Fisheries-independent trawl surveys have faced criticism for violating key assumptions of stratified random sampling, employing short tows which reduce capture efficiency, and causing mortality of the catch. These limitations are particularly consequential for the Western Gulf of Maine stock of Atlantic cod (Gadus morhua), a species of cultural significance, economic importance, and regulatory complexity due to its depleted status and classification as a choke species. In response to these challenges, the School for Marine Science and Technology (SMAST) developed a video trawl survey, which employs cameras mounted in the open cod end of a trawl to identify and enumerate groundfish. Initial pilot studies began in 2016, followed by the implementation of large-scale, semi-annual surveys of the Western Gulf of Maine in 2020.Accurate estimates of local density from the video trawl survey required estimates of catchability, capture efficiency and fish length. This project aimed to address these requirements through three experiments and then evaluate sampling methodology in a fourth experiment. First, a passive integrated transponder (PIT) tag detection system was developed, tested, and installed in the codend of the net. Second, a mark-recapture experiment to estimate the efficiency and catchability of Atlantic cod was conducted using this system. Third, the accuracy of length measurements derived from an off-the shelf stereoscopic camera mounted within the trawl was assessed. Finally, optical data from the survey were used to evaluate the effects of sampling design, tow duration, and sampling intensity on the variance of population estimates through a novel analytical approach. The custom-designed PIT tag detection system achieved an efficiency of 79%,with detection rates influenced by tag orientation and group size. A Petersen mark recapture model, based on 1,094 tagged fish and six recaptures, accounted for both discard mortality and reader efficiency, yielding an estimated door spread capture efficiency of 12% and a catchability coefficient of 0.0024 per hour of towing. The off the-shelf stereoscopic camera produced inaccurate length measurements. However, these findings helped inform the design of a custom imaging system to include a global shutter, rapid integration time, wider field of view, and larger sensor. Analysis of optical data indicated that stratified random sampling produced more precise biomass estimates than simple random sampling. Additionally, CPUE mean and variance increased with shorter tow durations. A 30-minute tow duration minimized within-tow variability and yielded the most precise abundance estimates, although this analysis lacked factors such as fish size and logistical constraints. Collectively, this research advances fisheries-independent survey methodology by addressing key limitations of new approaches. Although further work is needed to refine the 12% efficiency estimate, it would conservatively result in a 257% increase over previous video trawl survey cod biomasses when using the upper 95% confidence limit. These results highlight concerns in the assessment of Western Gulf of Maine cod, producing an estimate of SSB that is 160% of the stock assessment projection, for only 13% of the stock area. This SMAST video trawl survey technology represents an improvement upon traditional survey techniques and can provide model independent estimates of biomass to enhance the assessment of Western Gulf of Maine cod.
- Look but don't touch
- Nicholas M. Calabrese - University of Massachusetts Dartmouth, Department of Fisheries Oceanography
- 0000-0001-6798-8584
- Kevin D.E. Stokesbury (Advisor) - University of Massachusetts Dartmouth, School for Marine Science and TechnologySteven Cadrin (Committee Member) - University of Massachusetts Dartmouth, Department of Fisheries OceanographyPingguo He (Committee Member) - University of Massachusetts Dartmouth, Department of Fisheries OceanographyMichael J.W. Stokesbury (Committee Member)
- xiii, 170 pages
- illustrations (color)
- List of tables -- List of figures -- Introduction -- Video trawl survey methods -- Chapter 1. Designing and evaluating the efficiency of a codend PIT tag detection system -- Introduction -- Methods -- PIT tag detection system design -- Cod tagging protocol -- PIT tag detection system testing -- H01: The efficiency of the codend PIT tag detection system is 100% -- H02: The efficiency of the codend PIT tag detection system is unaffected by tag orientation -- H03: The efficiency of the codend PIT tag detection system is unaffected by multiple tags passing simultaneously -- H04: The codend PIT tag detection system only reads tagged fish within the codend -- Results -- H01: The efficiency of the codend PIT tag detection system is 100% -- H02: The efficiency of the codend PIT tag detection system is unaffected by tag orientation -- H03: The efficiency of the codend PIT tag detection system is unaffected by multiple tags passing simultaneously -- H04: The codend PIT tag detection system only reads tagged fish within the codend -- Discussion -- Chapter 2. A mark-recapture experiment to estimate efficiency and catchability of Atlantic cod in a bottom trawl -- Introduction -- Methods -- Fish collection for lab studies -- Discard mortality -- H01: The mortality associated with PIT tagging of cod is zero -- H02: The catchability of cod is equal before and after tagging -- H03: The tag shedding rate of the cod is zero -- H04: The tagged cod are randomly mixed with the population -- H05: The cod population in the survey area is closed for the duration of the mark-recapture experiment -- H06: The capture efficiency of cod in the SMAST video trawl survey is 100% -- Results -- H01: The mortality associated with PIT tagging of cod is zero -- H02: The catchability of cod is equal before and after tagging -- H03: The tag shedding rate of the cod is zero -- H04: The tagged cod are randomly mixed with the population -- H05: The cod population in the survey area is closed for the duration of the mark-recapture experiment -- H06: The capture efficiency of cod in the SMAST video trawl survey is 100% -- Discussion -- Chapter 3. Examining fish identification and measurement accuracy in stereoscopic video -- Introduction -- Methods -- Stereoscopic calibration and analysis -- H01: The counts of cod from the stereoscopic cameras GoPro and catch are similar -- H02: The length of cod measured from the stereoscopic cameras and on deck are similar -- Results -- H01: The counts of cod from the stereoscopic cameras GoPro and catch are similar -- H02: The length of cod measured from the stereoscopic cameras and on deck are similar -- Discussion -- Chapter 4. Examining variance and the allocation of resources in the SMAST video trawl survey -- Introduction -- Methods -- Survey design -- H01: The variance in biomass estimates from a simple random and stratified random video trawl survey are equal -- H02: There is no relationship between tow duration and the variance of cod CPUE -- H03: Tow duration and sampling intensity are optimized at 60 minutes -- Results -- H01: The variance in biomass estimates from a simple random and stratified random video trawl survey are equal -- H02: There is no relationship between tow duration and the variance of cod CPUE -- H03: Tow duration and sampling intensity are optimized at 60 minutes -- Discussion -- Conclusion -- Improvements to fisheries-independent surveys -- Impacts on the Western Gulf of Maine cod -- References.
- Includes bibliographical references (pages 116-138).
- University of Massachusetts Dartmouth
- Doctor of Philosophy (PHD)
- Department of Fisheries Oceanography
- English
- Dissertation
- Copyright 2025 Nicholas M. Calabrese
- https://doi.org/10.62791/20500
- 9914504462801301