Bigfin reef squid demonstrate capacity for conditional discrimination and projected future carbon dioxide levels have no effect on learning capabilities

Introduction

The coleoid cephalopods (squids, cuttlefishes, and octopuses), when compared to other invertebrates, show remarkably advanced learning and memory abilities (Dickel, Boal & Budelmann, 1999). Their complex nervous systems and highly diverse behaviours have been compared to those of lower vertebrates (i.e. fishes and amphibians), despite considerable evolutionary distance between these taxonomic groups (Boycott & Young, 1950; Boycott, 1961; Hanlon & Messenger, 2018). The morphology, physiology, ecology, and behaviours of coleoid cephalopods (henceforth referred to as cephalopods) were shaped by a coevolutionary arms race with modern teleost fishes (Mather & Kuba, 2013), resulting in many abilities and traits that both groups share (Packard, 1972; O’Dor & Webber, 1986). Among these are advanced eyes, large brains, and capacity for visual conditional discrimination. While conditional discrimination has been demonstrated in cuttlefish and octopuses (Hvorecny et al., 2007; Mather & Kuba, 2013), whether squid possess this trait has not been investigated.

Conditional discrimination is the ability to control discernment between different options through a sensitivity to context, which has significant benefits to individual performance (Mackay, 1991). For example, if reinforcement is delivered when an individual presses a square button and not a triangular button, discrimination is required for success. However, if reinforcement is contingent upon pushing the square button only after hearing the auditory stimulus ‘square’, conditional discrimination is necessary. In this example the auditory stimulus is the conditional stimulus, and the button shape is the discriminative stimulus. This allows for two initially unrelated stimuli to be associated with a single event, causing an emergent relationship to develop between those stimuli (Zentall, 1998). Conditional discrimination is considered ‘complex learning’ (Thomas, 1980, 1996) and has only been demonstrated in a small number of invertebrate species. Aside from in cuttlefishes (Sepia pharaonis and Sepia officinalis) and a species of octopus (Octopus bimaculoides), conditional discrimination has only been demonstrated in two other invertebrate classes, the sea hare, Aplysia californica (Colwill, Absher & Roberts, 1988), and the bees, Apis mellifera and Bombus impatiens (Couvillon & Bitterman, 1988; Brown et al., 1998; Mirwan & Kevan, 2015). The high level learning required for conditional discrimination undoubtedly contributes to the continued ecological success of the species, benefiting hunting, predator avoidance, social interactions, and navigation (Hanlon & Messenger, 2018). Thus, despite the paucity of studies, it is likely that conditional discrimination is common among a range of highly social or predatory animal taxa, such as squid.

Squid occupy an important mesopredator role in marine food webs as they both prey upon a wide range of marine species and are predated on by an even more diverse variety of species (Hanlon & Messenger, 2018). In mesopredators, the use of discrimination is important for both hunting and predator avoidance (MacDougall & Dawkins, 1998). In a predator capacity, all cephalopods detect prey predominately by sight and hunt with a wide array of techniques that vary depending on prey type (Boycott, 1954; Ross, 1971; Curio, 1976; Ross & Von Boletzky, 1979). Squid can use subtle visual cues to discern weaker or slower individuals within schooling fishes, ensuring greater capture success (Neill & Cullen, 1974). In a prey capacity, learning to identify predators can influence survival as predator recognition is not always innate knowledge (Brodie, Formanowicz & Brodie, 1991; Mitchell et al., 2011). Predator recognition also often involves fine tuning the efficiency of anti-predator responses by learning which predators are a substantive threat and which are not (Ferrari & Chivers, 2011).

Discrimination also supports spatial orientation, especially in benthic associated species such as octopuses and cuttlefishes (Hvorecny et al., 2007). Advanced conditional discrimination and long-term memory has been demonstrated in a number of cephalopod species (e.g. octopuses and cuttlefishes) using maze experiments (Schiller, 1949; Wells, 1964, 1967; Walker, Longo & Bitterman, 1970; Boal et al., 2000; Karson, Boal & Hanlon, 2003; Alves et al., 2006, 2007; Hvorecny et al., 2007). Compared to the mazes in laboratory experiments, conditional discrimination supporting spatial orientation in the wild often differentiates between much more specific and subtle cues within an unstable environment (Forsythe & Hanlon, 1997). The importance and level of complexity of spatial learning among species depends on their lifestyle and habitat type, but is expected to be widespread among animals (Capaldi, Robinson & Fahrbach, 1999).

Recently, projected future levels of carbon dioxide (CO₂) from increased anthropogenic activity have been shown to alter learning in some marine species (Ferrari et al., 2012; Chivers et al., 2014). Atmospheric concentrations of CO₂ have increased by more than 40% since the Industrial Revolution and the partial pressure of CO₂ (pCO₂) in the oceans rises at the same rate, a process known as ocean acidification (Dlugokencky & Tans, 2019). If CO₂ emissions continue at the current rate, atmospheric CO₂ could exceed 900 ppm by the end of this century (RCP8.5) (Collins et al., 2013). Furthermore, seasonal fluctuations in seawater pCO₂ are projected to be amplified with increasing CO₂ concentrations due to the increased Revelle (buffer) factor of acidified seawater (McNeil & Sasse, 2016). CO₂ concentrations matching these worst case scenarios for the end of this century elicit a range of altered behavioural responses in a variety of marine taxa (Clements & Hunt, 2015; Nagelkerken & Munday, 2016), including inhibited learning in fishes (Ferrari et al., 2012; Chivers et al., 2014).

Two species of tropical squid are among the animals that exhibit altered behaviour at elevated CO₂. The two-toned pygmy squid, Idiosepius pygmaeus, responds to elevated CO₂ with altered escape responses (Spady et al., 2014). Furthermore, I. pygmaeus and bigfin reef squid (Sepioteuthis lessoniana) respond to elevated CO₂ with altered predatory behaviours, such as increased latency to attack prey, and increased activity (Spady et al., 2014; Spady, Munday & Watson, 2018). Both in their predatory and anti-predator responses, these species displayed altered body pattern choice at elevated CO₂ (Spady et al., 2014; Spady, Munday & Watson, 2018). However, whether elevated CO₂ also affects learning in these species is unknown.

While the effects of elevated CO₂ on learning have not been investigated in any invertebrate species, research into the effects of elevated CO₂ on learning in fishes suggests a potential for similar adverse effects in other marine taxa. The tropical damselfish, Pomacentrus amboinensis, failed to learn to appropriately respond to a common predator fish after 4 days exposure to elevated CO₂ (~850 µatm) (Ferrari et al., 2012). In the temperate fish, Gasterosteus aculeatus, learning and memory is also affected by elevated CO₂ (~1,000 µatm), with reduced arena escape times over successive trials (Jutfelt et al., 2013). The effects of elevated CO₂ on learning in reef fishes have been linked to impaired function of GABA_A receptors, a major inhibitory neurotransmitter receptor (Chivers et al., 2014). In squid, GABA_A-like receptors and other inhibitory mechanisms are also important for learning processes (Conti et al., 2013). For example GABA_A is likely to be involved in the experience-dependent learning of prey capture and escape responses, which is dependent on the inhibitory control of the giant synapse output (Preuss & Gilly, 2000). While the relationship between CO₂ induced altered behaviour and GABA_A has not been investigated in squid, the effects of elevated CO₂ on the predator escape responses in the gastropod mollusc, the humpbacked conch Gibberulus gibberulus gibbosus, has been linked to interference with GABA_A-like receptors (Watson et al., 2014), which suggests that learning could also be affected by elevated CO₂ in molluscs.

In this study, we investigate the capacity for conditional discrimination in the bigfin reef squid and the effect of elevated CO₂ on their learning. Although conditional discrimination or associative learning has not been previously demonstrated in a species from the order of S. lessoniana (Teuthida), a wide range of cephalopod species appear to share advanced learning capabilities (Boal, 1996; Hvorecny et al., 2007; Karson, Boal & Hanlon, 2003; Zepeda, Veline & Crook, 2017). Furthermore, while conditional discrimination has not been demonstrated in many invertebrate species, it does occur in the relatively simple-“brained” sea hare, A. californica (Colwill, Absher & Roberts, 1988). Therefore, we expect that conditional discrimination is likely more widespread throughout the animal kingdom than is currently recognised, and we expect the trait is present in squid as well.

Sepioteuthis lessoniana was chosen due to their close association with coral reef habitats (Norman, 2003), indicating a high prioritisation of spatial orientation when compared to more pelagic squid species. The species is highly active and has one of the largest global distributions of any inshore squid, found throughout temperate and tropical waters from Japan to northern Australia and New Zealand and from Hawaii to the east African coast, north into the Red Sea and south to southern Mozambique and Madagascar (Jereb & Roper, 2005). The genus Sepioteuthis has been noted for their complex learned social and mating behaviours (Moynihan & Rodaniche, 1982; Hanlon & Messenger, 2018; Sugimoto & Ikeda, 2012). Furthermore, the previously demonstrated effects of elevated CO₂ on activity levels and predatory behaviours (Spady, Munday & Watson, 2018) make S. lessoniana an ideal subject for this study.

In the present study, squid were held at control (502 µatm) or elevated CO₂ (1,084 µatm) seawater treatments and subjected to repeated trials in a two-choice arena experiment with three separate tasks to complete, closely following the methods of Karson, Boal & Hanlon (2003) and Hvorecny et al. (2007) used for demonstrating conditional discrimination in cuttlefishes. Squid were trained using visual cues in repeated motivation-based trial and error tasks with the goal of using the designated exit. Learning was measured as improvements in trial completion over time as well as the number of trials needed to pass each task. By completing all three tasks, squid will have displayed a capacity for conditional discrimination. If learning or the potential conditional discrimination capabilities of S. lessoniana are altered by elevated CO₂, this could have far-reaching effects by adversely affecting the prey capture, predator avoidance, navigation, migration, or social interactions of an ecologically important species.

Materials and Methods

Testing arena

The testing arena and methods for the experiment described in this section closely resemble those of Karson, Boal & Hanlon (2003) and Hvorecny et al. (2007) designed for cuttlefish. A circular testing arena (56Ø × 40H cm) was constructed from dark grey PVC (Fig. 1). As with the experiments of Karson, Boal & Hanlon (2003) and Hvorecny et al. (2007), this is a two-choice exit arena, providing visual cues (object cues and exit frame) to teach individuals which exit is open and which is closed. A rectangular habituation chamber (27 × 15 cm) with a liftable gate, located at one end of the arena (equidistant and 90° from both exit points), was used to contain the animal before each trial. Upon entering the centre of the arena from the habituation chamber, the animal could view two circular exits (20 cm Ø and raised 10 cm from the base of the arena) located directly to the left and to the right. Visual object cues were placed directly opposite the habituation chamber gate except during preference testing trials. The arena and surrounding tank were filled with seawater to 29 cm deep (~500 L), leaving the top one cm of the exit above the water surface. This was done to avoid abrasions to the animals when passing through the exit as many individuals preferred to swim at the surface during trials.

Attached by Velcro around each exit was a square laminated exit frame (34 × 34 cm) (Karson, Boal & Hanlon, 2003; Hvorecny et al., 2007). One exit frame was solid yellow in colour while the other was yellow with black stripes. These two exit frames were alternated between the left and right exit for different individuals. A piece of clear Plexiglass™ that did not distort polarisation (Shashar, Rutledge & Cronin, 1996) was used to block the desired exit from the outside of the arena and was held in place by a brick at the base of the Plexiglass™, below the exit. Another brick was also placed in the same position outside of the unobstructed exit so as not to give any distinguishing characteristics to the blocked exit. Blocked exits were used to train squid to use a particular exit given a specific visual cue. In the experiments of Karson, Boal & Hanlon (2003) and Hvorecny et al. (2007), an arena with a transparent bottom was suspended above the tank floor in an attempt to encourage cuttlefish, which preferred staying on the substrate, to exit the arena to reach the floor. In this experiment however, preliminary tests of squid revealed an adverse reaction to the dark enclosure of the arena and a preference for areas of the tank with no obstructions above them. This instead was used as motivation for squid to escape the enclosed arena by navigating out one of the two exits. Therefore, a plywood board served as a lid to cover the top of the circular testing arena, while areas surrounding the arena were open and not covered. Trials were viewed via the display of a video camera (HC-V160; Panasonic, Osaka, Japan) which filmed downwards through a hole in the plywood board.

Results

All individuals

All 12 squid completed preference testing, taking between 9 and 30 trials. Four squid from the control treatment exhibited a strong preference towards one exit (5–6 uses of the same exit) and two squid from the elevated CO₂ treatment exhibited a strong preference towards one exit. There was no effect of elevated CO₂ on the preference strength of squid (GLM, χ² = 1.000, p = 0.317). All 12 animals met the passing criteria for Task 1 (testing against preference), taking between 7 and 38 trials. Seven individuals met the passing criteria for Task 2 (testing with preference) (between 7 and 26 trials). CO₂ treatment had no significant effect on their success in Task 2; four individuals were from the control treatment and three were from the elevated CO₂ treatment (GLM, χ² = 0.707, p = 0.401). Of the seven squid to reach Task 3 (testing for conditional discrimination), one of the four individuals from the control treatment failed to meet the passing criteria for Task 3, while all three squid from the elevated CO₂ treatment successfully met the passing criteria for this task; CO₂ treatment had no significant effect (GLM, χ² = 1.243, p = 0.265). Six of the twelve squid tested successfully completed all three tasks, indicating that S. lessoniana has a capacity for conditional discrimination. Of squid that demonstrated conditional discrimination, all individuals met the criteria for passing Task 3 in between 7 and 11 trials. Three of these squid were from the control treatment and three were from the elevated CO₂ treatment. Elevated CO₂ had no effect on the ability of squid to demonstrate conditional discrimination (GLM, χ² = 0.024, p = 0.876).

The mean time for individuals to exit the arena in Task 1 (testing against preference) had a clear positive relationship with their ability to reach Task 3 (testing for conditional discrimination) (GLM, χ² = 6.705, p = 0.010). Individuals that successfully reached Task 3 had a mean exit time in Task 1 trials of 32.1 s (±5.2 SE), whereas individuals that did not reach Task 3 spent only 10.1 s (±4.9 SE) in the arena during Task 1 trials (Fig. 2). There was also a positive relationship between the number of trials performed to pass Task 1 and the ability to pass Task 3 (GLM, χ² = 4.053, p = 0.044). The mean number of trials needed to pass Task 1 in individuals that passed Task 3 (20.8 ± 4.2 SE) was nearly double that of individuals that failed to pass Task 3 (11.3 ± 2.5 SE) (Fig. 3). All subsequent results focus only on the six individuals that successfully passed Task 3.

Individuals that demonstrated conditional discrimination

For individuals that successfully passed Task 3 and demonstrated conditional discrimination, the number of trials needed to successfully complete each task did not differ between CO₂ treatments (Fig. 4): Task 1 (GLM, χ² = 0.262, p = 0.609), Task 2 (GLM, χ² = 0.320, p = 0.571), Task 3 (GLM, χ² = 0.069, p = 0.793). There was also no difference between CO₂ treatments in the mean time to exit the arena within each task (Fig. 5): Task 1 (GLM, χ² = 0.012, p = 0.914), Task 2 (GLM, χ² = 0.433, p = 0.511), Task 3 (GLM, χ² = 0.371, p = 0.542). Improvements in exit times and correct exit choices of squid over successive trials within each task were observed in some tasks, but did not significantly differ between CO₂ treatments. In Task 1, the time for squid to exit the arena did not significantly decrease over successive trials (LME, χ² = 0.426, p = 0.514), and there was no effect of CO₂ treatment on this relationship (LME, χ² = 0.909, p = 0.340). There was an improvement in correct exit choice for squid over successive trials in Task 1 (GLMM, χ² = 5.806, p = 0.016), but this relationship was not influenced by CO₂ treatment (GLMM, χ² = 0.073, p = 0.787). There was no significant improvement in exit time in Task 2 over successive trials for squid (LME, χ² = 0.431, p = 0.511) and no effect of CO₂ treatment on this relationship (LME, χ² = 2.499, p = 0.114). Similar to Task 1, there was also a significant improvement in Task 2 in the correct exit chosen over successive trials (GLMM, χ² = 6.617, p = 0.010); CO₂ treatment had no effect on this relationship (GLMM, χ² = 1.960, p = 0.162). In Task 3, the time to exit the arena over successive trials for all squid did not improve (LME, χ² = 1.499, p = 0.221) and there was no effect of CO₂ treatment (LME, χ² = 2.005, p = 0.157). While there was a trend in Task 3 of increasing correct choices over successive trials, this was not significant (GLMM, χ² = 3.117, p = 0.078) and there was again no effect of elevated CO₂ (GLMM, χ² = 0.018, p = 0.892).

In individuals that successfully passed Task 3, there was a significant decrease in the number of trials needed to complete each task from Task 1 (mean of 21 trials) to Tasks 2 and 3 (mean of 8 and 10 trials, respectively) (GLMM, χ² = 40.842, p < 0.001) and elevated CO₂ had no effect on this relationship (GLMM, χ² = 1.249, p = 0.535). The mean time to exit the arena did not change significantly among Tasks 1, 2, and 3 (LME, χ² = 4.446, p = 0.108), and there was also no effect of elevated CO₂ (LME, χ² = 3.981, p = 0.137). The percentage of trials within each task in which individuals chose the correct exit increased from Task 1 (71% correct) to Tasks 2 and 3 (84% and 80% correct, respectively) but this was not significant (GLMM, χ² = 2.919, p = 0.232), and there was no difference between CO₂ treatments (GLMM, χ² = 2.103, p = 0.350) (Fig. 6). The probability of meeting the passing criteria for each task by chance alone (six correct exits out of seven trials) is 0.055. A power analysis revealed a power of 0.22 on the effect of CO₂ treatment on the ability to demonstrate conditional discrimination.

Discussion

Of the 12 bigfin reef squid (S. lessoniana) tested, six successfully completed the test for conditional discrimination (Task 3). This to our knowledge is the first demonstration of conditional discrimination in a cephalopod of the order Teuthida (i.e. squids). These six squid also showed significant learning abilities by demonstrating improvements in choosing the correct exit in subsequent trials within each task, and by completing Task 2 (testing with preference) and Task 3 (conditional discrimination) in significantly fewer trials compared to in Task 1 (testing against preference). The six squid that demonstrated conditional discrimination were represented by three individuals from control and three individuals from elevated CO₂, indicating that elevated CO₂ had no effect on the capacity of S. lessoniana to conditionally discriminate. Furthermore, elevated CO₂ did not have an effect on any of the traits measured within this experiment including the number of trials taken to complete each task, time to exit the arena, and percentage of correct exit choices within each task. This suggests that elevated CO₂ also has no effect on the learning or long-term memory of S. lessoniana. This experiment has brought to light some of the remarkable cognitive abilities of S. lessoniana, and suggests that seawater pCO₂ projected for the end of the century under a worst case scenario does not inhibit these abilities.

Squid in this experiment demonstrated individual variation in their learning and conditional discrimination abilities, and this individual variation was greater than any potential effect of CO₂ treatment. Importantly, individuals that passed all three tasks spent significantly more time in the arena during trials in Task 1 compared to individuals that failed to reach and pass Task 3. It is possible that spending more time within the arena gave the squid time to better assess their surroundings and commit visual cues to memory. Animals that failed to reach Task 3 were also more erratic in their Task 1 escapes in comparison to individuals that eventually demonstrated conditional discrimination. Throughout trials, squid that eventually demonstrated conditional discrimination typically slowly entered the arena and then made an exit choice after a brief pause somewhere between the two exits. In contrast, individuals that did not eventually demonstrate conditional discrimination typically moved more rapidly after exiting the habituation chamber, and jetted around the perimeter of the arena until reaching an exit. During Task 2, these rapidly jetting individuals had a mean percentage of correct exit choices of 51%, indicating complete randomness in their exit choice. Slower performing individuals that eventually demonstrated conditional discrimination chose the correct exit in Task 2 in a mean of 84% of their trials. This higher percentage of success, along with the significant increase in correct exit choice over successive trials, suggests that these individuals may have benefited from spending more time in the arena during Task 1 trials.

Individuals that were unable to demonstrate conditional discrimination completed Task 1 in approximately half as many trials compared to individuals that did demonstrate conditional discrimination by passing Task 3. It appears that performance during Task 1 was crucial for the squid to successfully reach and pass Task 3. A faster exit time and fewer number of trials for each task would seemingly indicate better performance. However, in this experiment it seems that it was important for individuals to spend sufficient time within the arena as well as to make a fair number of incorrect choices during Task 1 in order to understand the conditions of the arena. A drawback of this experiment is that once an individual passed Task 1, it was not presented that combination of visual cue and corresponding exit again until after it successfully passed Task 2 and reached the random trials. Animals that passed Task 1 in a very short number of trials (some in the minimum seven trials) may not have had sufficient trial and error experience in their learning process. Trial and error is an integral part of many learning processes, leading to sustained modifications of behaviour and improvements in complex tasks (Dayan & Balleine, 2002; Ruediger et al., 2012). Squid that completed Task 1 in a minimal number of trials may have benefited from a reintroduction to Task 1 if they were unable to complete Task 2 after a set number of trials.

Of the seven squid to reach Task 3, four were from the control treatment. However, one squid from control did not demonstrate conditional discrimination despite having reached Task 3. This squid, like individuals that failed Task 2, successfully chose the correct exit in only 50% of 26 total trials during Task 2. It seems likely that this individual was choosing the exit at random in Task 2 and met the passing criteria by chance. The lack of conditional discrimination among visual cues in this individual becomes apparent in the randomised trials of Task 3, in which only 35% of 60 trials did it choose the correct exit. Among animals that successfully demonstrated conditional discrimination, the mean percentage of correct exit choices was only slightly lower in Task 3 (80%) compared to in Task 2 (84%). A decrease in percentage of correct choices here was expected as the random aspect of Task 3 was a new concept for the squid. However, this difference was not as large as expected and shows advanced long-term memory, for up to 7 days, and context dependent discriminatory abilities.

The ability for cephalopods to conditionally discriminate and rapidly learn, especially during early life, is important in order to keep up with their fast-paced lifestyle (Dickel, Boal & Budelmann, 1999). Changes to these abilities could have adverse effects on squid predation, predator-avoidance, navigation/migration, and many other behaviours. This experiment provides no evidence that elevated CO₂ has any effect on the learning abilities of S. lessoniana. However, the small sample size, large variation in individual performance, and high difficulty of the passing criteria (50% pass rate) in this study may have masked any subtle effects of elevated CO₂ on learning. A power analysis revealed a probability of 0.22 of finding an effect of elevated CO₂ if one were present. Perhaps potential differences in performance between CO₂ treatments would be more detectible with a larger sample size, a more easily achievable associative learning test, and memory reinforcement in the first tasks. Nevertheless, the present study demonstrates advanced learning capabilities in squid, and no adverse effect of elevated CO₂ exposure on learning.

Adverse effects on learning arising from elevated CO₂ exposure has been demonstrated in damselfish and linked to inhibited GABA_A neurotransmitter function (Chivers et al., 2014). The effects of elevated CO₂ on GABA_A also appear to be responsible for a wide range of altered behaviours, such as in anti-predator responses in molluscs (Watson et al., 2014) and anxiety, lateralisation, and olfactory preference in fishes (Nilsson et al., 2012; Hamilton, Holcombe & Tresguerres, 2014; Heuer & Grosell, 2014; Lai, Jutfelt & Nilsson, 2015; Schunter et al., 2018). Previously demonstrated effects of elevated CO₂ on behaviour in S. lessoniana include increased activity and altered predatory behaviours (Spady, Munday & Watson, 2018), but the mechanism responsible for these changes has not been investigated. The importance of GABA_A receptors for learning in squid (Conti et al., 2013), and evidence of the lack of inhibited learning at elevated CO₂ in the current experiment, suggests that previously demonstrated altered behaviours in S. lessoniana at elevated CO₂ (Spady, Munday & Watson, 2018) may be due to a mechanism other than altered functioning of GABA_A receptors. It is important to investigate what mechanism is causing behavioural changes in order to understand why some important behaviours are altered, whereas other equally important behaviours are unaffected by elevated CO₂.

In addition to learning, the relatively large size of cephalopod brains are needed for processing large amounts of sensory information (Young, 1988). Cephalopods, for example can have 20 million receptor cells in the eye, requiring significant processing power from the brain and giving individuals a large amount of visual information to manage (Hanlon & Messenger, 2018). The ability to process complex visual cues, such as during male to male agonistic behaviour, is of great importance to their health and survival, as well as to the rewards of reproductive opportunities. As previous studies on fish have shown elevated CO₂ to inhibit learning via visual predator recognition (Ferrari et al., 2012), perhaps testing predator recognition learning at elevated CO₂ would be a valuable approach for S. lessoniana. Also, as Ferrari et al. (2012) used pre-settlement juveniles to test learning, it could be worthwhile to investigate effects of elevated CO₂ on learning in S. lessoniana as juveniles rather than adults. In this way, experiments could focus on naturally acquired learning during a life stage in which learning is more rapid and potentially more ecologically beneficial.

While the current study investigates learning in support of spatial orientation rather than in support of predator recognition, Jutfelt et al. (2013) tentatively reported inhibited learning in fish in an arena escape trial as a result of elevated CO₂ exposure. While both studies measured arena escape behaviours related to spatial orientation, there are a number of differences in the design of the studies and what was measured. Jutfelt et al. (2013) found reduced arena exit times in control fish from the initial trial to the second trial 20 days later, whereas fish held at elevated CO₂ showed no difference in exit times between the two trials. While this could be indicative of effects on learning, it could equally be a result of elevated CO₂ effects on exploratory behaviours (as the experiment was intended to measure), boldness, or activity levels. The present study is strictly an investigation into the learning of S. lessoniana with repeated motivation-based trial and error training, and provides no evidence of inhibited learning from elevated CO₂.

Conclusions

The importance of context learning in animals is of considerable ecological significance (Balsam & Tomie, 1985), and in squid, leads to the development of many important behaviours such as escape response, prey capture, body patterning, spatial orientation, and social behaviours (Hanlon & Messenger, 2018). The results from this study indicate for the first time that squid are able to conditionally discriminate, further expanding the known range of taxa with this ability. Furthermore, we found no evidence that elevated CO₂ levels projected for the end of this century and subsequent ocean acidification have any effects on learning behaviours of S. lessoniana. Future studies should investigate the effects of elevated CO₂ on a wider range of learning behaviours and in a variety of cephalopod species and life stages to ensure that these cognitive abilities will remain unaffected under future ocean conditions.

Supplemental Information