Peter C. Esselman | Search USGS

USGS2026

MusselFinder: Automated dreissenid mussel density estimation via robotic benthic imaging

Despite the widespread need to map the abundance of invasive quagga and zebr mussels ( Dreissena spp.) in lakes and reservoirs worldwide, current methods remain limited by data collection and analysis bottlenecks. Autonomous underwater vehicles (AUVs) combined with machine learning (ML) offer a scalable remote sensing approach for estimating mussel density (individuals m −2 ) across large spatial domains and a range of habitats that are difficult to sample with current techniques. We present MusselFinder: a deep neural network (DNN)-based pipeline that detects and enumerates live mussels in AUV-collected benthic images. MusselFinder estimates live mussel percent cover and density in nadir monocular images captured at known altitudes above the lakebed. In images of suitable clarity, MusselFinder achieved strong pixel-level localization (mussel-class Intersection over Union [IoU] > 0.76) and density prediction (log-density R 2 = 0.79–0.95) over a variety of bottom types. AUV-based image acquisition is more than three orders of magnitude faster than physical sampling by divers or ponar grabs, and DNN-based analysis can proceed at rates comparable to image acquisition. We used MusselFinder to analyze over 50,000 spatially referenced images at a spawning reef in Lake Michigan, representing a shift from small-scale, diver-collected datasets to large-scale monitoring in unconstrained environments. The method can be adapted to new imaging conditions using a modest training set and a user-friendly point-labeling technique, supporting broader-scale monitoring of one of the world's most invasive invertebrates.

Illinois, Michigan↗

USGS2026

Availability of dark daytime refuge may limit mysid abundance in the Laurentian Great Lakes

The zooplankton Mysis diluviana is a major component of the Laurentian Great Lakes food web and has recently declined in abundance in both lakes Michigan and Huron. Drivers of these declines are not well understood. Here, we explore the hypothesis that recent increases in water clarity have contributed to the decline of M. diluviana (mysids) by limiting the availability of daytime dark refuge from visual predators. Using Secchi depth data from 1996 to 2021, we estimate that dark refuge has decreased substantially in lakes Michigan and Huron where mysids have declined, but dark refuge has remained more stable in lakes Ontario and Superior where mysid populations did not decline. Results from a 2021 uncrewed surface vessel hydroacoustic survey and lake-wide, net-based sampling in lakes Michigan and Huron revealed significantly more mysids in areas with dark refuge, such as Lake Michigan’s northern basin. Conversely, Lake Huron contains sparse dark refuge consistent with low mysid densities in that lake. Higher water clarity leading to increased predation may be a primary driver of mysid declines in the Great Lakes.

Laurentian Great Lakes↗

USGS2026

Efficiency of down-looking cameras for detecting round goby (Neogobius melanostomus) over varying substrates in laboratory microcosms

Since invading the Laurentian Great Lakes in the late 1980s, round goby Neogobius melanostomus have become a dominant benthic prey species, resulting in a need to accurately monitor their population abundance to inform fisheries management. Camera-based methods for assessing round goby abundances have gained popularity, but their efficiencies for detecting round goby are poorly understood. We evaluated the efficiency of down-looking color monocular cameras for detecting round goby presence in microcosm enclosures with known numbers of fish. Detection efficiencies were compared between sand and cobble substrate during daylight hours using a generalized linear mixed-effects model. After accounting for trial effects, enclosure placement, and within-replicate temporal autocorrelation, round goby detection efficiency was 98.6% over sand (95% confidence limit (CL): 98.0–99.0%), and 55.3% over a single layer of cobble substrate (CL: 48.6–61.8%). The large difference in detectability between cobble and sand suggests that camera-based estimates of round goby abundances are likely to be biased low and have lower relative precision in high-structure habitats, but may be largely unbiased in low-structure habitats. Despite much lower relative detection efficiencies of cameras in high-structure cobble habitats, the efficiencies reported here still compare favorably to conventional methods like otter and beam trawling. Our results provide initial quantitative bounds on the possible degree of negative bias in camera-based estimates of round goby abundances in different substrates, but require further field verification across the diversity of substrates, interstitial infilling, and other structural attributes of habitat found in the Great Lakes.

Michigan↗

USGS2026

Lake Trout population dynamics on Six Fathom Bank and Yankee Reef in Lake Huron, 1992–2023

Objective Under the auspices of the Great Lakes Fishery Commission, the Lake Huron Committee and its technical committee initiated offshore stocking of Lake Trout Salvelinus namaycush beginning in 1985 as an additional step toward restoring a self-sustaining Lake Trout population to the lake, with stocking occurring first at Six Fathom Bank and then later at Yankee Reef. The goal of this study was to assess progress toward Lake Trout rehabilitation on Six Fathom Bank and Yankee Reef in Lake Huron through 2023. Methods We conducted annual (1992–2023) gill-net surveys in the spring–summer and the fall to assess juvenile and adult populations, as well as beam trawl surveys in the spring–summer to detect naturally reproduced age-0 Lake Trout. Result During 2000–2015, relative abundance of unclipped (wild) Lake Trout surged from levels of <2 fish · 305 m of gill net −1 · d −1 to levels exceeding 50 fish · 305 m of gill net −1 · d −1 on both Six Fathom Bank and on Yankee Reef, based on fall gill-net surveys, and has remained at a relatively high level through 2023. The proportion of wild fish in both the spring gill-net surveys and the fall gill-net surveys averaged well over 50% in the Six Fathom Bank–Yankee Reef (SFBYR) complex during 2014–2023. Sea Lamprey Petromyzon marinus wounding rate in the SFBYR complex decreased from 20 wounds/100 fish during 1994–2002 to just 7 wounds/100 fish during 2004–2023, based on the fall gill netting. Annual mortality rate estimates for SFBYR adult Lake Trout during the 2004–2023 period averaged 21%. Wild age-0 Lake Trout were captured in beam trawls in the SFBYR complex during the 1994–2022 period, providing further evidence that Lake Trout were actively spawning within this offshore reef complex. Conclusion Based on several metrics, the SFBYR Lake Trout population appeared to be self-sustaining. Lake Trout stocking, Sea Lamprey control, and restrictions on fishing for Lake Trout played important roles in the population recovery. The lessening of the negative effect of Alewives Alosa pseudoharengus on Lake Trout natural recruitment was likely a key factor in the restoration process.

Lake Huron, Six Fathom Bank, Yankee Reef↗

USGS2026

Design and function of the Autonomous Benthic Imaging and Surveying System (ABISS) for remote sensing of lake and seabed environments

Lake and seabed environments are home to fisheries and other biota that are important to ecosystems and economies, yet these environments and the species that use them are difficult to accurately assess and monitor. Traditional benthic survey techniques, like bottom trawling used by the U.S. Geological Survey, are limited by substrate constraints, poor spatial resolution and precision, and operational depth limits, hindering accurate assessment of benthic species and habitats. In response to these limitations, the U.S. Geological Survey developed the Autonomous Benthic Imaging and Surveying System, a camera system integrated into underwater vehicles, to capture high-resolution images of the lakebed. The system uses color and stereo cameras to collect imagery, which can be analyzed using computational methods to detect organisms and (or) characterize habitat features, such as geologic substrate types. The system has been integrated into autonomous underwater vehicles and into an underwater housing used by self-contained underwater breathing apparatus (SCUBA) divers. Although the engineering of the system was motivated by the need for data collection in the Great Lakes, it has potential to collect high quality data in any aqueous setting with sufficient water clarity and safe operating conditions. The Autonomous Benthic Imaging and Surveying System can operate across diverse depths and light conditions to map and quantify ecological patterns that were difficult or impossible to assess using traditional methods. The Autonomous Benthic Imaging and Surveying System offers the potential for accurate and precise monitoring and assessment of native benthic biota, invasive species, and habitat, potentially providing natural resource managers with improved information to support decision making about benthic resource management.

Great Lakes↗

USGS2026

Artificial intelligence strategy for the U.S. Geological Survey

Artificial intelligence (AI) can offer opportunities to enhance the science, science delivery, and business operations of the U.S. Geological Survey (USGS). Although USGS staff have proactively adopted AI into our workflows for many years, a comprehensive USGS strategy for AI has not previously been developed. The strategy described here is motivated by the acceleration of AI technological development, the benefits of increased AI adoption to USGS mission delivery as anticipated by USGS staff, rising public concern about the implications and trustworthiness of AI, and emerging Federal directives and guidance about AI. The USGS vision is to continue integrating AI to deliver valuable science for the public good while maintaining high ethical standards, scientific quality and integrity, and compliance with Federal and U.S. Department of the Interior requirements. To realize this vision, the USGS can take steps to (1) develop a strong AI workforce, (2) adapt our organizational approaches to include AI governance and communication, (3) ensure responsible and trustworthy use of AI, (4) modernize our computing and data infrastructure to support AI, and (5) accelerate AI adoption and innovation in the Bureau.

Circular↗

USGS2026

Evaluating the use of uncrewed surface vessels to enhance Lake Erie acoustic prey-fish surveys

Incorporation of autonomous uncrewed surface vessels (USVs) into large-scale acoustic surveys may enhance spatiotemporal extent and quality of fish density estimates. Lake Erie is currently surveyed by three motorized research vessels (RVs), which annually collect acoustic data and estimate prey-fish abundances. To evaluate the feasibility of incorporating a USV into the existing survey to increase spatial coverage, we compared paired acoustic measurements of target strength (TS), area backscattering coefficient (ABC), and areal fish density from four motorized RVs (three currently and one historically used) and one USV (Saildrone Explorer). Acoustic data were collected along 2-km transects during five sampling events distributed throughout Lake Erie. Despite observing event-specific differences that were likely impacted by sampling conditions, survey-level estimates of areal density between RVs and the USV were comparable (i.e., RV 6% lower than USV). RVs and the USV appear to react differently (speed and stability) to variable environmental conditions (wind and waves), which impacted data quality and must be mitigated in future applications. However, similar to previous comparisons in marine and freshwater environments, incorporating a USV into an acoustic survey could enhance annual fish density estimates by increasing effort, spatial coverage, logistical flexibility, and biological data collection opportunities from RVs.

Michigan, New York, Ohio, Pennsylvania↗

USGS2026

Status and trends of pelagic and benthic prey fish populations in Lake Michigan, 2025

Fall bottom trawl (fall BT) and lakewide acoustic (AC) surveys are conducted annually to generate indices of pelagic and benthic prey fish densities in Lake Michigan. The fall BT survey has been conducted each fall since 1973 using 12-m trawls at depths ranging from 9 to 110 m at fixed locations distributed across seven transects; this survey estimates densities of seven prey fish species [i.e., Alewife ( Alosa pseudoharengus ), Bloater ( Coregonus hoyi ), Rainbow Smelt ( Osmerus mordax ), Deepwater Sculpin ( Myoxocephalus thompsonii ), Slimy Sculpin ( Cottus cognatus ), Round Goby ( Neogobius melanostomus ), Ninespine Stickleback ( Pungitius pungitius )]. The AC survey has been conducted each late summer/early fall since 2004 (except 2020). The 2025 AC survey consisted of 26 transects [470 km total (292 miles)] covering bottom depths ranging from 5 to 259 m and 44 midwater trawl tows at 1.4 to 82.4 m fishing depth; this survey estimates densities of three prey fish species (i.e., Alewife, Bloater, and Rainbow Smelt). The data generated from these surveys are used to estimate various population parameters that are, in turn, used by state and tribal agencies in managing Lake Michigan fish stocks. For the AC survey, total biomass density of prey fish equaled 9.3 kg/ha, continuing a recent trend of index values above the long-term average of 5.4 kg/ha. For the fall BT, total biomass density of prey fish equaled 3.4 kg/ha, close to values observed since 2014 and well below historic numbers and those observed earlier in the 2000s. Over the period both surveys have been conducted (2004-2025), the total biomass density index had trended downward in the fall BT through the mid-2010s and appears to have stabilized at low values, while the AC survey biomass density index has remained relatively stable over the time series. Mean biomass of yearling and older (YAO) Alewife was 4.3 kg/ha in the AC survey and 0.45 kg/ha in the fall BT. Since 2014, annual survey results suggest that the catchability of YAO Alewife for the fall BT is substantially lower than the AC survey. The 2025 AC survey YAO Alewife biomass density estimate was 57% higher than the average from 2004-2024. The Alewife population of Lake Michigan appears to be composed mostly of young fish and the proportion of age-4 and older Alewife was ~5% in both surveys. Age-0 Alewife numeric density from the AC survey was 259 fish/ha in 2025, lower than the long-term mean (487 fish/ha). Biomass density of large (≥120 mm) Bloater was 3.5 kg/ha in the AC survey and 1.9 kg/ha in the fall BT. The density of small (<120 mm) Bloater was 540 fish/ha in the AC survey, the second highest value in the time series. Meanwhile, small Bloater density estimated in the fall BT was only 6.1 fish/ha. Biomass density of large Rainbow Smelt (≥90 mm) was 0.69 kg/ha in the AC survey and 0.04 kg/ha in the fall BT survey. Numeric density of small (<90 mm) Rainbow Smelt was 541 fish/ha in the AC survey, the highest value in the time series, and 41 fish/ha in the fall BT. All four prey fish species indexed only by the fall BT had below-average biomass densities. Deepwater Sculpin biomass density was 0.21 kg/ha, which makes 15 of the past 16 years with biomass <1 kg/ha. Slimy Sculpin was estimated to be 0.03 kg/ha, an order of magnitude lower than the long-term average from the fall BT. Round Goby biomass density was 0.44 kg/ha and Ninespine Stickleback density was 0.20 kg/ha, the highest value since 2007.

Lake Michigan↗

USGS2026

Status and trends of the Lake Huron prey fish community, 1976-2025

The U. S. Geological Survey-Great Lakes Science Center has monitored annual changes in the offshore (depth > 9m) prey fish community of Lake Huron since 1976. Assessment of prey fish populations in Lake Huron is based on a bottom trawl survey that targets demersal species (i.e., those predominantly or intermittently associated with the lake bottom) and an acoustic-midwater trawl survey that targets pelagic species and life stages. Prey fish abundance and species composition in 2025 was generally consistent with trends observed over the past decade. Overall status of main basin prey fish community was rated as ‘fair.’ Bloater ( Coregonus hoyi ) and Rainbow Smelt ( Osmerus mordax ) remain the first and second most abundant species in the main basin followed by Alewife ( Alosa pseudoharengus ), whose populations consist mainly of low numbers of age-0 individuals. Abundance of Rainbow Smelt is higher, and abundance of Bloater lower, in Georgian Bay and the North Channel than in the main basin, which suggests dynamics of these species vary by basin. Native sculpin communities consist primarily of Deepwater Sculpin ( Myoxocephalus thompsonii ) due to the reduced abundance and distribution of Slimy Sculpin ( Cottus cognatus ). In contrast, biomass of the exotic Round Goby ( Neogobius melanostomus ) in 2025 was the second highest observed in the bottom trawl time series. Evidence of strong Bloater and Rainbow Smelt year classes in 2025 and continued recovery by Cisco ( Coregonus artedi ) in the North Channel were interpreted as positive community trends, whereas growth and expansion of Round Goby populations and low species diversity of pelagic prey fish communities merit concern. Use of complementary surveys (bottom trawl, acoustics) remains important for evaluating prey fish status in Lake Huron, where prey fish community dynamics vary by basin and prey fish responses to changing environmental conditions depend on species and/or habitat.

Lake Huron↗

USGS2025

Improved ground-truthing and benthic habitat characterization with machine learning models and 3D photogrammetry

Benthic mapping relies on a range of methods to ground truth remotely sensed data to map physical and biotic features. Traditional methods are labor-intensive and time-consuming with limited scalability. Autonomous underwater vehicles (AUVs) and remote operated vehicles (ROVs) can collect high volumes of quality data but require automated processing routines and pose other logistic challenges. This paper explores the automated use of multiple machine learning (ML) models to detect and delineate benthic habitat components from structure-from-motion (SfM) orthomosaics. Overlapping AUV photos were processed into high-resolution, seamless orthomosaics for a portion of the Lake Michigan lakebed. ML models were trained to delineate rocks and live mussels and subsequently applied to the SfM orthomosaics. Results demonstrate that a robust pipeline with SfM and ML models is an effective approach to efficiently, accurately, and comprehensively map benthic habitats.

Lake Michigan↗

USGS2025

MIMAR-Net: Multiscale Inception-based Manhattan Attention Residual Network and its application to underwater image super-resolution

In recent years, Single-Image Super-Resolution (SISR) has gained significant attention in the geoscience and remote sensing community for its potential to improve the resolution of low-quality underwater imagery. This paper introduces MIMAR-Net ( M ultiscale I nception-based M anhattan A ttention R esidual Network), a new deep learning architecture designed to increase the spatial resolution of input color images. MIMAR-Net integrates a multiscale inception module, cascaded residue learning, and advanced attention mechanisms, such as the MaSA layer, to capture both local and global contextual information effectively. By utilizing multiscale processing and advanced attention strategies, MIMAR-Net allows us to handle the complexities of underwater environments with precision and robustness. We evaluate the model on three popular underwater image datasets, namely UFO-120, USR-248, and EUVP, and perform extensive comparisons against state-of-the-art methods. Experimental results demonstrate that MIMAR-Net consistently outperforms existing approaches, achieving superior qualitative and quantitative improvements in image quality, making it a reliable solution for underwater image enhancement in various challenging scenarios.

Electronics↗

USGS2025

Status and trends of the Lake Huron prey fish community, 1976-2024

The U. S. Geological Survey-Great Lakes Science Center has monitored annual changes in the offshore (depth > 9m) prey fish community of Lake Huron since 1973. Monitoring of prey fish populations in Lake Huron is based on a bottom trawl survey that targets demersal species (i.e., those predominantly or intermittently associated with the lake bottom) and an acoustic-midwater trawl survey that targets pelagic species and life stages. Prey fish abundance and species composition in 2024 was generally consistent with trends observed over the past decade. Bloater ( Coregonus hoyi ) remains the most abundant and widespread prey species, although its abundance is starting to decline owing to changes in demographic factors that are interacting to suppress reproduction. Rainbow Smelt ( Osmerus mordax ) remain widely distributed across Lake Huron but their dynamics vary by basin. Rainbow Smelt populations currently are larger in Georgian Bay and the North Channel than in the main basin where they have produced eight weak year classes over the past decade including in 2024. Populations of Alewife ( Alosa pseudoharengus ) continue to be comprised of low numbers of age-0 individuals, and sculpin communities consist primarily of Deepwater Sculpin ( Myoxocephalus thompsonii ) due to the reduced abundance and distribution of Slimy Sculpin ( Cottus cognatus ). In contrast, biomass of the invasive Round Goby ( Neogobius melanostomus ) in 2024 was the highest observed in the bottom trawl time series and was over four times greater than in 2023. Overall status of main basin prey fish community was rated as ‘fair.’ Stable dynamics of main basin Bloater populations and evidence of continued recovery by Cisco ( Coregonus artedi ) in the North Channel were interpreted as positive community trends, whereas growth and expansion of Round Goby populations and low species diversity of pelagic prey fish communities are inconsistent with fish community objectives. Use of complementary surveys (bottom trawl, acoustics) remains useful for evaluating prey fish status in Lake Huron, where prey fish community dynamics vary by basin and prey fish responses to changing environmental conditions depend on species and/or habitat.

Lake Huron↗

USGS2025

Understanding the influence of image enhancement on underwater object detection: A quantitative and qualitative study

Underwater image enhancement is often perceived as a disadvantageous process to object detection. We propose a novel analysis of the interactions between enhancement and detection, elaborating on the potential of enhancement to improve detection. In particular, we evaluate object detection performance for each individual image rather than across the entire set to allow a direct performance comparison of each image before and after enhancement. This approach enables the generation of unique queries to identify the outperforming and underperforming enhanced images compared to the original images. To accomplish this, we first produce enhanced image sets of the original images using recent image enhancement models. Each enhanced set is then divided into two groups: (1) images that outperform or match the performance of the original images and (2) images that underperform. Subsequently, we create mixed original-enhanced sets by replacing underperforming enhanced images with their corresponding original images. Next, we conduct a detailed analysis by evaluating all generated groups for quality and detection performance attributes. Finally, we perform an overlap analysis between the generated enhanced sets to identify cases where the enhanced images of different enhancement algorithms unanimously outperform, equally perform, or underperform the original images. Our analysis reveals that, when evaluated individually, most enhanced images achieve equal or superior performance compared to their original counterparts. The proposed method uncovers variations in detection performance that are not apparent in a whole set as opposed to a per-image evaluation because the latter reveals that only a small percentage of enhanced images cause an overall negative impact on detection. We also find that over-enhancement may lead to deteriorated object detection performance. Lastly, we note that enhanced images reveal hidden objects that were not annotated due to the low visibility of the original images.

Remote Sensing↗

USGS2025

Status and trends of pelagic and benthic prey fish populations in Lake Michigan, 2024

Fall bottom trawl (fall BT) and lakewide acoustic (AC) surveys are conducted annually to generate indices of pelagic and benthic prey fish densities in Lake Michigan. The fall BT survey has been conducted each fall since 1973 using 12-m trawls at depths ranging from 9 to 110 m at fixed locations distributed across seven transects; this survey estimates densities of seven prey fish species [i.e., Alewife ( Alosa pseudoharengus ), Bloater ( Coregonus hoyi ), Rainbow Smelt ( Osmerus mordax ), Deepwater Sculpin ( Myoxocephalus thompsonii ), Slimy Sculpin ( Cottus cognatus ), Round Goby ( Neogobius melanostomus ), Ninespine Stickleback ( Pungitius pungitius )] as well as age-0 Yellow Perch ( Perca flavescens ) and large (> 350 mm) Burbot ( Lota lota ). In recent years, wild juvenile (<400 mm) Lake Trout ( Salvelinus namaycush ) have also become more common in the fall bottom trawl. The AC survey has been conducted each late summer/early fall since 2004 (except 2020). The 2024 AC survey consisted of 24 transects [468 km total (291 miles)] covering bottom depths ranging from 16 to 173 m and 38 midwater trawl tows at 4 to 72 m; this survey estimates densities of three prey fish species (i.e., Alewife, Bloater, and Rainbow Smelt). The data generated from these surveys are used to estimate various population parameters that are, in turn, used by state and tribal agencies in managing Lake Michigan fish stocks. In spring of 2024, an additional spring bottom trawl survey (spring BT) was implemented across six of the transects sampled in the fall and sites ranged in depth from 9 to 237 m. The goal of the spring BT, conducted annually since 2021 with differing levels of effort, was to explore seasonal differences in biomass density and distributions of key prey species, most notably Alewife. Total prey fish biomass density from the spring BT was 5.7 kg/ha. For the AC survey, total biomass density of prey fish equaled 10.8 kg/ha, more than double the long-term average (20042023) of 5.1 kg/ha but 4.0 kg/ha lower than the 2023 estimate. For the fall BT, total biomass density of prey fish equaled 2.1 kg/ha, the lowest value since 2020 and 69% lower than the average from 2004-2023 (6.8 kg/ha). The 2024 fall BT biomass density was only 6.3% of the average over the entirety of the time series (1973-2023; 33.1 kg/ha). Over the period both surveys have been conducted (2004-2024), total biomass density has trended downward in the fall BT (despite a high 2022 estimate) and remained relatively stable in the AC survey. Deepwater Sculpin and Bloater were the most common species (by biomass) among prey fishes in the spring BT while the AC survey and fall BT reported co-dominance of Bloater and Alewife. Mean biomass of yearling and older (YAO) Alewife was 1.30 kg/ha in the spring BT, 4.7 kg/ha in the AC survey, and 0.68 kg/ha in the fall BT. Since 2014, annual survey results suggest that the catchability of YAO Alewives for the fall BT is substantially lower than the AC survey. Like previous spring surveys, Alewives were aggregated in deeper habitats, with 93% of biomass collected between 110 and 201 m. Results of the 2024 spring BT align with past spring surveys and do not suggest that spring bottom trawling provides a better index of age-2 and older Alewives than fall bottom trawling, even with adjustments for differences in habitat use. However, the spring BT does appear to index age-1 Alewives more effectively than the fall BT. The 2024 AC survey YAO Alewife biomass density estimate was 77% higher than the average from 2004-2023. The Alewife population of Lake Michigan appears to be composed mostly of young fish and the proportion of age-4 and older Alewives was <1.8% in each of the three surveys. Age-0 Alewife numeric density from the AC survey was 510 fish/ha in 2024, slightly higher than the long-term mean (486 fish/ha). Biomass density of large (≥120 mm) Bloater was 5.2 kg/ha in the AC survey and 0.76 kg/ha in the fall BT, while total Bloater biomass in the spring BT was 1.8 kg/ha - all three estimates were much lower than what was estimated by the fall BT between 1981 and 1998. The density of small (<120 mm) Bloater was 456 fish/ha in the AC survey, the second highest value in the time series and potentially reflective of an above average 2024 year-class. Meanwhile, small Bloater density estimated in the fall BT was only 16 fish/ha. Biomass density of large Rainbow Smelt (≥90 mm) was 0.21 kg/ha in the AC survey and 0.03 kg/ha in the fall BT survey, continuing the trend of low large Rainbow Smelt biomass observed since 2001. Numeric density of small (<90 mm) Rainbow Smelt was 31 fish/ha in the AC survey and 143 fish/ha in the fall BT. All four prey fish species indexed only by the fall BT had below-average biomass densities regardless of trawling season. Deepwater Sculpin biomass density was 0.26 kg/ha, which makes 14 of the past 15 years with biomass <1 kg/ha. Spring BT Deepwater Sculpin biomass density (2.0 kg/ha) was higher than any fall BT estimate since 2006, likely reflective of including bottom trawls at greater depths in the spring than the fall. Slimy Sculpin was estimated to be < 0.04 kg/ha in the spring and fall BT, an order of magnitude lower than the long-term average from the fall BT. Round Goby biomass density estimates were low and similar across seasons (0.43 kg/ha in the spring and 0.10 kg/ha in the fall). Ninespine Stickleback density was 3.9 fish/ha in the fall BT and no fish were collected in the spring BT.

Lake Michigan↗

USGS2024

A transferable approach for quantifying benthic fish sizes and densities in annotated underwater images

1. Benthic fishes are a common target of scientific monitoring but are difficult to quantify because of their close association to bottom habitats that are hard to access. Advances in image-acquisition technologies, machine vision, and deep learning have made capturing and quantifying fishes with cameras increasingly feasible. We present a method and open-source software called ‘FishScale’ to estimate benthic fish lengths, numeric abundance, and biomass density in underwater environments assessed with down-looking monocular images. 2. ‘FishScale’ estimates fish abundances and size frequencies from near-nadir monocular images where fish have already been semantically segmented. The software accounts for lens distortion, underwater magnification effects, and fish body curvature to automatically estimate fish lengths and the areas of images where they were captured. Numeric and biomass density are estimated through a deterministic machine vision algorithm that requires a user-provided length-weight relationship for species of interest and calibration images. 3. Results from validation studies show that lengths and weights can be estimated with high accuracy and precision for round goby ( Neogobius melanostomus ) captured in distorted action camera images, and from large-bodied lake trout ( Salvelinus namaycush ) imaged with a machine vision camera. The real-world utility of the approach is demonstrated in a case study estimating round goby abundances and size frequencies along a 10.7-km transect surveyed with an autonomous underwater vehicle in Lake Michigan, USA. 4. Our validation studies demonstrate that the approach estimates benthic and benthopelagic fish lengths and weights with little bias and good accuracy and precision for species with much different body shapes and sizes. The method is applicable to data collected using a variety of nadir imaging approaches with widespread applications to fisheries monitoring and quantification of any species or object for which nadir images and working distances between the camera and feature of interest are available.

Illinois, Indiana, Michigan, Wisconsin↗

USGS2024

Zebra and Quagga mussels in the United States—Dreissenid mussel research by the U.S. Geological Survey

The U.S. Geological Survey (USGS) delivers high-quality data, technologies, and decision-support tools to help managers both reduce existing populations and control the spread of dreissenid mussels. The USGS researches ecology, biology, risk assessment, and early detection and rapid response methods; provides decision support; and develops and tests control measures.

conterminous United States↗

USGS2024

Classification of lakebed geologic substrate in autonomously collected benthic imagery using machine learning

Mapping benthic habitats with bathymetric, acoustic, and spectral data requires georeferenced ground-truth information about habitat types and characteristics. New technologies like autonomous underwater vehicles (AUVs) collect tens of thousands of images per mission making image-based ground truthing particularly attractive. Two types of machine learning (ML) models, random forest (RF) and deep neural network (DNN), were tested to determine whether ML models could serve as an accurate substitute for manual classification of AUV images for substrate type interpretation. RF models were trained to predict substrate class as a function of texture, edge, and intensity metrics (i.e., features) calculated for each image. Models were tested using a manually classified image dataset with 9-, 6-, and 2-class schemes based on the Coastal and Marine Ecological Classification Standard (CMECS). Results suggest that both RF and DNN models achieve comparable accuracies, with the 9-class models being least accurate (~73–78%) and the 2-class models being the most accurate (~95–96%). However, the DNN models were more efficient to train and apply because they did not require feature estimation before training or classification. Integrating ML models into benthic habitat mapping process can improve our ability to efficiently and accurately ground-truth large areas of benthic habitat using AUV or similar images.

Michigan, Wisconsin↗

USGS2024

Predicting Lake Huron Dreissena spp. spatial distribution patterns from environmental characteristics

Invasive dreissenid mussels ( Dreissena polymorpha and Dreissena rostriformis bugensis ) have altered Great Lakes ecosystems through a multitude of effects on benthic habitats, food web structure, and nutrient cycling. This study explores whether spatially continuous geographic data of environmental factors can be utilized to predict Dreissena spp. spatial distributions on a lake-wide scale. Categorical variables were also assessed for significant relationships with Dreissena spp. biomass. Point observations from the 2017 Lake Huron benthic survey under the Cooperative Science and Monitoring Initiative (CSMI) were utilized for in situ measurements of dreissenid presence and biomass at 119 sites across Lake Huron. Basin, bathymetric zone, and tributary influence were found to have statistically significant relationships to dreissenid biomass. A boosted regression tree (BRT) model (ROC score 0.707) was developed to spatially predict dreissenid presence probability across Lake Huron from six environmental explanatory variables: April, May, and October chlorophyll, June dissolved organic carbon, January bottom temperature, and May bottom temperature. The importance of food availability and bottom temperature illuminated relationships between dreissenid mussels and periods of benthic-pelagic mixing in the spring and fall seasons. Future models could be improved through advancements in survey technology for improved geographic characterization of mussel habitat characteristics and environmental constraints.

Lake Huron↗