Integrating Advanced Technologies to Improve Data Quality and Reduce Bias in Fisheries and Wildlife Population Research and Management

As computing, network, software, and scientific technologies continue to advance and develop rapidly, so too do their applications to fisheries and wildlife research. As a result of this fast-paced growth, keeping atop of new technological developments which could benefit fisheries and wildlife research can be time-consuming and overwhelming. Additionally, use of new technologies can be daunting when specialized knowledge is required to implement or modify these new tools or approaches. With no assurance that there will be a positive cost-benefit to a project to use new technology(ies), this can further discourage researchers from undertaking projects which incorporate them. However, recent research is showing that these tools can provide scientists the opportunity to improve data quality, reduce survey bias, reduce cost, and sample increasingly remote areas around the globe. The focus of this symposium, hosted by the AFS Fisheries Information & Technology Section (FITS), is to highlight current projects which are integrating advanced technologies and how they are thereby achieving one or more of the aforementioned goals within the terrestrial, marine, and/or freshwater environment(s).

8:00AM Welcoming Remarks
8:20AM Using Feedback from Advanced Technologies to Improve Long-Term Population Assessments: A Case Study o Reef Fishes in the Gulf of Mexico
  Theodore S. Switzer, Matthew Campbell, Chris L. Gardner, Sean F. Keenan, Kevin A. Thompson, Kevin Rademacher, Brandi Noble, Ryan Caillouet, Kate Overly, Mary Christman
Most long-term monitoring surveys were developed solely to track population trends through time for a handful of priority taxa. As management needs expand and evolve, surveys must adapt to address emerging issues without sacrificing historical time series. We present early results from efforts to integrate three spatially-restricted surveys of reef-fish populations (conducted by NMFS-Pascagoula, NMFS-Panama City, and the state of Florida) under a unified survey design based on feedback provided by ongoing acoustic mapping and stationary video surveys. A constrained binary divisive clustering approach was used to delineate candidate spatial and habitat stratification schemes that captured statistically-significant differences in overall managed reef-fish assemblage structure. Importantly, habitat stratification was based on measures of both the spatial extent and relative relief of identified reef features definable from both multibeam and side-scan sonar data. Effort allocation among survey strata was primarily based on measures of multispecies diversity with some modification to ensure the ability of the survey to provide meaningful abundance data for key managed taxa. The resultant survey design represents an efficient and optimized survey with improved capabilities of assessing multispecies changes through time, yet also preserves much of our ability to include historical data into analyses of long-term population trends.
8:40AM Integrating Hydroacoustic and Telemetric Data to Estimate Fish Abundance: A New Approach to an Old Problem
  Brian J. Shuter, Scott Milne, Michael D. Rennie, Derrick T. de Kerckhove
We present an approach to abundance estimation that integrates hydro-acoustic echo counting with telemetry from fish implanted with responsive acoustic tags. These estimates are free of assumptions common to traditional methods. Fish are implanted with tags that broadcast unique identifying signals that can be located in three dimensions by a broad-swath multi-beam sonar. The multi-beam system detects and counts the volume-density of tagged fish. A narrow, split beam scientific sounder counts the volume density of fish-targets of similar size to the tagged fish. Concurrent collection and integration these two data streams generates abundance estimates based on mark-recapture principles. In lakes dominated by a large top predator, these data are sufficient to provide an abundance estimate for that top predator. In lakes where several species contribute to a size category, spatial habitat-use information generated by passive and/or active telemetry can be used to adjust the estimate of overall abundance to a species-specific estimate. In this paper, we describe the method in detail and illustrate its use on 4 wild lake trout populations varying in known abundance from 200 to 3000 individuals
9:00AM Comparative Utility of Stationary and Towed Video Imagery for Characterizing Reef-Fish Assemblages Associated with Habitats of Varying Complexity
  Sean F. Keenan, Theodore S. Switzer, Sarah Grasty, Steven Murawski, Chad Lembke, Alexander Ilich
Underwater imagery has become a commonly-used sampling tool to assess fish populations within sensitive areas (e.g., MPAs) as well as high-relief habitats to minimize gear loss and habitat damage. Improved technologies promote the collection of high-resolution optical data using various stationary and mobile sampling platforms, although important questions remain regarding relative strengths and limitations of these approaches. We present results from a pilot study to evaluate the detectability of economically and ecologically-important species among six natural reef habitats of varying relief and spatial scale between two survey platforms: stationary stereo-baited remote underwater video (S-BRUV) and the towed Camera-Based Assessment Survey System (C-BASS). During independent research cruises, C-BASS was towed across multiple transects covering 109.6 km, while 54 S-BRUV deployments were made along the C-BASS transects. The S-BRUV and C-BASS sets observed 77 and 22 taxa, respectively; C-BASS observations were generally limited to fish greater than 200mm. Species such as Gag, Lionfishes and Gray Snapper were commonly observed by both systems. Ultimately, each approach has advantages and disadvantages; we will discuss the relative utility of S-BRUV and C-BASS surveys and how these approaches may be integrated to improve assessments of reef fishes and associated habitats in the eastern Gulf of Mexico.
9:20AM Reef Fish Densities As Determined By Acoustics and Video Technologies – Comparative or Complimentary?
  Edmund Hughes, Sarah Grasty
The development of new fisheries survey technologies and methods to supplement traditional sampling methods has been a research priority for over a decade. Acoustic and visual technologies have several advantages over traditional sampling gears as their use is non-lethal and minimally invasive. They are particularly useful in regions considered untrawlable or which have fishing restrictions. Reef fish densities determined from concurrent acoustic (Simrad EK60) and towed video methods were compared in terms of correlation over varying bathymetry and relief on the West Florida Shelf in the Gulf of Mexico. Sampling was carried out along four different transects ranging 26-53 km kilometers in length between 2016 and 2017. Vertical distributions of biomass determined from acoustic techniques were examined and compared for each of the four regions while matched, concurrent video imagery was used to estimate fish density within three to five meters of the seafloor. Fish densities estimated from the two different approaches showed similar patterns however the level of correlation varied for the different transect regions. Distributions of biomass in the water column suggested that vertical distribution was influenced by the level of relief in the region and that relief likely influences the relative efficiency of each type of technology.
09:40AM Break
1:10PM Mapping Benthic Habitat and Estimating Reef Fish Abundance Using Multibeam Acoustics and Towed Underwater Video on the West Florida Shelf
  Alexander Ilich, Jennifer Brizzolara, Sarah Grasty, John Gray, Matthew Hommeyer, Chad Lembke, Stanley Locker, Alex Silverman, Theodore S. Switzer, Abigail Vivlamore, Steven Murawski
The West Florida Shelf (WFS) is an extremely important area for both commercial and recreational fisheries. However, the lack of habitat maps in this area makes planning fisheries independent monitoring surveys difficult, and hinders the ability to manage and monitor fish stocks and ecosystems over time. As of 2015, only 5% of the WFS had been mapped in high resolution using a multibeam echosounder with little effort expended to infer and verify habitat type. In 2015, The Continental Shelf Characterization Assessment and Mapping Project (C-SCAMP) began using a multibeam echosounder and towed underwater video to map benthic habitats and improve our understanding of fish-habitat relationships on the WFS. Progress and challenges in integrating these technologies to map habitats and assess fish population abundances will be discussed. For this study, high resolution multibeam bathymetry and co-registered backscatter data were collected in conjunction with towed underwater video. Habitat maps were created using a supervised classification model that predicts benthic habitat type based on the acoustic signature. Resultant habitat maps were then combined with habitat-specific fish densities determined from the towed underwater video transects to provide fish abundance estimates throughout several localized study areas.
1:30PM Obtaining in Situ Fish Morphology from a Demersal Survey Trawl Using Low-Cost Consumer-Grade Stereoscopic Action Cameras.
  Christopher Rillahan, Kevin D.E. Stokesbury, Pingguo He
Most fisheries surveys rely on the capture and physical handling of fish to obtain morphological information. The University of Massachusetts has been developing a video-based trawl survey which enumerates fish passing through the trawl extension without capture. To obtain fish length measurements two consumer-grade 3D action cameras (SID, Weeview Inc.) were placed on the top and side of the trawl extension. Sea trials were conducted in the Gulf of Maine during the winter of 2019 onboard the commercial fishing vessel ‘Justice’. Ten tows were conducted in which the codend was closed and the fish were brought aboard the vessel for measurement. This allowed for a direct comparison between the traditional method of data collection and the video-based method. Kernel density estimation was used to compare the length-frequency curves obtained from the two methods. Data obtained from this project indicates that reliable estimates of population structure can be obtained from low-cost consumer-grade 3D cameras. Developing methodologies to collect in situ fisheries data will help reduce the impact fisheries studies have on wild fish populations.
1:50PM Experiments in Using the Whooshh Fishl Recognition System to Investigate Biases in Adult Salmon Sampling at Bonneville Dam
  Jeffrey K. Fryer, Jon E. Hess, Vince Bryan, Janine Bryan, Jim Otten, Dan Schneider, Marc Bommarito
Sockeye, Chinook, and steelhead have been trapped at the Bonneville Dam Adult Fish Facility since 1985 to estimate age and length composition which is used in fisheries management, run reconstruction, and forecasting. Increasing restrictions on the use of the trap due to fish abundance and/or water temperatures have resulted in insufficient sample sizes. For example, in one week in September 2017 we only sampled 10 steelhead to represent 10.2% of the entire steelhead run that passed Bonneville Dam during that week. As part of their fish transport system (a.k.a. “The Salmon Cannon”), Whooshh has been developing a fish identification imaging system to allow for automatic sorting of fish to be transported by Whooshh from those that managers do not want to be transported. In 2019, a stand alone Whooshh FishL Recognition™ system will be tested at the Bonneville Dam Adult Fish Facility, capturing 18 images of each fish that bypass sampling due to insufficient capacity in our sampling tank. A subset of these images will then be analyzed to determine species, fork length, and fin clip information which can be used to supplement and compare with data provided by our project. Preliminary results from this work will be presented.
2:10PM Real-Time Visualizations from Coordinated Fish Monitoring Programs to Balance Fishery and Water Management in California’s Central Valley
  Joshua Israel, Arnold J. Ammann, James Anderson
Reclamation supports many fish monitoring projects from natal areas to the marine environment, which historically were only retrospectively available after water management decision occurred. Over the past five years, Reclamation has focused funding towards development on reporting and visualization systems of forecasting tools to evaluate and understand what incoming monitoring data may mean in terms of exposure and risk of ESA-listed fishes from real-time water operations. Two case studies from collaborative projects by federal and state agencies and universities will be presented to highlight how they have improved the tools and information available for ESA species population management. First, we examine a collaboratory (SacPAS: designed to provide web-based services to link data and science to in-season management. It includes a database, web-based visualization and analytical tools, and fish survival and migration management tools. Second, we examine the Central Valley Enhanced Acoustic Telemetry Program ( designed to improve monitoring of salmonid migration and survival in the Central Valley. It provides a database, web-based visualization, and modeled survival and entrainment estimates for the numerous releases of acoustically tagged juvenile Chinook salmon occurring annually.
2:30PM An Assessment of a Deep Learning Computer Algorithm to Count and Measure Atlantic Sea Scallops
  N.David Bethoney, Benjamin Woodward, Kevin D.E. Stokesbury
The Atlantic sea scallop (Placopecten magellanicus) is a marine bivalve distributed throughout the continental shelf of the northwestern Atlantic. The sea scallop fisheries in the United States and Canada are extremely valuable; annual combined landings over the past several years have approached $1 billion. The management approach of these fisheries requires spatially-specific information on scallop density and size, which is provided by several surveys including the University of Massachusetts Dartmouth School for Marine Science and Technology (SMAST) drop camera survey. A major survey cost, in time and money, is the counting and measuring of scallops within each image. To reduce this cost, SMAST collaborated with CVsion AI to develop an algorithm to count and measure scallops. Comparison of results suggest the algorithm can produce similar results at reduced costs to humans but revealed short comings in certain circumstances. Thus, a rigorous evaluation of the algorithm performance is required before utilizing automation as a scientific tool. By assessing performance on a curated set of data across a range of characteristics (e.g. scallop density, scallop size), expected error rates can be determined. These studies form a basis of utilizing automation in tandem with human review and expanded algorithm application.
2:50PM Refreshment Break
3:20PM Have Your Fire Hose and Drink from It, Too: An Expert-Approved Approach to Using Angler Apps to Generate Large Volumes of Usable Data
  Paul Venturelli, Christian Skov, Kieran Hyder
Angler smartphone applications (apps) are a new tool for efficiently collecting conventional and novel fisheries data. An important step in realizing the potential of angler apps is to develop standards that apply across apps to ensure a large and reliable data stream. We convened a workshop that was attended by representatives from 11 angler apps, and 22 experts in recreational fisheries, human dimensions, economics, data management, citizen science, and standards. An initial survey identified gaps between fisheries data needs and the data that angler apps were collecting. We addressed these gaps during the workshop by cataloguing the importance and specific needs associated with 49 data fields (i.e., standards), and then determining whether apps can deliver on these standards. We concluded that any standard can be met, but that anglers will only be willing to supply data for a subset of these standards. Therefore, we propose an initial set of standards that are important to fisheries and/or easy to obtain (e.g., via automation). These standards should be maintained by a science-based and international standards council that is also responsible for maintaining a repository of participating angler apps and prioritizing research.
3:40PM Implementing a Video Monitoring System to Estimate Buoy 10 Recreational Fishing Effort
  Natalie Scheibel
The Buoy 10 fishery in the Columbia River estuary depends on accurate effort and catch estimates to maximize valuable fishing opportunities while limiting impacts on ESA-listed salmonid stocks. Traditional instantaneous trailer and mooring basin counts have been the primary method to estimate fishing effort and are effective at developing peak effort counts. However, instantaneous counts do not provide adequate information to develop total effort estimates and are problematic if counts occur outside peak fishing times or if there are large shifts in fishing effort. Video boat counts (VBC) present an alternative method to estimate fishing effort. In order to transition from the instantaneous counts to VBC, a three-year comparison study was conducted to develop relationships between the two methods. Video cameras were installed at five access points throughout the estuary and recorded outgoing and incoming boats. Video was analyzed on a daily basis and VBC were adjusted from on-site creel surveys to exclude incomplete trips and to differentiate trip types. VBC counts were consistently higher than the traditional counts. Video cameras can provide a cost-effective way to monitor fishing effort 24 hours a day, which helps managers improve the accuracy of effort and catch estimates and better understand angler behavior.
4:00PM Assessing the Usage of Time-Lapse Digital Cameras to Obtain Angler Effort Estimates at Alabama Reservoirs
  Robert Eckelbecker, Matthew Catalano
We compared different creel survey methods at three Alabama reservoirs (Harris, Jordan, and Mitchell) to identify approaches that could improve precision or reduce costs. We were particularly interested in whether time-lapse photos taken at boat ramp parking lots could be used as an index of fishing effort to improve the temporal coverage of sampling at relatively low cost. Angler effort was estimated independently through the use of roving creels, access point creels, aerial census counts, and fixed-location digital camera images of boat ramps. Evaluation of the accuracy of angler effort from time-lapse photos was analyzed based on ramp-specific access point creel surveys that occurred simultaneously. We compared a reservoir-wide effort index from time-lapse photos of all boat ramps with aerial census counts using an analysis of covariance that varied by reservoir, season, and day type. If the relationships are strong between both the aerial counts and access point creel surveys to the camera images, then these angler effort estimates from remote cameras could reduce the amount of effort needed for on-site survey samples.
4:20PM A Perspective on Angling Pressure Estimates Generated from Tradition Creel Surveys and Unmanned Aerial Systems.
  Jared Engelbert, William Stark, David Spalsbury
Unmanned Aerial Systems or drones were used to estimate angling pressure over 3 months at Cedar Bluff Reservoir, Kansas, and were compared to those generated from a traditional creel survey. We used a boat to launch and retrieve a DJI Phantom 4 Pro on each leg of the sUAS survey. We detected more shoreline anglers (2.24 anglers/unit) and boat anglers (9.06 anglers/unit) compared to estimates from the traditional creel. On average the sUAS survey was completed within 1.5 hrs. of the 4-hour survey time period designated for the traditional creel survey. In addition, we were able to estimate angler use within specific areas of the reservoir through the season.
4:40PM Ensuring That Technological Advances Actually Advance the Fisheries Profession: Developing a Strategic Framework That Uses Advanced Technologies to Solve Persistent Fisheries Problems
  Martha E. Mather, Daniel E. Shoup, Quinton Phelps
Fisheries can benefit from advanced technologies (e.g., telemetry, remote sensing, social media, quantitative approaches). However, advanced technologies do not solve complex fisheries problems, improve data quality, or reduce bias by themselves. In fact, the tool utility hypothesis proposes four relationships between technology and successful problem-solving. (1) Initially, the perceived relationship is strong as general agreement exists that new tools could be valuable. (2) An assumed but often untested positive relationship follows as previously-unavailable technologies are developed and perfected. (3) However, this relationship can weaken with increased use of the technology if data collection objectives prioritize how to (rather than why) use the technology. (4) Nevertheless, benefits from advanced technology can accelerate if strategic planning structures data collection so that important questions drive technology use. Here we provide a framework to focus advanced technologies on common fisheries management problems (e.g., standard/standardized assessments, harvest regulation application/evaluation, habitat improvement, invasive species control, managing barriers). For these shared problems, an adaptive, iterative approach integrates questions, methods, and tool application in a way that increases data comparability across agencies and institutions. Thus, the technology we propose here (a strategic framework) can enhance the utility and increase the benefits of diverse advanced technologies.

Organizers: Sarah Grasty, Kayla Key, Julie Defilippi Simpson, Rebecca M. Krogman
Supported by: Fisheries Information and Technology Section (FITS)

Location: Reno-Sparks CC Date: October 1, 2019 Time: 8:00 am - 5:00 pm