Date Awarded


Document Type


Degree Name

Doctor of Philosophy (Ph.D.)


Virginia Institute of Marine Science


John M Hoenig

Committee Member

Andrew M Scheld

Committee Member

David S Johnson

Committee Member

Elizabeth A Babcock

Committee Member

Cindy A Tribuzio


In the United States, the Magnuson-Stevens Reauthorization Act mandates that all federally fished species must have catch limits, which can be challenging for data-limited species. One approach is to assess and manage a group of species with similar life history characteristics, vulnerability to the fishery, and overlapping geographic distributions in a single management unit, or a complex (i.e., stock or species complex). Using the Gulf of Alaska (GOA) Other Rockfish complex as a case study, the main goals of this dissertation are five-fold: 1) review species complexes in the United States; 2) compare multivariate techniques for assigning species to complexes; 3) group species based on spatial and temporal patterns using a new application of a species distribution model (i.e., Vector Autoregressive Spatio-Temporal model, VAST, model); 4) compare catch advice between existing assessment models used for species complexes with that from the new spatio-temporal modelling (i.e., VAST) application; 5) refine management advice on appropriate species groupings and associated catch limits for this complex. In Chapter 1 a review was undertaken of all managed and assessed complexes in the United States, thereby identifying regional differences in management strategies and assessment models used to set catch limits for established complexes. The remaining chapters focused on the GOA Other Rockfish, a group of 27 Sebastes species. In Chapter 2, a suite of multivariate methods (e.g., cluster analyses and ordination techniques) was developed and applied on an array of datasets (e.g., life history values, fishery-dependent catch, and fishery-independent surveys), to examine how species groupings can vary depending on the methods or data utilized. Results indicated that the species composition for the two main gear types, trawl and longline gear, were different. Chapter 3 addressed the complex membership using a spatio-temporal species distribution model, which was used to investigate the temporal and spatial relationships among species and compared with groupings based on harvest fractions and life history values. Main results for species groupings were consistent across methods from Chapter 2 and 3, suggesting that rockfish belonging to a sub-group of the GOA Other Rockfish (i.e., members of the Demersal Shelf Rockfish) should be removed and managed separately from the Other Rockfish complex throughout the GOA management area. Using the resultant complexes, Chapter 4 compared two assessment models for the GOA Other Rockfish: the currently used random effects model and a newly, developed spatio-temporal model (VAST). While the results of this research are specific to the GOA Other Rockfish, the lessons and recommendations are applicable to other complexes with similar data availability. Multiple data sources and a variety of methods should be used to identify or verify complex membership, where the best species groupings are those that are consistent across all analyses. Variation in groupings across analytical methods and data inputs can provide further insight into data needs or species that warrant careful monitoring. Additionally, new assessment models for species complexes should be explored and tested to ensure results adequately reflect the status of the complex and provide reasonable harvest limits.




© The Author