Determining the drivers of diversification: Evolutionary models of island biogeography, their robustness, and the effects of data uncertainty

Islands and archipelagos are often used by biologists to study evolution because they are isolated systems, making it easier to avoid confounding factors that are often found on continents. However, islands, archipelagos, and mainlands are not without their idiosyncrasies; they have many features that may be sources of bias if not accounted for, such as sea level changes, land connectedness, area etc. Additionally, biological data and the methods to collect it have limitations and may also introduce biases. Thus, I investigated and quantified the effect of ignoring these biases and sources of uncertainty for the island evolutionary model DAISIE. I did this by creating computer simulations that explicitly modelled these biases in a controlled manner and developed strategies to deal with uncertainty by interpreting data in several ways. I was able to determine the impact of ignoring these biases, and in some cases, suggest better alternatives to deal with them. I applied this approach to a diverse set of systems: to study the geological history of islands, the uncertainty of evolutionary datasets of spiders from the remote archipelago of Hawai’i, and a global dataset of how the areas of islands have changed over thousands of years. Unlike what is often assumed, existing models are robust (insensitive) to many island and archipelago features, and the impact of this data uncertainty can often be mitigated. I also outline specific cases where action must be taken to avoid drawing wrong conclusions and suggest strategies to avoid common pitfalls.