Spatial learning in intertidal gobies

Spatial learning and memory has been demonstrated in many terrestrial mammals, birds, reptiles and insects, however, considerably less is known about how/whether fishes use learning and memory in spatial orientation. The movements of fish within their natural habitats suggest impressive spatial abilities but this does not provide evidence for the role of spatial learning and memory or indicate what is being learned. Moreover, learning ability is likely to be shaped by natural selection, thus species may vary in their learning abilities depending on the environment they occupy. Intertidal Gobies in particular are ideal for exploring the relationship between spatial learning and memory capabilities and the environment. Importantly from a comparative perspective, they differ considerably, in their lifestyles and ecologies in response the extreme conditions in the intertidal zone (eg. shifting water levels, strong wave action and desiccation). Gobies respond to these fluctuating conditions in two main ways. Some species remain within the intertidal zone during low tide by seeking refuge from these adverse conditions in tide pools. Other species leave the intertidal zone completely during low tide only to return when conditions are more favourable (eg. high tide). Rock pools can be found in intertidal marine environments worldwide, however, there have been few studies exploring what drives their species composition, especially in Australia. This is quite disturbing considering the growing worldwide support for the preservation of marine biodiversity. Chapter 2 determined the major physical and biological factors influencing the distribution and abundance of rock pool fishes inhabiting the intertidal zone. This chapter evaluated the importance of within-pool shelters, algal cover, substrate type (rock or sand), rock pool depth, volume, and vertical position in limiting the distribution and abundance of these fishes. It was found that larger rock pools containing more algal and rock ledge cover hosted a larger and more diverse population of fish. Furthermore, certain species appear to only be found in pools with certain characteristics such as algal and rock cover. In intertidal fish, survival may be influenced by an individual’s ability to return to its home rock pool after high tide feeding excursions and thereby avoid being stranded in unsuitable areas at low tide. Chapter 3 tested site fidelity and homing ability for five species of intertidal rock pool fish by taging and displacing them to new rock pools at various distances from their ‘home’ rock pools. Three of the species were rock pool specialists whilst the remaining two spend a small proportion of their life in rock pools during early ontogeny. The three specialists showed strong site fidelity with > 50% of individuals found in the same pool 42 days after tagging. In contrast, the non-specialist species showed low fidelity and poor homing abilities. Homing success in the rock pool specialists remained relatively stable as displacement distance increased. The effect of body size on homing ability was species dependent, with only one species showing a significantly greater tendency to home with increasing size. Given that some rock pool species show acute homing abilities one would expect that they should also display spatial learning and memory capabilities on a smaller scale (within pool). It is unlikely that natural selection would favour spatial learning skills in fishes inhabiting sandy shores because they simply shift back and forth with the tides. Chapter 4 determined if inter-tidal gobies collected from rocky platforms and sandy beaches differ in their ability to learn and memorize the spatial locations of tide pools in a simulated rocky inter-tidal zone. Gobies were tested individually for spatial learning using a small spatial scale rendition of a rock platform (tanks =1162cm2), containing four artificial tide pools that retain varying depths of water at simulated low tide. The gobies were determined to have a capacity for spatial learning/ memory if they were able to repeatedly locate the tide pool that retains the most water at simulated low tide as the safest refuge. Rock pool species were able to relocate the correct pool to wait out low tide for ~95% of the trials, while species from sandy shores appeared to have ignored the rock pools completely and followed the tide out (only found in the correct pool for ~10% of trials). It is expected that rock pool species memorised the location of rock pools during simulated high tide so that they could be relocated at low tide. The ability to learn and recall the location of biologically important resources in a changing environment is essential for an animal’s survival and reproductive success. Chapter 5 investigated the cues gobies use when learning orientation routes, and determined if they varied depending on visual stability of their natural habitats. Gobies collected from rocky platforms and sandy beaches were trained to locate a hidden reward in a T-maze. Locating the reward required the fish to learn a body-centered pattern of movement (turn left or right) or to follow plant landmarks. This study found that rock pool species learnt the location of the correct shelter much faster, made fewer errors and used both types of cues available (plant landmarks and turn direction) to locate the shelter, while sand species relied significantly more on turn direction for orientation. For fish living in areas where predation is a high risk an effective strategy for escape, such as being able to relocate a refuge, should be of high importance. Chapter 6 investigated whether habitat stability influences spatial learning ability and cue use by comparing the performance of gobies from two contrasting habitats in their ability to locate shelter. Fish were trained to locate a shelter under the simulated threat of predation and asked weather they used local or global cues to do so. It was hypothesized that fish from rock pools would outperform fish from sandy shores and the two groups would differ in their cue use. This study found that rock pool species learnt the location of the correct shelter much faster, made fewer errors and used both types of cues available (local landmarks and global cues) to locate the shelter, while sand species relied significantly more on global cues for orientation. The results reported here support the hypothesis that fish living in complex habitats are more likely to rely on local landmarks as directional cues than fish living in mundane habitats, where local cues such as visual landmarks are unavailable. The ecological cognition hypothesis poses that brains and behaviour of individuals are largely shaped by environments in which they live and associated challenges they must overcome during their lives. Chapter 7 examined the effect of environmental complexity on relative brain size and structure in collected from rocky platforms and sandy beaches. Rock pool species are repeatedly found in their home pools at low tide, whereas beach dwelling species follow the tide out, hence it was predicted that these two groups would vary in brain structures as a function of habitat complexity. This study found that rock pool dwelling species had relatively larger brains and telencephalons in particular, while sand dwelling species had a larger optic tectum and hypothalamus. In general, it appears that various fish species trade off neural investment in specific brain lobes depending on the environment in which they live. We suggest that rock pool species require greater spatial learning abilities to navigate in their spatially complex environment, while sand dwelling species, on the other hand, likely have reduced need for spatial learning due to their spatially simple habitat and a greater need for visual acuity. When correlating brain structures with behavioural and environmental characteristics, a variety of techniques can be utilised. Chapter 8 quantitatively compared differing methodologies used to determine brain volumes using the conventional techniques of histology and approximating brain volume to an idealized ellipsoid (or half ellipsoid) and through the use of a 3D model acquired using a micro-CT scanner. It was found that MRI and micro-CT methods generated more reliable results than those of histology and the ellipsoid method, however they are more expensive to generate. Over 350 goby species have established themselves within Australian waters, however, very few of these species have been included in phylogenetic analyses. Chapter 9 used partial mitochondrial D-Loop sequences to infer phylogeny among nine Australian goby species. Phylogenetic analysis identified a number of interrelationships among Australian gobiidae and produced a depiction of lineage evolution that is mostly consistent with previous studies. Four main lineages were identified within the Gobiids analysed which corresponds to the habitats they utilise; sand gobies were represented by the genera Acentrogobius, Favonigobius and Istigobius. Rockpool gobies were represented by a single genus Bathygobius, and the inshore goby lineage combined Glossogobius and Psammogobius into a clade. The final lineage, burrowing mud dwelling gobies, was represented by the genera Scartelaos and Taenioides. The phylogeny provides a greater understanding of how particular morphological characteristics, habitat preferences and behaviours have coevolved. These results pave the way for comparative studies of these common intertidal fish species. In summary, this research highlights the relationship between a gobies’ capacity for spatial learning/ memory and the environment, and will not only greatly extend our understanding of the evolutionary forces shaping learning and memory formation in fish but also in vertebrates generally.