Predicting Invasion Risk in Coastal Marine Environments Utilizing Species Distribution Modeling

Download or read book Predicting Invasion Risk in Coastal Marine Environments Utilizing Species Distribution Modeling written by Raymond Eliot Crafton and published by . This book was released on 2015 with total page pages. Available in PDF, EPUB and Kindle. Book excerpt: Determining species' distributions and understanding the underlying factors that limit species' ranges is a central objective in ecological research. This pursuit is especially important to the study of invasive species and understanding where species are likely to be able to survive outside of their native range. Invasive species can cause significant ecological and economic damages, and opportunities for species introduction are becoming increasingly available in contemporary society. In order to minimize the impacts of invasive species, preventing species from being introduced and early detection are paramount. This dissertation presents research that utilizes species distribution modeling to identify potential ranges for non-native coastal marine species as a component of assessing invasion risk when combined with introduction likelihood. Three chapters are presented that examine different aspects of this research. In chapter 1, disparities between observed and modeled richness are identified on a global scale for a set of invasive ascidians. Several regions are highlighted as having lower observed richness than modeled richness, primarily in subtropical and tropical regions. While these regions are identified as potentially at risk from invasion, it is not possible to differentiate between under invaded and under sampled regions, which leads to a discussion on bias in sampling and observation data. Chapter two presents a direct assessment of invasion risk as the result of commercial shipping for five crab species globally. Using species presence observations and major world port locations, I calculate the introduction likelihood as a function of distance from a port. This is then overlaid on modeled environmental suitability for each species resulting in an assessment of relative invasion risk. Major ports are globally distributed, but they have uneven prevalence. Temperate locations primarily in the northern hemisphere have the highest introduction likelihood, and these regions represent hotspots for invasion risk for those species with environmental suitability in these areas. However, this chapter considers the impact of having an incomplete representation of suitable conditions in the training data and how this impacts the final invasion risk assessment. In chapter three, I look specifically at invasion risk in New Zealand for eight species of concern. I model environmental suitability for current and future conditions to see how climate change might impact expected distributions and discuss the implications this has on invasion risk. Several temperate species are modeled to lose potentially suitable environments in New Zealand with a warmer climate. For these species, invasion risk may decrease over time. In contrast, species that are tolerant of warmer climates, such as Caulerpa taxifolia, will likely have greater environmental suitability as the result of climate change and have a higher invasion risk in the future as more points of introduction overlap with suitable environments. This chapter also looks at the impact of model parameterization (including feature type, background, environmental layer, and threshold selection) on the model output and argues for an ecological sensible model despite this parameterization not having the highest fit according to common metrics (area under the curve and true skill statistic). Invasive species are a significant concern worldwide, and having a better sense of where species pose the greatest risk of invasion is helpful to aid in mitigating associated dangers. This research supports the use of species distribution models to aid in assessing invasion risk, but it highlights several important factors to consider when utilizing these models.