Project Title: Do deep-sea corals have an additive effect on fish occurrence and abundance when controlling for physical habitat?
Goal or Purpose

To provide improved statistical techniques for evaluating evidence for association of structure-forming deep coral with enhanced fish presence, abundance, diversity, and/or biomass, and to apply these techniques to assess fish-coral habitat associations using a large database of fish and coral ROV image surveys in the Southern California Bight.  

Anticipated Management Application(s)

Our results will improve the database of information available to make spatial management and conservation decisions with respect to deep-sea corals in the Southern California Bight, an important nexus of fishing activity and deep-sea coral abundance.  An improved understanding of the potential associations between deep-sea corals and sponges and managed fisheries will help managers conduct future EFH reviews, evaluate spatial management and conservation plans, and develop fishery regulations and technologies that reduce potential impacts on deep coral and sponge ecosystems.  If a direct link between coral habitat and fish response variables can be quantified and verified, it may also aid future ecosystem-based management measures and stock assessments. 

Fiscal Funding:
  • FY 2015 @ $75,000

  • Pacific Council
  •  None Defined.
Project Type:
  • Habitat Suitability Modeling
Point of Contact: Office of the Point of Contact:
Team Members:
  •  None Defined.
Project Title: Do deep-sea corals have an additive effect on fish occurrence and abundance when controlling for physical habitat?

Data: As described above, PI Stierhoff at SWFSC manages a large database of benthic ROV imagery (>26,000 unique images) from the Southern California Bight.  Methods involved in collecting such imagery and classifying fish and seafloor habitat are described in Stierhoff et al. (2013).  Seafloor habitat is classified using a system modified from Stein et al. (1992) and Greene et al. (1999), which is also compatible with and could be converted with relatively little effort into the CMECS version 4 classification system (FGDC 2012).  As part of a previous DSCRTP project, PI Etnoyer reviewed all images in the SWFSC database and identified and counted deep-sea corals and sponges.  Methods for this processing are described in Etnoyer et al (2013).  We now propose to thoroughly review this database, ensuring that a consistent seafloor classification system has been used and documented in a manner consistent with the CMECS classification scheme (FGDC 2012), QA/QC’ing all fish data, and merging and QA/QC’ing all coral and structure-forming sponge data.  We will then extract all relevant data from the image database, processing it into a flat file suitable for import into R statistical software and containing all relevant putative independent variables (seafloor physical habitat classification and coral presence, abundance, species richness, and size classes), and dependent variables (fish occurrence, abundance, species richness, length classes, and biomass).  

We will also engage the research coordinator at CINMS and obtain and process additional ROV still/video surveys conducted in 2012 in the CINMS.  Processed images will be entered into the SWFSC database with all relevant fish, seafloor habitat, and coral annotations.

We will also process oceanographic and high-resolution multibeam bathymetric data available in the immediate vicinity of each still image (available for a substantial percentage of the images) in order to derive larger-scale physical habitat metrics such as slope and rugosity, and annotate each database record with this information for use as a physical habitat covariate.


Analysis: In consultation with the SWFSC and CCEHBR PI’s, PI Kinlan and the CCMA team will lead development of a set of additive statistical models to quantitatively evaluate the question of whether coral and structure-forming sponge-related dependent variables have a significant effect, independent of the shared effect of common physical habitat, on fish response variables. Several statistical techniques will be explored and compared, including boosted regression trees (BRT), generalized additive models (GAM), and component-wise ensemble-boosted generalized additive models of location, scale, and shape (boosted GAMLSS).  PI Kinlan and his team are experts in these modeling techniques and have already developed a substantial library of code that will be modified and customized for this project.   The final model results, as applied to the SWFSC database, will be described in a report, presentation, and journal article as outlined below.

Project Results and Management Outcomes  None Defined
Project Title: Do deep-sea corals have an additive effect on fish occurrence and abundance when controlling for physical habitat?
Internal References:
  •  None Defined
  •  None Defined

There are currently no items to list.