Publication Date

Fall 2018

Degree Type

Thesis

Degree Name

Master of Science (MS)

Department

Moss Landing Marine Laboratories

Advisor

Scott L. Hamilton

Keywords

remote sensing, rockfish, species distribution models, species-habitat associations

Subject Areas

Conservation biology

Abstract

Seafloor maps are often used in species distribution modeling (SDM), where maps are paired with fish observations to create models predicting habitat suitability, species density, or species biomass. Problems with the current use of SDM include limited understanding of species relationships with benthic morphology, lack of practical model testing, and deficiency of information on the effects of map resolution on population estimates. A drop camera was used to gather observations of fishes along Central California and paired with remotely sensed bathymetry to create predictive models and maps of species density and biomass. I found that relationships with remotely sensed habitat variables are strong enough to create robust models. However, predictive maps at 10m resolution only gave a broad-scale picture of density distributions. Predictive maps consistently overpredicted species density, but often underpredicted peaks in density. Map resolution had a large effect on biomass predictions, where total predicted biomass was found to increase with increasing resolution. In conclusion, predictive maps seem to capture general patterns of species distributions; however, often peaks or hot spots in density are not captured. Predictive maps are very useful for understanding general patterns of species distributions, but one should be cautious when using them to obtain density of biomass estimates, especially when using estimates to inform management.

Share

COinS