Kristin April Meseck

Thesis title: "Habitat Distribution for Non-native Amazona viridigenalis Within San Diego County using Maxent Predictive Model"

Abstract

Human propagated changes to the environment have adversely effected certain species while advantaging other species. Psittacines, or species that fall within the parrot family, have been found to be well adapted to modified environments. Over time, transportation of various parrot species for use in the exotic pet trade has caused accidental releases of individual parrots, resulting in species groups forming and colonizing in new, non-native environments, specifically urban and suburban ones. Amazona viridigenalis, the Red-crowned parrot, is a species that has adapted to living in several regions within the United States including Texas, Florida, and California. This species is endangered within its home range (where it is), yet has the most numerous and established population of any other psittacine species in California but remains severely understudied in its new range. Using geographic information systems and Maxent predictive model, this research aims to achieve a greater understanding of the role that the Red-crowned parrot plays in its new environment of San Diego County and the habitat variables that enable its establishment success. Presence locations of where individuals of the species were using habitat were collected along with 12 important variables that represent Red-crowned parrot habitat elements. These were used in the creation of a predictive habitat model utilizing Maxent machine-learning technique. Three models were created using three different background extents from which the pseudo-absence points were generated. These models were tested for statistical significance and predictive accuracy. It was found that model performance slightly declined with a decrease in model extent. The largest extent was chosen to model habitat using the five variables that proved to be the least correlated, achieved the most gain, and contributed the most to earlier models. The final model generated depicts highly suitable habitat for A. viridigenlis along the entire coast of San Diego County with pockets of moderately suitable to highly suitable habitat infiltrating the interior. Some locations in East County were also identified as having moderate habitat suitability. It was found that A. viridigenlis requires habitat with elevations between sea level and 200 m., has slopes of zero to 11% grade, and large amounts of non-native vegetated ground cover. More suitable habitat was found at locations with moderate to high population density and in close proximity to roads, suggesting the importance of human-presence in shaping habitat for A. viridigenalis.