*Article* **Ocean Warming Amplifies the Effects of Ocean Acidification on Skeletal Mineralogy and Microstructure in the Asterinid Starfish** *Aquilonastra yairi*

**Munawar Khalil 1,2,3,\*, Steve S. Doo 1,4, Marleen Stuhr <sup>1</sup> and Hildegard Westphal 1,2,4**


**Abstract:** Ocean acidification and ocean warming compromise the capacity of calcifying marine organisms to generate and maintain their skeletons. While many marine calcifying organisms precipitate low-Mg calcite or aragonite, the skeleton of echinoderms consists of more soluble Mgcalcite. To assess the impact of exposure to elevated temperature and increased *p*CO2 on the skeleton of echinoderms, in particular the mineralogy and microstructure, the starfish *Aquilonastra yairi* (Echinodermata: Asteroidea) was exposed for 90 days to simulated ocean warming (27 ◦C and 32 ◦C) and ocean acidification (455 μatm, 1052 μatm, 2066 μatm) conditions. The results indicate that temperature is the major factor controlling the skeletal Mg (Mg/Ca ratio and Mgnorm ratio), but not for skeletal Sr (Sr/Ca ratio and Srnorm ratio) and skeletal Ca (Canorm ratio) in *A. yairi*. Nevertheless, inter-individual variability in skeletal Sr and Ca ratios increased with higher temperature. Elevated *p*CO2 did not induce any statistically significant element alterations of the skeleton in all treatments over the incubation time, but increased *p*CO2 concentrations might possess an indirect effect on skeletal mineral ratio alteration. The influence of increased *p*CO2 was more relevant than that of increased temperature on skeletal microstructures. *p*CO2 as a sole stressor caused alterations on stereom structure and degradation on the skeletal structure of *A. yairi*, whereas temperature did not; however, skeletons exposed to elevated *p*CO2 and high temperature show a strongly altered skeleton structure compared to ambient temperature. These results indicate that ocean warming might exacerbate the skeletal maintaining mechanisms of the starfish in a high *p*CO2 environment and could potentially modify the morphology and functions of the starfish skeleton.

**Keywords:** ocean acidification; ocean warming; echinoderm; starfish; mineralogy; skeleton; biomineralization

#### **1. Introduction**

The oceans are estimated to take up ~31% of the CO2 increase that is currently observed [1]. This is known to lead to lowered pH values of the seawater, resulting in a reduced carbonate saturation state (Ω), and changes in carbonate–bicarbonate ion balance, recognized as ocean acidification (OA) [2,3]. These changes in seawater chemistry lead to measurable reactions of marine species and ecosystems, and are known to have negative repercussions for many calcifying organisms [4–7], including certain scleractinian corals, bryozoans, molluscs, and echinoderms. Decreased carbonate ion concentration [CO3 <sup>2</sup>−] can disrupt the physiologically regulated biomineralization mechanism that generates, preserves and maintains calcium carbonate (CaCO3) structures, e.g., exoskeleton, test,

**Citation:** Khalil, M.; Doo, S.S.; Stuhr, M.; Westphal, H. Ocean Warming Amplifies the Effects of Ocean Acidification on Skeletal Mineralogy and Microstructure in the Asterinid Starfish *Aquilonastra yairi*. *J. Mar. Sci. Eng.* **2022**, *10*, 1065. https://doi.org/ 10.3390/jmse10081065

Academic Editor: Markes E. Johnson

Received: 31 May 2022 Accepted: 1 July 2022 Published: 3 August 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

spine, tube feet, teeth, pedicellariae and spicules [4,8–10]. This mechanism involves mineral formation, characteristics, morphogenesis, and organic molecules [9,11–13]. This is particularly the case for organisms precipitating a skeleton composed of high Mg-calcite (HMC) i.e., with a significant concentration of magnesium carbonate (MgCO3) in the carbonate (>4 mol% MgCO3) [14,15]. HMC is the most soluble of the polymorphs of crystalline CaCO3 and is thermodynamically metastable [14,16]. Furthermore, previous studies have shown that OA significantly affects the size and weight of shells or skeletons of many marine calcifiers [17–19]. In contrast, some calcifiers, mainly photosynthesizing or photosymbiotic ones, including some corals and algae, show positive responses in calcification and growth values as they benefit from increased CO2 concentrations by an enhanced photosynthesis rate, which provides additional potential energy for the calcification process [20].

At the same time, seawater temperature (ocean warming, OW) influences the ecophysiology of marine organisms [21], skeletal mineralogy [11,14,22–27], growth rate [28], and mineral growth control mechanisms [29]. Previous studies have found inconsistent reactions to increased temperature exposure, ranging from no effect to significant changes in skeletal mineralogy. For example, the skeletal Mg/Ca ratios of the scleractinian coral *Acropora* sp. [30], the sea urchin *Paracentrotus lividus* [25], the foraminifera *Planoglabratella operculari*, *Quinqueloculina yabei* [31] and *Ammonia tepida* [32] increase with temperature; conversely, the skeletal Mg/Ca ratios of the sea urchin *Lytechinus variegatus* was significantly lower in individuals kept at high temperature (~30 ◦C) compared to the ambient temperature (~26 ◦C) [26]. Skeletal Sr/Ca ratios decreases with increasing temperature in the scleractinian coral *Acropora* sp. [30,33], Sr/Ca ratios increases with increasing temperature in the foraminifera *A. tepida* [32] or Globigerinoides ruber [34], and is not significantly affected by temperature in the foraminifera *Trifarina angulosa* [35]. However, the effects of temperature on skeletal mineralogy and structure are still largely unclear and are likely influenced by phylogenetic factors, growth rate [22,23,36], latitude [24], biological 'vital effects' [37], and stage or species-specific differences [38].

The effects of combined OA and OW on marine calcifiers organisms are thought to be additive, antagonistic, or synergistic [39]. Previous meta-studies observed complex responses of calcifying marine organisms to combined OW and OA that vary from a significant negative effect to no effect on the organism [40]. Moreover, there is a trend toward enhanced sensitivity (i.e., the capability to sense and respond to environmental alterations) of biomineralization, growth, survival, and life stages development to OA in corals, echinoderm, and molluscs when being concurrently exposed to elevated temperatures [41]. Besides, there is growing evidence that elevated temperatures can exacerbate microstructure disruption caused by elevated *p*CO2 in ectotherm species, e.g., in some molluscs such as the giant clam *Tridacna maxima* [42] and the mussel *Mytilus edulis* [43,44], and echinoderms such as the sea urchin *Tripneustes gratilla* [45]. However, these synergistic effects of OA and OW seem to be complex, and the magnitude of effect sizes and organism response varies between taxa groups, trophic level, habitat and life stages [40,41,46].

Echinoderms comprise a wide variety of taxa, with a complex calcium carbonate biomineralization. Their skeleton is formed within the syncytium by progressive crystallization of a transient amorphous calcium carbonate phase (ACC) [47], through a biologically controlled intracellular mechanism within vesicles or vacuoles formed by fused cell membranes inside cells [11,22,48]. The echinoderm endoskeleton is composed of a complex three-dimensional porous microstructure (stereom) with connecting trabeculae (i.e., mesh-like interconnecting matrix rod of calcite skeleton) [11], which are composed of 99.8–99.9% weight/weight (*w*/*w*) HMC with Sr as the primary trace element [49,50], and the other 0.1–0.2% (*w*/*w*) of the skeleton being organic components consisting of proteins and glycoproteins, called the intrastereomic organic matrix (IOM) [51,52]. The IOM has critical functions in the biomineralization process during the transient ACC phase by stabilizing the skeleton, controlling mineral incorporation into the skeleton, and controlling the nucleation and morphology of the skeletal crystals [53,54]. The trace element concentration of biogenic CaCO3 is affected by biological factors, e.g., phylogeny, life stage, food supply, as well as by physical-chemical factors, e.g., temperature, salinity, seawater carbonate chemistry, concentration of Mg2+ and Ca2+ ions in the seawater, light, and hydrostatic pressure [14,27,37,49,55,56]. Seawater chemistry influences the mineralogy of the echinoderm skeleton by changing the physiological cost of sustaining the biological control of intracellular chemistry [57]. For echinoderms with their skeletons being composed of HMC (MgCO3 between 2.5% and 39% [8]), their skeletons become more soluble under OA conditions [16,50,58,59].

To gain a better understanding of the effects of combined OA and OW on biomineralization, systematic comparative studies across a phylogenetically diverse spectrum of taxa are needed [57]. The present study contributes to this goal by investigating the mineral composition and skeleton microstructure of *Aquilonastra yairi* (phylum Echinodermata, class Asteroidea, family Asterinidae) under controlled OA and OW conditions. This asterinid starfish thrives in the marine intertidal and coral reefs of the Red Sea and the Mediterranean Sea, where it lives in crevices and beneath corals or rocks [60]. It has an important ecological function as a grazer of marine algae, bacterial mats, detritus, and other fragments of food [61]. Previous studies have indicated deleterious effects of combined OA and OW on the physiological performances of asterinid starfish [62–65], while the effects of each stressor and their potential synergetic effects on skeletal mineral ratio and microstructure are currently still poorly understood.

#### **2. Materials and Methods**

#### *2.1. Experimental Design and Control of Seawater Chemistry*

In this study, starfish *A. yairi* was used as a model organism to investigate the combined effects of OW and OA. A total of 342 specimens of *A. yairi* (size 3–11 mm) from the cultivated stock of the MAREE (Marine Experimental Ecology facility) of ZMT, Bremen, Germany, were studied in the present experiment. Following acclimation procedures (see electronic Supplementary Materials), the starfish were cultured for 90-days in six different treatments, namely at two different temperatures (ambient temperature: 27 ◦C, and high temperature: 32 ◦C) crossed with three levels of *p*CO2 (low *p*CO2: 455 μatm, medium *p*CO2: 1052 μatm, and high *p*CO2: 2066 μatm). All six treatments were replicated in three aquaria (electronic Supplementary Materials Figure S1).

Target temperature and *p*CO2 levels were ramped up gradually over the first ten days to avoid physiological shock. Then the temperature (mean ± SE) in the treatment tanks was maintained at 27 ± 0.05 ◦C and 32 ± 0.08 ◦C, respectively, using a closed circle heating system (Heaters Titanium Tube 600 W, Schego Schemel & Goetz, Offenbach, Germany), controlled with a programmable thermostat. The mixture of the gas bubbled into the seawater in the bottom storage compartment sump was reached by blending compressed CO2-free air and compressed CO2 (pure CO2 provided by Linde GmbH, Pullach, Germany) using electronic solenoid-valve mass-flow controllers (HTK Hamburg GmbH, Hamburg, Germany) in accordance with the standard operating procedure (SOP) for ocean CO2 measurements [66]. Details of the seawater chemistry control and manipulation are provided in the electronic Supplementary Materials. The seawater parameters and carbonate chemistry for the experimental exposures are given in Table 1.

**Table 1.** Seawater chemistry values measured during a 90-day experimental period for *A. yairi* reared under two temperature levels (27 ◦C and 32 ◦C) crossed with three levels of *p*CO2 (455 μatm, 1052 μatm, and 2066 μatm). *AT*, total alkalinity; DIC, dissolved inorganic carbon; *p*CO2, partial pressure of CO2; [CO3 <sup>2</sup>−], carbonate ion concentration; [HCO3 −], bicarbonate ion concentration; [CO2], dissolved CO2; ΩCa, calcite saturation state; ΩAr, aragonite saturation state. Data are presented as mean values ± SE.


*2.2. Skeletal Mineral Composition Analysis*

On days 45 and 90 of the experimental treatment, six specimens from each replicate tank (i.e., 36 in total on each of those days) were randomly collected and rinsed in Milli-Q (18.2 MΩ) water before drying for ~48 h at 40 ◦C. In preparation for trace element analysis, to remove organic material from the skeletal matrix, dried starfish were soaked in hydrogen peroxide (H2O2) [67–69] for 24 h and subsequently cleaned mechanically, i.e., residual organic material was removed by forceps, and further potential contaminations were removed with deionized water in an ultrasonic bath. Then, the sample material was manually ground using a mortar and pestle. The powdered samples were kept at room temperature in sealed vials until analysis.

The element concentration of Ca, Mg and Sr in the skeleton was determined with a Spectro CIROS Vision (SPECTRO Analytical Instruments GmbH, Kleve, Germany) inductively coupled plasma optical emission spectroscope (ICP-OES). The samples (weighing 0.02–0.1 mg) were digested with concentrated nitric acid (HNO3) and H2O2 (high-purity of trace metal grade reagent). The solutions were then diluted to the acidity of 0.5 M HNO3 with aliquots of 0.1 mL and weighed again. Instrument calibration solutions (Inorganic VenturesTM 1000 ppm standard stock solution) were prepared using single-element standards in proportion to the *A. yairi* skeleton concentrations. Measurements of all starfish samples were done routinely against the international reference standard JLs−1, a coral in-house working standard (ZMT-CM1), and HNO3 blanks. Mg and Sr mineral elements are reported as a ratio over calcium (Ca), i.e., Mg/Ca, and Sr/Ca and over total skeletal material, i.e., Canorm, Mgnorm, and Srnorm, to account for minor organic material still present on or within the carbonate skeleton.

#### *2.3. Analysis of the Skeleton Microstructure*

One specimen from the 45-day and 90-day incubations from each treatment tank was randomly selected (*n* = 12), washed, and prepared for the SEM analysis. Each starfish was cut with dissecting scissors around the part of arms and cleaned of soft tissue. Organic material was dissolved using a 30% H2O2 solution buffered in NaOH (0.1 N) at room temperature for 24 h. Skeletons were then rinsed with distilled water repeatedly to

remove any remaining organic material and then air-dried for 48 h at room temperature. Skeleton plates were then mounted on a stub with carbon-based tape and gold-sputtered (Cressington Sputter Coater 108 auto, Cressington Scientific Instruments, Watford, UK) for 30 s. Secondary electron images (SE) were generated with a scanning electron microscope (SEM; Tescan Vega3 XMU, Brno-Kohoutovice, Czech Republic) to characterize the skeleton microstructure, using a beam voltage of 5 kV for a magnification of up to 3000×. All SEM micrographs were examined for any visible differences between treatments, including signs of dissolution, surface smoothness, the shape of stereom pores, and the shape of inner matrix aperture pores.

#### *2.4. Statistical Analysis*

Statistical analysis was performed using the software R, version 4.1.3 [70]. Normality of data distribution and homogeneity of variance was tested with the Shapiro–Wilk statistic *W* test (α = 0.05) [71] and Levene's test (α = 0.05) [72], respectively, and indicated that all data of skeletal mineral ratios were normally distributed and the homoscedasticity assumption for the data was equal. The effects of temperature, *p*CO2, incubation time and their interactions on skeletal mineral element to calcium ratios (Mg/Ca and Sr/Ca) and skeletal mineral element to total skeletal material ratios (Canorm, Mgnorm, and Srnorm) were examined using three-way analysis of variance (ANOVA), and Tukey HSD post hoc analyses were conducted using agricolae R-package 1.3-5 [73]. Temperature, *p*CO2, and incubation time were fixed factors, while skeletal mineral ratios were used as response variables. All statistics were evaluated with a significance level of α = 0.05.

#### **3. Results**

Over the duration of the experiment (90 days), the starfish mortality rate was low and only found in the high-temperature treatment. In general, the results indicate that elevated temperature and *p*CO2 changed the skeletal mineral composition, whereas elevated *p*CO2 affected skeletal microstructure in *A. yairi*.

#### *3.1. Elemental Composition of Skeletal Carbonate*

Overall, a relatively small range of Mg/Ca ratio values were observed across all our treatments (181.95–204.26 mmol/mol). The starfish had consistently higher Mg/Ca ratios in the 32 ◦C treatments (190.90 ± 1.41 mmol/mol, mean ± SE) than those held at 27 ◦C (187.59 ± 0.83 mmol/mol, mean ± SE) throughout all *p*CO2 concentration levels. Both incubation time and temperature had a main effect on skeletal Mg/Ca ratio (*p* = 0.049 and *p* = 0.033, respectively; Table 2). Inter-individual variability in skeletal Mg/Ca ratios was substantially higher in starfish subjected to high temperatures (32 ◦C) compared to those exposed to ambient temperatures (27 ◦C) in all *p*CO2 combined treatments (Figure 1A). No consistent *p*CO2 effect as the sole factor was found, and the Mg/Ca ratio displayed the typical parabolic responses to *p*CO2 (Figure 1A). Elevated *p*CO2 as the sole stressor did not significantly affect skeletal Mg/Ca ratios (*p* = 0.414, Table 2). The interaction of temperature: *p*CO2: incubation time on starfish skeletal Mg/Ca ratios was significant (*p* = 0.014, Table 2). However, Tukey's HSD post hoc analysis did not reveal any significant interactions in Mg/Ca ratios (Table 2 and electronic Supplementary Materials Table S1).

**Table 2.** Summary of three-way ANOVA results for the skeletal mineral ratios of *A. yairi* exposed to temperature (27 ◦C, 32 ◦C) crossed with elevated *p*CO2 (455 μatm, 1052 μatm, 2066 μatm) treatments for 45 and 90 days incubation time. Tukey HSD post hoc tests were performed where ANOVA results indicated significant effects of one or several factors (incubation time, *p*CO2, temperature) with *p*-values adjusted for multiple testing (padj). Bold terms indicate a significant difference (*p* < 0.05).


**Figure 1.** Changes in skeletal properties (**A**) Mg/Ca ratios (mmol/mol) and (**B**) Sr/Ca ratios (mmol/mol) of skeletal carbonate in *A. yairi* exposed to different temperatures (27 ◦C and 32 ◦C) crossed with different *p*CO2 concentrations (455 μatm, 1052 μatm, and 2066 μatm) measured after 45 and 90 days of incubation (*n* = 36). Dots represent individual skeletal mineral values, displayed with jitter to avoid overlap.

Sr/Ca ratios ranged from 2.52 mmol/mol to 2.76 mmol/mol across treatments (Figure 1B). The Sr/Ca ratio at 32 ◦C had the highest fluctuation in values compared to 27 ◦C, where Sr/Ca ratios at 27 ◦C treatments (2.63 ± 0.01 mmol/mol, mean ± SE) were slightly higher than for the 32 ◦C treatments (2.62 ± 0.01 mmol/mol, mean ± SE) (Figure 1B). Interindividual variability of Sr/Ca ratios were substantially higher for medium and high *p*CO2 treatments (1052 μatm, 2066 μatm) compared to low *p*CO2 (455 μatm) treatments; this was the case at both temperature levels (Figure 1B). No significant response of Sr/Ca ratio to differences in the *p*CO2 and temperature as combined or as sole stressors (Table 2). However, skeletal Sr/Ca ratios were significantly altered over incubation time (*p* = 0.006, Table 2), with increasing values from samples taken after 45 days compared to those collected after 90 days.

Canorm ratios showed relatively variable values, which ranged between 647.74 mg/g and 755.30 mg/g across all treatments (Figure 2A). Canorm ratios were changed over the incubation time (*p* = 0.036, Table 2). The Canorm ratios at 27 ◦C rose from 699.73 ± 5.41 mg/g (mean ± SE) at 45 days to 709.02 ± 4.33 mg/g (mean ± SE) at 90 days incubation time (enhanced 1.33%), whereas the Canorm ratios at 32 ◦C rose from 693.51 ± 8.93 mg/g (mean ± SE) at 45 days to 712.60 ± 7.48 mg/g (mean ± SE) at 90 days incubation time (enhanced 2.74%). Canorm ratios were not significantly affected by temperature (*p* = 0.838, Table 2) or *p*CO2 (*p* = 0.307, Table 2) as a single factor, nor as a combined factor (*p* = 0.250, Table 2).

**Figure 2.** Ratios between skeletal mineral element to total skeletal material (**A**) Canorm ratios (mg/g), (**B**) Mgnorm ratios (mg/g), and (**C**) Srnorm ratios (mg/g) in *A. yairi* exposed to elevated temperatures levels (27 ◦C and 32 ◦C) crossed with increased *p*CO2 concentrations (455 μatm, 1052 μatm and 2066 μatm) measured after 45 and 90 days of incubation (*n* = 36). Dots represent individual skeletal mineral values, displayed with jitter to avoid overlap.

Mgnorm ratios from the skeleton of *A. yairi* were significantly altered by incubation time (*p* = 0.001, Table 2). Mgnorm ratios increased from 45 days to 90 days in both temperature treatment conditions (Figure 2B). At 27 ◦C, the Mgnorm ratios increased from 109.57 ± 0.94 mg/g (mean ± SE) at 45 days to 112.98 ± 0.57 mg/g (mean ± SE) at 90 days incubation time (i.e., 3.11% increase), while at 32 ◦C, the Mgnorm ratios increased from 110.61 ± 1.04 mg/g (mean ± SE) at 45 days to 115.31 ± 0.63 mg/g (mean ± SE) at 90 days incubation time (4.25% increase). There was no significant effect of *p*CO2 nor any combined effect of the factors (Table 2). However, temperature led to a marginal increase in the Mgnorm ratios (*p* = 0.051, Table 2).

Srnorm ratios were altered over incubation time (*p* = 0.005, Table 2) at both temperature treatments (Figure 2B). The mean value of the Srnorm ratio at 27 ◦C increased by 2.38% from day 45 (3.36 ± 0.04 mg/g, mean ± SE) to day 90 (3.44 ± 0.02 mg/g, mean ± SD). Similarly, the mean value of Srnorm ratio at 32 ◦C increased from 45 days (mean ± SD, 3.30 ± 0.04 mg/g) to 90 days (3.47 ± 0.06 mg/g, mean ± SD) incubation time (5.15% increase). However, there was no significant effect of combined stressor factors nor solely stressor factors on Srnorm ratios (Table 2).

#### *3.2. Skeletal Microstructure*

High magnification SEM micrographs showed marked differences in skeletal structure between low *p*CO2 (455 μatm) compared to medium and high *p*CO2 treatments (1052 μatm, 2066 μatm, respectively) at both ambient and high temperatures (27 ◦C, 32 ◦C, respectively). The skeletal structure in low *p*CO2 crossed with ambient (27 ◦C: 455 μatm) and high temperatures (32 ◦C: 455 μatm) revealed no remarkable differences between 45-day and 90-day incubation time. The stereom pores were arranged equally in shape and the aperture pores of the inner matrix were relatively equal in shape, while the trabecular surface was smooth (Figure 2A,B,G,H and Table 3).

**Figure 2.** *Cont.*

**Figure 2.** *Cont.*

**Figure 2.** SEM micrographs of the skeletal microstructure of *A. yairi* cultured under two temperature levels (27 ◦C and 32 ◦C) crossed with three *p*CO2 concentrations (455 μatm, 1052 μatm and 2066 μatm). Stereom microstructure after 45 (**A**,**C**,**E**,**G**,**I**,**K**) and 90 days (**B**,**D**,**F**,**H**,**J**,**L**) of incubation time. *sp*: stereom pore; *st*: skeleton trabeculae; *imp*: stereom inner matrix pore; (a) the galleries stereom pores are less-equal in shape; (b) dissolution in calcium carbonate skeleton; (c) increased size of inner matrix aperture pores. Technical image acquisition, SEM mode: SE, SEM HV: 5.0 kV, SEM magnification: 3000×.

In contrast, after 45-day and 90-day incubation times, the skeleton from medium and high *p*CO2 treatments at ambient temperatures showed stereom structures that were more variable in shape compared to the control treatment, and signs of degradation, i.e., dissolution, were observed on the surface of the trabeculae (Figure 2C–F and Table 3). Furthermore, under high temperatures, these medium and high *p*CO2 treatments in addition result in signs of skeletal degradation observed at the trabeculae surface, while the apertures of the inner matrix pores were wider (i.e., un-equal in shape) compared to the control treatment and the ambient temperature crossed with medium and high *p*CO2 treatments (Figure 2I–L and Table 3).

**Table 3.** Skeletal microstructure characteristics of *A. yairi* under crossed temperatures and *p*CO2 conditions at different incubation times as observed with scanning electron microscopy (SEM). ND: the skeleton surface is smooth and has no signs of degradation; DS: the skeleton surface had degradation signs; HD: the skeleton surface had high degradation signs; ES: the stereom pores were equal in shape; US: the stereom pores were un-equal in shape; HU: the stereom pores were highly un-equal in shape; EP: the inner matrix aperture pores were relatively equal in shape; UP: the inner matrix aperture pores were relatively un-equal in shape (wider). (*n* = 12).


#### **4. Discussion**

Our results imply that temperature plays a primary regulatory role in Mg concentration in the skeletal carbonate of *A. yairi*, where the Mg/Ca ratios increase under hightemperatures. This corroborates results from previous studies that found higher Mg concentrations associated with higher temperatures in echinoid and asteroid species [23,24,74]. This temperature association was previously attributed to kinetic factors affecting ion discrimination [23] and also to the physiological mechanisms that control Mg absorption in cells [75] during the biomineralization process. The increased Mg content of HMC was connected to an amorphous calcium magnesium carbonate (ACMC) precursor [76,77]. At high temperatures, the aqueous Mg2+ solvation energy barrier becomes lower [78]; hence this condition might favor more Mg2+ to be incorporated into the calcite lattice, encouraging the formation of ACMC, which later transforms into HMC.

Echinoderms are generally considered relatively poor regulators of internal acid–base balance, where the range of regulatory capacities is species-specific [79]. In hypercapnic conditions (high *p*CO2, low pH), echinoderms increase the bicarbonate concentration in their coelomic fluid and practice passive skeletal dissolution to support internal acid– base regulatory functions due to acidosis [80]. Higher Mg concentrations in the skeleton in conjunction with a degradation in the inner skeleton (see Section 3.2) due to skeletal dissolution, as documented in the current studies, support the assumption of a tradeoff mechanism [65,81]. The released HMC mineral is then used as an active buffering mechanism to compensate for changes in internal pH that may help to avoid or reduce physiological impacts.

We noticed that no single (temperature or *p*CO2) nor combined stressor treatment affected skeletal Sr (Sr/Ca and Srnorm) in *A. yairi.* This contrasts with previous studies on the starfish *Asterias rubens* [49] and the sea urchin *Paracentrotus lividus* [25], which reported that the Sr/Ca ratio depended on temperature. These contrasting results indicate a speciesspecific skeletal Sr control mechanism in the echinoderm group that may respond to the stressors. The calcification rates might play an important role in skeletal Sr precipitation rather than direct dependence on temperature. It is difficult to discriminate the effects of temperature and *p*CO2 on skeletal Sr. Our data indicated that the skeletal Sr was increased over incubation time in all treatment combinations (electronic Supplementary Materials Figure S2) except the 27 ◦C: 455 μatm treatment, assuming that temperature and *p*CO2 might have an indirect effect on skeletal Sr through their control over the

calcification process [23,34,69,82], which are common in biogenic calcites [83,84]. However, the incorporation pathways of Sr into the echinoderm ACC are still poorly understood, so further studies are needed.

We found a strong correlation between skeletal Sr and skeletal Mg (R2 = 0.04, *p* < 0.001, electronic Supplementary Materials Figure S3), as Srnorm increases with increasing Mgnorm ratio. Hence, Mg might play a role in affecting the Sr precipitation. Sr ions cannot easily substitute for Ca ions due to differences in cations ions size and weight (Sr2+ ionic radius of 1.18 Å: Ca2+ ionic radius of 1.00 Å [85]), which makes them incompatible in calcite [86]. High concentrations of Mg2+ (ionic radius of 0.72 Å [85]) that are incorporated in inorganic calcite can distort the crystal lattice (i.e., lattice deformation), which increases the size of calcium lattice positions and thus allows for increased incorporation of Sr2+ [87].

Similar to skeletal Sr, we found no direct influence of temperature or *p*CO2 on the Canorm ratio (Table 2). However, the highest inter-individual variability of Canorm indicates that temperature and/or *p*CO2 considerably alter the production of skeletal Ca in *A. yairi*, where Canorm ratios exhibited an increase with incubation time, except for the treatment of 27 ◦C: 455 μatm (~703 mg/g) where the ratio was relatively unchanged (electronic Supplementary Materials Figure S4). CaCO3 is more soluble at lower temperatures [88] and high *p*CO2 [20]. Under OW conditions, starfish seem to be able to boost their capacity to control calcification through the modulation of the intracellular calcifying fluids pH to produce CaCO3. This might provide them with higher resistance and resilience towards the effects of OA. The increased energetic costs of this physiological response are indicated by elevated respiration rates [65]. Hence, in the long term, this phenotypic plasticity and biocalcification control mechanism might have further physiological implications for the starfish.

The hypercapnic conditions of the high *p*CO2 treatment in our experiment have resulted in visible adverse effects on the skeletal structures in both ambient as well as elevated temperatures (Figure 2, Table 3). We detected changes in the stereom pores, the inner matrix pores and degradation of the skeleton surface through reduced deposition or dissolution, which increased over incubation time in the starfish that were exposed to medium and high *p*CO2. These skeletal alterations might be due to the thin layer of epithelial tissue covering the starfish body wall [89], which acts as a protective membrane between the skeleton and the seawater [90]. Furthermore, the degradation of the epithelial tissue could also explain why the combination of OW with OA enhances the degradation of the skeletal structure that was observed in the high *p*CO2 treatment (32 ◦C: 2066 μatm). We hypothesize that the epithelial tissue might become weaker in its function due to degradation of the epithelium cells during long-term experiments. Epithelial tightness is controlled by a protein series that molds a seal among epithelial cells [90], which are sensitive to elevated temperature.

The seawater chemistry factor, which was underlying the altered skeletal structure in *A. yairi* might involve calcium carbonate saturation state (Ω). Skeletal degradation provides evidence that the skeleton of *A. yairi* was negatively affected by lowered calcite saturation state (Ωca) as a function of increasing *p*CO2 (Table 1), which directly affects the biomineralization and dissolution of the CaCO3 in skeletal structures [17,91]. Near Ωca ~3, we found stereom alteration and degradation signs in trabeculae, which indicates that *A. yairi* was facing difficulty in producing and maintaining their skeleton compared to low *p*CO2 treatments (Ωca > 5). Previous studies found that the skeletal morphometric development of echinoid (e.g., sea urchin *Lytechinus variegatus*) was significantly affected by Ωca 2.53 [92]. Furthermore, increased susceptibility of the skeleton to dissolution through OA was suggested to be due to the skeleton composed of HMC, which is 30 times more soluble than pure calcite [47,93]. When the qualitative approach depicts changes in the skeleton micro-morphology because of exposure to elevated *p*CO2 and temperature (Table 3, Figure 2), further investigation using quantitative analysis (e.g., stereom and inner matrix pore size) is required to quantify the alteration magnitude in skeleton microstructure as the deleterious impact of OA and OW [94,95].

Under OW and OA scenarios, calcifying marine organisms are expected to face unfavorable conditions to produce and maintain their skeletal HMC in order to sustain their biomechanical functions. Phenotypic plasticity requires re-allocation of energy as a tradeoff and may represent a potential key mechanism for species viability [96]. For *A. yairi*, this mechanism might reflect strategies to maintain their skeletal HMC in a low pH environment by modulating their cellular chemistry to create isolated microenvironments of deposition, producing the specific mineral they necessitate, which involves significant energy re-allocation [97]. Since the asterinid starfish produces HMC, which is more soluble than calcite and aragonite, it signifies that more energy will be required to preserve the calcification process as the mechanism in both constructing and maintaining their skeletal components, leading to energy trade-offs against other physiological processes [65,98].

Changes in the shape of the stereom pores and degradation of the starfish skeleton potentially have implications for skeletal strength, stiffness, and function. Weakening of the structure possibly will reduce locomotion performance and result in a lower ability to resist predators and to face ocean currents, which then conveys consequences to the benthic community structure.

**Supplementary Materials:** The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/jmse10081065/s1; Table S1: Tukey HSD post hoc test for the interactive effects of incubation time (45 and 90 days), *p*CO2 (455 μatm, 1052 μatm, 2066 μatm) and temperature (27 ◦C, 32 ◦C) on skeletal Mg/Ca ratio of *A. yairi* using the 'agricolae R-package' for multiple comparisons to interrogate the main effects of incubation time, temperature and *p*CO2 (incubation time: *p*CO2: temperature = 0.014, Table 2); Figure S1: Schematic of ocean acidification and ocean warming experimental design with a fully factorial combination of low *p*CO2 (455 μatm), moderate *p*CO2 (1052 μatm), and high *p*CO2 (2066 μatm) treatments with ambient temperature (27 ◦C) and high temperature (32 ◦C) treatments; Figure S2: Skeletal Sr values over incubation time in *A. yairi* exposed to elevated temperatures levels (27 ◦C and 32 ◦C) crossed with increased *p*CO2 concentrations (455 μatm, 1052 μatm, and 2066 μatm). (**A**) Sr/Ca ratio (mmol/mol), (**B**) Srnorm ratio (mg/g). Data were presented as mean (*n* = 36); Figure S3: Correlation of Srnorm ratios (mg/g) against Mgnorm ratios (mg/g) (R<sup>2</sup> = 0.44, F1, 34 = 26.78, *<sup>p</sup>* = 1.019 <sup>×</sup> <sup>10</sup><sup>−</sup>5) in *A. yairi* exposed to elevated temperatures levels (27 ◦C and 32 ◦C) crossed with increased *p*CO2 concentrations (455 μatm, 1052 μatm, and 2066 μatm). (*n* = 36); Figure S4: Skeletal Canorm (mg/g) ratio over incubation time in *A. yairi* exposed to elevated temperatures levels (27 ◦C and 32 ◦C) crossed with increased *p*CO2 concentrations (455 μatm, 1052 μatm, and 2066 μatm). Data were presented as mean (*n* = 36). For more details, please see [66,99–106].

**Author Contributions:** Conceptualized and designed: M.K. and H.W.; methodology: M.K., S.S.D., M.S. and H.W.; investigation: M.K.; resources: M.K. and H.W.; sample processing and analysis: M.K.; data curation: M.K.; writing original draft preparation: M.K.; visualization: M.K.; writing-review and editing: M.K., S.S.D., M.S. and H.W.; supervision: H.W. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research project was supported by the Academy Doctoral Research—Grant Leibniz Centre for Tropical Marine Research (ZMT), Germany and the Ministry of Education, Culture, Research, and Technology (MoECRT), Republic of Indonesia—Asian Development Bank (ADB) AKSI Project [grant number L3749-INO] to M.K.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** The datasets used and analyzed during the current study are present in the manuscript and/or supplementary materials. Data are available upon request from the corresponding author.

**Acknowledgments:** We would like to thank Andreas Kunzmann (ZMT, Bremen, Germany) for his advice on the starfish *A. yairi* as a model organism. Many thanks are due to the ZMT laboratory staff, namely Jule Mawick, Sebastian Flotow, Silvia Hardenberg, José Garcia, Nico Steinel, Matthias Birkicht and Fabian Hüge.

**Conflicts of Interest:** On behalf of all authors, the corresponding author states that there is no conflict of interest.

#### **References**


## *Article* **Responses of Freshwater Calcifiers to Carbon-Dioxide-Induced Acidification**

**Aaron T. Ninokawa 1,2,\* and Justin Ries 3,4**


**Abstract:** Increased anthropogenic carbon dioxide (CO2) in the atmosphere can enter surface waters and depress pH. In marine systems, this phenomenon, termed ocean acidification (OA), can modify a variety of physiological, ecological, and chemical processes. Shell-forming organisms are particularly sensitive to this chemical shift, though responses vary amongst taxa. Although analogous chemical changes occur in freshwater systems via absorption of CO2 into lakes, rivers, and streams, effects on freshwater calcifiers have received far less attention, despite the ecological importance of these organisms to freshwater systems. We exposed four common and widespread species of freshwater calcifiers to a range of pCO2 conditions to determine how CO2-induced reductions in freshwater pH impact calcium carbonate shell formation. We incubated the signal crayfish, *Pacifastacus leniusculus*, the Asian clam, *Corbicula fluminea*, the montane pea clam, *Pisidium* sp., and the eastern pearlshell mussel, *Margaritifera margaritifera*, under low pCO2 conditions (pCO2 = 616 ± 151 μatm; pH = 7.91 ± 0.11), under moderately elevated pCO2 conditions (pCO2 = 1026 ± 239 uatm; pH = 7.67 ± 0.10), and under extremely elevated pCO2 conditions (pCO2 = 2380 ± 693 uatm; pH = 7.32 ± 0.12). Three of these species exhibited a negative linear response to increasing pCO2 (decreasing pH), while the fourth, the pea clam, exhibited a parabolic response. Additional experiments revealed that feeding rates of the crayfish decreased under the highest pCO2 treatment, potentially contributing to or driving the negative calcification response of the crayfish to elevated pCO2 by depriving them of energy needed for biocalcification. These results highlight the potential for freshwater taxa to be deleteriously impacted by increased atmospheric pCO2, the variable nature of these responses, and the need for further study of this process in freshwater systems.

**Keywords:** ocean acidification; freshwater acidification; freshwater calcifier; carbon dioxide; calcification; biomineralization; Lake Tahoe

### **1. Introduction**

Increased partial pressure of atmospheric carbon dioxide (pCO2) drives large scale alterations to environmental systems. In particular, this additional CO2 can enter surface waters and perturb the aquatic carbonate system, lowering pH, carbonate ion concentration ([CO3 <sup>2</sup>−]), and the saturation state of water with respect to calcium carbonate (CaCO3). In marine systems, this process, termed ocean acidification, can impair tissue and shell growth and alter the behavior of many marine species, yielding ecological changes across a range of spatial, temporal, and trophic scales [1].

Calcifying organisms, those producing calcium carbonate shells or skeletons, are especially sensitive to CO2-induced changes in carbonate system chemistry, although the drivers of this sensitivity can be complex [2–4]. In many taxa, calcium carbonate material is produced more slowly under increased pCO2 [1,5], with shell and skeletal material produced under these conditions tending to be weaker [6,7]. Some species, however, grow faster under increased pCO2 or can maintain or enhance the quality of their shell or skeletal

**Citation:** Ninokawa, A.T.; Ries, J. Responses of Freshwater Calcifiers to Carbon-Dioxide-Induced Acidification. *J. Mar. Sci. Eng.* **2022**, *10*, 1068. https://doi.org/10.3390/ jmse10081068

Academic Editors: Maria Gabriella Marin and Ryan J.K. Dunn

Received: 21 May 2022 Accepted: 27 July 2022 Published: 4 August 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

material [5,8]. In many cases, the effects of pCO2 operate indirectly, often via pathways related to pH, bicarbonate, or carbonate ions [3,9]. Indeed, the diversity of responses to increased pCO2 complicates the process of generalizing how calcifying taxa will respond to CO2-induced acidification in the future [4].

Although there have been many studies examining effects of CO2-induced acidification on marine species, far fewer have considered responses of freshwater species. Freshwater typically has lower alkalinity, or chemical buffering capacity, which makes the CaCO3 saturation state of freshwater less than that of seawater for equivalent pCO2 conditions, while also rendering the carbonate chemistry of freshwater systems more sensitive to variations in atmospheric pCO2 than that of marine systems [10,11]. Freshwater species, for example, will experience larger shifts in pH and lower absolute CaCO3 saturation states than their marine analogs for a given increase in anthropogenic CO2. The lower alkalinity of freshwater systems also makes the carbonate chemistry of these systems more vulnerable to diurnal and seasonal cycles in photosynthesis and respiration than marine systems [10]. It is, therefore, possible that freshwater calcifiers will be more sensitive than marine calcifiers to equivalent shifts in atmospheric pCO2. Alternatively, the naturally higher variability in carbonate chemistry and lower absolute saturation state of freshwater systems could select for freshwater calcifying taxa that are relatively resilient to CO2-induced perturbations to carbonate system chemistry.

Although past research has examined responses of freshwater calcifying species to elevated aqueous CO2, most of this earlier work was conducted with the aim of understanding physiological acid–base dynamics within the organisms themselves, rather than understanding the impacts of CO2-induced acidification on shell and skeletal formation. Thus, much of this prior work employed unrealistically high pCO2 conditions that exceed projected end-of-century CO2 partial pressures by tens of thousands of microatmospheres (μatm), resulting in unrealistic pH reductions in the order of 2 units [12]. Furthermore, among the modest number of studies where pCO2 treatments reflected realistic future scenarios, prior work has focused on shifts in primary productivity by phytoplankton and changes in food quality [10] as driving calcifier performance, rather than on the direct effects of CO2-induced changes in freshwater carbonate chemistry. Therefore, to determine the effect of increased atmospheric pCO2 on calcifying invertebrates in the context of increasing anthropogenic pCO2, we reared four species of freshwater calcifiers under three atmospheric pCO2 conditions that bracket the range of conditions expected to occur over the next two centuries given the range of future emissions scenarios [13].

#### **2. Materials and Methods**

This project was conducted at the University of California, Davis, Tahoe Environmental Research Center (Tahoe City Field Station) and Lot #4 of Alpine Meadows, CA, USA between December 2020 and August 2021. The freshwater employed in the flow-through CO2-induced acidification experiments was sourced from Burton Spring, a tributary of Lake Tahoe. The carbonate chemistry of the treatments was controlled by maintaining constant pH in three 800 L sumps with a pH-stat system (American Marine Inc. Pinpoint pH Controller) that reduced pH by dosing pure compressed CO2 with a solenoid-valve controlled gas regulator (FZone Pro Series CO2 Regulator), or increased pH by dosing ambient air through an air pump (Simply Deluxe Electromagnetic Air Pump). The compressed CO2 and compressed air were sparged into the sumps and treatments tanks with flexible, microporous bubbling tubes that were designed to expedite equilibration between aqueous and gas phases. Spring water (filtered to 10 μm) entered the sumps at a rate of 1 L min−<sup>1</sup> to minimize accumulation of metabolic byproducts and prevent depletion of alkalinity and calcium ions through the calcification process. The sumps and experimental tanks were covered with plastic sheeting and plastic lids, respectively, to prevent room air from equilibrating with the water in the sumps and the experimental treatments. The water within each of the three sump systems was recirculated through 12–40 L tanks (3 sumps × 12 tanks = 36 tanks total), with the recirculating water passing through an

activated carbon filter (changed biweekly) before entering the tanks and through 10 L of loose activated carbon media (changed monthly) before re-entering the sump (Figure 1). Temperature varied naturally with the seasons but was maintained below 22 ◦C with chillers and the addition of frozen spring water.

**Figure 1.** Schematic diagram of one of the three identical pCO2 control systems used in the experiment. Fresh spring water flowed continuously through the main sump and recirculated amongst the experimental tanks. A pH stat controller added either pure CO2 or ambient air as required to maintain water pH within the target range. The image in the top left shows arrangement of all three pCO2 treatment systems in the laboratory.

Water chemistry was measured three times per week with a Thermo Scientific Orion Star A329 Multiparameter Meter. Oxygen (Orion RDO Optical Dissolved Oxygen Probe) and pH sensors (Atlas Scientific Spear-tip pH) were calibrated daily in air-equilibrated water and with NIST-traceable NBS buffers (pH 7 and 10), respectively. The conductivity probe (Duraprobe) was calibrated every other week with Oakton 84 and 1413 μS cm−<sup>1</sup> standards. Discrete 200 mL water samples were collected from the three sumps during each of these sampling events and frozen until they could be analyzed for alkalinity with a Metrohm 855 Robotic Titrosampler.

In addition to these higher frequency measurements, discrete 500 mL water samples were collected weekly from each of the 36 tanks and analyzed at the Tahoe Environmental Research Center (Incline Village Field Station) for dissolved inorganic carbon (DIC) on a Lachat IL 500 TOC instrument. The 'seacarb' package in *R* was used to calculate total alkalinity (TA) from these lower frequency DIC and pH measurements using constants from Waters et al. [14]. This information was used to define the linear relationship between specific conductance and TA for water employed in this experiment (Figure 2). We then used this relationship to estimate TA from the higher frequency measurement of waterspecific conductance (described above). The calculated TA was 96 ± 0.3% (mean ± se, n = 149) of the measured alkalinity described above. Given this good agreement between calculated and measured alkalinity, we then used these higher frequency measurements of pH and conductivity-estimated alkalinity, along with measured temperature and total dissolved solids, to calculate carbonate system parameters for each tank throughout the duration of the experiment. The saturation state of calcite (Ωcalcite) was calculated assuming that the calcium concentration was half the alkalinity as calcium concentrations in the Lake Tahoe area tend to be between 0.5 and 1 times the alkalinity [15]. Small deviations in Ca2+ concentration from this relationship will not materially impact the calculated saturation states or the interpretation of the results.

**Figure 2.** Relationship between specific conductance and total alkalinity for water samples obtained from this experiment. The linearity of this relationship allowed the calculation of total alkalinity (for carbonate system calculations) from higher frequency measurements of specific conductance.

The species investigated in this experiment were collected from the field and acclimated to laboratory conditions for at least 5 days prior to starting the experiments. Signal crayfish, *Pacifastacus leniusculus*, were collected from Pomin Park in Tahoe City, CA, USA, in late April 2021. Two cohorts of Asian clams, *Corbicula fluminea*, were collected from Marla Bay and Lakeside near Stateline, NV, on 15 April 2021 and 24 May 2021.

Two cohorts of pea clams, *Pisidium* sp., were collected from a spring feeding into the Truckee River near Polaris, CA, on 13 May 2021 and 19 June 2021. Pearlshell mussels, *Margaritifera margaritifera*, were collected from a lake near Jupiter, Florida, in early May 2021. Bivalves were held in three of the 40 L tanks per CO2 treatment while crayfish were held in a different set of three tanks (9 tanks total per species). Remaining tanks (6 per pCO2 treatment) were used for respiration and feeding trials described below. Bivalves were fed a commercially available Shellfish Diet twice daily at a concentration of 5 mL per 40 L of tank water. Crayfish were fed 1 × 1 cm dehydrated algae sheets and raw shrimp ad libitum every other day, with uneaten food removed at the time of feeding. All tanks were cleaned of accumulated solid waste three or four times per week.

Organisms remained within the tanks for the duration of the experiment and net calcification rates were calculated as the fractional change in shell mass determined from buoyant weights at the beginning and end of the experiment, normalized to a 30-day interval. Shell mass was calculated from buoyant weight measurements, where the density of the water was determined daily by the air and buoyant weights of glass bead standards (measured daily) with densities of either 2.55 or 2.23 g cm−<sup>3</sup> [16]. Daily buoyant and dry weights of four half shells of *Corbicula fluminea* and two half shells of *Margaritifera margaritifera* revealed that these species had shell densities of 2.81 and 2.71 g cm−3, respectively and these values were used for calculating shell mass from buoyant weight [16]. A shell density of 2.71 g cm−<sup>3</sup> was used for calculating the crayfish shell mass from their buoyant weight. Because crayfish transport calcium carbonate between their external shell and internal gastrolith during the molting process [17], crayfish that, at the conclusion of the incubation, had recently molted and had not yet remineralized their shell (i.e., the shell was soft) were excluded from further analysis. Due to their small size, net calcification rate of pea clams

was calculated as the fractional change in wet mass between the beginning and end of the experiment normalized to a 30-day interval, measured with a Sartorius 2120T microbalance after removing excess water from the shell with lint-free wipes and allowing the clams to air dry for precisely 15 min prior to weighing. Net calcification rates were analyzed with a maximum likelihood routine fitting various functional models identified previously [5], including linear, parabolic, and threshold (exponential) relationships with average pCO2 during the incubation period using individuals as replicates. The statistically significant model with the lowest AIC was selected as the best one to describe the relationship between pCO2 and net calcification rate.

At the conclusion of the growth experiments, additional experiments were conducted to test whether CO2-induced changes in carbonate chemistry impaired crayfish feeding. A new batch of live crayfish prey (pea clams) was collected and acclimated to the pCO2 treatments for three days. Five pea clams of comparable size were placed in each of six tanks per pCO2 treatment with a single crayfish per tank. The number of pea clams consumed after each 25 min trial was recorded. The respiration rates of the crayfish were also obtained to determine whether pCO2-induced changes in water chemistry altered metabolic rates and, therefore, feeding requirements. Before the feeding trials, we placed each crayfish in a sealed incubation vessel and measured the change in oxygen concentration during the incubation time with the Orion RDO dissolved oxygen probe. Respiration and feeding trials both took place at night in the dark, as this is when the crayfish are most active.

#### **3. Results**

The freshwater calcifying organisms in this study were exposed to pCO2 conditions approximately double and quadruple those of average present-day conditions. Mean daily pCO2 ± SD (calculated from pH, total alkalinity, TDS, and temperature) across all tanks in the control, intermediate, and high treatments were 616 ± 151 μatm, 1026 ± 239, and 2380 ± 693 (Figure 3d, Table 1, n = 35 sampling days), respectively, corresponding to mean daily pH ± SD of 7.91 ± 0.11, 7.67 ± 0.10, and 7.32 ± 0.12 (Figure 3b, Table 1, n = 35 sampling days). Daily mean alkalinity decreased throughout the course of the experiment from 1272 to 927 μmol kg−<sup>1</sup> (Figure 3c), probably due to increased contribution of snowmelt relative to groundwater through the spring–summer transition. This trend was accompanied by an increase in daily mean temperature from 12.2 to 21.4 ◦C (Figure 3a). The relationship between specific conductance (SC, μS cm−1) and total alkalinity (TA, <sup>μ</sup>mol kg<sup>−</sup>1) was TA = −75.46 + 10.38 ∗ SC (R2 = 0.96, F1314 = 6937, *<sup>p</sup>* < 0.001).


**Table 1.** Average treatment conditions during the experiment. Measurements are the mean (standard deviation) of 12 tanks per treatment over 35 sampling days.

**Figure 3.** Average daily conditions within the experimental replicate tanks throughout the duration of the experiment. Blue circles indicate the low pCO2 treatment, green triangles indicate the intermediate pCO2 treatment, and black diamonds indicate the high pCO2 treatment. Vertical bars show the standard deviation of all 12 tanks at each pCO2 treatment.

Net calcification rates of the four species as a function of pCO2 followed two general response types (Tables 2 and 3). Net calcification rates of *Pacifastacus leniusculus* (Figure 4a), *Corbicula fluminea* (Figure 4b), and *Margaritifera margaritifera* (Figure 4c) exhibited a linear decline in net calcification with increasing pCO2 (decreasing pH). Net calcification rate of *Pisidium* sp. exhibited a parabolic relationship with pCO2, in which maximum rates of net calcification were observed in the intermediate pCO2 treatment (Figure 4d).

**Table 2.** Net calcification rates (fractional change in buoyant or wet weight normalized to a 30-day growth interval) in each treatment for each species during the experiment. Values reported are mean (standard error), sample size.


**Table 3.** Parameter estimates for tested models describing the relationship between net calcification rate and pCO2. Linear models took the form net calcification rate = b0 + b1 ∗ pCO2. Parabolic models took the form net calcification rate = b0 + b1 ∗ pCO2 + b2 ∗ (pCO2) 2. Exponential models took the form net calcification rate = b0 + b1 <sup>∗</sup> <sup>e</sup>b2∗pCO2 . Bold rows indicate statistically significant model (α = 0.05) with the lowest AIC value for a given species.


**Figure 4.** Net calcification rates (expressed as fractional change in buoyant or wet weight normalized

to 30-day growth interval) as a function of water pH for (**a**) the signal crayfish, *Pacifastacus leniusculus* (n = 33, 2−6 individuals per tank); (**b**) the Asian clam, *Corbicula fluminea* (n = 95, 6−15 individuals per tank); (**c**) the Eastern pearlshell mussel, *Margaritifera margaritifera* (n = 62, 2−10 individuals per tank); and (**d**) the pea clam, *Pisidium* sp. (n = 85, 6−15 individuals per tank). All species shown exhibit a linear negative response in net calcification rate to increasing pCO2, except for the pea clam that exhibits a parabolic response. Data markers represent the average net calcification rates for all individuals in each of the three replicate tanks for each of the three pCO2 treatments. Vertical bars represent the standard error of the net calcification rates, while the horizontal bars represent the standard deviation of pCO2 during the incubations. Shaded regions represent the 95% confidence intervals of the best fitting model.

Metabolism and food consumption of the crayfish, *Pacifasticus leniusculus*, exhibited more complex responses to altered pCO2 treatments. Respiration rates averaged 14.7 μmol g−<sup>1</sup> h−<sup>1</sup> (n = 43 crayfish, 13 in low and high pCO2 treatments, respectively, 17 in the intermediate pCO2 treatment), but did not vary with pCO2 (Figure 5a, F1,41 = 0.281, *p* = 0.60). Feeding rates, however, were impacted by pCO2 (Figure 5b, ANOVA, F2,15 = 4.47, *p* = 0.030), such that feeding rates in the highest pCO2 trials were significantly less than in the intermediate pCO2 trials (post-hoc Tukey, *p* = 0.043), and with nearly significantly less than in the lowest pCO2 trials (post-hoc Tukey, *p* = 0.060).

**Figure 5.** (**a**) Respiration and (**b**) feeding rates of the signal crayfish as a function of pCO2. High pCO2 inhibits feedings rates for the crayfish, while respiration rates do not significantly vary across pCO2 treatments. Vertical bars represent standard error of respiration rates for of feeding rates for six crayfish per treatment; horizontal bars represent standard deviation of pCO2 for the respective treatments.

#### **4. Discussion**

Elevated pCO2, resulting in decreased pH, caused a reduction in net calcification rates in three species of freshwater calicifiers, the Asian clam *Corbicula fluminea*, the pearlshell mussel *Margaritifera margaritifera*, and the crayfish *Pacifastacus leniusculus*. In the pea clam, *Pisidium* sp., however, net calcification rate was highest under intermediate pCO2. The linearly negative response of the Asian clam, pearlshell mussel, and crayfish suggests that these species will be negatively impacted by CO2-induced acidification of freshwater systems over the coming decades. However, optimization of calcification under moderate acidification exhibited by the pea clam suggests that this species may be more resilient to CO2-induced acidification predicted for the near future, but this resilience may diminish at higher pCO2 conditions predicted for the next century or two. Although the high interspecimen variability in growth rates of the pea clam may argue for selection of a simpler

model (i.e., linear), the large number of individuals employed in the pea clam experiment (85) supports adoption of the more complex parabolic model.

There are various mechanisms that can lead to the observed functional relationships between net calcification rate and pCO2. Because we did not manipulate calcium ion concentrations, shell formation here is likely the culmination of two main processes the uptake of inorganic carbon, primarily in the form of bicarbonate ion and, to a lesser extent, aqueous CO2, and the efflux of protons (H+) from the site of calcification that effectively converts bicarbonate ion to carbonate ion available for calcification [18,19]. The linear reduction in net calcification rate with decreased pH (increased pCO2) exhibited by the Asian clam, pearlshell mussel, and crayfish suggests that these species are limited in their ability to control carbonate chemistry at the site of calcification in support of calcification, such as by removing H+ from their calcifying fluid.

This linear negative calcification response to elevated pCO2 exhibited by the crayfish is somewhat surprising given that various species of marine decapod crustacea are reported to calcify faster under elevated pCO2 [5,20], suggesting that freshwater crustacea utilize dissolved inorganic carbon in shell formation differently than marine crustacea and/or that other processes within their physiological repertoire are more sensitive to pH (see discussion below). The parabolic relationship between net calcification rate and pH exhibited by the pea clam suggests that it may be able to utilize the additional dissolved inorganic carbon (DIC) resulting from increased pCO2 by, for example, converting it to carbonate ions by removing protons from the site of calcification. Under the highest pCO2 condition, it is possible that the benefits of this process are overwhelmed by other processes that are more deleteriously impacted by the low pH conditions, such as the dissolution of pre-formed, exposed shell in water that is relatively undersaturated with respect to the clam's aragonite [21] shell mineral. This trend may be further reinforced by nonlinearities imposed by scaling differences between surface- vs. volume-related processes. For example, because *Pisidium* sp. broods its larvae [22], calcification occurs throughout the clam's body cavity as its offspring build their shells. This contrasts the calcification processes of *M. margaritifera* and *C. fluminea*, which do not brood their calcifying offspring. Dissolution, however, acts only on shell surfaces exposed to water with a low saturation state, meaning that the brooding larval *Pisidium* sp. are potentially shielded from this process, which may confer some resilience to calcification of the whole pea clam (i.e., both adults and brooded larvae) under intermediate pCO2 conditions, potentially resulting in their observed parabolic responses to elevated pCO2.

It is also possible that decreased water pH impacts rates of shell formation indirectly, potentially through physiological processes beyond biocalcification. For the crayfish, *P. leniusculus*, we observed that feeding rates were reduced at the lowest pH despite no pH-dependent variation in respiration rates. We did not test for specific mechanisms, but similar reductions in feeding rates, arising from modified behavior, have been observed for marine decapods exposed to increased pCO2 [23]. Alternatively, lowered pH has been shown to cause a biomechanical weakening of the exoskeleton in some species of marine crustacea [24], and it is possible that this could impair their ability to handle and consume prey. Regardless of the cause, these reduced feeding rates under CO2-induced reductions in water pH would yield less food and energy for shell production by the crayfish, potentially contributing to or driving their linear decline in net calcification rate with increasing pCO2. Although the feeding rates of the other species were not quantified in the present study, obtaining these types of measurements in future experiments should provide valuable insight into the mechanisms responsible for the observed reductions in net calcification rate of freshwater calcifiers under conditions of elevated pCO2.

Anthropogenic increases in atmospheric pCO2 and the resulting decrease in freshwater pH has the potential to negatively impact freshwater aquatic ecosystems by altering the net calcification rate and behavior of calcifying invertebrates [1]. Differential sensitivities to CO2-induced changes in water chemistry can further shape populations, communities, and ecosystems [25]. Given the ecological importance of freshwater calcifiers in lakes, ponds, rivers, and streams [26–28], further research is merited to determine the range of responses that such species exhibit to elevated atmospheric pCO2, and the mechanisms driving these diverse responses. A better understanding of how freshwater species perform under acidification can help inform strategies for managing freshwater systems amidst the various threats that they face, including species invasion, eutrophication, sedimentation, warming, and CO2-induced acidification.

#### **5. Conclusions**


**Author Contributions:** Conceptualization, A.T.N. and J.R.; data curation, A.T.N. and J.R.; formal analysis, A.T.N. and J.R.; funding acquisition, A.T.N. and J.R.; investigation, A.T.N. and J.R.; methodology, A.T.N. and J.R.; resources, A.T.N. and J.R.; writing—original draft, A.T.N. and J.R.; writing review and editing, A.T.N. and J.R. All authors have read and agreed to the published version of the manuscript.

**Funding:** A.T.N. was supported by a Russel J. and Dorothy S. Bilinski Fellowship at Bodega Marine Laboratory. J.R. and this research were supported by MIT Sea Grant award no. NA18OAR4170105, by Northeastern's Interdisciplinary Sabbatical Program, by J.R.'s overhead return fund at Northeastern, and by J.R.'s personal funds. J.R. was also supported by the University of California at Davis Tahoe Environmental Research Center, which provided space, equipment, analytical resources, utilities, and boat/dive time in support of this research.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Data available upon request from the authors.

**Acknowledgments:** We are grateful to numerous scientists and staff members at the UC Davis Tahoe Environmental Research Center (TERC), including Anne Liston, Brant Allen, Katie Senft, and Brandon Berry, for their assistance in the laboratory and with collections in the field, and to TERC Director Geoffrey Schladow for hosting J.R.'s sabbatical and A.T.N.'s visiting appointment and for generously providing resources in support of this work. We are also grateful to Brian Gaylord for advice and comments on previous versions of the manuscript and to Cary and Cindy Ninokawa for their assistance with crayfish collection and feeding trials.

**Conflicts of Interest:** The authors are not aware of any conflict of interest. Additionally, the external funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript; or in the decision to publish the results.

#### **References**

