Calcifying algae maintain settlement cues to larval abalone following algal exposure to extreme ocean acidification

Ocean acidification (OA) increasingly threatens marine systems, and is especially harmful to calcifying organisms. One important question is whether OA will alter species interactions. Crustose coralline algae (CCA) provide space and chemical cues for larval settlement. CCA have shown strongly negative responses to OA in previous studies, including disruption of settlement cues to corals. In California, CCA provide cues for seven species of harvested, threatened, and endangered abalone. We exposed four common CCA genera and a crustose calcifying red algae, Peyssonnelia (collectively CCRA) from California to three pCO2 levels ranging from 419–2,013 µatm for four months. We then evaluated abalone (Haliotis rufescens) settlement under ambient conditions among the CCRA and non-algal controls that had been previously exposed to the pCO2 treatments. Abalone settlement and metamorphosis increased from 11% in the absence of CCRA to 45–69% when CCRA were present, with minor variation among CCRA genera. Though all CCRA genera reduced growth during exposure to increased pCO2, abalone settlement was unaffected by prior CCRA exposure to increased pCO2. Thus, we find no impacts of OA exposure history on CCRA provision of settlement cues. Additionally, there appears to be functional redundancy in genera of CCRA providing cues to abalone, which may further buffer OA effects.


Methods
Algal collection and preparation. We collected algae crusts from cobbles at 9 m depth in kelp forests in the Monterey Bay, California, in February 2013. CCRA covered cobbles are the primary abalone settlement habitat 31 . Fifteen cobbles (1000 to 4000 cm 3 ) were collected at Lovers Point (N36°37′ 32.65, W121°54′ 54.55) and 15 from the Hopkins Marine Reserve (HMR, N36°37′ 10.57, W121°54′ 11.61) and stored in running seawater at the HMR under ambient Monterey Bay temperature (~11.7 °C) and pH (~7.81) until early March 2013. Fluorescent bulbs provided light on a 12 hour cycle at ~15 µmol/m 2 /s, approximating average light levels measured at midday under kelp canopy at 9 m depth at HMR in October 2011 and March 2012 (LI-COR PAR measurement unit with an LI-193 spherical quantum sensor). Thin rock pieces (30-405 mm 2 ) covered with a single algal morphology were cut from the cobbles using a rock saw, maintained in running seawater, and examined under an Olympus SZX-ILLD100 dissecting scope to ensure that no invertebrates were attached or embedded. Each piece was assigned to one of five morphologically distinct groups based on: type of reproductive structure (conceptacle versus nemathecium), size, and distribution; surface growth patterns; growth margins; crust thickness; and hypothallus cell arrangement (for CCA only).
To identify each morphological group, we sequenced 5-10 representatives per group. After extracting total genomic DNA for CCRA 32,33 , we amplified partial sequences from two plastid encoded genes (rbcL and psbA) used extensively to distinguish genera and species of CCRA 32,34 . Amplification and sequencing protocols were those of 33 . Sequences were obtained from an ABI 3100 Genetic Analyzer at the University of North Carolina, Wilmington (see 32 ). Identifications were based on sequencing ~20% of the rbcL gene and matching with sequences from curated type specimens. Reference samples of each sequenced crust have been deposited in the University of North Carolina Herbarium (NCU) herbarium. Four of the five morphological groups were taxonomically consistent to genus, and these were used in the experiments.
We created thirty replicate vials per algal taxon on March 2, 2013 by epoxying 1-3 pieces of alga into clear polystyrene vials (47.75 mm diameter ×102.12 mm height, 147 mL) using Z-spar epoxy compound. The living algal surfaces were flush with the epoxy, ~20 mm above the vial bottom, with space to grow laterally to at least double the initial area. Vials were open at the top and submerged in seawater in an upright position at the Hopkins Marine Station (conditions described above). Crust area was measured from digital photographs (Canon Powershot S100) using Image J (http://rsbweb.nih.gov/ij/). On March 7, 2013 the vials were transported in cooled seawater to the Monterey Bay Aquarium Research Institute (MBARI).  36 and the worst-case representative carbon pathway (RCP 8.5) 37 . In the Monterey Bay, this "high" level of pCO 2 can occur during upwelling events. The extreme treatment (2,012.9 µatm; pH 7.31; Ω aragonite 0.49) exceeded both predictions for the year 2100 33 and levels found in nearshore waters during strong upwelling 38,39 . We report Ω aragonite because the solubility of high-Mg calcite is closer to aragonite than calcite 24 (see Table 1a for Ω calcite ).
The water was prepared by passing through a series of partially recirculating tanks. Oxygen and CO 2 were stripped by bubbling nitrogen gas through membrane contactors 40 . The low-O 2 , low-CO 2 water was the source water for 3 gas-controlled tanks where specified pH and water chemistry was maintained, continuously monitored 40 , and delivered at ~50 l*h −1 to flow-through plastic aquaria (56 × 30 × 25 cm) containing the algae vials.
One plastic flow-through aquaria was used for each pCO 2 treatment, in a temperature controlled room with no outside light. Ten vials per algal taxon were assigned haphazardly to each of the three pCO 2 levels along with 10 control vials (containing epoxy but no algae) and submerged vertically in the aquaria (top of vials under 15 cm of water) with vial tops open. Aquaria contained 4.2 L of water and water delivery was at 50 l*h −1 , so water in aquaria and vials would be replaced every ~5 minutes. Vials were systematically mixed in the aquaria and rotated weekly. The vials were kept in the pCO 2 treatments from March 7 to July 10, 2013. Interior and exterior walls of the vials, algal surfaces, and plastic aquaria were gently brushed weekly with a soft toothbrush to remove any accumulated diatoms which can reduce light levels and are normally removed by grazers. Algal surfaces were examined via digital photographs during the first month of the experiment, and under an Olympus dissecting microscope (40x magnification) immediately upon removal from MBARI, and there was no evidence of algal surface damage.
Sensors in the reservoir tank and each gas-controlled tank measured temperature, oxygen and carbon dioxide concentrations, and pH every 60 seconds (Table 1a). Oxygen and temperature were sensed using oxygen optodes (Aanderaa Inc., model 3835) immersed in the reservoir tank and in the three gas-controlled tanks, and pH was measured with Honeywell Durafet pH sensors. Temperature in the aquaria was maintained at a constant average of 14.4 °C (Table 1) because during upwelling events (when surface pH is naturally reduced) temperature remains low 38 . Two fluorescent bulbs ~0.5 m above each of the aquaria provided 13-15 µMol m −2 s −1 irradiance on a 12-hour cycle, measured at the experiment start and end using the LI-COR instrument. These levels mimicked average mid-day light levels at 9 m depth under kelp canopy at Hopkins Marine Station (measured in October of 2011 and March of 2012).
We also sampled water chemistry in the experimental aquaria three times monthly during the course of the experiment (March-July 2013) to measure salinity, total alkalinity, and dissolved inorganic content (DIC). On each collection date, nine water samples (three for each parameter/bin) were collected in 30 ml Borosilicate Glass serum bottles from beneath the surface of each bin. However, due to malfunction of water testing equipment for alkalinity and DIC, we were not able to measure the water samples until 4 months after the experiment terminated and data was unreliable. Therefore, in April 2014, we recreated the conditions of the experiment at MBARI using the same set points for the gas-controlled tanks, the same bins, and the same mass of coralline in each treatment bin. We let the system equilibrate for one week, then measured pH using a Shimadzu UV-1601 spectrophotometer, and re-sampled the water as described above (but with six replicates/parameter/pCO 2 treatment) for salinity, total alkalinity, and DIC. For samples used to measure alkalinity and DIC, 10.9 μl HgCl 2 was added to kill living organisms. Each bottle was sealed, stored in a dark refrigerator, and processed two days after collection. Salinity was measured with a YSI 3200 Conductivity instrument with a YSI 3252 cell, total alkalinity using a SI Analytics Titroline 6000 titrator, and DIC using a UIC Inc. Model 5015 CO2 Coulometer with a CM5230 Acidification module. We used the program CO2SYS to calculate pCO 2 and Ω using the data from April 2014, with pH from the spectrophotometer readings and total alkalinity measured from collected water. The pCO 2 values calculated were almost identical to the measures recorded within the gas-controlled tanks that fed water into experimental bins (Table 1b).
On July 10, vials were transported in cooled seawater to the Hopkins Marine Station for immediate use in larval settlement experiments. Lateral algal crust area was compared in before-and after-treatment photographs taken   Table 1. (a) Water chemistry set points (mean ± 1 SE) in gas-control tanks at MBARI, measured continuously every 60 seconds between March 7 and July 10, 2013. (b) Water chemistry as measured in the bins containing coralline algae and calculations of pCO 2 and Ω (from pH and total alkalinity). Salinity was 33.9 ± 0.05 ppm.
of individual vials (Canon Powershot S100 using Image J) as an indication of the effects of the pre-conditioning treatments. We do not report vertical growth because crusts were thin (≤0.2 mm), exhibited no visual change in vertical growth across treatments, and surface area is more important for maintenance of larval settlement space.
Abalone settlement. Red abalone larvae from the Cayucos Abalone Farm (Cayucos, CA) were shipped overnight to the Hopkins Marine Station in seawater in a sterile Nalgene container surrounded by ice packs. Larvae were spawned on July 3, 2013, shipped on July 9 th , and arrived on July 10, 2013. Because we were interested in how OA might alter the role of calcifying algae in species interactions, we exposed only algae crusts, not the larvae, to the different pCO 2 treatments. To test whether OA affected algae cues for settling larvae, 95-110 red abalone larvae (Haliotis rufescens, 7-day old) were added to each of 7 haphazardly selected vials per taxon from each pCO 2 treatment on July 10 th within hours of removal of algae from pCO 2 treatments and immersion in ambient seawater (~14.2 °C and pH ~7.82). Abalone larvae were placed in sterile petri dishes and were inspected under an Olympus 40x dissecting microscope: all were actively swimming with no apparent abnormalities. Vials were filled with 65 mL of seawater (described above) leaving a 1 cm air space in each vial and 95-110 larvae were pipetted into each vial. Vials were closed with watertight polyethylene (LDPE) caps and put into a flowing seawater table to maintain temperature at 14 °C, under lights providing irradiance of ~15 µMol/m 2 /s on a 12 hour cycle. After 24-hours the number of settled (metamorphosed), swimming (veliger stage), and dead larvae were counted in each vial under a microscope. These counts were repeated 48 hours after abalone insertion.
Statistical Analyses. All analyses were conducted in SYSTAT 13. Prior to conducting the pCO 2 treatments, we assessed the surface area of CCRA assigned to the 3 treatments using analysis of variance (ANOVA). We used pCO 2 treatments to pre-condition algae for later species interaction trials, which were multiple independent replicates. Data were normally distributed for algal area prior to treatment and no data transformation was applied. We then tested for differences in growth to judge the effect of pre-conditioning. Because data were skewed, we used a logarithmic (base 10) transformation (plus the constant 20 to make all data positive as a few specimens had slight decline in healthy surface area). We compared the change in algal surface area between treatments and genera using a 2-way ANOVA with percent change in algal surface area (or cover) as the response variable, and pCO 2 (3 levels) and algal genera (4 levels) as fixed, independent and orthogonal factors, and with the interaction term pH*genera. We then used a posthoc Tukey's Honestly Significant Difference test to evaluate pairwise differences in growth between pCO 2 levels and between algal genera.
To evaluate differences in abalone settlement across CCRA genera, we first ran an ANCOVA considering only the vials with CCRA (excluding non-CCRA control vials from each pCO 2 treatment) so that we could include algal area as a predictor variable. We used the proportion of abalone that settled within 24 hours as the response variable and the predictor variables: algal pCO 2 treatment (3 levels), algal genera (4 levels), algal surface area (as a covariate to account for any differences in the amount of algae in each vial), and the interaction between algal genera, pCO 2 treatment, and algal area. We then ran an ANOVA with the same response variable and prior algal pCO 2 treatment, substrate type (algal genera type or control vials with no algae), and the interaction between substrate type and prior algal pCO 2 treatment. In both cases, the number of abalone settled was normally distributed and had homogeneous variances, so we did not use a transformation. Following the above analyses, where results were significantly different, we used a posthoc Ryan-Einot-Gabriel-Welsch (REGW-Q) test to evaluate which pCO 2 or genera were significantly different from each other in inducing settlement. The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.

Results
Algal Species Identity. Four of five morphological groups of CCRA collected from cobbles in shallow kelp forests were consistently identified visually to genus (confirmed genetically), and these were used in the experiments. Three groups (each containing 2-3 closely related species) are common genera of CCA in the Northeast Pacific: Lithothamnion spp. (2 species), Lithophyllum spp. (3 species), and Leptophytum spp. (2 species; Fig. 1). The fourth group is a subtidal, lightly calcified aragonitic Peyssonnelia species (order Peyssonneliales) that may be undescribed (Fig. 1). These were the four most common genera on the cobbles collected from sites in the Monterey Bay. In addition, based on preliminary data from samples collected every 2.5 m on 30 m benthic transects in central California (13 transects from 2 sites in the Monterey Bay) and northern California (29 transects at 6 sites in Mendocino and Sonoma Counties), the 3 CCA genera (Liththamnion, Lithophyllum, and Leptophytum) are also common on reef substrates in kelp forests, accounting for 52% of the algal crust from transects in central California and 76% of crusts from transects in northern California. Peyssonnelia sp. was not found on any of the samples identified from transects and may be found only on cobbles rather than on rocky reefs (JKO, unpublished data).
Algal Growth (Response to pre-conditioning period). Initial mean algal area (per vial) did not differ between pCO 2 treatments (2-way ANOVA df = 2,6, F = 1.05, p = 0.35). The pre-conditioning of algae with elevated pCO 2 negatively affected algal growth with significantly greater growth (+32.6 mm 2 ) under normal pCO 2 conditions compared to the high and extreme pCO 2 conditions (+22.3 mm 2 and + 17.8 mm 2 respectively; Fig. 2, Tables 2a and 3a). Regardless of pCO 2 treatment, there was variation in growth among the CCRA genera, with one genus, Leptophytum spp., growing significantly more than the others (Fig. 2, Tables 2a and 3b). All algal tissue appeared healthy at the end of the experiment with little bleaching (<7% of samples and with <2.6% of surface area affected).
Abalone Settlement. The majority of abalone that settled (94%) did so within 24 hours, so we used the 24-hour numbers in analyses. We found no significant difference in the number of abalone settlers based on  . This data is shown to demonstrate that the pCO 2 treatments had an effect on the CCRA. Letters above bars denote significant differences in growth considering all four genera. variation in CCRA surface area in vials, which ranged from 147 to 405 mm 2 (Table 2b). Treatments that had no CCRA (control vials) had only 11% settlement compared to 48-69% in vials with CCRA (Table 2c and Fig. 3). There were significant differences in settlement rates between the four CCRA genera (p < 0.0001, Table 2c and Fig. 3): two genera, Lithothamnion spp. and Leptophytum spp., had higher settlement with 62 and 69% of larvae settling respectively while the other two genera, Lithophyllum spp. and Peyssonnelia sp, each had 48% of larvae settle (REGW-Q Test). To our knowledge, this is the first observation that Peyssonnelia may be an important settlement substrate for abalone. However, there was no change in settlement rates with prior pCO 2 treatments of algae (the interaction term between algal genera and prior algal pCO 2 treatment was not significant; Table 2c). Thus, while there was no significant difference in settlement associated with prior algal pCO 2 treatment, there were strongly significant associations with presence or absence of CCRA as well as between algal genera.

Discussion
One of the major unknowns in OA research is how species will respond to OA in a community context via species interactions [8][9][10] . For other environmental impacts like global warming, changes in species interactions have fundamental impacts on communities 41 . CCRA cover a high proportion of marine benthic substrates, are generally thought to be highly susceptible to OA 42 , and mediate multiple species interactions 43 . We tested the response of abalone larval settlement to CCRA that were preconditioned in OA treatments. Contrary to our hypothesis (that pre-conditioning CCRA with OA would disrupt larval settlement), we found that CCRA maintained their ability to induce larval abalone settlement despite prior exposure to strong OA (pCO 2 of up to 2,013 µatm) for 4 months, at least when the algae but not the larvae were exposed.
Response of CCRA to OA treatments. CCRA generally show diverse responses to experimentally increased pCO 2 (Supplementary Table 1), likely reflecting the high diversity within this group. Prior OA experiments have found strong negative effects such as reduced growth and tissue integrity as well as increased likelihood of dissolution and tissue necrosis, but also sometimes no effect of increased pCO 2 on growth 10 . While our ability to discuss the generality of changes in growth due to OA is constrained by the design of our experiment (e.g. one tank per algal pCO 2 treatment), the level of CCRA growth reduction we observed is generally less severe  Table 3. Posthoc Tukey's Honestly Signficant Difference (HSD) tests for algal growth following pCO 2 treatments for the model in Table 2a with algal growth as the response variable and CCRA genera and pCO 2 as predictor variables. Growth changes have been back transformed. (a) Growth differences by pCO 2 (b) growth differences by CCRA genera.
than reported elsewhere. In our experiment, after 4 months, algal growth declined by 48% in the extreme pCO 2 treatment (2,013 µatm). In four (of 6) prior laboratory and field studies in the extreme pCO 2 range (1,400-2,200 µatm), coralline growth or cover was reduced by 93-100% (Supplementary Table 1). In our high pCO 2 treatment (1,176 µatm), CCRA growth reduced by 32%, whereas in prior studies, cover or growth loss in similar pCO 2 ranges varied from 22-100%, and most (9 of 13 studies) found >50% loss (Supplementary Table 1). Even in comparatively mild increased pCO 2 treatments (600-850 µatm), prior studies have found cover or growth losses ranging from 15-100%, with 8 of 14 studies exhibiting >50% loss (Supplementary Table 1). Further, in our study, the fastest growing CCRA genus (Leptophytum spp.) maintained growth under high pCO 2 and only reduced growth under the extreme pCO 2 treatment (Fig. 2), indicating the potential for some taxa to resist at least moderate pH changes, which should be tested further.
CCRA genus-specific settlement cues to abalone. Unlike corals, which have high species specificity for CCA 15,19,44 , red abalone larvae responded to cues from three genera of CCA and from the lightly calcified aragonitic crust, Peyssonnelia sp. To our knowledge, this is the first demonstration of Peyssonnelia providing abalone settlement cues. In a study on corals, Peyssonnelia was not found to be a particularly inductive substrate 45 .
Although some CCRA genera in our study induced 1.4 times more settlement than others (Fig. 3), these differences pale when considering that there was >5 times more settlement when CCRA were present compared to when they were absent (10% compared to 48-69% settled). This finding is similar to the results of previous experiments where red abalone settled on multiple genera including the CCRA species Lithothamnium californicum and L. glaciale, Lithophyllum spp., Clathromorphum circumscriptum, and a CCRA non-coralline Hildenbrandia dawsonii, but did not settle when provided with foliose red, green, or brown algae 23 . There thus appears to be some functional redundancy among CCRA in providing settlement cues to red abalone. The functional redundancy in settlement cues found with red abalone is not necessarily the case for other abalone species (e.g. Haliotis laevigata), which show distinct settlement preferences among CCA species 46 .
Abalone settlement on CCRA exposed to OA. We found that CCRA maintain settlement cues after prolonged and extreme OA exposure. This finding contrasts with previous studies on coral larvae showing settlement declines ranging between 20-86% on OA-treated CCRA 25,26 , and showing settlement reductions at milder algal pCO 2 treatments (600-1,300 µatm). The lack of OA-induced changes in larval settlement with elevated pCO 2 in our experiment could be due to conditions not being extreme enough, short treatment times, or lack of power, but these scenarios are unlikely. Our OA treatments represented strong scenarios, with the most elevated treatment (2,013 µatm) set to higher pCO 2 than global 2100 predictions 2, 36 . Further, pCO 2 treatments had the expected negative effect on growth, with growth slowing significantly at high and extreme pCO 2 treatments, especially for the fastest growing genus (Leptophytum spp.) in the extreme pCO 2 treatment. The duration of the exposure of the algae to pCO 2 treatments in this study was 126 days, well within the range of other similar laboratory studies (mean 138 days, range from 14-420 days; Supplementary Table 1), although one prior study found that CCRA showed a much stronger negative response at 420 days than at 90 days 47 . In California, upwelling of high pCO 2 water lasts hours to days, not months, so our treatments were sufficient to evaluate potential effects of prolonged upwelling followed by the return of lower pCO 2 water. It should be noted that many impacts of OA on chemical communication are due to acid-base disturbances and subsequent impairment of neuronal ion channel function 48 or due to changes in protonation of chemical signal compounds 49 that lead to disrupted chemical signaling. Both types of effects would only be visible during direct and potentially prolonged exposure of both abalone larvae and CCRA to ocean acidification during settlement assays. However, our study focused on potential changes to CCRA provision of cues, rather than cue reception by larvae.
Our findings indicate that under expected future changes in pH in the California current system, cues to larval settlers like abalone that rely on CCRA may be retained following the cessation of upwelling and high pCO 2 . We found that several CCRA genera induced settlement (regardless of OA exposure history), suggesting functional redundancy that may provide an additional buffer against the effects of OA on loss of benthic coverage of any one coralline taxon. There is also evidence that the presence of algae, rather than percent cover, is sufficient for settlement 50 . In this experiment, there was no change in abalone settlement with algal surface area varying between 147 and 405 mm 2 . Direct contact by the larvae with the inducing algal surface is necessary for induction 23 , so at least locating CCRA is a requirement for abalone settlement, and this should be influenced by CCRA surface area. Nonetheless, reliance on presence rather than area will further buffer against the effects of slowed CCRA growth under future OA scenarios.

Conclusions
Upwelling regions (like the California Current) have highly variable pCO 2 conditions that can result in periodic exposure to pH well below normal 38,39 . It has been suggested that for taxa that have evolved under conditions of pCO 2 variation, OA effects may be less pronounced due to population acclimatization or adaptation 8 than for species evolved under stable pCO 2 conditions 51 . Thus, while California CCRA are susceptible to growth reductions under OA, they may be better able to withstand significant and prolonged changes in pCO 2 than corallines from non-upwelling systems, and thus maintain chemical settlement cues. The hypothesis of environmental variability evolutionarily favoring physiological resistance to climate change is supported by studies showing that temperature variation is a key factor enhancing bleaching resistance in calcifying corals. This has been demonstrated in region-wide analyses 52 and in field and laboratory experiments showing both acclimation and adaptation 53,54 . Highly variable pCO 2 may operate similarly, and background environmental variability may be a generally important factor in determining when to expect resistance. For example, in a fjord system with wide pCO 2 fluctuations, barnacles showed a strong tolerance to high pCO 2 (1,000 µatm) 55 . Though few species from the California upwelling system have been studied in an OA context 56 , prior studies have found evidence of OA adaptation in sea urchin (S. purpuratus) gametes 56 and larvae 57 , and in mussel growth 56 . Abalone (H. rufescens) were found to have distinct differences in biomineralization genes between regions with different upwelling conditions, indicating local population adaptation 58 . Some calcifying species in California thus appear to tolerate broad pCO 2 fluctuations or have sufficient genetic diversity to allow rapid evolution. For CCRA, individuals from a naturally variable tropical environment calcified 42% more than individuals from a uniform environment when experimentally placed under increased pCO 2 59 . A study on a central California geniculate coralline, Corallina vancouveriensis, also found evidence of local adaptation, and it may be that the ability of CCRA spores to attach rapidly limits dispersal distance, restricts gene flow among populations, and increases the potential for local adaptation 60 .
Research is just revealing which taxa might be more vulnerable to OA and scaling up to species interactions and ecosystem functions is critical to predict and manage future changes. Species interactions of abalone larval settlers with CCRA in the California upwelling-dominated ecosystems appear to be resistant to the impacts of a four month history of OA exposure, and functional redundancy in settlement inducing CCRA may also buffer future OA impacts to abalone. The ability of CCRA to maintain chemical cues in this system may be due to periodic exposure to OA due to upwelling. If so, areas with variable OA environments like the California Current might represent areas with high resilience to OA in the future, compared with areas where pCO 2 is more stable. Regardless, our findings demonstrate maintenance of some level of ecosystem function under OA in the California Current and possibly other upwelling ecosystems.