A Predictive Model for Geographic Distributions of Mangroves

Climate change is an impending disaster which is of pressing concern more and more every year. Countless efforts have been made to study the long-term effects of climate change on agriculture, land resources, and biodiversity. Studies involving marine life, however, are less prevalent in the literature. Our research studies the available data on the population of mangroves (groups of shrubs or small trees living in saline coastal intertidal zones) and their correlations to climate change variables, specifically, temperature, heat content, various sea levels, and sea salinity. Mangroves are especially relevant to oceanic ecosystems because of their protective nature towards other marine life, as well as their high absorption rate of carbon dioxide, and their ability to withstand varying levels of salinity of our coasts. The change in global distribution was studied based on global distributions of the previous year, as well as ocean heat content, salinity, temperature, halosteric sea level, thermosteric sea level, and total steric sea level. The best performing predictive model was a support vector regressor, which yielded a correlation coefficient of 0.9998.

Mangrove Geographic Distribution Over Time

Climate change has always been an impending disaster, and is becoming an issue of pressing concern more and more every year. Countless efforts have been made to study the long-term effects of climate change, however this data is multifaceted in nature, and can be studied from a variety of angles. The goal of our research is to study the available data on the population of mangroves - groups of shrubs, or small trees, living in saline coastal intertidal zones, and studying the change in their distribution over time. The change in global distribution was studied based on distribution in the previous year, as well as ocean heat content, salinity, temperature, halosteric sea level, thermosteric sea level, and total steric sea level. The predictive model that performed significantly better than the rest was a support vector regressor, which yielded an $r^2$ value of 0.9998.