Abstract:

This dataset contains predicted marine mammal distributions for the Gulf of Mexico based on passive acoustic monitoring (PAM) [Gulf of Mexico (GOM); US exclusive economic zone (EEZ)]. Modeled species include sperm whales, Cuvier's beaked whales, pilot whales, Stenellid dolphins, Risso’s dolphin and Kogia spp. Model predictions are based on PAM conducted between 2010 and 2012 and have been validated using PAM data from 2013, and National Oceanic and Atmospheric Administration (NOAA) visual survey data from 2009. This dataset supports the publication: Frasier, K. E., Garrison, L. P., Soldevilla, M. S., Wiggins, S. M., & Hildebrand, J. A. (2021). Cetacean distribution models based on visual and passive acoustic data. Scientific Reports, 11(1). doi:10.1038/s41598-021-87577-1.

Suggested Citation:

Frasier, Kaitlin E, Lance P. Garrison, Melissa S. Soldevilla, Sean M. Wiggins, and John A. Hildebrand. 2021. Dataset for: Cetacean distribution models based on visual and passive acoustic data. Distributed by: GRIIDC, Harte Research Institute, Texas A&M University–Corpus Christi. doi:10.7266/n7-4xzx-d372

Publications:

Frasier, K. E., Garrison, L. P., Soldevilla, M. S., Wiggins, S. M., & Hildebrand, J. A. (2021). Cetacean distribution models based on visual and passive acoustic data. Scientific Reports, 11(1). doi:10.1038/s41598-021-87577-1

Purpose:

Output of a pilot study on integrating marine mammal observations from shipboard visual surveys and passive acoustic monitoring to predict spatial distributions of marine mammal populations in the Gulf of Mexico.

Data Parameters and Units:

Grids of predicted marine mammal species density in 10x10 km grid in units of animals per 100 km^2 are model output per species in each directory. Maps are provided for each month based on average monthly climatologies between 2004 and 2014. They do not represent occurrence in any particular year. Due to the dynamic nature of the oceanography of the Gulf of Mexico, these maps should not be to predict future occurrences at specific times. Habitat models are included for 7 species or species groups, denoted by the following abbreviations: Gg = Grampus griseus, Risso's dolphin; Gmsp = Globicephala spp., pilot whale, presumed short-finned based on region; Kspp = Kogia spp., pygmy/dwarf sperm whale; Me = Mesoplodon europaeus, Gervais' beaked whale; Pm = Physeter macrocephalus, sperm whale; Ssp = Stenella spp, offshore Stenella dolphins including clymene, attenuatta, longirostris, and coeruleoalba, but not frontalis, which is typically more coastal in the Gulf. For each species, models are produced by running _model_NN_density_GRIIDC.Rmd. These scripts can be compiled to HTML using Rmarkdown functionality, to reproduce _model_NN_density_GRIIDC.html. They are configured to load existing models, as the neural net training is the slowest step of the modeling process. Networks can be retrained by setting eval = TRUE on the following blocks in each script: r Ac only model; r Vis only model; r save models. All other files are inputs or outputs required or produced by these primary scripts. Please note that these models are the product of a pilot study comparing spatiotemporal predictions of models trained using visual survey and passive acoustic monitoring data. Model predictions are illustrative only, and should not be interpreted as a representation of the true distribution or density of animals in any given year. In particular, although climatologically-based average distributions are calculated by month, many of the models show an affinity of these species for mobile features such as eddies which are obscured by climatological averages. Please refer to the HTML documents included within the dataset for more information on each model.

Methods:

Maps were generated using habitat models trained using both visual and passive acoustic observations of three marine mammal species: Cuvier's beaked whale, sperm whale, and Risso's dolphin. Passive acoustic data were collected from 2011-2014 at five long term monitoring sites in the Gulf of Mexico. Shipboard visual survey data were collected by NOAA in the Gulf of Mexico from 2003-2014. Habitat variables included sea surface temperature, sea surface height anomaly, sea surface salinity, chlorophyll concentrations, mixed layer depth, upwelling velocity, surface current magnitude, and distances to cyclonic and anticyclonic eddies. Monthly predictions are based on mean climatologies for each of these habitat variables from 2004-2014 and should be considered preliminary. All code and outputs are included in this dataset.

Individual Files: