Characterizing distributions and drivers of emergent aquatic vegetation in Minnesota
Date
2020
Authors
LaRoe, Jillian, author
Vogeler, Jody C., advisor
Tinkham, Wade, committee member
Evangelista, Paul, committee member
Journal Title
Journal ISSN
Volume Title
Abstract
The emergent aquatic vegetation (EAV) communities across the lakes of Minnesota serve critical functions within ecosystems by providing habitat and forage for native waterfowl and fish species, moderating water chemistry, and serving as a cultural and economic resource. Communities of EAV are changing dramatically in response to alterations in hydrologic flow regimes, nutrient availability, biological homogenization, and near-shore development. To address the conservation of these communities at a spatial scale relevant for landscape management, the changes need to be evaluated at local and regional scales. Previous efforts to map and monitor EAV have utilized field surveys, aerial imagery, multispectral imagery, and synthetic aperture radar (SAR). However, it is difficult to apply the findings of previous studies to broader spatial scales because they lack field surveys, clear or repeatable methodologies, rigorous validation, and/or applying methods to broad spatial extents, all of which are all necessary for providing direct implications for landscape level management. The first chapter of this thesis aimed to overcome these challenges and create statewide maps of EAV in Minnesota at a spatial resolution relevant to landscape management at both broad and local scales. We paired detailed field surveys of EAV communities with Sentinel-1 SAR and Sentinel-2 Multispectral Imager to create annual maps of EAV across the lakes of Minnesota at a 10 m spatial resolution in 2017 and 2018. We created two random forest models, a species model predicting general classes of EAV and a water model identifying open water regions across hydrologic features in Minnesota. We validated both classification models using withheld field sample locations to measure overall accuracy as well as individual class user's and producer's accuracies. The species and water map predictions were combined into a final map representing water and EAV classes each year. We also evaluated each map by the area-based percentage of overlap between model predictions and field surveys which ranged from 54.5 to 90.1% agreement. The 2017 map was further evaluated using an area-based weighted probability with an overall accuracy of 89.9% (±0.7%). The methods and promising results highlighted by this study set the stage for subsequent analyses at broader spatial scales to quantify temporal shifts or trends in EAV communities. The combination of these diverse and detailed datasets provides methods for generating annual maps of EAV distribution across Minnesota, and ultimately provide a tool to support landscape-scale conservation efforts of EAV communities in Minnesota. The second chapter investigated the influence of systemic drivers related to the decline of northern wild rice (Zizania palustris L.) over the last century. Wild rice is an environmental indicator species that is sensitive to hydrologic changes and disturbances and serves an essential role in ecological, cultural, and economic systems in Minnesota. Due to the previous lack of comprehensive information regarding its extent and distribution, previous efforts to study its decline have been limited to small regions or small samples of lakes across the state. We utilized 2018 presence maps of wild rice from the first chapter and summarized wild rice cover across 366 lakes. Then, we employed a suite of spatial, hydrological, ecological, and environmental variables summarized at a variety of spatial scales within a three-step modeling framework to select the most significant drivers of wild rice cover, explore interactions between drivers, and account for inherent spatial autocorrelation in the datasets. A final spatial lag model revealed that dispersal and population connectivity had the strongest relationships with wild rice cover on each lake. While further exploration may better quantify this relationship, land managers should consider the degree of connectivity between wild rice lakes and their spatial configuration on the landscape during conservation planning to maximize population resilience. Our results suggest that it may be more suitable to approach populations as connected habitat regions, in contrast to the more widely accepted notion that wild rice lakes are self-contained or independent populations.
Description
Rights Access
Subject
Minnesota
Sentinel-2
emergent aquatic vegetation
wild rice
Sentinel-1