Defective skeletogenesis and oversized otoliths in fish early stages in a changing ocean

Atmospheric carbon dioxide (CO₂) concentration has increased from pre-industrial levels of 280 μatm to present-day levels of 394 μatm, and it is expected to rise to 730–1000 μatm by the end of the century (Caldeira and Wickett, 2003; Meehl et al., 2007). Continuous CO₂ uptake by the world's oceans is changing the seawater chemistry and is estimated to lead to a drop of 0.4–0.5 units in seawater pH (Caldeira and Wickett, 2005). Concomitantly, the temperature of the oceans is rising, and global sea surface temperature is expected to increase ~4°C by 2100 (Meehl et al., 2007), leading to profound impacts on marine ecosystems. In fact, the predictable rapid rate of climate change will induce thermal stress to coastal marine biota as their thermal tolerance limits are reached or even exceeded. Beyond a certain thermal limit, biological processes such as metabolism, growth, feeding, reproduction and behavior may be affected (Carmona-Osalde et al., 2004; Pörtner and Knust, 2007; Nilsson et al., 2009; Byrne, 2011; Pimentel et al., 2012; Rosa et al., 2012), thus compromising the overall fitness and survival of species. Additionally, under higher temperatures, marine organisms are likely to be more vulnerable to other environmental stressors such as ocean acidification (Pörtner, 2008; Byrne et al., 2010; Findlay et al., 2010; Parker et al., 2010; Sheppard Brennand et al., 2010; Byrne, 2011; Rosa et al., 2013; Rosa et al., 2014).

Ocean acidification is considered a major threat to marine organisms as it may lead to acid–base balance disturbances, protein biosynthesis decrease, metabolic depression and growth reduction (Seibel and Walsh, 2001; Pörtner et al., 2004; Langenbuch et al., 2006; Rosa and Seibel, 2008; Baumann et al., 2012). Exposure to elevated CO₂ particularly affects calcifying organisms (Orr et al., 2005; Dupont et al., 2008; Fabry et al., 2008; Talmage and Gobler, 2010), although detrimental effects on survival, growth and respiratory physiology of non-calcifying marine animals have also been observed (Seibel and Walsh, 2001; Rosa and Seibel, 2008; Munday et al., 2009b).

Fish have developed an effective acid–base regulatory mechanism, which allows them to accumulate bicarbonate and exchange ions across gills under hypercapnic conditions (Pörtner et al., 2005; Ishimatsu et al., 2008; Melzner et al., 2009). While this is true for adult organisms, early life stages may not benefit from it, as they lack well-developed and specialized ion-regulatory mechanisms to regulate and maintain their internal ionic environment (Morris, 1989; Sayer et al., 1993). Therefore, early life stages are expected to be the most vulnerable to ocean climate-change-related conditions and their eventual inability to cope and adapt may constitute a bottleneck for species persistence in a changing ocean (Bauman et al., 2012; Frommel et al., 2012). Until now, only a few studies have scrutinized the impact of ocean climate change on fish larvae performance. While some report negligible effects of ocean acidification on fish larvae (Munday et al., 2011b; Hurst et al., 2012; Harvey et al., 2013; Hurst et al., 2013; Maneja et al., 2013), others demonstrate that ocean warming and acidification may have a direct impact on embryonic development, larval growth, metabolism, behavior and survival (Bauman et al., 2012; Franke and Clemmesen, 2011; Frommel et al., 2012; Bignami et al., 2013; Pimentel et al., 2014). More recently, it has also been shown that larval otoliths can be affected by changes in seawater carbonate chemistry (Checkley et al., 2009; Munday et al., 2011a; Bignami et al., 2013), but the impact of hypercapnia on larval fish skeletogenesis still remains unclear.

In the present study, we investigated how the combined effect of warming (+4°C) and high partial pressure of CO₂ (pCO₂; 0.16% CO₂, ~1600 μatm, ΔpH=0.5) affects the hatching success, larval survival, growth, metabolic rates, thermal tolerance limits and skeletogenesis of early life stages of a flatfish, Solea senegalensis Kaup 1858, with major commercial importance. This teleost fish is an environmentally resilient species that inhabits the Western Iberian Upwelling Ecosystem, the northern limit of the Canary Current Upwelling System, one of the four major eastern boundary currents of the world, where pCO₂ levels may reach up to ~500 μatm (Álvarez-Salgado et al., 1997; Pérez et al., 1999; Borges and Frankignoulle, 2002). Thus, organisms inhabiting such upwelling ecosystem are commonly exposed to seasonal high pCO₂ events, because of the emergence of deep hypercapnic water masses. In these regions, the future pCO₂ levels are thus expected to exceed the forecasted 1000 μatm for 2100 (Meehl et al., 2007).

The impact of high pCO₂ and environmental warming on the hatching success, survival, length and growth of S. senegalensis larvae is shown in Fig. 1 (see also supplementary material Table S1). Warming had a significant negative impact on the hatching success of sole larvae (P<0.05), but neither hypercapnia (P>0.05) nor the interaction between hypercapnia and warming had a significant effect (P>0.05). The hatching rates decreased from 86.7±5.8% at the present-day scenario to 70.0±10.0% under the future hypercapnic and warming conditions (Fig. 1A).

Survival rates of 30 days post hatching (dph) larvae were also significantly affected (Fig. 1B). Both temperature and pCO₂ had a significant effect (P<0.001) on survivorship, which decreased from 45.7±1.9% under control conditions to 32.7±2.6% in the future scenario. However, the interaction between the two variables was not significant (P>0.05). The mean length of 30 dph larvae under control conditions was 13.2±1.5 mm (Fig. 1C). Larval growth increased significantly with warming (P<0.05), but decreased significantly under acidified conditions (P<0.05), with an observed significant interaction effect between these two variables (P<0.05). Warming was responsible for increasing length by 48.6 and 46.5% under normocapnic and hypercapnic conditions, respectively. Regardless of temperature, S. senegalensis larvae became nearly 22% smaller with increasing CO₂. As a result, the highest length value (19.4±1.1 mm) was observed under the warming and normocapnic scenario, while the lowest length (10.3±0.9 mm) was found at lower temperature and hypercapnic conditions. An almost identical trend was observed for specific growth rate, which presented a 23.7–28.4% increase with warming and an 11.9–15.1% decrease with acidification (Fig. 1D). No significant effect of the interaction between these two factors was observed (P>0.05).

The effect of warming and high pCO₂ on the metabolic rates and thermal tolerance limits of S. senegalensis larvae is presented in Fig. 2 (see also supplementary material Table S2). Temperature had a positive effect (P<0.05) on oxygen consumption rates (OCR), upper thermal tolerance limits (LT₅₀) and critical thermal maximum (CT_max), while hypercapnic conditions promoted a significant reduction (P<0.05) of these physiological parameters. Even so, no significant effect of the interaction between these two factors was observed (P>0.05). The OCR of 30 dph larvae increased with temperature from 23.1±3.2 to 34.8±3.5 μmol O₂ h⁻¹ g⁻¹ and from 16.8±3.8 to 25.3±1.5 μmol O₂ h⁻¹ g⁻¹ under normocapnic and hypercapnic conditions, respectively (Fig. 2A). These findings represent a decrease of 27.3% under acidified conditions. The LT₅₀ of 30 dph larvae increased with temperature from 37.5±0.1 to 37.7±0.0°C under normocapnia, and from 36.1±0.1 to 38.8±0.3°C under hypercapnia conditions (Fig. 2B). The CT_max of 30 dph larvae followed a similar pattern as for OCR and LT₅₀, increasing with temperature from 37.0±0.9 to 38.3±0.5°C under normocapnia, and from 35.5±0.6 to 37.3±0.7°C under hypercapnia (Fig. 2C). Additionally, the development stage had a significant effect (P<0.05) on metabolic rates and thermal tolerance limits. Solea senegalensis hatchlings presented higher OCR and lower LT₅₀ and CT_max values in comparison to 30 dph larvae (Fig. 2).

Thermal sensitivity (Q₁₀) of S. senegalensis larvae between 18 and 22°C ranged between 1.89 and 2.79 (Table 1). Q₁₀ values decreased under acidified conditions and increased with fish age.

Several types of skeletal anomalies were found in 30 dph S. senegalensis larvae (Table 2, Fig. 3). Skeletal deformities consisted mainly of vertebral abnormalities, such as fusions (Fig. 3C–G), body malformations (Fig. 3C,D), and vertebral curvatures such as scoliosis, lordosis and kyphosis (Fig. 3I,J). Structures such as haemal and neural spines and arches were some of the most affected structures across treatments (Fig. 3C–G).

Future ocean warming and high pCO₂ conditions had a significant effect on the incidence of skeletal deformities in S. senegalensis larvae (Figs 4, 5; see also supplementary material Table S3). Rising temperature and CO₂ levels increased the frequency of total skeletal deformities (Fig. 4A), from 70.9±2.6% at the present-day scenario to 93.2±2.7% under the future conditions (P<0.05), an increase of 31.5%. No cranium or pectoral fin deformities were observed under control temperature and pCO₂ rearing conditions. Under the future scenario, caudal vertebra was the most affected region (Fig. 4D), followed by cranium (Fig. 4B), caudal fin (Fig. 4E), abdominal vertebra (Fig. 4C), pelvic fin (Fig. 4H), dorsal fin (Fig. 4F), and finally the pectoral fins (Fig. 4H). In what concerns severe skeletal deformities, pCO₂ was the main factor contributing to the higher proportion of deformities observed in the future scenario (Fig. 5). Under present-day conditions, less than 1.9% of the larvae presented severe vertebral curvatures such as scoliosis (Fig. 5B) or lordosis (Fig. 5C), and no kyphotic larvae were observed (Fig. 5D). In contrast, all types of severe anomalies significantly increased (P<0.05) with future environmental predictions, especially with high pCO₂. The interaction factor between temperature and pCO₂ did not have a significant effect (P>0.05) on the incidence of skeletal deformities (including the severe ones), except for abdominal vertebra and dorsal fin deformities.

Otolith size was also greatly affected by future warming and hypercapnia conditions (Fig. 6; see also supplementary material Table S1). Solea senegalensis larvae experienced a 109.3% increase in otolith area with increasing temperature and pCO₂ (P<0.05). Otolith area increased from 1063.6±398.8 mm² under the present-day conditions to 1994.5±234.5 mm² under warming, and then to 2226.2±187.0 mm² under the combined effect of rising temperature and pCO₂. However, the interaction of both factors was not significant (P>0.05).

The future predictions of ocean warming and acidification were revealed to have a negative impact on several aspects of the early ontogeny of the environmentally resilient flatfish S. senegalensis. Despite the short embryonic development time of this species (less than 2 days), the warming experienced during egg incubation was enough to elicit a negative effect on hatching success. Hatching rates decreased 16.7 percentage points with warming and acidification, in comparison to the present-day conditions. Moreover, the high temperature and pCO₂ levels had a further negative effect on larval survival, representing a decrease of 28.4 percentage points in relation to the present scenario.

As expected, larval growth greatly increased with warming. Increased temperature was responsible for increasing length by 46.5–48.6%. Nevertheless, it is important to keep in mind that this increment does not reflect differences in size at a specific stage of development, as development is accelerated at higher temperatures. In contrast, larval growth decreased under high pCO₂ levels. Contrary to some studies that have shown that larvae can become bigger under high pCO₂ conditions (Munday et al., 2009a; Hurst et al., 2012; Hurst et al., 2013), S. senegalensis larvae became almost 25% smaller with increasing pCO₂.

An almost identical trend was observed for larval metabolic rates and thermal tolerance limits. While temperature had a positive effect on OCR (within normal Q₁₀ values) and thermal tolerance limits, hypercapnic conditions triggered a significant reduction in such physiological parameters. Additionally, and as expected, mass-specific metabolic rates decreased with development, while thermal tolerance limits revealed an opposite ontogenetic trend, i.e. older larvae revealed higher thermal tolerance limits than newly hatched ones. We presume that exposure to higher pCO₂ might have impaired acid–base balance regulation, which directly affects the efficiency of cellular activities (Pörtner et al., 2005; Perry and Gilmour, 2006) and may cause deleterious effects on larval physiology and growth.

Faster growth at higher temperatures could have some advantages, because slower growing larvae are potentially more vulnerable to predators and may thus experience greater mortality (Anderson, 1988). Nevertheless, growth enhancement with temperature might also present some disadvantages, because faster larval growth was accompanied by an increase in the incidence of skeletal deformities. Indeed, temperature is known to be one of the most important environmental factors that can induce morphological deformities during fish development (Aritaki and Seikai, 2004; Georgakopoulou et al., 2010; Dionísio et al., 2012). Additionally, pH may also affect the prevalence of fish skeletal deformities (Lall and Lewis-McCrea, 2007). Although fish skeleton is predominantly composed of calcium phosphate (in the form of hydroxyapatite and cartilaginous material) (Lall and Lewis-McCrea, 2007), additional buffering of tissue pH with bicarbonate and non-bicarbonate ions is expected under acidified conditions, which may interfere with larval skeletal development. In this study, the future warming and high pCO₂ scenario was responsible for increasing the incidence of total skeletal deformities by 22.2 percentage points, affecting 93.1% of the larvae. Moreover, high pCO₂ was the main factor responsible for the increase of severe skeletal deformities in flatfish larvae. Under the present-day conditions, less than 1.9% of the larvae presented vertebral curvature deformities such as scoliosis or lordosis, and no kyphotic larvae were observed. In contrast, more than 50% of the larvae under the future environmental scenario presented vertebral curvature deformities. These findings, however, are in disagreement with a recent study that found no effects of CO₂ on the skeletal development of a reef fish (Munday et al., 2011b).

However, the higher incidence of malformations under the future scenario should be carefully interpreted. The high percentage of skeletal deformities found in S. senegalensis under control temperature and pCO₂ conditions (70.9±2.7%), although similar to the values commonly found for this species under intensive rearing conditions (Fernández et al., 2009; Dionísio et al., 2012), may indicate that fish were potentially stressed in captivity and would, therefore, be more susceptible to the negative effects of higher temperature and CO₂ levels. Nevertheless, this fact does not exclude the amplifying effect that warming and hypercapnia had on the incidence of skeletal deformities. Even though the increase may be overestimated, the higher rate of malformations in captive larvae under high temperature and pCO₂ conditions may provide an insight into how future warming and acidification may impact the development of wild flatfish larvae and their future performance in a changing ocean.

Skeletal deformities may impair the ecophysiological performance of fish larvae in many different ways. Vertebral curvatures and fin deformities may affect larval swimming behavior, feeding efficiency and the capacity to maintain their position in a current (Powell et al., 2009). Additionally, larvae with cranium deformities, such as ocular migration anomalies, probably will have their capability to feed, attack prey and avoid predators affected. Larvae with operculum deformities may increase gill's susceptibility to fungus, bacteria and amoebic parasitic infections (Powell et al., 2008) and, as a result, their swimming and cardiovascular performance might be compromised (Powell et al., 2008; Lijalad and Powell, 2009; Powell et al., 2009). Additionally, fish with dental, premaxillar or maxillar deformities cannot adduct their mandible and, besides having potential feeding restrictions, the buccal-opercular pumping of water across gills is also likely to be impaired and compromised (Lijalad and Powell, 2009).

In addition to skeletal deformities, S. senegalensis larvae under this future climate change scenario will also be affected by changes in otolith size. Solea senegalensis larvae experienced a 109.3% increase in otolith area with rising temperature and pCO₂. Although otoliths are calcified structures composed of aragonite–protein bilayers, recent studies revealed that pH regulation in otolith endolymph may lead to increased precipitation of calcium carbonate in otoliths of fingerlings exposed to elevated CO₂ (Checkley et al., 2009; Munday et al., 2011a; Bignami et al., 2013). However, this is not a rule among fishes. In at least one coral reef fish species, otolith size was not affected by exposure to elevated pCO₂ (Munday et al., 2011b). Otoliths are used by fish for orientation, perception and acceleration, and to maintain postural equilibrium. Thus, changes in otolith size may have implications for their ecological performance, behavior and individual fitness (Gagliano et al., 2008; Bignami et al., 2013).

In conclusion, the results presented in our study provide comprehensive insight into the combined effects of ocean warming and hypercapnic conditions on S. senegalensis larval development. Fish larval stages represent a critical life phase for species' ecological success. Therefore, climate-change-related impairments in metabolism, thermal tolerance, growth, skeletal development and survival may lead to substantial declines in adult populations, putting in jeopardy the species' persistence under a climate change scenario.

Solea senegalensis eggs were obtained from a wild-caught broodstock of four females and two males, under natural spawning conditions at Instituto Português do Mar e da Atmosfera (IPMA), Centro Regional de Investigação Pesqueira do Sul (CRIPSul, Olhão, Portugal), during June 2012. After collection, eggs were transported and immediately transferred, under environmentally controlled conditions, to the aquaculture facilities in Laboratório Marítimo da Guia (Cascais, Portugal). To estimate the potential physiological responses of early life stages to climate change, S. senegalensis eggs and larvae were acclimated for 1 month at: (1) 18°C – control temperature, the mean sea surface temperature in summer (sSST) – and normocapnia (0.04% CO₂, pCO₂=~400 μatm, pH=8.0); (2) 18°C and hypercapnia (0.16% CO₂, pCO₂=~1600 μatm, ΔpH=0.5, pH=7.5); (3) 22°C – the future sSST warming scenario for the western coast of Portugal in 2100 [+4°C above the average sSST (Meehl et al., 2007)] – and normocapnia; and (4) 22°C and hypercapnia. Prior to releasing the eggs in the rearing tanks, a 2 h thermal and chemical acclimation was performed.

Eggs and larvae were reared in 12 individual recirculating systems (i.e. three systems per treatment), filled with filtered (series of 20, 10, 5 and 0.35 μm) and UV-irradiated natural seawater. Each system comprised a 19 l cylindrical shaped tank (larval rearing tank) connected to a 100 l sump. All rearing tanks were placed inside 400 l water bath tanks (see supplementary material Fig. S1), where temperatures (18.0±0.2 and 22.0±0.2°C) were maintained and controlled via seawater chillers (HC-1000A, Hailea, Guangdong, China), in order to ensure thermo-controlled conditions.

The photoperiod was set at 14 h:10 h light:dark. Water filtration was performed through mechanical (glass wool), physical (protein skimmer, Schuran, Jülich, Germany) and biological (ouriço® bioballs, Fernando Ribeiro, Portugal) filters, as well as UV sterilization (TMC, Chorleywood, UK). Throughout the experiment, ammonia and nitrite levels were monitored daily and kept below detectable levels. Temperatures were controlled via seawater chillers (Frimar, Fernando Ribeiro, Portugal), while pH was adjusted automatically via a Profilux system (GHL, Kaiserslautern, Germany) connected to pH probes (WaterTech pH 201S) in the rearing tanks and to a standard solenoid valve system connected to a CO₂ tank. Any seawater pH modifications initiated CO₂ addition (if the pH increased) or CO₂ filtered air injection (if the pH decreased), until pH returned to the set value. Additionally, temperature and pH were controlled daily using a digital thermometer (Ebro thermometer TFX430) and a portable pH meter (SevenGo pro™ SG8, Mettler Toledo). Mean values were 18.0±0.2 and 22.0±0.2°C for temperature and 8.02±0.05 and 7.51±0.05 for pH. Salinity was kept at 35.4±0.4. Seawater carbonate system speciation (Table 3) was calculated weekly from total alkalinity [determined according to Sarazin (Sarazin et al., 1999)] and pH measurements. Bicarbonate and pCO₂ values were calculated using the CO2SYS program (Lewis and Wallace, 1998), with dissociation constants from Mehrbach et al. (Mehrbach et al., 1973) as refitted by Dickson and Millero (Dickson and Millero, 1987).

Newly hatched larvae were randomly placed into rearing tanks (19 l volume each) at a stocking density of 70 larvae per liter. All larvae were reared until 30 dph under the different experimental conditions. The feeding schedule was based on larval development under each set of experimental conditions. Larvae opened their mouth at approximately 2 dph and started to feed on rotifers, Brachionus plicatilis, at a density of 5 to 10 rotifers ml⁻¹. Live enriched (AlgaMac-3050) Artemia metanauplii were introduced at 5 dph and their proportion was gradually increased from 0.5 to 12 metanauplii ml⁻¹, becoming the only prey offered at 8 dph. Frozen metanauplii were also introduced as feed after larval settlement.

Hatching success was analyzed in small rearing boxes placed inside the rearing tanks (one per rearing system). In the beginning of the experiment, a total of 10 eggs (per box) were randomly placed inside each of the 12 boxes (three per treatment), and these were followed throughout the embryonic development. The hatching success was calculated as the percentage of eggs that hatched to normal larvae.

The survival rate was calculated as the percentage of surviving fish by the end of the experiment, with respect to the number of larvae at the beginning of the trial minus those individuals removed for sampling.

To identify and quantify larval skeletal deformities, 20 larvae per rearing tank (60 larvae per treatment) were randomly sampled and fixed in 4% (v/v) buffered paraformaldehyde for 24 h and then transferred to 70% ethanol until double stained. Larvae were stained for bone and cartilage using a modification of the method described by Walker and Kimmel (Walker and Kimmel, 2007), and observed under a stereoscopic microscope (Leica S6D, Leica Microsystems) in order to identify skeletal deformities. Skeletal deformities were defined according to previously established methods (Wagemans et al., 1998; Gavaia et al., 2002; Deschamps et al., 2008; Fernandez et al., 2009; Dionísio et al., 2012). Deformities were divided into several categories according to the affected structure (e.g. cranium, abdominal vertebra, caudal vertebra, caudal fin, dorsal fin, pectoral fin and pelvic fin), and are described in Table 2. Skeletal deformities such as scoliosis, lordosis, kyphosis, multiple vertebral fusions or more than three anomalies per individual were considered severe deformities. Skeletal deformities were quantified as the percentage of fish exhibiting a specific deformity.

In order to analyze otolith area, 20 larvae per rearing tank (60 larvae per treatment) were randomly selected, measured and preserved in absolute ethanol. The left and right sagittal otoliths of each individual were removed and photographed under a stereoscopic microscope (Leica S6D, Leica Microsystems). Otolith area was measured using the ImageJ program. Otolith area was calculated as the mean of the right and left otoliths, and normalized to fish length.

ANOVA was used to test for significant differences between the tanks of each experimental treatment. As no differences were found between tanks, all of the samples from the same treatment were pooled and analyzed together. Two-way ANOVAs were then conducted in order to detect significant differences in hatching success, larval survival, standard length, SGR, skeletal deformities and otolith size between temperature and pCO₂ treatments. Three-way ANOVA were applied to detect significant differences in OCR, LT₅₀ and CT_max between temperature and pCO₂ treatments and development stage (0 and 30 dph). Subsequently, post hoc Tukey's honest significant difference tests were performed. All statistical analyses were performed using a significance level of 0.05, using Statistica 10.0 software (StatSoft Inc., Tulsa, OK, USA).

We thank CRIPSul for supplying fish eggs, and Oceanário de Lisboa and Aquário Vasco da Gama for supplying rotifers and microalgae. We also thank Lloyd Trueblood for helpful suggestions and critically reviewing the manuscript.