150 bp SE) at
the QB3 Vincent J. Coates Genomics Sequencing Laboratory at the
University of California, Berkeley.
2.8 | Read processing and transcriptome assembly
We removed adapters and trimmed/removed low quality reads
using the Trimmomatic software (v0.36; Bolger et al., 2014; parame-
ters = LEADING:2 TRAILING:2 SLIDINGWINDOW:4:2 MINLEN:25)
and quality checked the trimmed sequences using FastQC (version
0.11.7; Andrews, 2010). We used the trimmed reads from all se-
quenced samples from both experiments to assemble a brain/muscle
tissue combined transcriptome for E. jacksoni using the genome-
guided TopHat/Cuffmerge/Cufflinks pipeline (default parameters;
TopHat version 2.1.1, Cufflinks version 2.2.1; Trapnell et al., 2012).
This pipeline creates separate assemblies for each sample, which are
then merged. A draft, scaffold- level E. jacksoni genome assembly
was used as the reference (see Supporting Information Materials).
We annotated the transcriptome assembly by running a blastx
query (e- value cutoff = 1e- 3; NCBI, Altschul et al., 199 0) against
the SwissProt database (uniprot_sprot.dat.gz downloaded 25 April
2020; The UniProt Consor tium, 2021).
2.9 | Multivariate analysis of gene expression
In Experiment 1, we sequenced transcripts from both muscle
and brain tissue, but only brain gene expression will be discussed
here. To characterize global gene expression of individuals,
trimmed reads were aligned and quantified into gene- level ex-
pression data using bowtie (version 1.2.3; Langmead et al., 2009)
and RSEM (version 1.3.3; Li & Dewey, 2011) within the Trinity
software package (version 2.9.1; Haas et al., 2013). Raw read
counts were then filtered to remove genes with low expression
using the default parameters of the filterByExpr function (min.
count = 10, min.total.count = 15, large.n = 10, min.prop = 0.7)
in the R package, edgeR (version 3.34.0; R Core Team, 2021;
Robinson et al., 2010). We normalized the read counts using the
TPM method, as implemented by the calcNormFactors function
in the edgeR package, then log2- transformed the data using the
cpm function (prior.count = 2). The transformed data were di-
mensionally reduced through multidimensional scaling (metric
MDS in Experiment 1, nMDS in Experiment 2) using Manhat tan
distances, as implemented through the wcmdscale and metaMDS
functions in the vegan package for R (version 2.5.7; Oksanen
et al., 2020).
TABLE 2 Carbonate chemistr y and environmental parameters for the headers of each treatment in Experiment 2
Treatment
pHT (Durafet)
(hourly)
pCO2
(μatm) Ω
TA (μmol/kg)
(bottle samples)
Temp (°C)
(Durafet)
Salinity
(ppt) (YSI)
pHT (YSI)
(daily)
Ambient 8.0 0 ± 0.04 452 ± 55 2.36 ± 0.23 2266 ± 3 17.5 ± 0.06 34.3 ± 0.1 8.01 ± 0.08
Target pH 7.85 - Static 7.9 0 ± 0.01 586 ± 14 1 .94 ± 0.07 2268 ± 3 17. 6 ± 0.06 34.3 ± 0.1 7. 90 ± 0.05
Target pH 7.85 - Variable 7.89 ± 0.08 614 ± 136 1.93 ± 0.34 2268 ± 4 1 7.5 ± 0.06 34.3 ± 0.1 7.8 8 ± 0.11
Target pH 7.70 - Static 7.76 ± 0.04 848 ± 64 1.46 ± 0.17 2268 ± 5 17.6 ± 0.06 34.3 ± 0.1 7. 75 ± 0.08
Target pH 7.70 - Variable 7.76 ± 0.09 870 ± 199 1.47 ± 0.30 2267 ± 4 17. 6 ± 0.06 34.3 ± 0 .1 7.74 ± 0.10
Note: Aragonite saturation state (Ω) and pCO2 were calculated with the R package seacarb (Gattuso et al., 2021) using the Durafet pH and
temperature values, average TA values from discrete bottle samples, and salinity values from YSI readings. All values are means ± SD. Mean pHT
(Durafet) values were calculated using hourly averaged pH readings (from headers) that were calibrated using discrete (bottle) water samples and
include only the time period of the first two pH c ycles (1/2 Septemb er– 17/18 September). Mean pHT (YSI) values were calculated using daily re adings
(from replicate containers) that were calibrated using bottle- calibrated Durafet values (taken simultaneously) from the 1/2 September– 17/18
September date range and include YSI readings from the entire length of the experiment (1/2 September– 23/24 September).
6
|
TOY eT a l.
To test whether global gene expression profiles differed among
treatments, we ran a permutational multivariate analysis of vari-
ance (PERMANOVA; Anderson, 2001, 2017) on the transformed
expression data using the adonis func tion of the vegan package
(method = “manhattan”, perm = 1,000,000). In Experiment 1, the
sole model factor was pH level (7.85, 7.30). In E xperiment 2, the
model was run with two factors: pH level (7.85, 7.70) and pH vari-
ability (static, variable). For Experiment 2, pairwise comparisons
between treatments were conducted using the pairwise.adonis
function of the pairwiseAdonis package (sim.method = “manhattan”,
perm = 1,000,000) (Martinez Arbizu, 2020).
2.10 | Differential gene expression analysis
Using the gene- level counts matrix created by RSEM, we identi-
fied differentially expressed genes (DEGs) between all pairwise
treatment comparisons using the edgeR package, as implemented
through Trinit y. To buffer against false positives and noise due to the
experimental conditions described above, we used a conservative
FDR cutoff value of 0.001 (- P parameter) and a fold- change cutoff of
1.5 (- C parameter) to create the final list of DEGs for each treatment
comparison. We then repeated MDS procedures and PERMANOVA
tests as described above, using only these DEGs.
2.11 | Functional enrichment analysis
To identify gene sets (groups of functionally related genes) that were
significantly enriched in a given treatment comparison (e.g., pH 7.85
vs. pH 7.30) we used the threshold- free analytical method, gene set
enrichment analysis (GSEA; Subramanian & Tamayo et al., 2005) as im-
plemente d through the FGS EA package for R (Koro tkevich et al., 2021).
Given a ranked list of genes derived from differential expression anal-
ysis, this method yields a list of gene sets from user- supplied gene
set databases - in this case gene ontology (GO; The Gene Ontology
Consortium, 2020), KEGG (Kanehisa & Goto, 2000), and the MSigDB
Hallmark collection (Liberzon et al., 2015) - that are enriched among
up- and downregulated genes. Enrichment analysis was completed for
Experiments 1 and 2 separately, and the resulting enriched gene sets
from analogous treatment comparisons were then contrasted across
experiments to identify commonly enriched gene sets.
3 | RESULTS
3.1 | Seawater pH manipulation
Mean seawater pH levels were maintained near their target set
points in each experiment. Seawater parameters and information on
how they were c alculated are given in Tables 1 and 2. Note that for
consistency, we continue to use target pH levels to refer to each
treatment.
3.2 | Sequencing and transcriptome assembly
The TopHat/Cufflinks pipeline yielded a transcriptome made up of
71,933 assembled transcripts grouped into 39,258 putative genes.
Aligning the trimmed reads back to the assembled transcriptome re-
sulted in an 82.98% alignment rate. A blastx search of the transcrip-
tome against the SwissProt database revealed that 8836 transcript s
represented nearly full- length transcripts (>80% alignment cover-
age), and that 16,574 proteins were represented in the transcrip-
tome at some level of alignment coverage. We used BUSCO (version
4.0.6; Simão et al., 2015) to quantify the completeness of our tran-
scriptome and found that of the 3640 BUSCO orthologs in the
Actinopterygii data set, 76.9% were found complete (41.2% single-
copy, 35.7% duplicated), 6.5% were found fragmented, and 16.6%
were missing. For more statistics on the assembly, see Table S2.
Total sequenced reads per sample are listed in Tables S3 and
S4. Of the 39,258 putative genes in the assembled transcriptome,
22,961 (Experiment 1) and 33,597 (Experiment 2) remained in each
dataset af ter filtering for genes with low expression.
3.3 | Multivariate analyses of global
gene expression
Single- factor PERMANOVA analysis identified a strong and signifi-
cant effect of pH level on global gene expression in both Experiment
1 (single- factor; r2 = 0.811, F = 25.766, p = .029) and Experiment 2
(two- factor, ambient excluded; r2 = 0 .159, F = 2.53, p = .021), with
pH explaining 81 and 16% of the obser ved variation, respectively
(Tables S5 and S6). In Exp eriment 2, we did not detect an effect of pH
variability on global gene expression (r2 = 0.037, F = 0. 59, p = .890)
or an interaction of pH level and variability (r2 = 0.052, F = 0.84,
p = .524). Pairwise comparisons of all treatments (including ambi-
ent) revealed two comparisons were nearly significantly different
(Table S7): ambient versus 7.70 static (r2 = 0.230, F = 1.79, p = .086)
and 7.85 static versus 7.70 static (r2 = 0.292, F = 2 .47, p = .057).
We visualized the differences in global gene expression patterns be-
tween treatments using MDS (Figures S3 and S4).
3.4 | Differential gene expression analysis
We found 10,656 DEGs between the treatments in Experiment 1
(Figure 2). In Experiment 2, we found a total of 200 DEGs across
all treatment comparisons (Table 3; Figure S5). The 7.85 static ver-
sus 7.70 static comparison produced the majority of DEGs (159) in
this experiment. The 7.85 static versus 7.70 variable and 7.85 vari-
able versus 7.70 static comparisons produced 11 DEGs each and six
genes each were differentially expressed in the st atic versus vari-
able comparisons of both the 7.85 and 7.70 pH levels (one gene was
consistently differentially expressed across the two comparisons).
Since we did not expect transcriptome- wide shifts in gene ex-
pressio n across pH variabili ty treatment s in Experiment 2 ( Figure S4),
|
7
TOY eT al.
and were instead interested in how the expression of acidification
response genes was affected by the introduction of environmental
variability, we repeated our multivariate analyses of gene expression
for only the DEG subset of Experiment 2. For consistency, analo-
gous analyses were also performed for the Experiment 1 DEG subset
(Table S8; Figure S6).
DEG expression differed among pH levels in Experiment 2
(r2 = 0.388, F = 10 .7 7, p = .004), with pH explaining 39% of the ob-
served variation (Figure 3). We did not detect an effect of pH vari-
ability (r2 = 0.041, F = 1.15 p = .291). The interaction of pH level and
variability, however, was marginally significant (r2 = 0.139, F = 3.85,
p = .052; Table S9). Pairwise comparisons of all treatments revealed
a significant difference between the pH 7.85 and 7.70 static treat-
ments (p = .029) and nearly- significant differences between the am-
bient and 7.85 static treatment s (p = .057) and between the ambient
and 7.70 static treatments (p = .0 86; Table S10). The comparison of
7.85 static versus 7.85 variable was also nearly significant (p = .057),
but the comparison of 7.70 static versus 7.70 variable was less so
(p = .171).
3.5 | Analysis of within- treatment variances
To test the effect of pH variability on within- treatment variabilit y
in gene expression, we calculated the variance of normalized gene
expression for each DEG within each treatment in Experiment 2 (ex-
cluding a mbient) and average d the variance ac ross all DEGs (Figure 4).
For each pH level (7.85, 7.70), we then calculated F- ratios by dividing
the mean variance of the variable treatment by that of the static
treatment. We log- transformed these variances and compared the
distributions to a t- distribution centred at 0 using one- tailed t- tests
(Table S11). We found that at both pH levels, the average variance
in DEG expression was greater in the variable pH treatment than in
the static pH treatment (pH 7.85: p = .0001; pH 7.70: p = .034 87).
3.6 | Functional enrichment analysis
In Experiment 1, gene set enrichment analysis using FGSEA revealed
240 enriched gene sets among the upregulated genes and 343 en-
riched gene sets among the downregulated genes in the pH 7.30
treatment compared to the 7.85 treatment (FDR <0.05; Table S12).
In Experiment 2, 61 gene sets were enriched among upregulated
genes and 71 among downregulated genes in the 7.70 static treat-
ment compared to the 7.85 static treatment (Table S13 ). At the
7.85 pH level, we found 4 4 enriched gene sets among upregulated
genes and 202 among downregulated genes in the variable treat-
ment compared to the static treatment (Table S14). At the 7.70 pH
level, 115 and 22 gene sets were enriched among the upregulated
and downregulated genes, respectively, in the variable treatment
compared to the static treatment (Table S15). To aid interpretation,
the enriched gene sets for the 7.85/7.30 comparison in Experiment 1
and the 7.85 static/7.70 static comparison in E xperiment 2 were fur-
ther collapsed into clusters of gene sets (using a gene set similarity
coefficient) using the AutoAnnotate and clusterMaker2 applications
for the Cytoscape software platform. These clusters were manually
summarized based on their constituent gene sets (Tables 4 and 5).
To assess consistency of response to acidification across exper-
iments, we determined the overlap in enriched gene sets between
the pH 7.85/pH 7.30 comparison of Experiment 1 and the pH 7.85
static/pH 7.70 static comparison of Experiment 2 (Figure 5, Table 6).
To assess consistency in expression response across the two static
versus variable treatment comparisons of Experiment 2 (7.85
static/7.85 variable, 7.70 static/7.70 variable), we determined the
overlap in enriched gene sets between these comparisons (Figure 6,
Table S16).
4 | DISCUSSION
4.1 | Impacts of static acidification
Numerous studies have demonstrated impaired behaviour and sen-
sory function in fish and other marine organisms when exposed
to low pH/high pCO2 (Domenici et al., 2012; Hamilton et al., 2017,
FIGURE 2 Heatmap of gene expression profiles for each
individual in Experiment 1. Each column represents an individual
fish, and each row represents a differentially expressed gene.
Yellow colours represent upregulation in a given treatment and
purple colours represent downregulation. Brighter hues represent
larger differences in relative gene expression across the treatments.
TABLE 3 Number of DEGs detected across all treatment
comparisons in Experiment 2
Ambient
pH 7.70
static
pH 7.70
variable
pH 7.85
static
Ambient
pH 7.70 static 5
pH 7.70
variable
8 6
pH 7.85 static 3159 11
pH 7.85
variable
611 9 6
8
|
TOY eT a l.
Hamilton et al., 2013; Munday et al., 2010; Pistevos et al., 2015,
but see Clark et al., 2020a, 2020b). Though the mechanisms be-
hind these changes are still poorly understood, significant effects of
low pH/increased pCO2 on brain gene expression have been docu-
mented in a few marine fish species that demonstrate associated im-
pairments in behaviour (Lai et al., 2016; Schunter et al., 2016, 2018,
2021). In this study we tested whether brain gene expression is simi-
larly impacted in a temperate reef fish that experiences prolonged
periods of natural acidification. Across both experiments presented
here, we found that global gene expression was significantly af-
fected by acidification (high vs. low pH). Comparing results across
experiments, the number of detected DEGs increased with more
extreme acidification, as did the number of enriched gene sets. A
similar increase in DEGs with increased intensity of acidification has
also been reported in the olfactory bulb of coho salmon (Williams
et al., 2019). This marked increase in ef fect size indicates that further
acidification past the already- low pH of 7.70 can have a substantial
additional impact on the physiology of marine fish. This pattern may
have important implic ations for the management of marine ecosys-
tems and the services they provide as our global societ y struggles to
control CO2 emissions.
Although a greater number of gene sets were enriched in
Experiment 1 than in the comparison of the static treatments of
Experiment 2, similar enrichment themes emerged. In both exper-
iments, static acidification led to the upregulation of gene sets re-
lated to turnover in the proteome and transcriptome that may reflect
ongoing physiological adaptation to altered environmental condi-
tions (Tables 4 and 5). Additionally, static acidification in both exper-
iments led to the downregulation of gene sets related to the MAPK
cascade, G protein- coupled receptor signalling pathways, plasma
membrane components, secretory vesicles and granules, neuroac-
tive ligand- receptor interaction, and calcium ion binding, indicating a
general reduction in cell signalling, including neuroactive signalling,
in response to high pCO2. In general, this is the opposite of the re-
sponse seen in similar gene sets in spiny damselfish (Acanthochromis
polyacanthus) (Schunter et al., 2018) and the olfactory bulb of coho
salmon (Williams et al., 2019). Schunter et al. (2019) proposed that
high pCO2- induced changes in electrochemical gradients across
GABAergic neuron membranes may initiate a “vicious cycle” of
feedbacks and ultimately an increase in excitatory activity in the
brain that may explain behavioural changes seen in other species.
If this is indeed the case, the downregulation of gene sets related
to neuroactive signalling seen here may represent a species- specific
FIGURE 3 nMDS plot of DEG
expression in Experiment 2. Points
represent single individuals. Ellipses are
95% confidence ellipses.
FIGURE 4 Box plot of within- treatment variances in Experiment
2 (DEGs only, outliers removed for clarity). Diamonds mark the
mean for each treatment. Notches represent a roughly 95%
confidence interval around the median. Removed points lie outside
of 1.5 times the IQR of each hinge.
|
9
TOY eT al.
TABLE 4 Summary of upregulated and downregulated gene set clusters in Experiment 1
Upregulated in pH 7.30 treatment Downregulated in pH 7.30 treatment
Categorical cluster
Number of gene sets
in each cluster Categorical cluster
Number of gene sets
in each cluster
Mitochondrion, aerobic respiration, mRNA
export from nucleus
44 Transmembrane ion transport, regulation
of synaptic signalling, ligand- gated ion
channel activity, behaviour, cognition
and sensory perception
90
RNA metabolism, processing, splicing,
modification, tRNA biosynthesis;
ribosome biogenesis
41 Regulation of nervous system development
and growth
60
Translation and protein localization 39 Synaptic vesicle membrane, regulation of
clathrin- dependent endocytosis
22
Muscle development 22 Axo- dendritic transport 20
Organic acid catab olism 15 Synaptic membrane and synapse 19
Muscle contraction and adaptation,
myogenesis
14 G protein- coupled receptor signalling 15
Energy reserve and carbohydrate metabolic
process
10 Exocytosis and secretion 14
Proteolysis, mRNA catabolism, negative
regulation of cell cycle G2/M phase
transition
10 Central nervous system development 12
Peroxisomal organization and transport,
protein localization to organelle
8 Regulation of pH and iron ion transpor t 9
Innate immune response 6Aminoglycan and glycoprotein metabolic
process
8
Telomere maintenance via lengthening and
organization
6Calcium- dependent phospholipid binding
and cell– cell adhesion
8
RNA polymerase II 5 Dopamine secretion and transport 7
Protein modification by small protein
conjugation or removal
3 Axon, distal axon and terminal bouton 6
Actin filament binding 2Dendritic tree and neuron spine 6
Alpha actinin binding 2GTPase activator activity 6
Cytoplasmic stress granule 2Positive regulation of MAPK cascade 6
DNA polymerase activity 2Receptor localization to synapse 6
Mitochondrial matrix and nucleoid 2Regulation of vesicle fusion 6
Ribosome binding 2Dendrite membrane 5
RNA helicase activity 2Ephrin receptor signalling pathway 5
adipogenesis 1 Ex trinsic component of cy toplasmic side of
plasma membrane
5
ADP binding 1Microtubule polymerization 5
Allograft rejection 1 Regulation of protein localization to
membrane
5
Androgen response 1Synaptic vesicle transport and localization 5
Cell substrate junction 1Glycosphingolipid biosynthetic process 4
Cysteine- type endopeptidase activity 1Cortical Actin cytoskeleton 3
Fatty acid metabolism 1 Regulation of cell shape 3
Ficolin- 1- rich granule lumen 1Vascular transport 3
General transcription initiation factor binding 1Intrinsic component of Golgi membrane 2
Interferon alpha response 1 Long term depression and vascular smooth
muscle contraction
2
Lysine degradation 1Negative regulation of secretion & transport 2
(Continues)
10
|
TOY eT a l.
adaptive response aimed at combating maladaptive runaway exci-
tation in acidic waters (see discussion of GABAA receptor related
genes below). Finally, downregulation of gene set s related to grow th
and morphogenesis, cell– cell adhesion, and the cytoskeleton indi-
cate potential disruption of cell growth and development due to in-
creased cellular stress. Similar themes of upregulated transcription
and cellular stress response have also been documented in the mus-
cle tissue of Pacific rockfish (Hamilton et al., 2017 ).
We also identified divergent sets of genes enriched be-
tween the moderate (Experiment 2, target pH 7.70) and extreme
(Experiment 1, target pH 7.30) acidification treatments, indicative
of a potential threshold ef fect as static pH decreases. In compari-
son to the static acidification in Experiment 2, static acidification in
Experiment 1 resulted in the up- and downregulation of additional
gene sets related to metabolic processes (Table 4). These changes
may again indicate further shifts to the synthesis of stress re-
sponse proteins, or to isoforms that are better suited to an altered
cellular environment. Because, at least in humans, there can be
interaction/crosstalk between cellular stress response pathways
and the innate immune system signalling pathways (Muralidharan
& Mandrekar, 2013), the upregulation of an additional six gene
sets related to the innate immune response may further indicate
increased cellular stress. The acidification in Experiment 1 also
resulted in the downregulation of broad categories of gene sets
related to basic neurological functions, behaviour, and cognition,
which supports the hypothesis that acidification can lead to be-
havioural impairment in this species though, as mentioned above,
the specific mechanisms through which OA induced alterations in
neurobiology might impact fish behaviour are still not well under-
stood (Tresguerres & Hamilton, 2017).
Upregulated in pH 7.30 treatment Downregulated in pH 7.30 treatment
Categorical cluster
Number of gene sets
in each cluster Categorical cluster
Number of gene sets
in each cluster
MYC target s version 1 (Hallmark) 1 Neuron apoptotic process 2
MYC target s version 2 (Hallmark) 1 Regulation of amyloid precursor protein
catabolic process
2
Platelet morphogenesis 1Regulation of neurotransmitter receptor
activity
2
Positive regulation mitotic cell cycle 1Regulation of small GTPase- mediated signal
transduction
2
Receptor signalling pathway via STAT 1 Response to catecholamine 2
rRNA binding 1Synaptic vesicle recycling 2
Sarcolemma 1Vesicle docking 2
Sarcoplasm 1Amyotrophic lateral sclerosis 1
Starch & sucrose metabolism 1Anchored component of membrane 1
Viral myocarditis 1Cyclic nucleotide- mediated signalling 1
developmental maturation 1
Endocytosis 1
Gap junction 1
Genes upregulated by KRAS activation 1
Kinesin binding 1
Long- term potentiation 1
Neuron migration 1
Perinuclear region of c ytoplasm 1
Phosphoprotein binding 1
Phosphoric diester hydrolase activity 1
Protein serine threonine kinase inhibitor
activity
1
Regulation of neuron differentiation 1
Renal system process 1
Tau protein binding 1
Note: Enriched gene set s (GO, KEGG, hallmark) were clustered by similarity using the AutoA nnotate and clusterMaker2 applications for the
Cytoscape software platform. Clusters were then manually examined and name d. See Table S12 for the full list of enriched gene sets in this
experiment.
TABLE 4 (Continued)
|
11
TOY eT al.
TABLE 5 Summary of upregulated and downregulated gene set clusters in Experiment 2 (comparison of static treatments
Upregulated in pH 7.70 treatment Downregulated in pH 7.70 treatment
Categorical cluster
Number of gene sets in
each cluster Categorical cluster
Number of gene sets
in each cluster
RNA processing & splicing, histone
methyltransferase complex
26 Immune response 34
Epigenetic regulation of gene expression
and chromatin organization
8Lymphocyte proliferation, differentiation and
activation
26
DNA repair, recombination and replication 7 Secretory granule and myeloid leucocyte
mediated immunity
13
mRNA export from nucleus 6 Endothelial cell migration and blood vessel
morphogenesis
10
E- box binding 4JAK– STAT signalling pathway 9
Ubiquitin- mediated proteolysis 4Neuropeptide/G protein- coupled receptor
signalling pathway
7
Ubiquitin ligase complex 4Cellular ion homeostasis 6
RNA phosphodiester bond hydrolysis 3Positive regulation of MAPK cascade 6
Gene silencing 2Cell– cell junction assembly 5
Nuclear speck 2 Developmental growt h involved in
morphogenesis
5
A band 1 Regulation of cytoskeleton and
supramolecular fibre organization
5
Cell cor tex region 1 Wound healing and regulation of body fluid
levels
5
Inositol phosphate- mediated signalling 1 Leucocyte migration and regulation of
chemotaxis
4
Regulation of long- term synaptic
potentiation
1External side of plasma membrane 3
Single- stranded RNA binding 1Leading edge membrane 3
Structural constituent of cytoskeleton 1Plasma membrane signalling receptor complex 3
Transcription coregulator activity 1Positive regulation of phagocytosis 3
Protein complex involved in cell adhesion and
integrin- mediated signalling pathway
3
Regulation of cytokine production 3
Cilium movement and cell motility 2
Collagen- containing extracellular matrix 2
Endocytic vesicle 2
Positive regulation of cell- substrate adhesion 2
Receptor- mediated endocytosis 2
Regulation of peptidyl- tyrosine
phosphorylation
2
Guanyl nucleotide binding 1
Calcium ion binding 1
Allograft rejection 1
Membrane microdomain 1
Complement system 1
Positive regulation of cell population
proliferation
1
Smooth muscle contraction 1
Superoxide metabolic process 1
Response to organophosphorus 1
(Continues)
12
|
TOY eT a l.
In Experiment 2, additional gene set s related to the regulation of
gene expression (including epigenetic regulation) were upregulated,
again indicating a systemic shift in gene expression and response
to cellular stress. We also found a unique downregulation of a large
number of gene sets related to immune response, which may in part
reflect the external conditions of this experiment, particularly the
unusually warm ambient temperatures. The combination of physi-
cal and chemical stressors may have led to the suppression of the
immune system in fish in the acidified treatment. Suppression or
dysregulation of immune function is a well- established response to
stress (Dhabhar, 2014), and heat stress- induced immunosuppres-
sion, specifically, has been noted across various animal systems
(Nardone et al., 2010).
The biological themes of the enriched gene sets in both experi-
ments are consistent with enriched categories identified in previous
studies in tropical reef fish (e.g., Schunter et al., 2016, 2018) and
salmon (Williams et al., 20 19). Interestingly, however, the pattern of
enrichment in E. jacksoni under acidif ied conditions is ge nerally oppo-
site to the enrichment pattern found by Schunter et al. (2018) when
comparing acute or developmentally (together: cis- generationally)
exposed spiny damselfish to control (untreated) individuals, but
closely resembles the pattern of gene set enrichment that Schunter
et al. found when comparing transgenerationally exposed A. poly-
acanthus to those that were developmentally exposed to acidified
conditions (not the control treatment). The contrast of our results
may reflect the transgenerational and evolutionary exposure history
of E. jacksoni populations to naturally acidic environments. While
A. polyacanthus on coral reefs may experience diurnal pCO2 fluctua-
tions on the scale of ±50– 150 μatm (Schunter et al., 2021 and refer-
ences therein), E. jacksoni in upwelling regions are likely to regularly
experience prolonged increases in pCO2 (days to weeks) from as low
as ~300 to >1000 μatm (Chavez et al., 2018; Donham et al., 2022).
Upregulated in pH 7.70 treatment Downregulated in pH 7.70 treatment
Categorical cluster
Number of gene sets in
each cluster Categorical cluster
Number of gene sets
in each cluster
Ras protein signal transduction 1
Response to dopamine 1
Inflammatory response 1
Odontogenesis 1
Coagulation 1
Leucocyte transendothelial migration 1
Pigment granule 1
Cell adhesion molecule binding 1
Ciliary plasm 1
Note: Enriched gene set s (GO, KEGG, hallmark) were clustered by similarity using the AutoA nnotate and clusterMaker2 applications for the
Cytoscape software platform. Clusters were then manually examined and name d. See Table S13 for the full list of enriched gene sets in this
experiment.
TABLE 5 (Continued)
FIGURE 5 Overlapping enriched gene
sets across both experiments. “Up” and
“down” refer to gene sets that were up- or
downregulated in the lower pH treatment
relative to the higher pH treatment in
each experiment (i.e., pH 7.85 treatments
are treated as baseline in both cases).
Only static treatments are included for
Experiment 2.
|
13
TOY eT al.
Those living in estuaries can experience even greater shifts in car-
bonate chemistry over even shorter timescales (Duarte et al., 2013;
Hofmann et al., 2011). It is therefore likely that the juvenile E. jack-
soni in our experiments were transgenerationally exposed to acidi-
fied conditions in situ. Additionally, while both species lack a pelagic
larval stage, A. polyacanthus is a substrate spawner and E. jacksoni is
a live- bearing species. This means that the E. jacksoni used for this
experiment may also have developmentally experienced their moth-
ers' natural environmental exposures prior to their birth. Because of
this potential in- situ transgenerational exposure, our experimental
design may be more comparable to the transgenerational treatment
used by Schunter et al. (2018).
As mentioned above, it has been suggested that the cause of
previously documented acidification- induced behavioural changes
in fish may not only be due to a reversal of electrochemical gradi-
ents that flip the nature of GABAergic neurons from inhibitory to
excitator y (Nilsson et al., 2012), but also due to a positive feedback
cycle that may develop as a response to this increase in excitatory
activity in the brain (Schunter et al., 2018, 2019). This proposed
response consists of an increase in GABA release and in the abun-
dance of GABAA receptors, which under nonacidified conditions
would serve to reduce overactivity in the brain, but under acidi-
fied conditions probably act to exacerbate the overactivity. Some
previous studies in fish have seen changes in expression consistent
with this response, such as increased expression of GABAA recep-
tor subunits and transporter genes (e.g., Lai et al., 2016; Schunter
et al., 2016, 2018). However, the fish in Experiment 1 showed the
opposite response in GABA- related genes. In Experiment 1, GABA A
receptor subunit isoforms α (1– 6) , β (1– 3) , γ (1– 3), ρ (2), and π were
TABLE 6 Overlapping enriched gene sets between high pH
versus low pH comparisons in Experiments 1 and 2 (upregulated vs.
downregulated gene sets in both experiments)
Upregulated in both experiments
Downregulated in both
experiments
GOBP ribonucleoprotein complex
biogenesis
GOBP MAPK cascade
GOBP ncRNA processing GOCC side of membrane
GOCC ribonucleoprotein complex GOBP receptor mediated
endocytosis
GOMF catalytic activity acting
on RNA
GOCC cell surface
GOBP translational termination GOBP positive regulation of
protein kinase activity
GOBP RNA export from nucleus GOBP positive regulation of
MAPK cascade
GOBP RNA processing GOCC cell leading edge
GOBP RNA phosphodiester bond
hydrolysis
GOBP cell– cell junction
organization
GOBP mRNA export from nucleus GOBP cell– cell adhesion
GOBP nuclear export GOBP endocytosis
GOBP mRNA metabolic process GOBP exocytosis
GOCC U2 type spliceosomal
complex
GOBP cell– cell junction
assembly
GOBP RNA 3′- end processing GOBP cell growth
GOBP nucleic acid
phosphodiester bond
hydrolysis
GOBP taxis
KEGG spliceosome GOCC secretory granule
membrane
GOCC transferase complex GOBP regulation of
anatomical structure
morphogenesis
GOBP protein modific ation by
small protein conjugation
GOCC secretory vesicle
GOBP RNA localization GOMF neuropeptide receptor
activity
GOCC spliceosomal complex GOBP cell junction assembly
GOBP protein modific ation by
small protein conjugation or
removal
KEGG cell adhesion molecules
cams
GOBP RNA splicing GOCC plasma membrane
protein complex
GOBP mRNA processing GOMF calcium ion binding
GOCC nuclear protein- containing
complex
GOCC cell projection
membrane
GOCC intracellular protein-
containing complex
GOCC plasma membrane
signalling receptor
complex
GOBP cell– cell adhesion
via plasma membrane
adhesion molecules
GOBP developmental growth
involved in morphogenesis
(Continues)
Upregulated in both experiments
Downregulated in both
experiments
GOBP developmental cell
growth
GOBP neuropeptide signalling
pathway
GOBP adenylate cyclase
inhibiting G protein-
coupled receptor
signalling pathway
GOCC leading edge
membrane
GOCC vesicle membrane
GOMF G protein- coupled
receptor activity
GOCC receptor complex
KEGG neuroactive ligand
receptor interaction
GOBP G protein- coupled
receptor signalling
pathway
GOMF molecular transducer
activity
TABLE 6 (Continued)
14
|
TOY eT a l.
all downregulated in the pH 7.30 treatment, along with many other
GABA signalling genes, including glut amate decarboxylases gad1
and gad2 (L ai et al., 2016) and gabarapl2. Interestingly, a similar gen-
eral downregulation of GABAergic signalling pathways was recently
noted in A. polyacanthus at CO2 seeps, but not in other reef fish spe-
cies (Kang et al., 2022). A study on Pacific coho salmon (Williams
et al., 2019) also found no changes in GABAA receptor subunit ex-
pression in the olfactory bulb under increased pCO2, but did find
an increase in the expression of a GABAB receptor subunit ( gabbr2),
which was instead downregulated in E. jacksoni in our Experiment
1. Williams et al. (2019) also found significant changes in the ex-
pression of other genes associated with GABA signalling, includ-
ing downregulation of the slc6 a13 gene involved in GABA uptake,
which we also saw downregulated in E. jacksoni in Experiment 1
(see Table S17 for all differentially expressed GABA- related genes).
These divergent responses between E. jacksoni, Pacific salmon, and
tropical fish species could represent species- specific adaptation to
differing environmental conditions. In the case of E. jacksoni, which
frequently experiences periods of high pCO2, the downregulation
of GABA- related genes under high pCO2 may be an adaptation that
prevents or interrupts the excitator y positive feedback cycle pro-
posed by Schunter et al. (2019). Previous studies have also noted
opposite responses in gene expression across species of the same
taxa (Kang et al., 2022; Strader et al., 2020), and even across popu-
lations of the same species (Goncalves et al., 2016), but the extent
of the role that transgenerational effects play in creating divergent
responses is still unclear (but see Goncalves et al., 2016; Schunter
et al., 2018). Importantly, however, our seemingly species- specific
results may indicate that E. jacksoni is preadapted to acidified con-
ditions, whether through long- term local adaptation or transgener-
ational plasticity. Because of its limited adult dispersal and lack of a
pelagic larval phase, E. jacksoni may be more likely to be genetically
adapted to local conditions than other species (Warner, 1997), and
its live- bearing reproduction may also facilitate adaptation through
maternal effects. Kang et al. (2022) recently proposed a similar hy-
pothesis to explain why A. polyacanthus (which also lacks a pelagic
larval stage) differed from other co- occurring damselfish species in
its molecular response to elevated pCO2.
Interestingly, the response of GABA- related genes to acidification
varied between Experiments 1 and 2 (which used different levels of
acidification). In response to the more moderate static acidification
in Experiment 2, E. jacksoni showed an upregulation of two subunits
of the GABA A receptor ( gabra6 and gabrb3), which were instead
downregulated in Experiment 1. Interestingly, the gabra6 subunit is
also upregulated in spiny damselfish transgenerationally exposed to
high pCO2 when compared to those that were only developmentally
exposed, an effect opposite to that seen in the expression of other
GABAA subunits in the same experiment (Schunter et al., 2018). No
other GABA- related genes were significantly affected by this treat-
ment. In Experiment 1, the greater magnitude change in pH resulted
in an opposite and much broader response of GABA- related genes.
These conflicting responses in the transcription of GABAA receptor
subunits and other GABAA- related genes indicate that in addition to
varying across species, the response of the GABA signalling pathway
to acidification/high pCO2 may also depend on the magnitude of the
environmental change. Further study is needed to determine how
the divergent transcriptomic response of E. jacksoni seen in our ex-
periments translates to behaviour and overall fitness, how the mag-
nitude of any emergent effects compare to those observed in other
species, and the role transgenerational exposure plays in E. jacksoni
response to acidification.
In both Experiments 1 and 2, an additional group of gene sets
related to muscle tissue were identified as enriched in the acidified
treatment. In Experiment 1, this included the upregulation of gene
sets related to muscle development, contraction, and adaptation
and muscle cell components, as well as the downregulation of the
vascular smooth muscle contraction GO gene set. In E xperiment 2,
the A band GO gene set was upregulated, while the smooth muscle
FIGURE 6 Overlapping enriched
gene sets across static versus variable
comparisons in Experiment 2. “Up” and
“down” refer to gene sets that were up- or
downregulated in the “variable” treatment
relative to the “static” treatment for a
given pH level (7.85 or 7.70; i.e., static
treatments are treated as baseline).
|
15
TOY eT al.
contraction gene set was downregulated. While smooth muscle is
present in blood vessels in the brain, it is possible that the identi-
fication of some of these pathways (such as those related to stri-
ated muscle tissue) as enriched is in part due to the misannotation
of genes in this nonmodel species to orthologous reference genes.
Alternatively, because our transcriptome was assembled using both
brain and muscle tissue, it is possible that some brain transcripts
were misaligned to muscle- exclusive reference transcripts during
differential expression analysis.
4.2 | Impacts of pH variability
Overall, we found that variability in pH moderated the differential
gene expression seen under static acidification. pH variability de-
creased the number of DEGs detected by the edgeR analysis be-
tween the pH 7.85 and pH 7.70 treatments in Experiment 2 (from
159 genes when treatments were static to nine genes when both
treatments were variable). This aligns with two previous studies
that found that effects of pH on fish gene expression and behaviour
were diminished by the incorporation of diel pH fluctuations (Jarrold
et al., 2017; Schunter et al., 2021).
Functional enrichment analysis revealed many up- and downreg-
ulated gene sets between static and variable treatments at each pH
level, though there were more enriched gene sets in the more mod-
erate pH 7.85 comparison (329 gene sets) than the pH 7.70 compar-
ison (139 gene sets). This difference may represent an acidification
threshold nearer to the 7.85 treatment, where the majority of tran-
scriptional adaptation to acidification is activated. Such a threshold
effect in gene expression patterns has also been observed in the gill
tissue of spider crabs exposed to two levels of acidification (Harms
et al., 2014), as well as in the muscle tissue of blue rockfish (Sebastes
mystinus; Hamilton et al., 2017), and thresholds in OA response have
been noted across taxa (Bednaršek et al., 2021; Castillo et al., 2014;
Wittmann & Pörtner, 2013). Interestingly, although 33 gene sets
were commonly enriched across the pH 7.85 and 7.70 static- variable
comparisons, the majority of them (30) were enriched in opposite
directions depending on the pH level, with variability at pH 7.85
eliciting a directional response mirroring that of static acidification,
and variability at pH 7.70 eliciting the opposite response (Figure 6;
Table S16). For example, at the 7.85 pH level, variability led to a
downregulation of gene sets related to morphogenesis, develop-
ment, cell differentiation, exocy tosis, cell– cell adhesion, molecular
transducer activity, and leucocyte mediated immunity, while vari-
ability at pH 7.70 led to upregulation in these gene sets compared to
the static treatment. These contrasting responses indicate that pH
variability can have opposing effects on brain physiology depending
on the underlying mean pH level. This interactive effect of acidi-
fication and variability may again reflect a threshold in the neural
response of fish to acidification. It may be that at more moderate
pH levels, variability exacerbates the negative effects of acidifica-
tion by temporarily dropping the pH further below the average, but
under more extreme acidification, perhaps past a biological tipping
point, any negative effects of further acidification introduced by
temporary oscillations may be outweighed by the temporary relief
provided by the upswing of the oscillations above the mean pH.
It is impor tant to note that our interpretation of these re-
sults could be limited by the scope of our experimental design. In
Experiment 2, we sampled tissue from individuals in each treatment
when the variable treatments were increasing in pH and intersecting
their corresponding static treatments. While this design keeps the
pH at the time of sampling consistent between the static and vari-
able treatments, it assesses expression at only a single time point,
and therefore does not account for likely divergent expression pat-
terns at dif ferent positions in the pH cycles of the variable treat-
ments. Additional experiments are necessar y to determine if and
how gene expression differs in E. jacksoni depending on the trajec-
tory and value of the pH at the time of sampling.
4.3 | Interindividual variability in gene expression
A particularly striking finding from our experiments is the observa-
tion that gene expression variability across individuals was greater
in the variable pH treatments of Experiment 2 than in the static
treatments (Figure 4). This pattern indicates that the environmen-
tal variability introduced by the pH oscillations may be revealing
significant “cryptic variation” (Rutherford, 2000, 2003; Ruther ford
& Lindquist, 1998) in the transcriptomic response of E. jacksoni to
acidification. In the context of climate change, such phenotypic
variation, if beneficial and heritable, could represent potential adap-
tive variation on which selection may ac t, allowing populations to
adapt to ongoing changes in environmental conditions (Rutherford
& Lindquist, 1998; Rutherford, 2000, 2003; Queitsch et al., 2002;
reviewed in Ghalambor et al., 2007).
Patterns of expression across individuals within static treat-
ments, and across functionally related genes within individuals,
were notably consistent. This consistency provides evidence of a
conserved stress response as described by Kültz (2005), and may
again reflect a biochemical “switch” type response, activated at a
certain environmental threshold. This idea is further supported in
Experiment 2 by the similarity of expression profiles of some indi-
viduals in the pH 7.85 variable treatment to the expression profiles
exhibited by those in the pH 7.85 static treatment, while others in
the variable treatment exhibited expression profiles similar to those
in the pH 7.70 static treatment (Figure S5; Figure 3).
Because we did not use the whole brain, and instead arbitrarily
subsampled brain tissue from each individual, some of the interin-
dividual variability in expression profiles may be the result of vari-
ability in the exact section(s) of the brain that was sampled for each
individual. Conversely, it is possible that this sampling method could
introduce treatment- level bias in the brain region sampled that could
lead to misleading signals of differential expression between treat-
ments. However, expression profiles within static treatments were
remarkably consistent across individuals, especially in Experiment
1 (Figure 2), indicating a low probability of sampling bias, and all
16
|
TOY eT a l.
individuals were subsampled in an arbitrary manner by a single re-
searcher for each experiment. We therefore maintain that alterna-
tive sampling methods would have been unlikely to change the major
patterns and conclusions presented here.
5 | CONCLUSIONS
Overall, our results indicate that both acidification and pH/pCO2
variability can have significant impacts on the brain gene expres-
sion of a nearshore temperate fish species. Given recent debate
regarding the generality of neurological impacts of OA on marine
fish (Clark et al., 2020a, 2020b; Munday et al., 2020), our study
provides evidence of neurological impacts, even in a species with
a high likelihood for local adaptation to naturally low pH/pCO2. We
found a significant effect of acidification on global gene expression
in E. jacksoni brain tissue, and that the transcriptomic response was
similar to a previous experiment that compared transgeneration-
ally exposed tropical damselfish to individuals that were develop-
mentally exposed (Schunter et al., 2018). These results suggest that
the E. jacksoni in our experiments were exposed to ecologically rel-
evant pH/pCO2 variability in situ, which may have influenced their
response to acidification in the laboratory. Additionally, our results
demonstrate that the incorporation of upwelling- scale pH variability
into acidification treatments has a substantial impact on the number
of DE genes detected between moderate and low levels of acidi-
fication, indicating that temporal pH variability can moderate the
impacts of acidification. Interestingly, we also found that the direc-
tion of the ef fect of variability on gene expression in certain genes
depended on the degree of acidification. These opposing patterns
of gene expression indicate that the impact of pH variability on fish
brain physiology may be context- dependent, perhaps ser ving as an
additional stressor at more moderate levels of acidification, but as
an ameliorating factor when the mean pH is more extreme. Finally,
we observed significant variation in gene expression across individu-
als, and found that upwelling- scale pH variability revealed additional
cryptic phenotypic variation. This finding indicates that studies em-
ploying only static treatments may underestimate standing genetic
variation in traits related to the response of fish to acidification. This
cryptic variation may provide additional genetic variation on which
selection may act and therefore increase the likelihood of successful
adaptation of fish populations to acidification. In summary, our re-
sults emphasize the importance of considering environmental vari-
ability in global change experiment s and demonstrate that a species
with an evolutionary history of exposure to acidified and variable
conditions exhibits a distinctive transcriptomic response in gene
sets similar to those affected in species that have shown behavioural
impairment.
AUTHOR CONTRIBUTIONS
JAT, KJK, GB, CAL, and YT conceptualized and designed the experi-
ments. JAT and YT designed and tested the pH- manipulation system.
GB carried out and led the sequencing project for E xperiment 1.
JAT carried out Experiment 2 and led the sequencing project with
laboratory assistance from CAL. GCL led the reference genome se-
quencing project. JAT conducted all data processing and analyses
with input from KJK, GB, CAL, and YT. JAT drafted the paper. All
authors reviewed and edited the drafts.
CONFLICT OF INTEREST
The authors declare no competing interests.
ACKNOWLEDGEMENTS
Work was funded by a fellowship from the David and Lucile Packard
Foundation to KJK, as well as a Packard Endowment Award ad-
ministered through UCSC to GB, KJK, CAL and YT, and a UCSC
Committee on Research (COR) grant to GB. YT was also supported
by the Packard Foundation. We would like to thank the many labo-
ratory members, staff, and volunteers who helped carry out these
experiments, including Sarah Lummis, Tye Kindinger, Emily Donham,
Evan O'Brien, Hector Alvarado, Jim Freed, Melissa Gutterman,
Savannah Mangold, Cassandra Powell, Nicole Lenoski, Jake Cline,
Jacoby Baker, Eric Garcia, Nicholas Toy, Joseph Warren, Nate
Moore, and Randolf Skrovan. We also thank Malin Pinsky and Pete
Raimondi for insightful comments, discussion, and assistance with
statistical analyses.
DATA AVAIL ABILI TY STATEMENT
The raw data supporting the conclusions of this article are available
through online data repositories. Raw sequence data are deposited
in the SRA (BioProject PRJNA757398). The annotated E. jacksoni
transcriptome assembly (FASTA file) and environmental data from
both experiments are available on Dr yad (https://doi.org/10.7291/
D19H5 G). An alternatively trimmed version of the transcriptome
assembly is also deposited in the TSA (BioProject PRJNA757398,
accession GJIV00000000). The scaffold- level reference genome
assembly for E. jacksoni is also available on GenBank (BioProject
PRJNA810428, accession JALAZG000000000). Scripts for read
processing and data analyses are available at https://github.com/
jt oy7.
ORCID
Jason A. Toy https://orcid.org/0000-0002-2126-7826
Kristy J. Kroeker https://orcid.org/0000-0002-5766-1999
Cheryl A. Logan https://orcid.org/0000-0001-7639-4956
Yuichiro Takeshita https://orcid.org/0000-0003-1824-8517
Gary C. Longo https://orcid.org/0000-0002-4282-9111
Giacomo Bernardi https://orcid.org/0000-0002-8249-4678
REFERENCES
Altschul, S. F., Gish, W., Miller, W., Myers, E. W., & Lipman, D. J. (1990).
Basic loc al alignment search tool. Journal of Molecular Biology,
215(3), 403– 410.
Anderson, M. J. (2001). A new method for non- parametric multivariate
analysis of variance. Austral Ecology, 26 (1), 3 2– 46.
|
17
TOY eT al.
Anderson, M. J. (2017). Permutational multivariate analysis of variance
(PERMAN OVA). Wiley StatsRef: Statistics Reference Online. ht t p s: //
doi .org/10.10 02/97811 18445 112.stat0 7841
Andrews, S. (2010). FastQC: a quality control tool for high throughput
sequence data [Online]. http://www.bioin forma tics.babra ham.
ac.uk/proje cts/fastq c/
Bednar šek, N., Calosi, P., Feely, R. A., Ambrose, R ., Byrne, M ., Chan,
K. Y. K ., Dupont, S., Padilla- Gamiño, J. L ., Spicer, J. I., Kessouri,
F., Roethler, M., Sutula, M., & Weisberg, S. B. (2021). Synthesis
of thresholds of ocean acidification impact s on echino-
derms. Frontiers in Marine Science, 8. https://doi.org/10.3389/
fmars.2021.602601
Bernardi, G. (2000). Barriers to gene flow in Embiotoca Jacksoni, a ma-
rine fish lacking a pelagic larval stage. Evolution, 54(1), 226. ht t p s : //
doi.org/10.1554/0014- 3820(2000)054[0226,BTGFI E]2.0.CO;2
Bernardi, G. (2005). Phylogeography and demography of sympatric sis-
ter surfperch species, Embiotoca Jacksoni and E. Lateralis along
the California coast: Historic al versus ecological factors. Evolution,
59(2), 386– 394. https://doi.org/10.1554/04- 367
Bolger, A. M., Lohse, M., & Usadel, B. (2014). Trimmomatic: A flexi-
ble trimmer for Illumina sequence data. Bioinformatics, 30 (15),
2114– 2120.
Castillo, K. D., Ries, J. B., Bruno, J. F., & Wes tfield, I. T. (2014). The reef-
building coral Siderastrea siderea exhibits parabolic responses to
ocean acidification and warming. Proceedings of the Royal Society B:
Biological Sciences, 281(1797), 20141856. http s://doi.or g/10.1098/
rspb.2014.1856
Chan, F., Barth, J. A., Blanchette, C. A., Byrne, R. H., Chavez, F., Cheriton,
O., Feely, R. A ., Friederich, G., Gaylord, B., Gouhier, T., Hacker, S.,
Hill, T., Hofmann, G., McManus, M. A., Menge, B. A., Nielsen, K .
J., Russell, A., Sanford, E., Sevadjian, J., & Washburn, L. (2017).
Persistent spatial structuring of coastal ocean acidification in the
California current s ystem. Scientific Reports, 7(1), 1– 7. ht t p s : //d o i.
org /10.1038/s 4159 8- 017- 02777 - y
Chavez, F. P., Sevadjian, J., Wahl, C., Friederich, J., & Friederich, G. E.
(2018). Measurements of pCO2 and pH from an autonomous
surface vehicle in a coastal upwelling system. Deep Sea Research
Part II: Topical Studies in Oceanography, 151, 137– 146. h t t p s: // do i .
org/10.1016/J.DSR2.2017.01.001
Clark, T. D., Raby, G. D., Roche, D. G., Binning, S. A., Speers- Roesch, B.,
Jutfelt, F., & Sundin, J. (2020a). Ocean acidification does not im-
pair the behaviour of coral reef fishes. Nature, 577(7790), 370– 375.
https://doi.org/10.1038/s4158 6- 019- 1903- y
Clark, T. D., Raby, G. D., Roche, D. G., Binning, S. A., Speers- Roesch, B.,
Jutfelt, F., & Sundin, J. (2020b). Reply to: Methods matter in re-
peating ocean acidification studies. Nature, 586(7830), E25– E27.
https://doi.org/10.1038/s4158 6- 020- 2804- 9
Cline, A. J., Hamilton, S. L ., & Logan, C. A. (2020). Effect s of multiple
climate change stressors on gene expression in blue rockfish
(Sebastes mystinus). Comparative Biochemistry and Physiology Part
A: Molecular & Integrative Physiology, 239, 110580.
DelValls, T. A., & Dickson, A. G. (1998). The pH of buffers based on 2- am
ino- 2- hydroxymethyl- 1,3- propanediol (‘tris’) in synthetic sea water.
Deep Sea Research Part I: Oceanographic Research Papers, 45(9),
1541– 1554. https://doi.org/10.1016/S0967 - 0637(98)00019 - 3
Dhabhar, F. S. (2014). Effects of stress on immune function: The good,
the bad, and the beautiful. Immunologic Research, 58(2), 193– 210.
https://doi.org/10.1007/s1202 6- 014- 8517- 0
Dick son, A. G., Sabine, C. L., & Christian, J. R. (2 007). Guide to best prac-
tices for ocean CO2 measurements. PICES Special Publication 3,
3(8), 191. htt ps://doi .org /10.1159/0 0033 1784
Domenici, P., Allan, B., McCormick, M. I., & Munday, P. L. (2012).
Elevated c arbon dioxide affect s behavioural lateralization in a coral
reef fish. Biology Letters, 8(1), 78– 81. https://doi.or g/10.1098/
rsbl.2011.0591
Donham, E. M., Strope, L. T., Hamilton, S. L., & Kroeker, K. J. (2022).
Coupled changes in pH, temperature, and dissolved oxygen impact
the physiology and ecology of herbivorous kelp forest grazer s.
Global Change Biology, 28, 3023– 3039. https: //doi .org/10.1111 /
gcb .1612 5
Duarte, C. M., Hendriks, I. E ., Moore, T. S., Olsen, Y. S., Steckbauer, A.,
Ramajo, L., Carstensen, J., Trotter, J. A., & McCulloch, M. (2013).
Is ocean acidification an Open- Ocean syndrome? Understanding
anthropogenic impacts on seawater pH. Estuaries and Coasts, 36(2),
22 1– 23 6. https://doi.org/10.1007/s1223 7- 013- 9594- 3
Gattuso, J.- P., Epitalon, J.- M., Lavigne, H., & Orr, J. (2021). seacarb:
Seawater Carbonate Chemistr y. R package version 3.3.0. h t t p s ://
CRAN.R-proje ct.org/packa ge=seacarb
G h a lamb o r, C . K ., Mc K ay , J. K . , Car roll , S. P. , & Re znic k , D. N. (2 0 07 ) . Ada ptiv e
versus non- adaptive phenotypic plasticity and the potential for
contemporary adaptation in new environments. Functional Ecology,
21(3), 394– 407. https://doi. org/10.1111/j.1365 - 2435. 2007.0128 3.x
Goncalves, P., Anderson, K., Thompson, E. L., Melwani, A ., Parker, L.m.,
Ross, P. M., & Raftos, D. A. (2016). Rapid transcriptional acclima-
tion following transgenerational exposure of oysters to ocean
acidification. Molecular Ecology, 25(19), 4836– 4849. ht t p s: //d o i .
org /10.1111/mec.1380 8
Griffiths, J. S., Pan, T.- C. F., & Kelly, M. W. (2019). Differential responses
to ocean acidification between populations of Balanophyllia ele-
gans corals from high and low upwelling environments. Molecular
Ecology, 28(11), 2715– 2730.
Gruber, N., Hauri, C., Lachkar, Z., Loher, D., Frolicher, T. L., & Plattner,
G.- K. (2012). Rapid progression of ocean acidification in the
California current s ystem. Science, 337(6091), 220– 223. h t t ps : //
doi.org/10.1126/scien ce.1216773
Haas, B. J., Papanicolaou, A., Yassour, M., Grabherr, M., Blood, P. D.,
Bowde n, J., Couge r, M. B., Eccles , D., Li, B., Li eber, M., Macma ne s,
M. D., Ott, M., Orvis, J., Pochet , N., Strozzi, F., Weeks, N.,
Westerman, R., William, T., Dewey, C. N., … Regev, A. (2013).
De novo transcript sequence reconstruction from RNA- seq
using the trinity platform for reference generation and analy-
sis. Nature Protocols, 8(8), 1494– 1512. ht tps: //doi. org/10 .1038/
nprot.2013.084
Hamilton, S. L., Logan, C. A., Fennie, H. W., Sogard, S. M., Barry, P.,
Makukhov, A. D., Tobosa, L. R., Boyer, K., Lovera, C. F., & Bernardi,
G. (2017). Species- specific responses of juvenile rockfish to el-
evated pCO2: From behavior to genomics. PLoS One, 12(1),
e0169670. https://doi.org/10.1371/journ al.pone.0169670
Hamilton, T. J., Holcombe, A., & Tresguerres, M. (2013). CO2-
induced ocean acidification increases anxiety in rock fish via
alteration of GABAA receptor functioning. Proceedings of the
Royal Societ y B: Biological Sciences, 281, 20132509. ht t ps : //d o i .
org /10.1121 /1.492 9899
Harms, L ., Frickenhaus, S., Schif fer, M., Mark, F. C., Storch, D., Held, C .,
Pörtner, H.- O., & Lucassen, M. (2014). Gene expression profiling
in gills of the great spider crab Hyas araneus in response to ocean
acidification and warming. BMC Genomics, 15(1), 789. ht t ps : //d o i .
org /10.1186/1471- 2164- 15- 789
Hauri, C ., Gruber, N., Vogt, M., Doney, S. C., Feely, R. A ., Lachkar,
Z., Leinweber, A., McDonnell, A. M. P., & Munnich, M. (2013).
Spatiotemporal variability and long- term trends of ocean acidifi-
cation in the Califor nia current system. Biogeosciences, 10 (1), 193–
216. https://doi.org/10.5194/bg- 10- 193- 2013
Heuer, R . M., & G rosell, M. (2014). Physiological impac ts of ele-
vated carbon dioxide and ocean acidification on fish. American
Journal of Physiology - Regulatory Integrative and Comparative
Physiolog y, 307(9), R1061– R108 4. htt ps://doi.o rg/10.1152/ajpre
gu.00064.2014
Hirsh, H. K., Nickols, K . J., Takeshita, Y., Traiger, S. B., Mucciarone, D. A.,
Monismith, S., & Dunbar, R. B. (2020). Drivers of biogeochemical
18
|
TOY eT a l.
variability in a Central California kelp Forest: Implications for local
amelioration of ocean acidification. Journal of Geophysical Research:
Oceans, 125(11), 1– 22. https://doi.org/10.1029/2020J C016320
Hixon, M. A. (1981). An experimental analysis of territoriality in the
California reef fish Embiotoca jacksoni (Embiotocidae). Copeia, 3,
653– 665. ht tps ://doi.org/10.2307/1444571
Hofmann, G. E., Smith, J. E., Johnson, K. S., Send, U., Levin, L. A., Micheli,
F., Paytan, A ., Price, N. N., Peterson, B., Takeshita, Y., Mat son, P.
G., de Crook, E., Kroeker, K. J., Gambi, M. C., Rivest, E. B., Frieder,
C. A., Yu, P. C., & Martz, T. R. (2011). High- frequency dynamic s of
ocean pH: A m ulti- ecos ystem comparison. PLoS One, 6(12), e28983.
https://doi.org/10.1371/journ al.pone.0028983
Jarrold, M. D., Humphrey, C., McCormick, M. I., & Munday, P. L. (2017).
Diel CO2 cycles reduce severity of behavioural abnormalities in
coral reef fish under ocean acidification. Scientific Reports, 7(1),
10153. https://doi.org/10.1038/s4159 8- 017- 10378 - y
Jarrold, M. D., & Munday, P. L. (2019). Diel CO2 cycles and parental
effec ts have similar benefits to growth of a coral reef fish under
ocean acidification. Biology Letters, 15(2), 20180724. h t tp s : //d o i.
org/10.1098/rsbl.2018.0724
Kanehisa, M., & Goto, S. (2000). KEGG: Kyoto encyclopedia of genes
and genomes. Nucleic Acids Research, 28(1), 27– 30. h t t p s :// do i .
org /10.1093/nar/28.1. 27
Kang, J., Nagelkerken, I., Rummer, J. L., Rodolfo- Metalpa, R., Munday, P.
L., Ravasi , T., & Schunter, C. (2022). Rapid evolution fuels tra nscrip-
tional plasticity to ocean acidification. Global Change Biology, 28(9),
3007– 3022. h ttp s://doi.org /10.1111/gcb.16119
Kapsenberg, L., Bockmon, E. E., Bresnahan, P. J., Kroeker, K. J., Gattuso,
J. P., & Mar tz, T. R. (2017). Advancing ocean acidification biology
using Dur afet® pH electrodes. Frontiers in Marine Science, 4, 1– 9.
https://doi.org/10.3389/fmars.2017.00321
Korotkevich, G., Sukhov, V., Budin, N., Shpak, B., Artyomov, M. N., &
Sergushichev, A. (2021). Fast gene set enrichment analysis. BioRxiv.
https://doi.org/10.1101/060012
Kroeker, K. J., Bell, L. E., Donham, E. M., Hoshijima, U., Lummis, S., Toy,
J. A., & Willis- Norton, E. (2020). Ecological change in dynamic en-
vironments: Accounting for temporal environmental variability in
studies of ocean change biology. Global Change Biology, 26(1), 54–
67. htt ps://doi.org/10.1111 /gcb.14 868
Kroeker, K. J., Kordas, R. L., Crim, R., Hendriks, I. E., Ramajo, L., Singh, G.
S. , Dua rt e, C . M., & Gat tuso, J.- P. (2013 ). Im pacts of ocean acidif ica-
tion on marine organisms: Quantifying sensitivities and interaction
with warming. Global Change Biology, 19(6), 1884– 1896. h t t ps : //d o i .
org /10.1111/gcb.12179
Kültz, D. (2005). Molecular and evolutionary basis of the cellular stress
response. Annual Review of Physiology, 67(1), 225– 257. h t t p s :// do i .
org/10.1146/annur ev.physi ol.67.040403.103635
Kwan, G. T., Hamilton, T. J., & Tresguerres, M. (2017). CO2- induced
ocean acidification does not affect individual or group b ehaviour
in a temperate damselfish CO 2 - induced ocean acidification does
not affect individual or group behaviour in a temperate damselfish.
Royal Societ y Open Science, 4(7), 170283. https://doi.org/10.1098/
rsos.170283
Lai, F., Fagernes, C. E., Jutfelt, F., & Nilsson, G. E. (2016). Expression of
genes involved in brain GABAergic neurotransmission in three-
spined stickleback exposed to near- future CO2. Conservation
Physiolog y, 4(1), 1– 15. https://doi.org/10.1093/conph ys/cow068
Langmead, B., Trapnell, C., Pop, M., & Salzberg, S. L. (2009). Ultrafast and
memor y- efficient alignment of shor t DNA sequences to the human
genome. Genome Biology, 10(3), 1– 10.
Laur, D. R., & Ebeling, A. W. (1983). Predator-prey relationships in surf-
perches. In D. L. G. Noakes, D. G. Lindquist, G. S. Helfman, J. A .
Ward (Eds), Predators and prey in fishes: Proceedings of the 3rd bien-
nial conference on the ethology and behavioral ecolog y of fishes, held
at Normal, Illinois, U.S.A., May 19– 22, 1981 (pp. 55– 67). Springer.
Li, B., & Dewey, C. N. (2011). RSEM: Accurate transcript quantifica-
tion from RNA- seq data with or without a reference genome.
Bioinformatics: The Impact of Accurate Quantification on Proteomic
and Genetic Analysis and Research, 12(323), 1– 16. ht t p s :// d oi .
org/10.1201/b16589
Liberzon, A., Birger, C., Thorvaldsdóttir, H., Ghandi, M., Mesirov, J. P.,
& Tamayo, P. (2015). The molecular signatures database Hallmark
gene set collection. Cell Systems, 1(6), 417– 425. h t tp s : //d o i .
org/10.1016/j.cels.2015.12.004
Lowe, A. T., Bos, J., & Ruesink, J. (2019). Ecosystem metabolism drives
pH variability and modulates long- term ocean acidification in
the Nor theast Pacific coast al ocean. Scientific Reports, 9(1), 1– 11.
htt ps://doi.o rg/10.10 38/s4159 8 - 018- 37764 - 4
Martinez Arbizu, P. (2020). pairwiseAdonis: Pairwise multilevel compar-
ison using adonis.
Martz, T. R., Conner y, J. G., & Johnson, K. S. (2010). Testing the
Honeywell Durafet ® for seawater pH applications. Limnology
and Oceanography: Methods, 8, 172– 184. https://doi.org/10.4319/
lom.2010.8.172
Munday, P. L., Dixson, D. L., Mccormick, M. I., Meekan, M., Ferrari, M . C.
O., Chivers, D. P., & Karl, D. (2010). Replenishment of fish popula-
tions is th reatened by oce an acidific ation. Proceedings of the National
Academy of S ciences of the United Sta tes of America, 107(29), 12930–
12934. https://doi.org/10.1073/pnas.10045 19107/ - /DCSup pleme
ntal.www.pnas.org /cgi/doi/10.1073/pnas.10045 19107
Munday, P. L., Dixson, D. L., Welch, M . J., Chivers, D. P., Domenici, P.,
Grosell, M., Heuer, R. M., Jones, G. P., McCormick, M. I., Meekan,
M., Nilsson, G . E., Ravasi, T., & Watson, S.- A . (2020). Methods mat-
ter in repeating ocean acidification studies. Nature, 586(783 0),
E20– E24. https://doi.org/10.1038/s4158 6- 020- 2803- x
Muralidharan, S., & Mandrekar, P. (2013). Cellular stress response and
innate immune signaling: Integrating pathways in host defense
and inflammation. Journal of Leukocyte Biology, 94(6), 1167– 1184.
https://doi.org/10.1189/jlb.0313153
Nagelkerken, I., & Munday, P. L. (2016). A nimal behaviour shapes the
ecological effec ts of ocean acidification and warming: Moving from
individual to community- level responses. Global Change Biology,
22(3), 974– 989. https://doi .org/10.1111/gcb.13167
Nardone, A., Ronchi, B., L acetera, N., Ranieri, M. S., & Bernabucci, U.
(2010). Effects of climate changes on animal production and sus-
tainability of livestock systems. Livestock Science, 130(1), 57– 69.
https://doi.org/10.1016/j.livsci.2010.02.011
Nilsson, G. E., Dixson, D. L., Domenici, P., McCormick, M. I., Sørensen,
C., Watson, S.- A., & Munday, P. L. (2012). Near- future carbon di-
oxide levels alter fish behaviour by interfering with neurotrans-
mitter function. Nature Climate Change, 2(3), 201– 204. h t t p s :// do i .
org /10.1038/n clim a te1352
Oksanen, J., Blanchet, F. G., Friendly, M., Kindt, R., Legendre, P., McGlinn,
D., Minchin, P. R., O' Hara, R. B., Si mpson, G. L., S olymos, P., Stevens,
M. H. H., Szoecs, E., & Wagner, H. (2020). vegan: Community ecol-
ogy package. https://cran.r- proje ct.org/packa ge=vegan
Pistevos, J. C. A., Nagelkerken, I., Rossi, T., Olmos, M., & Connell, S. D.
(2015). Ocean acidification and global warming impair shark hunt-
ing behaviour and growth. Scientific Reports, 5(1), 16293. h t t p s: //
doi.org/10.1038/srep1 6293
Queitsch, C., Sangs ter, T. A., & Lindquist, S. (2002). Hsp90 as a capacitor
of phenot ypic variation. Nature, 417(6889), 618– 624. h t t ps : //d o i .
org /10.1038/natur e749
R Core Team. (2021). R: A language and environment for statistical comput-
ing. R Foundation for Statistical Computing. https://www.r- proje
ct.org/
Robinson, M. D., McCarthy, D. J., & Smyth, G. K. (2010). edgeR: A
Bioconductor package for differential expression analysis of digital
gene expression data. Bioinformatics, 26(1), 139– 140. h t t ps : //d o i .
org/10.1093/bioin forma tics/btp616
|
19
TOY eT al.
Ruther ford, S. L. (2000). From genotype to phenotype: Buffering mech-
anisms and the storage of genetic information. BioEssays, 22(12),
1095– 1105.
Ruther ford, S. L. (2003). Between genot ype and phenotype: Protein
chaperones and evolvability. Nature Reviews Genetics, 4(4), 263–
274. htt ps://doi.o rg/10.10 38/nrg1041
Ruther ford, S. L., & Lindquist, S. (1998). Hsp90 as a capacitor for mor-
phological evolution. Nature, 396(6709), 336– 342. h t t p s :// do i .
org /10.1038/24550
Schunter, C., Jarrold, M. D., Munday, P. L., & Ravasi, T. (2021). Diel pCO2
fluctuations alter the molecular response of coral reef fishes to
ocean acidification conditions. Molecular Ecology, 30(20), 5105–
5118 . ht tps://doi.o rg/10 .1111/me c.16124
Schunter, C., Ravasi, T., Munday, P. L., & Nilsson, G. E. (2019). Neural
effec ts of elevated CO2 in fish may be amplified by a vicious cycle.
Conservation Physiology, 7(1), 1– 8. https://doi.org/10.1093/conph
ys/coz100
Schunter, C., Welch, M. J., Nilsson, G. E., Rummer, J. L., Munday, P. L .,
& Ravasi, T. (2018). An interplay between plasticity and parental
phenot ype determines impacts of ocean acidification on a reef fish.
Nature Ecology & Evolution, 2(2), 334– 342. https://doi. org/10 .1038/
s4155 9- 017- 0428- 8
Schunter, C., Welch, M. J., Ryu, T., Zhang, H., Berumen, M. L., Nilsson,
G. E., Munday, P. L., & Ravasi, T. (2016). Molecular signatures of
transgenerational response to ocean acidification in a species of
reef fish. Nature Climate Change, 6(11), 1014– 1018. h t t ps : //d o i .
org /10.1038/NCLIM ATE3087
Simão, F. A., Waterhouse, R . M., Ioannidis, P., Kriventseva, E. V.,
& Zdobnov, E. M. (2015). BUSCO: Assessing genome assem-
bly and annotation completeness with single- copy orthologs.
Bioinformatics, 31(19), 3210– 3212. https://doi.org/10.1093/bioin
forma tics/btv351
Strader, M. E., Wong, J. M ., & Hofmann, G. E. (2020). Ocean acidifica-
tion promotes broad transcriptomic responses in marine metazo-
ans: A liter ature survey. Frontiers in Zoolog y, 17(1), 7. ht t p s : //d o i.
org/10.1186/s1298 3- 020- 0350- 9
Subramanian, A., Tamayo, P., Mootha, V. K., Mukherjee, S., Ebert, B. L.,
Gillette, M. A., Paulovich, A ., Pomeroy, S. L., Golub, T. R., Lander, E.
S., & others. (2005). Gene set enrichment analysis: A knowledge-
based approach for interpreting genome- wide expression pro-
files. Proceedings of the National Academy of Sciences, 102(4 3),
15545– 15550.
Takeshita, Y., Frieder, C. A., Martz, T. R., Ballard, J. R., Feely, R. A., Kram,
S., Nam, S., Navarro, M. O., Price, N. N., & Smith, J. E. (2015).
Including high- frequency variabilit y in coastal ocean acidification
projections. Biogeosciences, 12(19), 5853– 5870. ht t p s : //d oi .
org/10.5194/bg- 12- 5853- 2015
The Gene Ontology Consortium. (2020). The gene ontology resource:
Enriching a GOld mine. Nucleic Acids Research, 49(D1), D325– D334.
https://doi .org/10.10 93/nar/gkaa1113
The UniProt Consortium. (2021). UniProt: The universal protein knowl-
edgebase in 2021. Nucleic Acids Research, 49(D1), D480– D489.
Trapnell, C ., Robert s, A., G off, L., Per tea, G., K im, D., Kelley, D. R., Pi mentel,
H., Salzberg, S. L., Rinn, J. L ., & Pachter, L. (2012). Differential gene
and transcript expression analysis of RNA- seq experiments with
TopHat and cufflinks. Nature Protocols, 7(3), 562– 578.
Tresguerres, M., & Hamilton, T. J. (2017). Acid– base physiology, neuro-
biology and behaviour in relation to CO2- induced ocean acidifica-
tion. Journal of Experimental Biology, 220(12), 2136– 2148. h t t ps : //
doi.org/10.1242/jeb.144113
Warner, R. R. (1997). Evolutionary ec olog y: How to reconc ile pelagic dis-
persal with local adaptation . Coral Reefs, 16(Suppl. 1), S115– S120.
https://doi.org/10.1007/s0033 80050247
Williams, C. R., Dittman, A. H., McElhany, P., Busch, D. S., Maher, M. T.,
Bammler, T. K., MacDonald, J. W., & Gallagher, E. P. (2019). Elevated
CO 2 impairs olfactory- mediated neural and behavioral responses
and gene expression in ocean- phase coho salmon (Oncorhynchus
kisutch). Global Change Biology, 25(3), 963– 977. h t t p s: //d o i .
org /10.1111/gcb.14532
Wittmann, A . C., & Pörtner, H.- O. (2013). Sensitivities of extant animal
taxa to ocean acidifi cation. Nature Climate Change, 3(11), 995– 1001.
htt ps://doi.o rg/10.10 38/ncl im ate1982
SUPPORTING INFORMATION
Additional supporting information can be found online in the
Suppor ting Information section at the end of this article.
How to cite this article: Toy, J. A., Kroeker, K. J., Logan, C. A.,
Takeshita, Y., Longo, G. C., & Bernardi, G. (2022). Upwelling-
level acidification and pH/pCO2 variability moderate effects
of ocean acidification on brain gene expression in the
temperate surfperch, Embiotoca jacksoni. Molecular Ecology,
00, 1–19. http s://doi.org/10.1111/mec .16611
