Cannibalism of small numbers of offspring by a parent has been proposed as an adaptive parental strategy, by providing energy to support parental care. However, there are few empirical studies to support this hypothesis. We conducted field and laboratory experiments to investigate partial filial cannibalism in Stegastes leucostictus, a coral reef fish with paternal care. Partial cannibalism was shown to be common, and males were found to remove developing embryos from throughout a clutch in a random pattern, rather than in the more aggregated pattern seen during embryo predation. Males that received a diet supplement grew faster than control males, but did not engage in less cannibalism. Also, males did not concentrate cannibalism on early embryonic stages with the highest energetic value. Experimental reduction of embryo densities was found to significantly increase embryo development rate and survival from egg deposition to hatching, and experimental reduction of oxygen levels significantly increased rates of partial filial cannibalism by males. Artificial spawning sites with low oxygen levels were avoided by spawning females, and cannibalism rates by males were higher. We propose that partial filial cannibalism serves as an adaptive parental strategy to low oxygen levels in S. leucostictus by increasing the hatching success of embryos.

No abstract

Parental male beaugregory damselfish Stegastes leucostictus commonly cannibalize entire egg clutches under natural conditions. Clutch size was experimentally reduced, but a relationship between the extent of clutch reduction and the frequency of total cannibalism was not detected.

The interactions between beaugregory damselfish and ophiuroids, potential damselfish egg predators, were investigated using several field studies. A survey of back reef areas showed that the two most common ophiuroid species, Ophiocoma echinata and Ophioderma appressum, were significantly associated with the territories of beaugregory males within the coral rubble habitat where both ophiuroids and beaugregory's can be found. Feeding experiments showed that three ophiuroid species (Ophiocoma echinata, Ophioderma appressum and Ophiocoma wendtii) and Echinometra viridis consumed late development stage beaugregory eggs, whereas earlier stages were only consumed by Ophioderma appressum and Echinometra viridis. Manipulation of ophiuroid densities had no significant effect on the survival of beaugregory eggs and had no measurable impact on female mate choice. The aggressive response of male beaugregory damselfish towards three ophiuroid and one echinoid species showed significant differences among species but were lower than those shown to bluehead wrasse, Thallasoma bifasciatum; the principal daylight predator of eggs. The intensity of attacks by male beaugregory damselfish towards Thallasoma bifasciatum increased significantly if males were guarding eggs, but did not appear to change for the four echinoderm species. Male habitat selection was shown to be unaffected by the presence of ophiuroids.

Linking models of animal behaviour and habitat management: Atlantic salmon parr and river discharge Understanding preferences of animals is of fundamental importance for modelling habitat quality and quantity. Important theoretical developments, for example using ideal free and ideal despotic distributions (IDD), have enabled biologists to build conceptual frameworks for relating habitat preferences of individual animals to distributions and dynamics of populations. At the same time, managers of natural resources have established predictive empirical models as a basis for understanding habitat quality. For example, the Physical Habitat Simulation model (PHABSIM) has been widely applied for managing river flows. The aim of this study was to conduct experiments, using Atlantic salmon parr, to test whether observed population distributions could be predicted using simple behaviour theory and PHABSIM. We show that predictions from PHABSIM depend crucially on population density, discharge and the interaction between density and discharge at the time when the model is parameterized. These findings can, in part, be explained by consideration and application of the IDD. However, the results of the experiments also suggest that models derived from first behavioural principles may need to be unexpectedly complex and species-specific if they are to capture the population response to variations in water discharge effectively.Fish cognition: what can we learn, and what do they need to learn?Fish provide a wonderful opportunity to explore processes that shape and select cognitive ability. In this presentation, I will illustrate three aspects of work that my colleagues and I have used to investigate fish learning and memory over the last decade. First, I will discuss how comparing different populations sampled from contrasting habitats allows differences in cognitive ability to be related to the evolutionary ecology of the fish. I will use examples that have investigated how differences in learning ability between populations of the same species can arise. Here, the examples will be taken from the ubiquitous three-spined stickleback, and a Panamanian poecilid, Brachyraphis.The second approach has used fish cognition as a tool to quantify behaviour to enable assessment of different aspects of fish welfare. For example, the recent work investigating pain perception in trout required the use of a learning task to quantify how fish behaviour was modified after noxious stimulation. Ways in which these, and similar, processes can be used in future studies of fish welfare will be discussed.The final part of the presentation will consider recent work that addresses the problems of releasing hatchery-reared fish for restocking purposes. Although a common practice, most of the hatchery-reared fish die shortly after they are released. Much of the observed mortality apparently stems from the fishes' inexperience with a variable environment. Experiments with juvenile cod and brown trout suggest that both age, and the early rearing environme...