A holistic model of the factors controlling reef-fish populations, including both pre- and post-settlement processes

The first thing that anyone diving on a coral reef will notice is the multitude of fishes. More careful observation of fishes will also highlight that each species is more abundant in some places than others. What controls the number of fish of each species at different sites? Like any species, the number of fishes is dictated by rates of births, deaths, immigrations, and emigrations. However, the processes that control these rates are numerous, and difficult to study simultaneously. For example, death rates might be affected by the number of fishes being eaten by other fishes, the amount of fishing, and the number that die because of competition with other species or because there is not enough food. Furthermore, birth rates are difficult to estimate because most fish larvae are small, develop away from reefs, and are hard to count when they do arrive at a particular site. Despite these problems, if we could make a realistic model of the processes controlling the number of fishes on a reef, it would provide important insights into reef ecology, and aid conservation work. There have been lots of studies on one or a few factors affecting fish populations, and this means it is possible to link them together to construct an overall theoretical model. For example, we know that predation by carnivorous fishes significantly affects the number of juvenile fishes on a reef. We also know that habitats with more coral will support more fishes because there are more places for fishes to hide and avoid predators. I have developed a theoretical model that summarises all these interactions. Structural equation modelling (SEM) is a powerful analytical technique that allows complex theoretical models to be tested with data collected during surveys. SEM provides results that show whether the model adequately explains the patterns seen in the data, which interactions in the model are most important, and what changes can be made to improve the model. Although the model could be used to test the factors controlling any fish species, I will focus on six species that have been selected because they have quite different life histories, are common, and are relatively well studied. A potential problem with SEM is that you need lots of data (surveys from at least 100 different sites) to test the theoretical model. This much data is rarely available to ecologists, but I will be able to get sufficient data by combining new surveys undertaken in The Bahamas during my fellowship with existing data from a project that already has data from 75 sites. My new surveys, from another approximately 75 sites from six islands in The Bahamas, will consist of fish counts and video surveys of the reef itself. Once the new and existing data have been combined, I will have data on the abundances of each target species at the 150 sites, along with details on a range of important variables such as the number of predators, the amount of fishing pressure, and the amount of coral. I will also work with a researcher at the University of Miami who has a state-of-the-art model for predicting the dispersal of larvae around The Bahamas. This model will provide an estimate of the number of larvae likely to arrive at each of my sites. These data will give an unprecedented opportunity to consider the importance of larval supply (effectively 'births') relative to a range of other factors. By combining theoretical models with all the data, I will gain a range of insights into the process controlling the number of fishes seen at any given reef site. In addition to a better ecological understanding of reefs, these findings will help conservation efforts. By knowing the relative importance of key factors that affect different fish populations, and the interactions among them, managers will be able to design more effective marine reserves. For example, is it better to place a marine reserve where lots of fish larvae arrive, or where there are large areas of nursery habitat?