ailich
commented
7 years ago My master's research focuses on studying fish-habitat relationships so I'd be interested in seeing how habitat affects the presence of marine mammals. It might be useful to narrow it down to one (or a few) species for modelling. Also, I believe OBIS-SEAMAP has data regarding the presence of marine mammals. Roberts et al 2010 (http://www.sciencedirect.com/science/article/pii/S1364815210000885) used a spatial toolbox they created for ArcGIS to model spotted dolphin presence, which could potentially provide some ideas/guidance.
tttang0602
Author: Kaitlin Frasier Topics: Habitat models, marine mammals
Summary of Synthesis
Habitat models are predictive tools that use information about where and when marine mammals were observed in the past to predict where they will be in the future based on oceanographic conditions, season, and species characteristics. This information is helpful for mangers seeking to understand and mitigate impacts of various human activities (e.g. shipping, fishing, natural resource exploration) on protected marine mammal populations. This project synthesizes data from visual marine mammal surveys (good spatial resolution) and passive acoustic monitoring (good temporal resolution) to develop models capable of predicting species distributions across seasons and oceanographic regimes. Model output is typically in the form of color-coded maps indicating areas of high and low animal density across the GoM region. Models can be used to predict across seasonal averages (e.g. create a map of typical GoM sperm whale densities for the month of May), or to estimate responses to specific oceanographic conditions.
A few challenges arise from this synthesis project: 1) Models should not be static. Survey data continue to be collected, therefore an ideal model framework could accept future datasets to continually improve model predictions. 2) Models (implemented in R using open source packages) are complex programs that would benefit from collaborative and ongoing development to remain relevant. 3) Model predictions need to be accessible and usable by non-scientists. A hosting mechanism is needed to make the data available to those who need the information, with temporal and spatial scales relevant to their applications.
The proposed group synthesis project involves developing solutions for these three challenges based on the core training topics addressed in the NCEAS Open Science course. At a minimum, the project would outline roadmaps for implementing solutions to the challenges. One or more solutions could be more deeply investigated through implementation if desired. This work combines data and methodologies from researchers at Scripps Institution of Oceanography (SIO), NOAA Southeast Fisheries Science Center (SEFSC), and Duke University. It also requires perspectives of stakeholders at BOEM, state level managers and industry partners.
Data Needs
There are no immediate data needs for this project. NOAA SEFSC visual survey data and SIO passive acoustic monitoring data are available for 2002 - 2014. Environmental covariates drawn from satellite data repositories hosted by institutions including NOAA, NASA, and AVISO, as well as hindcasts of oceanographic conditions from the HYbrid Coordinate Ocean Model (HYCOM) are available for the time and region of interest, and freely accessible.
Analytical approaches
Habitat models are built using Generalized Additive Models (GAMs), and incorporate environmental covariates to predict marine mammal distributions. Covariates can be drawn from online repositories using an existing toolkit (MGET; Roberts et al. 2010) implemented in Python. A working implementation of a GoM habitat model framework in R associated with this project can be found at https://github.com/kfrasier/AcoustoVisualDE. This code base is uses standard, freely available packages for distance sampling (MRDS; Laake 2016) and GAMs (MGCV; Wood 2017).
With the data and preliminary statistical analysis methodologies in place, this project provides a unique opportunity to focus on issues directly related to the NCEAS core topics including: 1) Development of a strategy for hosting and maintaining the data and code base in a way that encourages collaboration and updates. This might be through wikis and code repositories, or via a more sophisticated framework in collaboration with existing data hosting entities. 2) Reaching out to the anticipated user base to better understand their needs and use cases. A survey of potential users at government and private institutions could be developed to design an appropriate interface for accessing useful model outputs.
Impact and Significance
The GoM is one of the most heavily utilized marine ecosystems in the world. Managers in the region make complex decisions about shipping, fishing, and natural resource extraction that require consideration of protected marine resources, but data are often unavailable, sparse, or opaque. Through this project, the GoM could become a model for making key research findings available and usable by the non-scientific community. This tool would also be valuable in the future for emergency response, damage assessment, and mitigation efforts. Further, marine mammals are both “charismatic megafauna” capable of raising awareness about intractable offshore conservation issues, and “ocean sentinels” whose population status provides a window into overall ocean health. Creating tools to protect these species helps conserve the rich and complex ecosystems that support them.